ii I ill j!jj' M 1 1 U ' i I L UNIVERSITY OF CALIFORNIA LIBRARY THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW I 1M2 7 '53^v MAR 1 1 J968 D 5?n-10,'22 INTERNATIONAL CHEMICAL SERIES H. P TALBOT, PH.D., Sc.D.. CONSULTING EDITOR QUANTITATIVE ANALYSIS Trie Qmw'3/ill Book Ca 1m PUBLISHERS OF fc O O K. S FO R^ Coal Age Electric Railway Journal Electrical World v Engineering News -Record American Machinist v Ingenien'a Internacionai Engineering Mining Journal ^ Power Chemical 6 Metallurgical Engineering Electrical Merchandising QUANTITATIVE ANALYSIS BY EDWARD G. MAHIN, PH. D. Professor of Analytical Chemistry in Purdue University SECOND EDITION REVISED AND ENLARGED FIFTH IMPRESSION McGRAW-HILL BOOK COMPANY, INC. NEW YORK: 370 SEVENTH AVENUE LONDON: 6 & 8 BOUVERIE ST., E. C. 4 1919 COPYRIGHT, 1914, 1919, BY THE MCGRAW-HILL BOOK COMPANY, INC: 1'KKSS YORK PA PREFACE TO THE SECOND EDITION Since the publication of the first edition of this book numerous changes have been made in the standardized methods of analysis of certain industrial materials, as adopted by official committees of various scientific societies. In the present edition the dis- cussions and detailed procedure have been modified to conform to the revised methods, wherever possible. This statement applies to the analysis of coal, water, fertilizers, dairy products and insecticides. A number of other analytical methods are now described, of which may be mentioned the gravimetric determination of the chloride, sulphate and phosphate radicals and the in- direct determination of the halogens; the perchlorate method for potassium; the determination of chromium and vanadium in steel and the glyoxime method for nickel in steel; the analysis of brass and of soft bearing metals; and the volumetric determina- tion of zinc. A part of the discussion of metallography and treatment of steel has been rewritten and several new sections have been added to this portion of the book. New photomicrographs have been substituted for the old ones and about fifteen new figures have been used at various points throughout the book, the latter being the work of the author's students in Chemical Engineering, to whom grateful acknowledgment is due. Finally, a new system of chapter division has been introduced in order to give greater emphasis to the sections dealing with industrial analysis. In the process of making these changes a considerable poition of the book has been rewritten and the discussions have been amplified, in the interest of added clearness. It is hoped that these changes will serve to enhance the usefulness of the book as a college text and that the favor with which the first edition was so generously received will not be undeserved by the present edition. December, 1918. PREFACE TO THE FIRST EDITION This book cannot be classed as a complete reference work on quantitative analysis, neither is it a bare outline of laboratory exercises. The author has felt a desire that has probably been felt by every teacher of quantitative analysis, to produce a book that would cover the ground that he wishes to cover in the college courses, providing a reasonable degree of latitude in the selection of exercises for other possible users of the book, and at the same time to present a theoretical and practical discussion of the sub- ject, sufficiently simple to be comprehended by the average stu- dent but not so elementary as to destroy his self-respect. One of the most difficult tasks connected with the teaching of quantitative analysis is to produce in the mind of the student a clear comprehension of the scientific development of quantitative methods. There seems to be a more or less unconscious tendency toward the acceptance of the present laboratory method of pro- cedure as a gift of Providence. How well this situation has been met in the present volume must be shown by the test of experience. The general discussions have been given a large share of attention although elaborate or involved theoretical discussions have been, so far as possible, avoided. References to original papers have been carefully selected with a view to actual reading by the stu- dent and such references are, in nearly all cases, to discussions that will serve either to impress more clearly upon the reader's mind the principles mentioned in the text or to bring to his mind a realization of the labor involved in the development of the finished method. It is believed that careful and systematic reading and discussion of such original papers by the student with his in- structor is a most valuable aid in the understanding of quanti- tative analysis as a truly scientific study. The mathematical development of quantitative calculations in this book is somewhat unusual, in that the "rule of three" has been carefully excluded. It is the firm belief of the author, after several years of experience in teaching quantitative analysis, that the use of this rule of proportion has produced much harm and has been the greatest of all obstacles to the student in his attempt to grasp the principles of quantitative calculations, and par- vii viii PREFACE ticularly those of volumetric analysis. The ideas involved in the solution of proportions are so labored and so unnecessary and require such cumbersome solutions of problems involving them that it is difficult to see why emphasis has so generally been placed upon this rule in chemical calculations. Other teachers will, no doubt, differ with the author upon this point. It is desired only that the method of presentation involved in these pages be tested, not in part but in whole, before final judgment is given. Most of the calculations involved in the laboratory exercises have been left to the student. Principles of such calculations are first fully explained but ready-made calculations that leave nothing to the ingenuity of the student furnish poor preparation for scientific analysts. Acknowledgment is here gladly expressed to Mr. H. C. Mahin for all of the original drawings in this book, also to Wm. Ainsworth and Sons, Bausch and Lomb Optical Company, the Bureau of Standards, Eimer and Amend and The Scientific Materials Com- pany for several cuts which they have loaned. EDWARD G. MAHIN. LAFAYETTE, IND., December, 1913. CONTENTS PAGE PREFACE TO THE SECOND EDITION v PREFACE TO THE FIRST EDITION vii INTRODUCTION . xiii PART I GENERAL QUANTITATIVE ANALYSIS CHAPTER I GENERAL PRINCIPLES Cleanliness and care Limit of accuracy Classes of methods. CHAPTER II GRAVIMETRIC ANALYSIS 5 Factors Factor weights. General operations: Preparation of samples Solution Precipita- tion Colloids Enlargement of particles Filtration Washing Drying of precipitates Ignition Fusion Weighing Cali- bration of weights. Reagents Glassware Records. CHAPTER III EXPERIMENTAL GRAVIMETRIC ANALYSIS 76 Calcium Silver Chlorides, bromides and iodides Aluminium Barium Sulphates Strontium Potassium and sodium Re- covery of platinum from waste Magnesium Phosphates Manganese Halogen compounds Carbonates and caibon dioxide. CHAPTER IV ELECTRO -ANALYSIS 138 Nature of electrolyte Solvent Temperature Electrolytic pres- sure Current density Nature of electrodes Other apparatus. Copper Silver Iron Lead Nickel Separations Moving elec- trodes Mercury cathode. CHAPTER V VOLUMETRIC ANALYSIS 168 Apparatus Units of volume Calibration by weighing Calibra- tion by standardized bulbs Calculation of results Weight of X CONTENTS PAGE one substance equivalent to a stated weight of another Stand- ard solution for titration of but one substance Burette reading a percentage reading No system Normal system Decimal system Choice of system Temperature correction for standard solutions Adjustment to exact concentration. CHAPTER VI COLOR CHANGE OP INDICATORS 207 lonization theory Theory of chromophors Classification of indicators Description of indicators. CHAPTER VII STANDARDIZATION 216 Direct weighing Weighing a substance produced by a measured volume of solution Measuring the volume of solution required to react with a known weight of a substance of known purity Titration against another standard solution Primary standards. CHAPTER VIII EXPERIMENTAL VOLUMETRIC ANALYSIS 221 Standard acids: Materials for standardization Standardization by direct weighing Preparation of pure sodium carbonate. Standard hydrochloric acid: Soda ash Mixtures of carbonates and bases Mixtures of carbonates and bicarbonates Hardness and alkalinity of water. Standard bases: Selection of base for standard solutions Stand- ardization. Standard sodium hydroxide: Concentration of the laboratory acids Citric acid Vinegar Boric acid. Use of two standards : Limestone for agricultural purposes. CHAPTER IX OXIDATION AND REDUCTION 240 Apparent valence. Potassium permanganate: Iron Reduction of permanganate by chlorides Primary standards Reduction of iron solutions Calcium Manganese Available oxygen. Potassium dichromate: Iron Chromium. Iodine and sodium thiosulphate : Oxidizing power of peroxides Copper Bleaching powder Standard iodine solution Arsen- ical insecticides Total arsenic and copper in Paris green. CHAPTER X TITRATIONS INVOLVING FORMATION OF PRECIPITATES 276 Silver Halides and cyanides Zinc. CONTENTS xi PART II ANALYSIS OF INDUSTRIAL PRODUCTS AND RAW MATERIALS CHAPTER XI PAGE ROCK ANALYSIS 284 Carbonate minerals : Carbon dioxide Silica Iron and aluminium Calcium Magnesium Sodium and potassium. Silicate minerals: Moisture Silica Iron and aluminium Man- ganese Calcium Magnesium Sodium and potassium. CHAPTER XII FUELS 297 Coal: Proximate analysis Fusing point of ash Ultimate analysis Calorimetry. Gas mixtures: Apparatus Absorbents Illuminating gas Chim- ney gases. CHAPTER XIII OILS, FATS AND WAXES 345 Burning oils: Specific gravity Flash point Burning point Fractional distillation. Lubricating oils: Viscosity Specific gravity Separation of saponifiable from mineral oils Chill test Cold test. Edible fats and oils : Composition Specific gravity Melting point of fats Iodine absorption number Acid value Saponification number Insoluble and soluble acids Reichert-Meissl number Polenske value Acetyl value Maumene" number Specific temperature reaction Qualitative reactions Fish and marine animal oils in mixtures with vegetable oils Examination of an unknown oil Hardened oils. CHAPTER XIV WATER 391 Industrial analysis: Corrosives Incrustants Foam producers Hypothetical compounds Treatment. Sanitary examination : Potability Collection of samples Physical examination Chlorine Organic nitrogen Nitrogen in am- monia Nitrites Nitrates Required oxygen. CHAPTER XV STEEL AND ALLOYS 432 Steel and cast iron: Sampling Standard methods Solution and evaporation Standard samples Carbon Silicon Sulphur xii CONTENTS PAGE Phosphorus Titanium Manganese Tungsten Chromium Nickel Vanadium Oxygen. Treatment of steel: Thermal changes Allotropism Proximate constituents of slowly cooled steel Ferrite Cementite Pearlite Relation between structure and carbon percent Austenite Relation of proximate constituents to critical points of steel Steel as a solid solution Martensite Critical points related to rate of cooling or heating Hardening and annealing Troostite Sorbite Quenching media Tempering Granula- tion Overheating Uneven carbon distribution Streaks Sulphur prints Case hardening Effect of working Fatigue Slag Apparatus for metallographic work Experiments. Brass and bronze: Tin Lead Copper Zinc. Anti-friction metals : Tin Antimony Lead Copper. CHAPTER XVI AGRICULTURAL MATERIALS 510 Fertilizers: Moisture Nitrogen Availability of nitrogen Phos- phorus Potassium. Milk: Adulteration of milk Specific gravity Total solids Ash Total nitrogen Casein Albumin Lactose Fat. Cream: Ash Fat. Condensed milk: Total solids Ash Nitrogen Lactose Fat Sucrose. Butter and substitutes: Moisture Fat Casein Salt Examina- tion of fat Coloring matter. CHAPTER XVII THE FIRE ASSAY 564 Gold and silver ores: Sampling Weighing Crucible process Fluxes Reducing agents Oxidizing agents Crucible charge Reducing ores Ores containing copper, arsenic or antimony Cupellation Inqu art ation Parting Annealing Scorific at ion . TABLE OF LOGARITHMS AND ANTILOGARITHMS 586 INDEX ... . 593 INTRODUCTION In the study of such divisions of chemistry as naturally pre- cede quantitative analysis the work has been mainly descriptive. The chemical and physical properties of elements and compounds have been determined. Laws of chemical action have been devel- oped and theories have been evolved for the explanation of such action and as generalizations upon which to plan further studies and investigations. In the courses in qualitative analysis an effort was made to detect and recognize elements and their compounds by certain characteristic reactions of these sub- stances and to separate more or less complicated mixtures into their constituents. Quantitative analysis is the next logical step in the study of the composition of matter. The qualitative analysis should pre- cede the quantitative, unless the nature of the substance is already known, because in nearly all cases the presence of sub- stances other than those whose percent is being determined will make necessary certain modifications in the method to be em- ployed. Whether or not a qualitative analysis is made it is a fact to be constantly kept in mind that an intelligent understand- ing of quantitative processes can be obtained only by a continued application of the facts and laws earlier learned to the newer processes which are being studied. The industrial development of the world, as well as the evolution of chemistry as a science, would be a more rapid and substantial change were it not for the numbers of inadequately trained chemists who have found a place in industrial work and who have been content to allow their study to go no farther than the routine of following direc- tions without understanding. This is the inevitable end of the student who does not begin his scientific study with the definite determination patiently and persistently to think out each prob- lem to its logical conclusion as it presents itself, and who does not continue this process in his work in quantitative analysis, re- viewing his earlier work until the principles that were imperfectly understood expand and illuminate the newer problems. xiii QUANTITATIVE ANALYSIS PART I GENERAL QUANTITATIVE ANALYSIS CHAPTER I GENERAL PRINCIPLES Cleanliness and Care. In no other line of scientific work are neatness and care more essential than in quantitative analysis. It is desirable that the student should acquire speed so that his efficiency may be increased by his ability to accomplish much work in the time at his disposal but speed attained through care- less manipulation or through a sacrifice of a close study of the analytical process is of little avail because the results, when ob- tained, are not dependable. Speed should rather be acquired by an intelligent application of methods that are thoroughly under- stood. Several experiments may be performed at the same time without confusion if the analyst has cultivated the habit of delib- erate and clear thinking. It is highly important that as far as possible the operations should be systematized so that the mind may be left free for the more important problems that are not of a routine nature. For example a complete system of marking vessels and materials makes impossible errors due to confusion and lack of identification. Apparatus should be kept scrupu- lously clean, note books and other records with perfect system and the desk always cleared of apparatus that is not in use. Re- agents should be added carefully and, whenever practicable, should be measured approximately so that in case of error the defect may be corrected. A disregard of this precaution often makes it necessary to begin over again a determination that might have been saved. 1 2 QUANTITATIVE ANALYSIS All of these points will be amplified as the work proceeds and as occasion offers. They are here mentioned in order to give some idea of the requirements of the work that is to follow. Limit of Accuracy. A question frequently asked by beginners in quantitative analysis concerns the degree of agreement that is regarded as reasonable and necessary for the results of deter- minations made in duplicate. This question can never be answered without qualification. The ideal of all scientific work is absolute accuracy. How closely this ideal may be ap- proached will depend partly upon the individual analyst, but also upon the possibilities of the method with which he happens to be working. The latter is a variable quantity. Certain inorganic methods are capable of giving results that are quite reliable to within one hundredth of one percent while other (particularly organic) methods give only an approximation of whole percents. Thus no definite limit to the accuracy of methods in general can be set. Each case must be considered by itself and, while the student will use the utmost possible care in all cases, he must, for a time, be content to allow his instructor to be the judge of the accuracy that may be reasonably required. A very common fallacy is to the effect that no part of the work need be performed more carefully than that part which is necessarily least accurate. For example, it is said that if a certain part of an analytical process involves an unavoidable error of 0.10 percent, it is a waste of time to attempt to avoid errors in other parts of the work amounting to 0.05 percent or even 0.09 percent. This is a most unfortunate attitude for the analyst. This logic would lead one to the conclusion that if a method cannot give results nearer than 0.10 percent to the truth it should be given an excel- lent chance to depart 0.10 percent +0.09 percent from the truth or 0.10 per cent +nX 0.09 percent, if there are n other places where errors may occur. Of course it is not necessarily true that all errors will have the same algebraic sign and they may, to a certain extent, counteract each other in effect. It is important to note, however, that there is no assurance that they do counteract each other instead of accumulating. This misconception, no doubt, arises from the unquestioned fact that when a certain minimum error is unavoidable it is not wise to expend an undue amount of time in rectifying other possible GENERAL PRINCIPLES 3 errors where the ratio of these errors to the larger error is very small, because in such a method the analytical results have little significance as expressed in small fractions. For example, in the determination of volatile combustible matter in coal it is difficult to obtain results from analyses performed in duplicate, agreeing more closely than 0.2 to 0.5 percent, because of the impossibility of exactly duplicating heating conditions. It is then not profitable to weigh the sample so accurately as to justify an agreement within 0.005 percent unless there are several other places where rectification of similar errors is possible, simply because the results will have no significance in this remote decimal place and will not be expressed to the third or even the second place. If, however, there are five other points in the method where errors may be made as small as 0.005 percent but by undue haste or neglect might be as large as 0.05 percent, then the argu- ment already referred to would lead one to commit an error as large as 0.5 percent +5X0.05 percent = 0.75 percent where it could have been kept as low as 0.5 percent+5X 0.005 percent = 0.525 percent. There still remains to be answered a legitimate question re- garding the number of decimal places to be reported. The opera- tions of multiplication and division that are involved in calcula- tions of analyses often give five or more figures in decimal places and yet even the novice understands that these figures have no significance since the probability of their correctness is very slight. A rule that is very generally followed is to report one decimal place farther than the one that is considered to be certainly correct. Usually the student has little evidence concerning the accuracy of his work other than the agreement of his duplicate determinations. If two analyses of an iron ore give 52.75 percent and 52.71 per- cent, respectively, it would be considered, from this standpoint alone, that 52.7 percent is certainly correct and that the average of 52.73 percent is a close approximation for the second decimal place. This average would therefore be reported. If the re- sults were 52.75 percent and 52.67 percent the disagreement would be 0.08 percent and there is again agreement in the first decimal place although, at first sight, this would not appear to be true. The average 52.71 percent would then be reported. If the results were 52.75 percent and 52.45 percent the disagree- 4 QUANTITATIVE ANALYSIS ment would be 0.30 percent. Evidently the first decimal place is but an approximation and the average 52.6 percent would be reported. The rule for reports will then be as follows: Determine the disagreement of the two determinations by subtracting the smaller percent from the larger. Calculate the average and report as far as the first significant figure in the difference. Classes of Methods. Quantitative analysis is, for the sake of convenience, divided into two general classes which are desig- nated as Gravimetric and Volumetric. The general nature of the work is indicated by these terms, since the first class consists of analyses made by means of measurements of weight while the second class consists of those made by means of measurements of volume. The discussion of volumetric analysis will be left for a later section of the book because the complete volumetric proc- ess cannot be carried on without a knowledge of gravimetric methods. Gravimetric analysis will therefore comprise the first part of the course. The problem, as it presents itself to the quantitative analyst, is to take a substance whose qualitative composition is at least partially known and so to treat it that a part or all of the constitu- ents may be expressed in terms of percents. Thus we speak of the "analysis" of a substance and of the "determination" or "estimation" of the constituents, the last two terms being used with practically the same meaning. CHAPTER II GRAVIMETRIC ANALYSIS The gravimetric determination of any constituent of a com- pound or mixture usually depends upon the precipitation of this constituent in such a form that it may be separated and weighed. This makes it necessary that the following conditions shall be fulfilled: (a) The precipitate must have an extremely small solubility in order that the amount lost in the filtrate may be negligible, (b) The precipitate must be of such a nature that it may be readily retained upon the filter and washed free from all impurities, (c) The precipitate must be susceptible to drying or strong heating, changing its composition either not at all, or in a perfectly definite and well-known manner, (d) The pre- cipitate should, if an electrolyte, be a strongly ionized one. The last condition will be understood when the nature of precipita- tion is considered. The determination of calcium in a calcium compound may be taken as an illustration of the gravimetric process. A definite weight of sample is taken, dissolved and the calcium precipitated as oxalate. This is separated by filtration, washed free from soluble salts and then strongly heated and weighed as pure cal- cium oxide. It is known from the formula for calcium oxide (expressing as it does the results of many careful analyses) that 40 07 56 7)7 ^ ^ S we ^h"t is calcium. That is, 40.07P 5607 = weight of calcium, where P = weight of calcium oxide found. Since this experimentally determined weight of calcium was ob- tained from a known weight of the calcium compound it follows that weight of calcium X 100 , g - = percent of calcium in the sample, S representing the weight of sample used. 5 6 QUANTITATIVE ANALYSIS Combining these two expressions: 40.07 X 100P = ent of ca i c i um i n the sample. (2) 56.07S Factors. It is seen from an inspection of the above formula that so long as determinations of calcium are being made in this manner the constant fraction - ' 56 07 will enter into a11 cal ~ culations. The calculation will be somewhat shortened if these factors are collected into the decimal fraction 71.464. Expression (2) then becomes 71 464P - = percent of calcium. (3) fe This is a typical form of expression for the experimentally deter- mined percent of any element, radical or compound. The deci- mal fraction 71.464 is known as a gravimetric factor, which may be defined as the percent of a given element or group of elements in or equivalent to a pure compound weighed. The pure substance weighed usually contains the element or radical which is being determined but does not always do so. If the factor in general is represented by F, equation (3) may be written TTP -F-= required percent. (4) Atomic weights have been calculated from the results of the most careful experimental work of modern investigators. For this reason the factor is the most reliable of all of the experimental data used by the analyst and the calculation may be carried far- ther than that of analytical percents. Five significant figures should usually be recorded. Factor Weights. In equation (4), F is a constant for all deter- minations of the particular element or group of elements for which it has been calculated. It is possible so to choose the weight of sample taken as to simplify the calculation represented by equa- tion (4) . For instance, by taking a sample weight equal in grams F to the value of the factor, g = 1 and equation (4) becomes P = required percent. GRAVIMETRIC ANALYSIS 7 In such a case the weight of the precipitate, expressed in terms of grams or fractions, would indicate, without calculation, the required percent of the constituent being determined. A weight of sample equal in grams to the value of the factor is usually too large a quantity to be handled readily and a definite fraction of this weight (0.5, 0.2, 0.1, 0.01, etc.) may be used in- stead. Any such weight is called a factor weight, which may be denned as a quantity equal in weight units to the value of the gravi- metric factor, or to some simple fraction of this value. The use of such weights involves adjusting the quantity of the sample on the balance to a certain specified value. If this device is to result in the saving of time it is obvious that the weighing must not require any extra time. Therefore the use of factor weights is generally confined to the determination of such constituents as occur in small amounts in a sample that is to be analyzed, so that relatively large amounts of the sample may be used and the accuracy of the weighing may be correspondingly less without impairing the accuracy of the determination. Use of Logarithms. Most of the results of quantitative analy- sis involve chiefly multiplications and divisions. On this account a four or five place table of logarithms will be found extremely serviceable. Many students hesitate to use such tables because they find that mistakes are more readily made and that the cal- culations require more time than is the case when ordinary arithmetical calculations are employed. If this is true it is be- cause of a lack of facility in handling such tables and this comes through lack of practice. The beginner in this work is strongly urged to make use of his tables even though this should, at first, involve more work and care than direct arithmetical solutions. Wherever a constant occurs in a solution the logarithm of this constant may be recorded and used in all similar calculations. This is particularly helpful in the case of factors. Chemists* Slide Rules. A slide rule may be substituted for the logarithmic table but it is not accurate beyond the third significant figure and should not be used for the calculation of careful analyses. Chemists' slide rules have been devised, having marked at the proper points the constants that represent the analytical factors. By the use of such a rule a percent may be read from one adjustment of the slide. QUANTITATIVE ANALYSIS FIG. 1. Chemist's slide rule. Following is the international table of atomic weights for 1918. INTERNATIONAL ATOMIC WEIGHTS, 1918 Symbol Atomic weight Symbol Atomic weight Al 27 1 Molybdenum Mo 06 Antimony Sb 120 2 Neodymium Nd 144 3 A 39 88 Ne 20 2 Arsenic As 74 96 Nickel Ni 58 68 Barium Ba 137 37 Niton Nt 222 4 Bismuth Bi 208 Nitrogen N 14 01 Boron B 11 Os 190 9 Bromine Br 79 92 o 16 00 Cadmium Cd 112 40 Palladium Pd 106 7 Caesium. Cs 132 81 p 31 04 Calcium Ca 40 07 Pt 195 2 Carbon c 12 005 J on in Cerium Ce 140 25 Pr 140 9 Chlorine - Cl 35 46 Ra 226 Chromium Cr 52 Rh 102 9 Cobalt . Co 58 97 Rb OK AK Columbium Cb 93.1 Ruthenium Ru 101 7 Copper Cu 63 57 Sa 150 4 Dysprosium Dy 162 5 Sc 44 i Erbium Er 167 7 7Q 9 Europium Eu 152 Si 00 <3 Fluorine P 19.0 Silver ACT 107 RS Gadolinium Gd 157.3 Sodium Na 23 00 Gallium Ga 69 9 q_ Germanium Ge 72 5 g oo nc Glucinum Gl 9 1 Ta Gold Au 197 2 TP Helium He 4 00 Th Holmium Ho 163 5 T1 Hydrogen H 1.008 Th ooo 4 Indium In mo Iodine I 126.92 Tin Tm Sn 168.5 Indium Ir 193 1 Ti Iron Fe 55 84 Krypton Kr 82 92 Lanthanum La 139.0 Vanadium. . v 51 Lead Lithium Pb Li 207 . 20 6.94 Xenon Xe Vh 130.2 Lutecium Lu 175.0 Yttrium Yt 88 7 Magnesium Mg 24.32 Zinc 7n Manganese Mercury Mn Hg 54.93 200.6 Zirconium Zr 90.6 GRAVIMETRIC ANALYSIS Problem 1. Calculate the factors and their logarithms indicated in the follow- ing table, recording in the proper spaces for future use. The atomic weights as recorded in the preceding table should be used and the factors should be calculated in five significant figures. TABLE OF GRAVIMETRIC FACTORS AND LOGARITHMS Found Required Factor Log Factor CaO Ca AgCl Ag Cl AgBr Br Agl I A1 2 0, Al BaS0 4 Ba S0 4 S0 3 s SrS0 4 Sr K 2 PtCl 6 K KC1 K 2 S0 4 KC1O 4 K KC1 K 2 S0 4 Na 2 SO 4 Na NaCl Na Mg 2 P 2 O 7 Mg P 2 6 . P Mn 2 P 2 O 7 Mn Mg 2 As 2 O 7 As GENERAL OPERATIONS Certain operations enter into the majority of gravimetric analyses and the general conduct of these will be discussed briefly, principles and precautions being indicated. It will be readily understood that variations in standard procedure will be de- pendent upon the nature of the substance being analyzed and such variations will be discussed as the necessity arises. Preparation of Samples. The object of all preliminary work with samples is to make it possible to obtain, for the actual analy- 10 QUANTITATIVE ANALYSIS sis, a portion that shall truly represent the average composition of the entire material at hand. This matter is likely to be treated lightly by the beginner, but proper sampling is often one of the most difficult problems of quantitative analysis. It is generally necessary to use a quantity of 1 gm or less and if the substance is not homogeneous this small quantity may have an average com- position that is very different from the average composition of the entire material being investigated. No matter how carefully an analysis may be performed or how accurate the results ob- tained, if the substance used does not represent the average of the substance originally at hand the results become nearly or en- tirely valueless. If the substance is practically homogeneous the operation of sampling involves nothing more difficult than grind- ing down to a degree of fineness required for the work. This is the case when the substance is an approximately pure chemical compound, such as will be used for the earlier exercises. The gross sample, as the analyst receives it, may be in the form of lumps, as is frequently the case with minerals, or it may be in the form of small pieces, crystals, powder, or solution. In any case except that of liquid samples, the object is to reduce the size of pieces to that required for the analysis (usually a rather fine powder) and at the same time to select from the total mass such a quantity as is required for the experimental work* The original sample is often quite large. It may vary from a few grams, or less, to many pounds. It is obviously unnecessary and practically impossible to grind the entire amount into a fine powder. The operation then resolves itself into a thorough mix- ing and progressive grinding and dividing. Many forms of both hand and power grinders are in common use. For the first exercises nothing more complicated than a porcelain mortar and pestle will be required. Mixing and Dividing. The mixing and dividing is best carried out by using a sheet of oilcloth or paper and a spatula. In many laboratories it is customary to use oilcloth, particularly for mixing minerals. This is convenient but offers the possibility of con- tamination ("salting") of one sample by the remnant of one that has preceded it. It is better to use a large sheet of tough, flexible paper, which can be discarded after using. The sample, after having been broken down to the proper maximum size of pieces, GRAVIMETRIC ANALYSIS 11 is placed on the paper and thoroughly mixed by rolling diagonally across the paper and alternating the direction of rolling as illus- trated in Fig. 2. The proper rapid manipulation of the paper is attained only after considerable practice. One precaution is essential: the corner of the paper that is lifted must be drawn across, low down, in such a manner that the pile of material is not caused to slide along the paper but is turned over upon itself so that the bottom is brought entirely to the top. In this way only can a segregation of larger and smaller particles be prevented. Since the larger and smaller particles usually have different com- position it is essential that they should be thoroughly mixed. FIG. 2. Manipulation of paper when mixing samples. The number of times that the sample is rolled before dividing will depend upon the degree of homogeneity and the accuracy re- quired in the analysis. In the assaying of gold and silver ores it is not unusual to require one hundred times. Quartering. When the first mixing is finished the pile is made approximately circular and is then divided, by means of a spatula, into quarters. Opposite quarters are carefully scraped to another sheet of paper, ground finer if necessary, remixed and quartered as before. This process of grinding, rolling, and quar- tering is continued until a sample is finally obtained, small enough in quantity and fine enough in texture to serve the pur- pose of the final weighing and analysis. The maximum size of 12 QUANTITATIVE ANALYSIS particles to be allowed in any particular mixing and quartering will depend upon the total quantity of material being handled in this operation. No particle should be so large that its inclusion in any quarter would cause the average composition of this quar- ter to be appreciably different from the average composition of the entire pile. This means that the ratio of the size of the largest particle to the size of the quarter should not be greater than a certain maximum value. What this maximum value shall be must be arbitrarily determined by the nature of the sample and the degree of accuracy required in the analysis. It is obvious that the part can perfectly represent, in composition, the whole only when the largest particle is infinitesimal. It is equally obvious that this limit is impossible and unnecessary in practice and we may say that, in general, the ratio of the largest particle to the portion that includes it should not be greater than 0.01 percent. If this condition is met, then, after thorough mixing of the sample, the chance inclusion or exclusion of any given particle cannot modify the results of the analysis to any appreciable extent. The maximum size of the particles to be obtained in the final portion that is to be weighed and used in the analysis must be determined, not only from the above considerations, but also by the nature of the operation to follow the weighing. This is usually solution or fusion. If the substance is considered to be almost absolutely homogeneous and if it is easily soluble (as, for example, a crystal of cupric sulphate) then the grinding need be carried no farther than is necessary to permit the easy adjust- ment, between fairly narrow limits, of the weight taken for analy- sis. In such a case, if a sample of 0.3 to 0.5 gm is required, then no particle should weigh more than about 0.1 gm. If, however, the process of solution or fusion is a difficult one to accomplish or if the material is far from being homogeneous, the grinding is carried much farther, in order to provide a very large surface of contact between the particle and the solution or flux, or in order to conform to the rule of maximum size of particles, stated above. In many cases, as with minerals, the maximum size of particles is fixed by causing the sample to pass through a sieve having meshes of stated dimensions. A gold ore may be ground to pass a sieve having 100 or 200 meshes to the linear inch. In GRAVIMETRIC ANALYSIS 13 such a case one should not make the mistake of grinding and sifting a portion until a sufficient quantity is passed, discarding the remainder. This would cause an error because the particles FIG. 3. Vertical section of a pile of heterogeneous material, b FIG. 4. Division by quartering. that resist grinding longest are less brittle and have a composition different from that of the particles which pulverize easily. The reason for dividing into quarters after each mixing and for selecting opposite quarters will be understood from the following: 14 QUANTITATIVE ANALYSIS Close examination of the pile of unmixed material will reveal the fact that, even after the most thorough and careful mixing, it is not entirely homogeneous. Around the circumference of the base the particles are coarser and they may be gathered toward one side. Around the apex of the conical pile there is a collection of coarser particles. If we simply dig in at random for the por- tion to be removed the lack of homogeneity will alter the char- acter of the portion removed. Fio. 5. Riffle; for automatic division of dry samples. Fig. 3 shows how quartering will properly divide a pile with coarse material over the top. Fig. 4 shows how the opposite quarters, no matter in what direction the cuts be made, will ob- tain the average of a non-homogeneous pile, while a cut into halves will do so only in case the cut is made in the direction ab. In these diagrams the conditions are purposely exaggerated. An automatic sampler, called a riffle, is illustrated in Fig. 5. The nature of its action is easily seen. In case the substance to be analyzed is a liquid, the operation GRAVIMETRIC ANALYSIS 15 of sampling is usually a simple one, consisting of thorough mixing before the removal of the proper quantity for the analysis. Solution. After the sample of solid substance has been prop- erly selected and weighed, the next operation is usually one of solution. What the solvent shall be is determined by the nature of the case. It may be water, concentrated or dilute acids, bases or salts, organic solvents, or a solid that is to act as a dissolving flux when heated with the substance. In any event it is desirable to use a relatively small amount of the solvent, not only because it must finally be entirely removed, but also because all precipitates dissolve to some extent and it is only by keeping the amount of solvent down to the least quantity that is workable that the loss of precipitate is reduced to the practicable minimum. Precipitation. The process of producing precipitates for use in quantitative analysis is one that has received much study from the quantitative standpoint. Every substance has a more or less definite solubility, under definite conditions, and it is necessary to reduce this (already small) solubility to the least possible quantity. According to the ionic hypothesis reactions between most inorganic and many organic compounds are reactions of the ions of the compounds that change. Mass Law. The reaction between potassium chloride and silver nitrate is, under proper conditions, nearly complete nearly enough, in fact, that we regard it, for practical analytical purposes, as complete. Both of these salts are highly ionized in moderately concentrated solutions. When the solutions are mixed, silver chloride is formed and precipitated. The usual expression for the reaction is KCl+AgNO 3 <=AgCl+KNO 3 . This really expresses merely the substances as originally taken for making the solution and the products as they are when removed from solution. The equation that expresses the real reaction is Ag+Cl^AgCl, (1) also, to some extent, (2) Reaction (1) is the one that is important in this instance. 16 QUANTITATIVE ANALYSIS ' The principle known as the Mass Law, as formulated in 1867 by Guldberg and Waage, 1 states that the velocity of any chemical reaction is directly proportional to the active masses of the re- acting substances. In a reversible reaction this applies to the reverse as well as to the direct reaction, and in the case already cited this means that the velocity is proportional to the concen- trations of the reacting bodies, whether these are ions or molecules. Stated symbolically for reaction (1), we have vel = kC A + g C c -i and vd=k'C Agcl * 4 where vel and vel are the velocities of the reactions4aking place in the direction of the arrows, k and k' are constants and C^g, C^ and C Agc i the concentrations, stated in weight equivalents per unit volume, of silver ions, chlorine ions and silver chloride mole- cules at any given instant. At the moment of adding the silver nitrate to the potassium chloride C Ag and C^have their maximum values while C Agc i=0. As the reaction proceeds C Agcl increases while C A + a and Cji decrease, owing to the formation of silver chloride molecules. For this reason vel decreases while vel in- creases. Equilibrium occurs when the direct and reverse re- actions proceed with equal velocity. Therefore at equilibrium vel = vel and Therefore C; g C c i = ^C Agcl = KC Agcl , where K-p ( 3 ) Solubility Product. Equation (3) is an expression of the fact that at equilibrium there is a definite and constant relation between the concentrations of the reacting bodies and of the bodies formed. The laws of solubility tell us that there is simi- larly a constant relation between the concentration of dissolved substance and that of the same substance undissolved after pre- 1 Ostwald's Klassiker, No. 104. GRAVIMETRIC ANALYSIS 17 cipitation has begun and the stable condition known as saturation is reached. In the case of silver chloride, If concentration is taken to mean the ratio of the weight of solute (expressed in gram equivalents) to the weight of solution, then C Agcl _ solid must equal. 1, no matter how much or how little solid may be present. Therefore K'C Agcl _ aolid has a constant value : K' and K'' (4) Equation (4) is an expression of the fact that in a solution satu- rated with a precipitated substance, the product of the concen- tration of the ions of this substance is a constant quantity. This product is known as the solubility product and it is of great impor- tance in quantitative analysis. This product is a maximum value for the condition known as saturation. It is exceeded when the solution is supersaturated as every solution is before precipi- tation begins. Such precipitates as calcium oxalate and silver chloride attain the solubility product comparatively quickly, magnesium ammonium phosphate slowly. According to condition (a) on page 5 it becomes necessary that the solubility product as well as the concentration of dis- solved molecules shall be extremely small, if the substance is to be useful for quantitative purposes, as otherwise the results of the analysis will be vitiated by an appreciable loss of the substance being precipitated. It should be noticed also that it is the solu- bility product that is constant and not the concentration of the individual ions. It is usually true that a single gravimetric determination concerns a single radical and not the entire com- pound; thus, in the case already considered, if the chlorine radical were being quantitatively determined the chloride would be weighed and silver nitrate used as the reagent. It is possible so to increase the factor representing the concentration of the ion which is not being determined that the other factor must assume a very low value. As an illustration, suppose that a sample of a soluble chloride is weighed, dissolved in water and a solution of silver nitrate added until all of the chlorine is precipitated except 18 QUANTITATIVE ANALYSIS that represented by an ordinary saturated solution of^silver chloride. At 18 100 gm of water will dissolve 1.7 XlO" 4 gm or 1.19X10" 6 gram equivalents of silver chloride. 1 In such a highly dilute solution ionization is nearly complete (i.e., K is very large) and since there is present practically the same number of gram equivalents of Ag and Cl as represents the number of gram equivalents of dissolved silver chloride, C A + g Cci = (practi- cally) (1.19 X10~ 6 ) 2 = 1.42 X10- 12 . This is the solubility pro- duct of silver chloride. If to this solution 0.001 gram equivalent of silver ions is added (a condition approximately attained, with- out appreciable dilution, by the addition of 3.5 cc of 5 per- cent silver nitrate solution) ^AgCci =(1.19X10" 6 ) 2 becomes upon substitution of c; g =ixio- 3 , 10- 3 Cci=1.42X10" 12 , whence C c i= 1.42X10-' Thus by adding the precipitating reagent but 3.5 cc in excess of the amount required to form silver chloride completely, the amount of chlorine remaining in solution has been reduced from 1.19 X 10~ 6 gram equivalent to 1.42 X 10~ 9 gram equivalent; stated in terms of the actual weight of chlorine, 0.00004 gm remaining in solution has been reduced to 0.00000005 gm, a quantity so small that it has absolutely no significance in any but the most refined methods of analysis, such, for instance, as are involved in atomic weight determinations. From this reason- ing follows the general rule that the precipitating reagent should always be added in a certain slight excess over the amount equiva- lent to the substance being precipitated. Of course this is done also as a matter of convenience, since it is obviously impossible to add exactly the equivalent amount of the reagent and a defi- ciency would be fatal to the success of the experiment. Attention has already been called (page 5) to the desirability of having highly ionized electrolytes for precipitates. This can now be understood since the molecular form of the dissolved sub- stance constitutes the irrecoverable portion. The effect of an excess of reagent upon the solubility of the precipitate becomes l Kohlrausch and Rose, Z. physik. Chem., 12, 234 (1893). GRAVIMETRIC ANALYSIS 19 less as we pass to precipitates whose ionization is small and becomes negligible or zero with non-electrolytes. Colloids. There is a certain class of precipitates with which we cannot deal in so definite a manner. This is the class known as colloids. Certain important distinctions are to be made between the colloids and the crystalloids. While the latter have certain very definite effects upon the freezing-point, vapor pressure and osmotic pressure of solvents, the former have little that is appre- ciable. The solubility of the crystalloids in various solvents is a tolerably definite and constant quantity, temperature being speci- fied, while there is no definite solubility for the colloids. The solubility of those crystalloids which ionize is affected in a definite and calculable manner by other electrolytes. The solubility of the colloids is affected by the presence of electrolytes but not in the same definite way. Examples of well known colloids are ferric, chromium and aluminium hydroxides, arsenic sulphide, and many organic compounds such as the albumens. It is now tolerably well established that the colloidal "solutions" contain molecular aggregates that are large enough to exhibit more of the properties of invisible mechanical suspensions than of true solutions, although there is little doubt that the difference between the pseudo solution of a colloid and the true solution of a crystalloid is one of degree of molecular association rather than of kind, since most true solutions contain associated molecules. In order to distinguish the true solution from the pseudo solution of a colloid the latter is called a "sol" and if water is the solvent the sol is a "hydrosol." The importance of colloidal sols to the analyst lies in the fact that they do not respond to the effect of excess of reagent when attempting a precipitation and that the colloid will remain in invisible suspension to an extent that would normally represent a greatly supersaturated solution. The absence of effect of excess of reagent is, no doubt, due to the absence of any but a very slight concentration of ions in the solution of the precipi- tating colloid. Colloidal sols are broken down with precipitation (flocculation) of the colloid by the addition of certain electrolytes. The flocculated colloid may, under certain conditions, return to the sol. Certain colloids will not so return after drying. Colloids of the former class are known as "reversible" and those of the 20 QUANTITATIVE ANALYSIS latter class are called " irreversible." Some examples of reversi- ble colloids are dextrine, gums, albumens and many organic colloids, also silver chloride and molybdic acid. Examples of irreversible colloids are metal hydroxides, stannic acid, arsenic sulphide and colloidal metals. Reversible colloids may often, by certain treatment, be changed into irreversible colloids. Thus strong heating causes the change of reversible silica into irre- versible silica. Enlargement of Particles of the Precipitate. Complete separa- tion of a precipitate by filtration requires that the smallest indi- vidual particles of the precipitate shall be larger than the largest pores of the filter, although any further considerable increase in size is unnecessary and undesirable, because thorough washing is then more difficult. Amorphous precipitates offer* some diffi- culty, on account of the fact that sols are likely to form and these cannot be separated by the finest filters. Many crystalline pre- cipitates, notably barium sulphate and calcium oxalate, have a tendency to precipitate in very fine crystals so that, even when the finest grades of filter paper are used, appreciable quantities of the precipitate are lost. This tendency can be partly over- come by observing certain precautions during the process of precipitation, such as adding the reagent slowly, stirring the solution vigorously, heating the solution while adding the pre- cipitant, etc. It has already been noticed that a state of supersaturation always precedes precipitation. The extent to which super- saturation occurs depends, among other things; upon the rate at which the precipitating substance is formed in the solution and this, in turn, depends upon the rate of addition of the precipitant. The formation of crystal nuclei requires an appreciable amount of time. If the reagent is added very rapidly, a relatively large degree of supersaturation is produced before the substance which is being formed begins to crystallize. If, on the other hand, the reagent is added slowly and thoroughly mixed by stirring, crystals begin to form before any considerable supersaturation occurs. This condition is very important to the formation of large crystals. If there is but slight supersaturation at the beginning of precipitation and the solution is vigorously stirred as more reagent is added, a relatively small number of crystal nuclei GRAVIMETRIC ANALYSIS 21 may serve to maintain a condition of approximate equilibrium with the solution. If large supersaturation occurs before pre- cipitation begins, and if the rapid addition of reagent is continued many crystal nuclei will form at all parts of the solution. If we regard the quantity of precipitate ultimately formed as the same in either case it is evident that the solutions from which large numbers of crystals form will produce the finer precipitates. From this follows the rule that the precipitating reagent should always be added slowly and the solution should be stirred during the addition of the reagent. Even after the most careful attention to the method of precipi- tating it occasionally happens that a precipitate is produced in such a fine state of division that a portion passes through the paper. When this occurs it is usually possible to enlarge the crystals to a size that will admit a ready separation by filtration, by allowing to stand for some time in contact with the mother liquor, with or without the application of heat. This enlargement is not merely a coherence of small particles to form larger ones, but is a real enlargement of individual crystals, the number of particles decreasing as the average mass increases. Ostwald 1 demonstrated that this change in the form of the crystalline aggre- gate is due to the fact that small crystals have a greater solution tension than the large ones. The result of digestion in contact with the solvent is that the solution containing a definite con- centration of the dissolved precipitate cannot remain in equilib- rium with both large and small crystals. If in equilibrium with the larger ones, and thus saturated with respect to these, it is under-saturated with respect to the smaller ones so that these dissolve to some extent. The solution then becomes supersatu- rated with respect to the larger crystals, and some of the sub- stance precipitates (crystallizes) on these. The end of such a process is the total disappearance of the small particles and the appearance of a smaller number of particles having a larger average size. Hulett 2 showed that the solubility of barium sulphate could be increased from 0.00229 gm per liter (the solubility of the precipitated salt) to 0.00618 gm per liter by pulverizing the crystals. 1 Z. physik. Chem., 34, 495 (1900). 2 Ibid., 34, 69 (1900). 22 QUANTITATIVE ANALYSIS Filtration. Materials generally used for filtering are combus- tible (organic) or non-combustible (inorganic). The precipitate is to be weighed later and it is necessary that some means be found for either entirely destroying the filtering medium or discounting its weight. Paper filters are, at present, used for the majority of gravimetric .analyses although several inorganic substitutes are being used to a greater extent than formerly because they pos- sess the advantage of having no action on most precipitates at high temperatures. Filter Paper. Paper for qualitative work is necessarily made with great care since it must combine the quality of considerable strength with that of a uniform porosity. For quantitative purposes the paper must have even greater uniformity of texture; it is nearly always removed by burning, after the precipitate has been thoroughly washed, and it is essential that the inorganic matter which is always found in organic fibers and left as ash shall be either in sufficiently small quantity to be negligible or that its quantity shall be uniform so that a definite weight may be subtracted from the weight as found for each precipitate. The better grades of quantitative papers have been washed with acids so that much of the inorganic matter has been removed. This results also in softening the fiber, making it a better filtering medium although more frail and subject to rupture if suction is applied. For the highest grade of paper hydrochloric acid and hydrofluoric acid are used and the weight of ash for a paper 9 cm in diameter is reduced to as low as 0.00011 gm. On account of the fact that this weight is negligible in ordinary determina- tions such paper is often erroneously called "ashless" paper. Much trouble will be obviated in the use of paper filters if some care is exercised in the selection of funnels. A circular paper is usually folded into quadrants, then opened out to form a cone, the sides of which include an angle of 60. Funnels as purchased for chemical work are supposed to be made with a 60 angle but com- paratively few are exactly of this form. Others have the correct angle between the sides but near the apex of the cone the shape is' irregular so that the corresponding part of the paper is not prop- erly supported. Reduction by Burning Paper. Certain precipitates are changed by ignition in contact with organic matter, in such a manne GRAVIMETRIC ANALYSIS 23 manner that the composition of the resultant substance is un- certain. Examples are compounds of easily reducible metals, such as silver, platinum, lead, etc. If silver chloride is heated in contact with burning paper it is partially reduced to metallic silver but the amount of reduction is not known in any given case, so that no factor can be used to obtain the weight of either silver or chlorine. In such a case the precipitate may be merely dried without strong heating, or it may be treated by a process such that the altered portion will be reconverted into the original form, or, finally, it may be filtered on a filter of some inorganic material that is unaltered by high temperatures. The first method of treatment is objectionable because it involves drying and weighing the paper both before and after filtration and wash- ing. The cellulose fiber is somewhat soluble in almost any liquid, also it is practically impossible to dry it to the same degree of hydration. This leaves the second method of treatment (re- conversion of the changed precipitate) as the only desirable one, if paper is to be used as the filter. In the case of silver chloride the paper and precipitate are dried and most of the precipitate removed and preserved. The paper is then burned in air, this reducing a compartively small part of the precipitate. This small amount of reduced silver is treated in the crucible with nitric acid, which redissolves it, and with hydrochloric acid, which reconverts the silver nitrate to silver chloride. The acids are evaporated, the remainder of the (unchanged) silver chloride is added and the whole is heated and weighed. Reduction of Pressure for Filtration. In most cases the only pressure needed or desired for causing the liquid to filter is that due to gravity. If the funnel is properly made and the paper fits well the stem will fill with liquid and this increases the speed of filtration on account of the length of the column. Some precipitates, particularly those of a colloidal nature, clog the pores of the filter and render filtration a slow and tedious process. In such cases it is necessary to use some means for diminishing the pressure beneath the liquid. The funnel is inserted in a rubber stopper, placed in the top of a pressure flask or bell jar, to which is attached a suction pump. Since there is no support for the apex of the filter paper, a supporting cone of platinum or strong paper is placed in the funnel under the paper. The cone need 24 QUANTITATIVE ANALYSIS not be used if the paper is one that has been partially parch- mentized or " hardened" by treatment with sulphuric acid. Inorganic Filters. Many chemists prefer to use inorganic materials for such precipitates as are reduced by contact with burning paper, and for this purpose the crucible devised by Gooch 1 is widely used. This piece of apparatus takes the place of both filter paper and crucible, the precipitate being either simply dried or strongly heated directly in the filter, which has FIG. 6. Filtering crucible and bell jar. previously been weighed. It consists of a tall crucible of platinum, with a bottom having fine perforations. This is placed in a holder which can be fitted to a flask or bell jar for use with diminished pressure. Into the crucible is poured a small amount of finely shredded asbestos fiber suspended in distilled water, the water is drawn out and the asbestos sucked down over the perforated bottom making a thin felt which is an admirable substitute for the usual paper. The crucible may 1 Proc. Am. Acad., Feb. 13, 1878 GRAVIMETRIC ANALYSIS 25 then be rinsed with alcohol to promote rapid drying; it is dried and weighed and is then ready for use as a filter. In using the Gooch crucible it is essential that the suction pump be running when any liquid is poured into the crucible as otherwise the felt will be stirred up and disintegrated and some of the fiber will pass through the perforations. The loss of asbestos is the most frequent source of error and even with the greatest care it occasionally happens that small amounts are lost by washing through. The asbestos for this purpose should be as nearly pure silicate as possible and free from oxides of iron or other metals. It is first thoroughly shredded, then is digested with concentrated hydrochloric acid to remove all soluble material and is finally washed free from hydrochloric acid and soluble chlorides. The purified material is kept in bottles covered with distilled water, ready for use. Asbestos cannot be used in any case where the solution to be filtered is basic because it is somewhat soluble in bases. In case strong ignition is required the crucible of platinum is fitted with a cap which covers the bottom portion, thus pre- venting any loss of asbestos during heating. The porcelain modification of the Gooch crucible was devised by Caldwell. 1 It is not well adapted to strong heating because of its liability to crack and also because the asbestos felt curls up and partly disintegrates and will inevitably cause a loss of the fiber. A method has been devised 2 for making a platinum sponge of such texture that it is suitable for filtering. This has, as yet, not found a very extended use. Experiments have also been conducted with a view to adapting porous material, such as unglazed porcelain, "alundum" (vitreous aluminium oxide), etc., to the purpose of quantitative filters. Such materials may be used extensively in the future. They possess the very decided advantage that there is no possible loss of loose material such as is liable to occur during the use of the ordinary Gooch filter. Any filtrate obtained in the processes of quantitative analysis should be received in a beaker or other vessel which has* been 1 J. Am. Chem. Soc. 13, 105 (1891). 2 Munroe: Chem. News, 68, 101 (1888) and Snelling: J. Am. Chem. Soc., 31, 456 (1909). 26 QUANTITATIVE ANALYSIS thoroughly cleaned. It is often thought that the nature of the receiver is unimportant because the filtrate is to be finally dis- carded. It frequently happens, however, that a filter paper breaks, allowing the precipitate to escape, or the precipitate is so fine as to run through the pores, or it is discovered that precipita- tion from the filtrate is not complete. In any of these cases the filtrate must be returned to the original precipitating vessel and if the receiving vessel were not clean the determination is invalidated. Washing. The soluble products of reactions of precipitation, as well as soluble impurities originally present, must be washed away from the precipitate on the filter. If the precipitate itself exerted no action upon the dissolved substances washing would be comparatively easy as will be evident if the process is considered in detail. If we assume that the filtrate is allowed to drain away from the precipitate until a definite small quantity a remains, that the wash water is then added to make a volume 6, stirred up with the precipitate and allowed to drain until volume a again remains and that this process of dilution and draining is repeated with each additional washing, then each addition of wash water reduces the concentration of dissolved matter by the fraction r of the previous concentration. If the concentration in the original mother liquor is represented by c the first addition of water reduces this to-^Xc, the second to T X^Xc = (~\ c, the nth (a\ n kj c. After draining the last wash water the quantity of soluble impurity remaining with the precipitate is(^) ca. If, for example, c = 5 percent (which is greater than the usual con- centration of dissolved impurities) a = l cc and 6 = 10 cc, then after one washing the amount of impurity remaining is j^X 0.05 XI =0.005 gm; after two washings the amount is / 1\ 2 \TO/ X0 - 05><1 =0.0005 gm; after three washings there remains 0.00005 gm. The last is a quantity that would not be appreci- able to the ordinary analytical balance. GRAVIMETRIC ANALYSIS 27 Interference by Adsorption. The above method of reasoning is not strictly valid because the precipitate itself exercises an influence upon the solution, resulting in diminished efficiency of the washing. DeVille 1 first demonstrated the fact that wherever a solution is in contact with a solid, the former is slightly more concentrated in the region adjoining the surface of contact. This difference in concentration is due to a mutual attraction between the molecules of solid and those of solute. Where this surface is merely the wall of the containing vessel the difference in concentration is not made evident by any ordinary means of measurement because the thickness of the more concentrated layer is very slight and the interior surface of the containing vessel is not large. A precipitate, on the other hand, has a very much larger surface, owing to the usual fine state of division, the same statement being true with regard to the filter paper. The portion a of the solution is therefore always one of greater concentration than c or vj n c, and the amount of impurity remaining is greater than (r) ca. This action, known as "ad- sorption," does not greatly obstruct the washing of precipitates that are decidedly crystalline in character but causes much trouble in the case of amorphous and colloidal precipitates, pos- sibly because of the very great surface possessed by these bodies. The great surface exposed by hydrosols is illustrated by the figures of the following table adapted from the work of Wo. Ostwald. 2 A cube having a length of side equal to 1 cm is subdivided into smaller cubes, with this result: Length of side Number of cubes Total surface 1 cm . . 1 6 sq cm 0.1 cm . . 10* 60 sq cm 0.01 cm 10 600 sq cm 0.001 cm (Diam. of blood corpuscles = about 0.007 cm). 1 . 0(0 . 0001 cm; Diam. of small coccus) 0.1 ft ... 10* 10 1 ' 10 1 * 6,000 sq cm 6 sq meters 60 sq meters .01 ft (limit of ultramicroscopic -visibility) 1.0 ttft ( = 0.000001 mm; Diam. of smallest colloid particles) . 1 nfi (Diam. of elementary molecules) . . 10" 10" 10* 600 sq meters 6,000 sq meters 60,000 sq meters 1 Ann. chim. phys., 38, 5 (1853). * Grundriss der Kolloidchemie, 85. 28 QUANTITATIVE ANALYSIS While colloidal particles are not to be regarded as cubes, their surface would vary with continued subdivision in the same way. At the moment of precipitation a substance having a surface, relatively so enormous, may show the effect of adsorption to a marked degree, much of soluble salts being carried out of the solution. Flocculation no doubt diminishes the surface consid- erably but the flocculated colloid still possesses a much greater surface than the same weight of a crystalline solid may have. The hydroxides of iron, aluminium and chromium retain dis- solved salts or bases with great tenacity and are extremely diffi- cult of purification by washing. The number of washings neces- sary for the satisfactory purifica- tion of a given precipitate will depend upon the nature of both precipitate and dissolved sub- stance and must be learned by . experience. A safe plan to fol- low is that of testing the wash- ings until they are found to be practically free from the sub- stance in question. In deciding for what substance the test shall be made in the washings one must be guided by the reactions that are known to take place no. 7,-Wash bottle for distilled water, during precipitation and by a knowledge of what other sub- stances may have been present with the element or radical being precipitated. This will be dealt with in each specific case in the exercises that follow. Wash Bottles. The simplest apparatus for use in washing pre- cipitates is shown in Fig. 7. It consists of an ordinary flask of convenient size fitted with tubes and rubber stopper as shown. The tube (a) may be continuous, but the flexibility produced by a rubber connection is advantageous in directing the stream of water. For use with hot water the neck of the flask should be wrapped with cork, paper, or heavy twine for the protection of the hands. The usual equipment includes at least two wash GRAVIMETRIC ANALYSIS 29 bottles, one being for hot water and one for cold water. A mis- take that is often made is that of allowing the hot water bottle to remain over a flame or hot plate when not in use, keeping the water boiling meanwhile. Boiling promotes solution of the glass of the flask so that the water may become unfit for use. The nozzle (b) of the wash bottle may be made in either of two ways. A glass tube may be drawn out until a capillary tube is produced and then cut off where the bore is such as to give a fine stream of proper dimensions. The edges are then rounded. A better method is to cut off a piece of tubing and fuse one end until it has contracted to the proper diameter. This tube possesses the advantage that it is not easily broken by contact with the funnel. For use with organic solvents that dissolve rubber a wash bottle is used, having a ground glass stopper instead of a rubber one and the delivery tube is of one piece, omitting the rubber connection. A wash bottle of any design should be so con- structed as to furnish a very fine stream of the wash liquid. The stream is directed against the upper part of the paper in such a way as to wash thoroughly both paper and precipitate. It is never directed against the funnel above the paper as the precipitate will almost invariably creep up the glass. Drying of Precipitates. Unless the precipitate is to be removed from the paper it is generally unnecessary to dry it completely before placing in the crucible for ignition. It should be allowed to drain thoroughly before removing from the funnel, after which the paper may be folded and placed directly in the crucible. It sometimes happens that the precipitate is reduced or otherwise affected by contact with carbon or reducing gases from the burn- ing paper; such precipitates must be largely removed before burning the paper and this involves previous drying in order to prevent the sticking of the precipitate to other materials with which it may come in contact. Ovens. A glance at the pages of the catalogues of dealers in chemical apparatus will impress one with the fact that there are available many types of ovens for such purposes. These FIG. 8. Incor- rect (a) and cor- rect (b) forms for nozzles. 30 QUANTITATIVE ANALYSIS types need not be described here. It is sufficient to notice that the oven must possess at least two features: circulation of air through the drying chamber in order to remove the water vapor as it is formed, and a fairly accurate means of controlling the temperature. Electrically heated ovens are more convenient than those heated by gas and, considering the length of life, are probably not more expensive. There are cases where the precipitate is affected by oxygen or carbon dioxide. Such a precipitate must be dried in an atmosphere of some gas, such as hydrogen or nitrogen, toward which it is chemically inert, and the oven must be provided with means for passing a current of gas through it. FIG. 9. Wash bottle for organic solvents. Desiccators. In addition to devices for drying at elevated temperatures we have also those for drying at the ordinary temperature of the room. Substances that are not definitely and decidedly hygroscopic will lose most of their moisture by simple exposure to the air but this is obviously an inconvenient procedure and involves much loss of time. Evaporation of moisture can be hastened in one or both of two ways without GRAVIMETRIC ANALYSIS 31 raising the temperature: (a) by enclosing the moist substance in a vessel which also contains a strongly hygroscopic material and (b) by keeping the atmosphere which surrounds the material practically free from water vapor by mechanical means, such as exhausting by means of an air pump or by passing a dry gas through the vessel. The vessel known as a " desiccator" is of such form that the drying agent, such a sulphuric acid or calcium chloride, may be contained in the lower portion, while the substance to be dried is placed above. A desiccator for drying under reduced pressure is shown in Fig. 10. FIG. 10. Desiccator for drying under reduced pressure. In order to understand the action of the various forms of desiccators it is necessary to recall the physical law that any moist substance, if confined in a vessel at a given temperature will continue to lose moisture until a definite pressure of water vapor is established, when equilibrium between liquid and vapor phases is accomplished. If the pressure of the vapor is reduced by extraneous means evaporation begins in an attempt to re- establish equilibrium and so long as the vapor pressure is kept 32 QUANTITATIVE ANALYSIS reduced evaporation continues. It is important to note, however, that the vapor pressure to be considered is not the total pressure (such as that of the atmosphere) but is the partial pressure of the vapor of water. The same result is therefore finally accomplished by pumping out the mixture of air and water vapor, by simply dis- placing this mixture by means of any gas that has been freed from water vapor or by having present and in contact with the confined atmosphere any substance that readily absorbs moisture. The simplest desiccators involve no principle other than that of con- finement with a drying agent. A small desiccator of this descrip- tion like Fig. 11 is used by the analyst, not for drying precipitates FIG. 11. Ordinary desiccator. but for keeping crucibles, precipitates and small amounts of mate- rials in a dry atmosphere, previous to weighing. In such cases the materials have already been dried and the only function of the desiccator is to prevent the absorption of moisture. The desiccator is prepared for use by partly filling the lower chamber with the proper drying agent, a triangle or perforated porcelain plate for supporting small objects being then placed upon the shoulder above. The ground-glass joint of the cover is lightly smeared with vaseline to make it impervious to air. If calcium chloride is used as the drying agent a small piece of GRAVIMETRIC ANALYSIS 33 sodium hydroxide may be added to keep the atmosphere free from carbon dioxide. Drying Agents. The drying agents commonly used in desic- cators and for drying gases are sulphuric acid, calcium chloride and phosphorus pentoxide. The efficiency of any drying agent will depend upon the pressure of water vapor that is maintained when equilibrium between the agent and the surrounding atmos- phere is established. Every definite chemical substance which combines with water maintains a definite tension of water vapor at a definite temperature. If this tension is large the compound is a poor drying agent. If as large or larger than that of the substance to be dried it does not act at all or even adds moisture to the latter. If the aqueous tension is exceedingly small the substance is a good drying agent. The rapidity of action depends also upon the relative surface exposed to the atmosphere. Thus a granular or porous solid will absorb moisture more rapidly than a liquid, its aqueous tension being the same. Phosphorus pentoxide is the most hygroscopic of the three substances mentioned above and is therefore the most efficient drying agent. In addition to its greater cost its use is also limited by the fact that it becomes viscous when moist and that a large amount of heat is evolved when it combines with moisture. Sulphuric acid ranks next to phosphorus pentoxide in efficiency but it is not much used in desiccators that are to be carried about the laboratory because of its tendency to splash against crucibles or other articles carried in the desiccator. Calcium chloride, although the least efficient of all, is the most convenient for many purposes and is generally used for desiccator's and for the drying of gases in many analytical processes. Ignition of Precipitates. The term " ignition" is used in this connection in a sense somewhat beyond its ordinarily accepted meaning, since it is applied to the heating to high temperatures of substances that are entirely incombustible. The purposes of ignition are to destroy the filter, if paper has been used, to expel the last traces of moisture and volatile impurities that have not been removed by washing and to cause the precipitate to change in a definite manner, if a change is to be made. If a paper filter has been used it is carefully removed from the funnel by slipping up the side. It is then folded as indicated in Fig. 12, the object 34 QUANTITATIVE ANALYSIS being so to enclose the precipitate that loss is impossible. If it is to be dried and removed it is then placed in the oven on a cover glass. Reducible Precipitates. The method of treating a precipi- tate that is affected by burning paper is as follows: After drying completely the paper and cover glass are placed on a sheet of FIG. 12. Folding a filter paper for ignition. glazed paper. This paper is to prevent possible loss of traces of the precipitate during removal from the filter and should be black if the precipitate is white, or white if the precipitate is dark in color. The paper is carefully opened and, by use of a spatula of horn, steel or platinum, the precipitate is carefully removed and placed in the cover glass. In doing this the pre- FIG. 13. Paper held by platinum wire for ignition. cipitate must be removed as completely as possible but the paper must not be scraped, as otherwise fiber will be removed with the precipitate. The paper is now re-folded and placed in a crucible which has been ignited and weighed, where it is carefully burned. The objectionable action of the paper, whereby some of the pre- cipitate is changed, is usually that of reduction. It is obvious GRAVIMETRIC ANALYSIS 35 that, if the small amount of precipitate remaining on the paper is to escape this action to some extent, the burning of the paper must be performed under conditions favorable to vigorous oxida- tion. Burning in an atmosphere of pure oxygen would seem to be the best remedy but this is usually impracticable. Slow combus- tion, at as low a temperature as will support combustion and with an excess of air, is easily carried out. One method is that of burning on a platinum wire. The paper is tightly rolled and held by a heavy platinum wire over the crucible which is placed on the FIG. 14. Crucible inclined for accelerating combustion. glazed paper. The paper is fired by touching with the outer edge of the burner flame and is allowed to burn slowly, the ash drop- ping into the crucible. After treating the partially changed pre- cipitate to convert it into the original condition the main portion is brushed in, using a small pencil brush of camel's hair, and the whole is heated to the desired temperature, then cooled in the desiccator and weighed. If burning on the wire is considered unnecessary the paper is placed directly in the crucible and ig- nited by placing the burner under it. The proper method of 36 QUANTITATIVE ANALYSIS heating crucibles in order to oxidize the contents is described below. The brush already mentioned should be free from loose hairs and should be rounded on the end. In using it is drawn sidewise in such a manner that very little of the precipitate enters the brush itself. Oxidation in the Crucible. The crucible is almost invariably heated by means of a naked flame, being supported on a tri- angle by means of some kind of stand. When the object is to oxidize the paper or precipitate the crucible is placed on its side and the cover leaned against it as shown in Fig. 14. The burner FIG. 15. Correct position FIG. 16. Incorrect position of inclined crucible. of inclined crucible. is placed under the bottom of the crucible in such a position that the gaseous products of the burner cannot enter the crucible. The uprising current of warm air strikes the cover and is deflected into the crucible, thus providing an oxidizing atmosphere about the paper. If the flame from the burner is applied only enough to keep the paper burning the desired condition is attained. No harm results if the volatilized combustible material from the paper burns with a flame above the crucible. After the paper is thoroughly charred the temperature is gradually raised to complete the combustion. The proper position of the crucible on the triangle is shown in Fig. 15. If placed as in Fig. 16 the crucible is liable to fall back and it may even sometimes fall through and cause a loss of the determination. Even in cases where the burning paper has no reducing action upon the precipitate it is still desirable to complete the com- bustion of the paper at a comparatively low temperature. Crys- talline precipitates that are ordinarily regarded as infusible will often undergo softening at the sharp corners of the crystals. GRAVIMETRIC ANALYSIS 37 This causes a certain sticking together which results in the enclo- sure of a small amount of carbon in such a way as to make its oxidation extremely difficult. If the paper containing the precipitate is heated to a high temperature at the very beginning it is often almost impossible to make it white. One of the best examples of this action is in the ignition of magnesium ammo- nium phosphate to convert it into magnesium pyrophosphate. Premature heating of this substance to very high temperatures will frequently result in a black or gray material that cannot be whitened by long ignition. Decomposition in the Crucible. After oxidation of the paper is completed the temperature is raised in order to volatilize completely any volatile impurities that may remain and to cause whatever decomposition is desired. Since oxidation is no longer an object the crucible is placed in an upright position and the cover is placed over the top. This gives an opportunity for the flame to bear directly on the bottom of the crucible where the precipitate lies. The cover also largely prevents loss of heat due to convection currents of air within the crucible. Porcelain Crucibles. The most commonly used crucibles are made of a high grade of porcelain. Such a crucible will withstand temperatures as high as can be attained by use of the ordinary air and gas blast lamp without more than a trifling loss of weight and they possess the decided advantage of low cost. They cannot be used for fusions because most fluxes, particularly those of a ba- sic nature, attack the glaze as well as the porcelain itself. It is because of this susceptibility to the action of fusible salts which act as fluxes that the life of a porcelain crucible is limited. In spite of the most careful washing there will always remain with the precipitate traces of fusible materials and these will, in time, cause destruction of the glaze lining of the crucible. After the crucible is thus roughened it is difficult to clean it and it becomes unsuitable for further use. Porcelain for Chemical Uses. The cutting off of imports by the war left America in a situation with regard to chemical porcelain, similar to that which existed in the case of glassware. This country had depended very largely upon foreign porcelain for chemical uses and German and Austrian ware had practically monopolized the field. American industry has since developed 38 QUANTITATIVE ANALYSIS an excellent porcelain and the Bureau of Standards has made comparative tests 1 of two American makes (Coors and Guernsey) one Japanese ware, marked "S. C. P." and two German wares (Royal Berlin and Bavarian). The tests included resistance to sudden heating and cooling, also treatment with a number of acids and bases, in solution and fused. It was concluded that the Japanese porcelain was fully equal to the Royal Berlin in every respect. American ware suffered in the heat tests, although it was stated that later samples had withstood the test satisfactorily. Platinum Crucibles. Crucibles of platinum are very desirable for ignition and are almost essential for fusions. Platinum fuses at 1770 and does not soften enough to preclude its use much below this temperature. It resists the action of all single acids if these are pure. It is readily dissolved by solutions of chlorine and, on this account, by aqua regia. Even "chemically pure" hydrochloric acid often contains traces of chlorine and will slightly attack platinum. Platinum easily alloys with most metals and so should not be heated in contact with compounds of easily reducible metals, particularly if carbon be present. When heated for a long time in contact with carbon it slowly dissolves this and becomes brittle because of the presence of the carbide of platinum. This is noticed when the crucible is heated in a reducing flame and on this account it is necessary carefully to adjust the flame so that the tip of the inner cone is below (not against) the bottom of the crucible. A. flame showing yellow must never be used. Platinum crucibles cannot be heated in contact with alkali hydroxides although they are not attacked by alkali carbonates. Compounds of phosphorus are reduced to some extent by heating with carbon and the phosphorus readily combines with platinum, causing destruction of the crucible. Ignition of phosphates thus requires especial care if this action is to be prevented. Formerly platinum crucibles and dishes contained a small percentage of iridium, added for the purpose of hardening the metal and giving greater resistance to mechanical wear. Such an alloy, however, is more readily attacked by reagents and it is more volatile and practically pure platinum, is now used almost 1 Bur. Stand. Tech. Paper, 105. GRAVIMETRIC ANALYSIS 39 exclusively. It was also formerly the custom to form crucibles and dishes by spinning the metal. This gives an article with a fine surface as apparent to the naked eye but it results in the formation of minute surface cracks and scales and disintegration is aided by this process. The crucible of the present day is given its surface by hammering. This results in a more uneven surface but the particles of metal are firmly welded together and the hammered crucible has a life appreciably longer than that of the spun crucible. Very great differences are observed in the platinum ware as furnished by different refiners and manufacturers and many troubles are experienced in the use of such ware for accurate work. The continued advance in the cost of platinum has made it correspondingly difficult to obtain satisfactory ware and a com- mittee was appointed by the American Chemical Society to investigate the subject and to make recommendations to the Society. Two reports of this committee have been made. 1 An investigation of the quality of platinum ware has been made also by the Bureau of Standards. 2 In these reports the objections to inferior platinum ware are summarized as follows: 1. Undue loss of weight on ignition. 2. Undue loss of weight on acid treatment, especially after strong ignition. 3. Unsightly appearance of the surface of the ware after strong ignition, especially in the early stages of heating. 4. Adherence of dishes and crucibles to triangles. 5. Basicity of the surface of the ware after strong ignition. 6. Blistering. 7. Development of cracks after continued heating. So far as these matters have been studied the causes are stated as follows: Loss on Ignition is chiefly due to the presence of iridium which is added to stiffen and harden the ware. The loss at 1200 is more than twice as great as at 1100 for nearly all ware used in the experiments. Appearance of the Surface after Ignition. Even with good ware the surface is frosted after strong ignition. This is due to sur- 1 J. Ind. Eng. Chem., 3, 686 (1911), Ibid., 6, 512 (1914). 2 Bur. Stand. Tech. Paper, 254. 40 QUANTITATIVE ANALYSIS face crystallization, but the crystallization should be fine and evenly distributed. Poor ware becomes, when ignited, unevenly coated with a whitish layer and with brown stains, the latter being due to small amounts of iron. Sometimes the entire inner surface becomes stained brown from the latter cause. Adherence of Crucibles and Dishes to Triangles is caused by welding of the platinum of the ware with the triangle. This cannot well be overcome but is not serious at ordinary blast-lamp temperatures unless platinum triangles are used. Basicity of the Surface after Ignition is due to the presence of traces of calcium alloyed with the platinum. Calcium oxide is produced when the crucible or dish is heated and this is made evident by testing with moist red litmus paper. Blistering is found to occur appreciably only witji ware of earlier manufacture and is not now an important objection. Cracking. No cause has been determined for this form of deterioration. Specifications. It is recommended that purchasers specify that platinum ware must show no marked uneven discoloration on heating, must give no test for iron after heating for two hours and that the rate of loss per hour at 1100 over a period of four hours shall not exceed 0.2 mg and that 5 percent of rhodium be substituted for iridium as a hardening agent. While platinum possesses properties that make it an extremely valuable metal for the chemist, its use is greatly limited by its present high cost. Because of this fact, available substitutes are always in demand. Care of Platinum. Platinum ware will deteriorate rapidly unless precautions are taken in its use and care. 1. Handle carefully to avoid bending. Use clean crucible tongs and do not allow the tongs to come into contact with fused materials within the crucibles or dishes. 2. For cleaning apply the appropriate solvent, according to the nature of the material to be removed. Chromic acid is suitable for removing organic matter, hydrochloric or nitric acids for insoluble carbonates or metallic oxides and fusing with sodium carbonate for silica or silicates, or with sodium pyro- sulphate for such metals or metallic oxides as resist the action of acids. GRAVIMETRIC ANALYSIS 41 3. Do not heat platinum in contact with the inner cone of the laboratory burner, as brittleness results from such exposure. 4. Do not heat compounds of lead, tin, bismuth, arsenic, antimony or zinc in contact with platinum. Reduction may occur, the reduced metal alloying with the platinum. 5. Do not attempt to remove fusions from platinum crucibles or dishes by means of files, glass rods or other hard tools. Use a rubber-tipped rod or solvents. 6. Dull surfaces should be polished lightly with wet emery slime or fine carborundum. Platinum Substitutes. The increasing scarcity of platinum has made the introduction of substitutes a practical necessity. While it is true that pure platinum possesses certain properties that cannot be duplicated by any other metal or alloy, yet certain alloys have been found to be suitable for making into crucibles and dishes that will serve for many of the operations of the analytical laboratory, in place of the platinum that has been in use. Two of these will be mentioned. Palau. This is a trade name for an alloy containing about 80 percent gold and 20 percent palladium. The alloy is some- what darker in color than platinum but resembles it otherwise. The melting point is 1370. This is higher than the melting point of gold but it is 400 lower than that of platinum. The Bureau of Standards found that a crucible of this material was comparatively free from iron and that its loss on heating to 1200 was less than that of a platinum crucible containing 2.4 percent of iridium. The resistance to acids, sodium hydrox- ide and ferric chloride solutions and to fused sodium carbonate is comparable with that of platinum. The ware is therefore suitable for fusions with sodium carbonate but it is decidedly attacked by fused pyrosulphates. Crucibles of palau could not be used for work at high temperatures on account of its relatively low melting point. Rhotanium. Fahrenwald 1 described a series of six gold- palladium alloys, some of which compare very favorably with platinum in many of the essential properties. Unfortunately the percentage composition of these alloys is not definitely stated but it is to be inferred that gold ranges from 60 to 90 1 J. Ind. Eng. Chem ., 9, 590 (1917). 42 QUANTITATIVE ANALYSIS percent and that rhodium is contained in some members of the series. The melting points range from 1150 to 1450 for the series. Presumably the alloy having the highest melting point contains the highest proportion of palladium but increasing palladium also increases the rate of attack by some reagents. It would appear that rhotanium of properly chosen composition might well replace platinum for most analytical purposes, excluding processes where hot concentrated nitric acid is used. The alloys cannot be used as anodes in electroanalysis. Gold shows about the same resistance to the action of reagents as does platinum, but its relatively low melting-point (1035) makes it unsuitable for crucibles or other articles that must be strongly heated. Silica. Crucibles have recently been made of fused silica. This resists the action of chemicals better than does glass and the melting-point is such that no ordinary air-gas flame will fuse the crucible. Pure quartz fuses at 1600 and amorphous silica at 1750-! 780. Moreover the coefficient of expansion is very small (5.4X10~ 7 at temperatures between and 1000) so that sudden heating or cooling does not cause cracking. Silica cru- cibles have not yet taken the place of those of porcelain, largely because of their higher cost. Alundum. Recently a highly refractory form of aluminium oxide has been utilized for crucibles. This is commercially known as "Alundum." Alundum does not fuse below 2050, does not much soften at 1775 and its coefficient of expansion is 7.8 X10- 6 . 1 Triangles. The discussion concerning the relative merits of various materials used for crucibles will apply equally well to those used for supporting triangles. Porcelain is the cheapest serviceable material and will do for supporting crucibles of any other material as well as those of porcelain itself. The familiar "pipe stem" triangle is constructed of three tubes of refractory, unglazed porcelain, held together by a frame of iron wire (Fig. 17). The chief objection to this form lies in the fact that the relatively large tubes, lying on three sides of the crucible, obstruct the flame and cause a very noticeable decrease in efficiency. An improved form is shown in Fig. 18. The projections serve to sup- 1 Saunders: Trans. Am. Electrochem. Soc., 19, 333 (1911). GRAVIMETRIC ANALYSIS 43 port the crucible and allow much better contact with the flame. Any porcelain triangle becomes practically useless if cracked because the supporting iron frame is thus exposed to the flame and soon oxidizes and breaks. FIG. 17. Common "pipe stem" triangle. Platinum triangles offer the great advantage of long life and they also obstruct the flame very little on account of the small size of the framework. Here again the high cost of platinum pre- cludes its extensive use. A common form of platinum triangle FIG. 18. Triangle with projections. is shown in' Fig. 19. A less expensive form can be constructed by stretching heavy platinum wire on an iron framework, like Fig. 20 or Fig. 21. This is not very satisfactory because the weight of the crucible causes lengthening and sagging of the wire 44 QUANTITATIVE ANALYSIS at high temperatures. The screws shown in Fig. 21 can be used in taking up this slack but the wire eventually weakens and breaks. Alloys of nickel and chromium having high fusing points and considerable resistance to oxidation at high temperatures have been adapted to the construction of chemists' triangles. The Fio. 19. Heavy platinum triangle. proportion of the two elements may be varied somewhat but the presence of iron, even in small amounts, makes the alloy oxidizable in the flame and the resulting oxide combines with the silicious glaze of porcelain crucibles, thus changing the weights of the latter. As it is becoming increasingly difficult to obtain Fio. 20. Triangle of platinum wire. nickel-chromium triangles that are free from iron it is likely that their use must soon be restricted to qualitative work or other work at not very high temperatures. Probably the best triangle that is now obtainable at a moderate cost is one of the pipe-stem form, made from silica tubes on a GRAVIMETRIC ANALYSIS 45 frame of nickel wire. These may be used with either porcelain or platinum crucibles and they are not easily cracked by sudden heating or cooling. FIG. 21. Another form of platinum wire triangle. Burners. The burner that is to be used by the analyst may be anything from the cheapest and simplest burner of the Bunsen type to the most expensive and complicated burner obtainable. The purchaser has his choice and probably certain advantages FIG. 22. Alloy triangle. are possessed by each burner. The only feature that is really essential is independent regulation of air and gas supply. The requirements are quite different in different cases and the analyst must have at his disposal all kinds of flame, from the yellow illumi- 46 QUANTITATIVE ANALYSIS nating flame to the most intensely hot and oxidizing flame, and he requires very small and very large flames of each class. In order to obtain this variety of flame there must be some method of regulating the gas supply without changing the pressure at the gas valve, since this also changes the amount of air drawn in at the mixer. The simplest form of Bunsen burner does not permit this gas regulation without unscrewing the upper tube and chang- ing the gas jet by the use of pliers. Such regulation is not pos- sible in practice. FIG. 23. Section of E. and A. burner. FIG. 24. Section of Teclu burner. Two forms of adjustable burners are shown in Figs. 23 and 24. In the E. and A. form the gas flow is regulated by the large milled disc near the bottom of the burner and the air supply by the ring which screws up and down at the base of the burner tube. In the Teclu burner (Fig. 24) the gas is controlled by the screw on the side of the base while the disc at the bottom of the cone controls the air supply. It is important to note that in both of these burners the regu- lation of gas flow is not accomplished by altering the pressure GRAVIMETRIC ANALYSIS 47 under which it is delivered but by changing the size of the orifice in the burner. The maximum pressure is thus used at all times and the result is a better mixture of gas with air than is obtainable by regulating the gas cock of the supply line. A very common error on the part of students lies in careless- ness with regard to the regulation of flames. If a relatively cool flame is required and if a deposit of carbon is not objection- able the air should be excluded from the mixer. If, on the other hand, the highest efficiency of the burner is desired, careful regulation of the air and gas is necessary. The inner blue cone should be well defined and it should not show a yellow tip. If air is admitted more than that required to completely burn the gas with production of a blue flame, the result is a roaring and fluttering flame. This means that more air is being admitted than can be used and this air, in being heated by the flame, lowers the temperature of the latter. Blast Lamps. The blast lamp is used for obtaining tempera- tures higher than are attainable by use of the ordinary burner. In this burner the gas is used in large quantities and air is delivered under pressure so that a flame is produced more intensely hot than the slower burning flame of the common burner. Where extremely high temperatures are necessary the oxyhydrogen blow- pipe is used but this is rarely the case for analytical operations. Meker Burner. A somewhat radical departure from the older types is found in the Meker burner. This is shown in section in Fig. 25. The air is drawn in through several holes in the base of the tube. The delivery of the gas under pressure into the inverted cone which forms the burner tube causes a greater reduction of pressure within the tube than is the case with burners having cylindrical tubes. The result is a greater inflow of air, making possible the combustion of a greater amount of gas in a given space, and also more complete mixing of gas and air. The nickel grid through which the mixture flows at the top of the burner causes the gas to burn exactly as though each mesh were a small individual burner. The tip of the inner reducing cone of each small flame is usually about one millimeter above the top of the burner and, as all of the small flames unite to form one large one, the result is a highly concentrated flame, every 48 QUANTITATIVE ANALYSIS part of which is oxidizing in character except a zone of about one millimeter in depth, immediately above the top of the burner. This is a distinct advantage, especially in heating platinum articles, since platinum is easily damaged by heating in a re- ducing flame. FIQ. 25. Section of M6ker burner. The flame of the Me*ker burner is nearly as hot as that of the ordinary blast lamp using the same gas and it may be substi- tuted for the blast lamp in many cases. There is also a Me"ker blast lamp, similar in construction to the one already described but using air under pressure. Fusion. For the purposes of quantitative analysis the fusion of materials is almost always accomplished with the end in view of producing more soluble substances through the interaction of GRAVIMETRIC ANALYSIS 49 an added agent, called a flux, and the refractory material. For instance, most of the natural silicates are practically insoluble in water and all ordinary reagents and therefore they cannot be analyzed by ordinary methods. By a preliminary heating to a high temperature in contact with a basic substance like sodium carbonate, a fusible mixture of new compounds is formed and these will, for the most part, be soluble in water and hydro- chloric acid so that the solution may be submitted to precipita- tion and filtration processes for the separation and determination of the elements. Similarly, refractory and insoluble metallic oxides may be heated with sodium pyrosulphate with the for- mation of a fused mass consisting of soluble sulphates of the metals. The necessary qualities of any useful flux are (1) that it must be of such a nature as to be capable of reacting with the re- fractory body when heated with it and (2) that the resulting compounds shall fuse at the prevailing temperature. To these the analyst adds a third requisite: (3) that the resulting com- pounds shall be soluble in water or in the laboratory reagents. The first condition is met by choosing as the flux a substance of opposite nature to that of the refractory sample. That is, if the latter is of an acid nature (as silica and polysilicates) the flux should be basic, and conversely. No general statement can be made with regard to the relative fusibility of various compounds, as based upon the chemical composition of these compounds. It may be noted that re- fractory silicates are usually made more readily fusible by reduc- ing the ratio of silica to metal oxide by introducing more metals, and particularly by the introduction of the alkali metals. Both of these points are made by using alkali metal carbonates as fluxes, since the net result of the reaction at high temperatures is to expel carbon dioxide and to combine the alkali metal oxide with the refractory silicate. This will explain why these carbonates are almost always chosen as fluxes for silicates. A reaction such as the following may occur when orthoclase is fused with sodium carbonate: 2KAlSi 3 O 8 +5Na 2 CO 3 -+K 2 SiO3+5Na 2 SiO 3 +2NaAlO 2 +6CO 2 , 50 QUANTITATIVE ANALYSIS a more or less complicated mixture of aluminates and silicates of the alkali metals being formed. Basic Fluxes. Sodium carbonate, potassium carbonate and the double sodium-potassium carbonate are the most important of the basic fluxes that are used for analytical purposes. These are used chiefly for fusion with silica and the refractory silicates. Such fluxes as calcium oxide, used for fluxing silicates in the blast furnace for iron, are of little use for analytical purposes, partly because the resulting compounds are not soluble and partly because metals that are to be determined in the sample are introduced by the use of such materials. Acid Fluxes. Fluxes of an acid nature are valuable chiefly for forming fusible, soluble compounds when heated with metallic oxides or salts that are over-saturated with metallic oxides. The most useful of such fluxes are the pyrosulpHates and the biborates of sodium and potassium. Acid sulphates are often used instead of pyrosulphates. When the former are heated they give off water and they are completely converted into pyrosulphates by heating to higher temperatures : 2NaHS0 4 ->Na 2 S 2 7 +H 2 O. Because of the excess of sulphur trioxide in the pyrosulphate, this readily reacts with metallic oxides when heated with them: The biborates likewise combine with metallic oxides because of their excess of boric anhydride. Fe 2 O3+3Na 2 B 4 O 7 ->2Fe(BO 2 )3+6NaBO2. Weighing. The balance is the only kind of weighing apparatus that is independent of the numerical value of the force of gravity so that it actually measures mass and not weight. In its simplest form it consists of a beam, supported at its middle point on a fulcrum, having suspended at its ends, at equal distances from the middle fulcrum, two platforms for holding, respectively, the objects to be weighed and the weights. It is thus seen to be an apparatus for comparing masses and if it is mechanically perfect and if the standard weights are correct, then the mass of the object is given by the mass of the weights that counter- balance it when the balance is at equilibrium. GRAVIMETRIC ANALYSIS 51 Description of Balance. The balance as used by the analyst is a much more sensitive and carefully built piece of apparatus than the ordinary balance and must be capable of giving weights that are correct to within 0.0001 gm. In order to reduce the friction of the bearings to the lowest possible value these are made in the form of a knife edge supported by a polished block-, the material of the bearings being some substance that is quite hard, this usually being agate. The method of weighing is to place the FIG. 26. Analytical balance. object to be weighed in one pan and to add weights to the other until equilibrium is attained, judging this condition by observing when a pointer, attached to the beam, swings equal distances to the right and left of an experimentally determined "zero point" on a scale. The whole balance is enclosed in a glass case in order to exclude dust and to prevent the interference of air currents. The case is provided with levelling screws and a spirit level or plumb bob. To prevent needless wear on the knife edges the beam and pans should be provided with suitable rests so that the knife edges may 52 QUANTITATIVE ANALYSIS not be in contact with their bearings when the balance is not being used. These rests are necessary also for the proper control of the action of the balance, as will be noticed when exercises with the balance are described. The rests for the beam and for the pan supports are usually operated by one piece of mechanism. The exact construction of this varies in different balances but the action of the beam rests belongs to one of two classes, the vertical and the circular. In the first (Fig. 27) the rests move vertically upward. The chief defect of this action is the fact that if the beam is caught at any position except a horizontal one the knife FIG. 27. Beam rests with vertical action. edges are caused to slide upon their bearings, causing unnecessary wear. In the circular action (Fig. 28) the arms of the arrests have the same length as the arms of the beam and they move about the same axis. No matter in what position the beam is caught or how sudden the motion of arrest, no damage results to the knife edges. All designs of mechanism for this purpose include devices for automatically placing beam and pan bearings in their proper position in case they have been accidentaUy twisted out of posi- tion. These are shown at (a), Fig. 28. The rests under the pans are not for the purpose of lifting knife edges from their bearings, but merely for steadying the pans and for controlling the move- ments of the balance when it is in use. In some balances these GRAVIMETRIC ANALYSIS 53 are operated by the same mechanism that operates the other rests, while in others a separate knob is provided. Sensibility. The sensibility of the balance is stated in terms of the displacement of the beam by a given excess of load in one pan. More specifically, it is the number of scale divisions of displace- ment of zero point by an excess of 1 mg in one pan. The sensibility is affected by a number of factors. The bal- ance, in order to possess stability, must have the center of gravity of the moving parts slightly below the point of the middle knife edge. This distance is one of the determining factors of the sen- FIG. 28. Beam rests with circular action. sibility. If large, the sensibility must be relatively small, since a given displacement of the zero point will involve a relatively large displacement of the center of gravity and will, in conse- quence, require a greater difference in load. Every balance has some provision for arbitrarily altering the sensibility by altering the location of the center of gravity. The most common device is a weight that can be moved up and down the pointer. The three knife-edge bearings must lie in the same plane, as well as parallel to each other. Since the pans swing freely upon their own bearings, the whole load of the pans is applied at these points. If the plane of these bearings were below that of the middle bearing it could easily be that the center of gravity would lie between these planes and then an increase in load 54 QUANTITATIVE ANALYSIS would lower the center of gravity with reference to the central bearing and thus decrease the sensibility. If their plane were above that of the middle bearing an increase in load would raise the center of gravity to a point above the middle point of support and give instability to the balance so that if dis- placed from its normal horizontal position the beam would not return. When the three bearings lie in the same plane an in- crease in load will raise the center of gravity but can never raise it to the level of the middle knife edge. The above method of reasoning supposes that the balance beam is perfectly rigid, a property that is never attained in practice. Increase in load, FIG. 29. Illustrating the principle of moments. therefore, does actually cause decrease in sensibility because the beam is somewhat distorted, causing the center of gravity to be lowered. In order to combine great strength with lightness of weight and so minimize the distortion of the beam, makers have tried many designs and many alloys in the manufacture of balance beams. Another property that has a large influence upon the sensibility is the length of the arms of the beam. It is a well known principle of physics that when a balance is in equilibrium the product of the weight of one side into the length of the corresponding arm must equal the product of the other weight into the length of its arm. This is the "principle of moments." In Fig. 29, aw = a'w'. The greater the inequality of the statical moments (aw; and a'w') when an excess of weight lies on one side the greater GRAVIMETRIC ANALYSIS 55 will be the displacement of zero point and this inequality will be greater when a and a' (the lengths of the arms) are large. Lengthening the arms also causes slower swinging so that sensi- bility gained by this means results in loss of time in weighing. This fact sets a practical limit to the length of the beam. The following properties have been discussed as having an influence upon the sensibility of the balance: 1. Distance between the center of gravity and point of support. 2. Coincidence of the planes of the three bearings. 3. Length of the arms of the beam. 4. Reduction of friction to a minimum by finely ground knife edges. In addition to these should be mentioned the weight of the beam. A heavy beam makes a balance that is less sensitive than a balance having a light one. In addition to those features which affect the sensibility, certain others are essential if the balance is to weigh accurately. It is extremely difficult to construct a beam having absolutely the same distance between the central knife edge and the two end ones. Obviously any difference involves a slight difference in the two weights required to bring the balance to equilibrium. From the equation aw = a'w', if a^a' then w^w' and the weight of the substance which is being weighed is not the same as that of the weights which counterbalance it. The relative lengths of the arms must be determined for each balance and the observed weights corrected if these lengths are appreciably different. Even if the discrepancy in lengths is sufficiently small to be negligible it may be magnified by a change in temperature or by a change in load, if the beam is not absolutely uniform in material and structure. This last condition is impossible of attainment in practice. It will be made clear in the course of the work in gravimetric analysis that it is not the absolute weight that is important in most cases but only relative weights because the object of quantitative analysis is to determine the proportionate parts of the constituents of a compound or mixture. From the equation aw = a'w', if a and a' are not equal the error in w' bears in all cases a definite ratio to w 1 '. While the balance gives the true mass of the object and is in- dependent of the magnitude of the force of gravity, this expres- sion is true only if the buoyancy of the air acts upon weights and objects alike. This can be the case only when the density of 56 QUANTITATIVE ANALYSIS weights is the same as the density of objects, a condition that is not fulfilled in the majority of cases. A correction must there- fore be introduced in such cases in order to find the true weight of the object. The amount of correction is negligible in gravi- metric analysis and becomes serious only when the total weight is considerable. The method for applying this correction will be explained in the section dealing with volumetric analysis. (See p. 180.) Weights. Sets of analytical weights as purchased frequently include weights as small as 1 mg. These are rarely used because the balance provides a more convenient method for making the final adjustment, in the form of a " rider" or small weight of fine platinum or aluminium wire which may be shifted to various positions on the beam. The manner in which the FIG. 30. Balance beam. beam is divided varies with the balances of different manufac- turers. The lowest subdivision should be at most 0.1 mg. The weight of the rider will depend upon the manner of numbering the milligram divisions and the weight will be represented by the number which is directly over the terminal knife edge. This is because when the rider is placed on this division it is essentially the same as though it were in the pan below. To correctly indicate weights it must then weigh the number of milligrams indicated by this division, which may be 5, 6, 10, 12, or any other number. A mechanism is provided for lifting and ad justing the rider with- eniout opng the balance case. This may be a simple sliding hook, or an elaborate carrier, such as are found in more expensive GRAVIMETRIC ANALYSIS 57 balances. It is essential that the carrier be capable of quickly and easily shifting the rider without danger of throwing it from the beam. The Chain Rider. The "Chainomatic" balance entirely dis- penses with a separate rider. One end of a small gold chain is permanently attached to the balance beam. The other end of this chain is fastened to a hook which may be moved up and down a scale, this action being controlled by a knob outside the balance case. Movement of the hook on the scale varies in a definite manner the length of side of the loop which is supported Fia. 31. Usual form of a set of analytical weights. by the beam and this may be adjusted while the beam is in motion. This is a distinct advance in balance design, although this improvement adds considerably to the cost of the balance. Every balance is rated for a certain maximum load, it being understood that this is the load for each pan and not the total load. The normal load is fixed by the strength of the knife edges and by the capacity of the beam to resits deformation under stress. If the knife edges are short and ground to exceeding fineness they are injured more readily by a load than if they are slightly more blunt. If the beam is overloaded it is temporarily deformed to 58 QUANTITATIVE ANALYSIS such an extent that there is an unusual loss of sensibility, due to the excessive lowering of the center of gravity. It is thus evi- dent that the weighing of heavy objects requires correspondingly more sturdy balances and these will, of course, be less sensitive. The usual form of a set of metric weights is shown in Fig. 31. The largest weight should not be heavier than the maximum load for which the balance is rated and the least weight should be such that, used in conjunction with the rider, 10 mg may be made up. The larger weights are constructed of brass or bronze, plated with platinum or gold to prevent corrosion. The fractional pieces are of platinum in the better sets or of aluminium in the cheaper ones. Even with weights plated with platinum or gold it is comparatively easy to damage the surface by careless handling or by allowing chemicals to touch the weights. Printed directions for setting up always accompany a new balance. The following rules deal only with the balance set up and ready for use. Cleanliness. The pans, beam, bearings and all other parts inside the glass case must at all times be kept free from dust and chemicals. A camel's hair brush 1 inch wide should be provided for this purpose. It is not permissible to weigh any soluble mate- rial in direct contact with the pans because some of this will in- variably stick to the pan and eventually it may cause corrosion. Volatile acids must never be brought inside the balance case un- less securely stoppered in an air-tight container. Adjustment. The balance is levelled by means of the screws provided for that purpose. Examination is made to determine whether the knife edges are in the proper position with respect to their bearings. The pan rests are released to determine whether the pans hang vertically from the stirrups, or whether they swing horizontally when released. If so this swinging is stopped by momentarily touching the pans by the pan rests, repeating the operation until the pans hang quietly upon release. To Set the Beam in Motion. Various methods are used for starting the oscillation of the balance about the central bearing. One pan may be lightly touched with a small camel's hair brush. This is not an easy process to carry out properly because it is difficult to control the impulse given to the pan. Another method is to raise the door and fan one of the pans slightly with GRAVIMETRIC ANALYSIS 59 the hand. This is open to the same objection as is the first method and, in addition, it defeats the primary aim of the glass case which is to prevent the interference of air currents. Even the slight current started by the hand does not at once die and it must be a disturbing influence for some time after the balance case is closed. A better method than either of the above mentioned is to lower the rider to the beam just before releasing the latter, then to catch up the rider with the carrier after releasing and allowing the proper start. A short practice will enable the operator to give just the desired impulse to the beam to make the pointer swing over from five to ten divisions on either side of the zero of the scale. The rests should be so adjusted that the three knife edges are lifted from their bearings when the rests are raised, but the distance between the edge and bearing should be barely perceptible. If this distance is unduly large the shock to the delicate knife edge is so great that this edge is soon dulled or chipped with a consequent loss of sensibility. To Determine the Zero Point. The zero point may be de- fined as that point on the scale at which the pointer would eventually come to rest from swinging over the scale. It is never observed by allowing the pointer to actually come to rest because such in- fluences as minute air currents would either prevent this con- summation entirely or would cause the observation of a fictitious zero point. The effect of these influences is counteracted by allowing the pointer to swing a number of times to the right and to the left, taking the average of the indications. Proceed as follows : If the balance operates all of its rests by one mechanism care- fully lower these rests and set the beam in motion as directed above. If the pan supports are controlled by a separate button lower the beam and stirrup rests first, then the pan rests. Allow the pointer to swing three or four times in each direction and re- cord the number of scale divisions over which it swings, taking the last reading on the same side as the first. Record in two columns and take the average of each column. Subtract the less average from the greater and divide the remainder by 2. This gives the zero point if the proper direction is assigned to it. 60 QUANTITATIVE ANALYSIS Example: Left Right 8.25 7.75 7.25 7.00 . 6.75 6 50 6.00 5.50 5.00 4.75 4.50 Average 7 . 25 Average 5.15 rj rtr r I r - = 1.05. Therefore the zero point is 1 division to 2 the left of the zero of the scale. Two methods of procedure are now open to the operator. He may either make his weighings with reference to this observed zero point or he may adjust the balance so that the observed zero point is the actual zero of the scale, using for this purpose the small screws provided on the ends of the beam. The first method is preferable for the beginner because any attempt to change the adjustment will probably result in more serious derangement. After skill has been gained by practice time will be gained and much calculation will be saved if the observed zero point is adjusted to coincide with the ideal zero of the scale. A method for making a close approxi- mation of the zero point without resorting to calculations on paper will be explained in a later paragraph. The zero point changes and it must be determined each day, or more often if necessary. To Determine the Sensibility. The sensibility in the case of the analytical balance has already been denned as the number of scale divisions that the zero point is displaced by an excess in weight of 1 mg on one side. That the sensibility varies with the total load has already been explained. To determine the sensibility with zero load, first determine the zero point of the balance. Place the rider on the beam at the division marked 1 and redetermine the zero point. The difference is the sensi- bility. Determine the sensibility when both pans are loaded with 5, 10, 25 and 50 gm, respectively. Record the results on a card and place this in the balance case for future reference. GRAVIMETRIC ANALYSIS 61 To Determine the Relative Lengths of the Arms. Weigh a small object, such as a crucible, placing it on the left pan and the weights on the right, then place the object on the right pan and the weights on the left. If the arms of the balance are of unequal length these weights will not be the same. Let W = the true weight of the object, TF' = the sum of the weights when the object is on the left pan, a = the weight added to W when the object is placed on the right pan, r = the length of the right arm and Z = that of the left arm. From the principle of moments W'r = Wr=( WW'r 2 =W(W'+a)l 2 This is the proper method for testing the equality of arms. If the arms are equal then a = and 7 = 1. If a is negative, < 1 and the right arm is therefore shorter than the left, while if a is positive -7 > 1 and the right arm is the longer. As has already been explained, the question of equality has little or no importance for purely analytical work. Inasmuch as the chemist has uses for his balance in other lines of work it should be tested. To Weigh. With all of the rests raised the object to be weighed is placed on the left pan by means of the crucible tongs or by some other method that avoids contact with the fingers. The pan rests are lowered and raised momentarily until the pans stop swinging on their bearings, then these rests are lowered and fastened down unless all of the rests are governed by one mechanism. By means of the weight forceps place one of the weights, judged to be somewhat heavier than the object, on the right pan. Lower the beam rest slowly until the pointer just starts to move to the right or left. If to the right the weight is too light, if to the left it is too heavy. If it is too light raise the 62 QUANTITATIVE ANALYSIS beam rest and exchange the weight for the next heavier one and repeat the trial until the first weight is found that is too heavy. Remove this and replace by the next lighter one. Continue the addition of weights, trying after each addition and adding always the next consecutive weight lighter than the one used in the preceding trial. When the weights below 1 gm (milligram pieces) are reached the difference between the loads on the two pans is so small that the pan rests will now readily control the movements of the balance. If these rests are operated by a separate knob they are then raised and the beam and stirrup rests lowered, and the process of trial is continued until the range covered by the rider is reached. The balance case is now closed, and the pans again steadied if they have shown a tendency toward lateral swinging. Make the preliminary ^ rider trials by placing the rider on the division estimated to be nearest the proper one, slightly releasing the pan rests until the pointer starts to move. Arrest the motion and move the rider by whole milli- gram divisions to the right or left, as may be required, until the milligram nearest to the correct weight is reached. Estimate the proper fraction of a milligram, set the beam in motion as already directed and drop the rider on the estimated division. By observing the swinging determine the zero point and calcu- late from this and the sensibility with this load what change should be made in the position of the rider to bring the balance to equilibrium on the zero as determined with no load. Repeat the trial with the rider on this calculated position and shift if necessary to obtain the exact weight. Raise all of the rests and read the weights as follows: Observe the empty places in the box. If the weights have been systematically placed in the box, none being removed except those on the pan, the empty places will give the correct weight. Record this weight in the data book. Confirm by counting the weights as they lie on the pan. Reconfirm by counting as they are removed and replaced in the box. This gives three readings and if these are carefully made a mistake is practically impossible. In recording the weights mental addition may be made if they are taken in order, pro- ceeding from the larger ones to the smaller ones. This is because, as the sets of weights are made, no one order of digits can total more than 9. Each order can be mentally added and recorded GRAVIMETRIC ANALYSIS 63 with the certainty that no other order will change the one read. Thus, if there are on the pan the following pieces : 10 gm, 5 gm, 1 gm, 200 mg, 100 mg, 50 mg, 20 mg, 10 mg, and 5 mg, and on the rider 3.2 mg, we should read and record thus: Of whole grams 16, of tenths (100 mg) 3, of hundredths 8, of thousandths 8, of ten-thousandths 2. Writing in the same order we should have 16.3882 gm. Use of Rests. The beam and stirrup rests must be used when changing weights heavier than 1 gm for two reasons: First, the shock of the weight against the pan must not be allowed to communicate itself to the knife edges when on their bearings and second, the pan rests are held up by a spring that will not support an excess load of more than 1 gm in one pan. In other words, these rests would not control the balance at this point in the experiment. When the milligram pieces are being ex- changed the shock of impact with the pan is so small that the knife edges are not damaged and the pan rests offer an easier method of control. Trial of Weights. In making a trial of a weight the pointer should be allowed to move only far enough to indicate the direc- tion of motion. This indicates the proper change to be made in weights as well as if it were allowed to move half way across the scale and it does not derange the balance. Estimation of Zero Point. It has been stated that in later work extended calculations of zero point would not be made. On account of the resistance due to friction with the air and in the bearings, any balance decreases the amplitude of vibration with each successive j ourney . The amount of such decrease varies with different balances but a close approximation can be made by simple observation. If the zero point of the unloaded balance has been adjusted to coincide with that of the scale, in the final adjustment of weights the loaded balance can also be brought to this ideal zero point without the necessity of extended cal- culations, by simply noting that the distance to which the pointer swings in one direction is a certain (approximate) fraction of a division less than the distance in the opposite direction on the next preceding journey. In many cases of quantitative analysis, if the longer (even though slightly more exact) method involving calculations of zero point were followed, the time consumed 64 QUANTITATIVE ANALYSIS in weighing would be so great that the weight of the object would change appreciably while on the balance, owing to absorption or evolution of water, carbon dioxide, etc. To Obtain a Specified Weight of Sample. It is often desirable, in order to simplify calculations, to use a certain specified weight of sample, as 1 gm, 0.5 gm, 5 gm, etc. This is a difficult operation if the ordinary method of weighing by difference is used, because the sample that is to be used is poured from a weighing bottle and if too much is inadvertently poured out it is not easy to return the excess without loss. If the sample is dry and is unaffected by free contact with air one can dispense with the weighing bottle and weigh directly on a watch glass or scoop. For this purpose one may obtain "counterpoised watch glasses," which are pairs of glasses the members of which possess so nearly the same weight that they can easily be balanced by means of the rider so that their weight does not enter into the calculation. The method of obtaining a predetermined weight of a sample is as follows: The glasses are placed on the pans and exactly balanced, if necessary, by using the rider. Weights totaling the desired quantity are placed on the right glass and the pan rests lowered, steadying the pans at the time if necessary. The beam rests are now lowered just enough to allow the central knife edge to come to its bearing and the pointer to perceptibly move to the left. The balance door is lowered about half way and then the sample, in a fine state of division, is carefully poured on the left glass until but a slight excess is obtained, as evidenced by the swing of the pointer to the right. (The total length of wing as allowed by the beam rests should nob exceed two scale divisions as otherwise the adjustment of the balance may be disturbed.) Using the spatula, sufficient sample is now removed from the glass to make this side slightly too light. This is held over the glass and, by gently tapping the spatula, the sample thus held is gradually dropped to the pan until equilibrium of the balance is nearly or quite attained. The excess is dis- carded. By repeating this process once or twice apparent equilibrium is quickly attained and this is confirmed by closing the case and determining the zero point by the usual method. In this way any desired weight of sample can be obtained in a comparatively short time. The degree of accuracy with which it is fina GRAVIMETRIC ANALYSIS 65 it is finally weighed will depend, as in all other cases, upon the nature of the sample, the total weight being taken and the degree of accuracy possible at other points in the analysis. To Correct the Observed Weight for Inequality of Arms. Certain investigations in physics and physical chemistry require that the weight found shall be the absolute weight, correcting for the inequality of arms and for buoyancy of the air. The former correction need not be made for analytical work but where necessary either of the following methods may be used. Method of Gauss. Weigh the object first on the left pan and then on the right. Let W be the true weight, a the weights required to counterbalance when the object is on the left pan and b the weights when the object is on the right. By the principle of moments: Wl = ar U = Wr W 2 lr = ablr Therefore the true weight is the square root of the product of the two observed weights. Where the inequality of arms is very slight (the usual case) the arithmetical mean of the two weights is a sufficiently close approximation to the square roo't of the product. Method of Borda. This is also known as the method of substitution. Place the object to be weighed on one pan and counterbalance with any other material, such as similar weights or dry sand. Remove the object and substitute accurate weights until the balance is again in equilibrium. These weights are necessarily the same in value as the object for which they substitute, irrespective of the relative length of the arms. Use of Arm Ratio. Weigh the object, as usual, on the left pan with the weights on the right. Multiply the observed weight by the ratio v- This gives the true weight. If the arms of the balance are equal in length, v = 1 and the observed weight is the true weight. 5 66 QUANTITATIVE ANALYSIS Calibration of Weights. For analytical purposes it is not necessary that the various pieces in a set of weights shall have the exact values indicated by the stamp. This is because an analysis is always reported as a percent or as some similar ratio. The only requirement is, therefore, that the pieces shall have the correct relation to each other. That is, the piece marked "1 gm" need not weigh exactly one gram. Indeed it might conceivably have any other value that is reasonably near to one gram; but it is necessary that its weight shall be one-tenth of that of the piece marked " 10 gm, " ten times that of the piece marked "100 mg," etc., or that the deviation from these ratios be known and corrected in calculations of the weights of objects. Commercial weights are seldom accurately adjusted unless the cost is high. Therefore a calibration should always be made and correction applied in such cases as are made necessary by errors in manufacture. Also, if standard pieces are available it is desirable to correct each set to true gram values, rather than to merely relative values, since the chemist frequently uses his weights for other than analytical purposes. Before beginning the calibration see that in all cases where there is more than one piece of a given denomination the different pieces bear some distinctive mark. A good plan is to make small dots by means of a prick punch. This marks without damaging the plate or changing the weight of the pieces. One of the 10-gram pieces may be marked () and the other (); the three 1-gram pieces (), ("), ()> an d similarly with other dupli- cate or triplicate pieces. Calibration of weights is essentially a comparison of the dif- ferent pieces of a set with each other and with a standard, followed by a calculation of either relative or absolute values. The method to be followed in making this comparison will depend upon whether the arms of the balance have been found to be of equal length, within reasonable experimental limits. The analytical problem sometimes requires an accuracy represented by a maximum error of 0.001 percent, although the maximum permissible error is usually larger than this. In order to meet these requirements the weights should be calibrated with the same degree of accuracy and if the ratio of arm lengths, -,- is GRAVIMETRIC ANALYSIS 67 not greater than 1.00001 nor less than 0.99999, calibration by direct comparison of weights on the two sides of the balance may be employed. If the ratio of arms is not within these limits the method of direct comparison may still be used, pro- vided that a correction is applied. The apparent weight of the piece on the right side, in terms of the piece on the left, is mul- tiplied by the ratio j> this giving the true comparative values. The principle of Gauss may also be used in making the com- parison on a balance having unequal arms but this involves so many calculations that it is generally better to use the method already described, or else the method of substitution, the latter being based upon the principle of Borda. Since these methods are independent of the arm ratio the latter need not be determined. Finally it may be stated that the most accurate values are ob- tained by comparing each piece with a standard piece of the same denomination. But as this involves the use of an accurately standardized complete set and as such a set is necessarily quite expensive this method is generally impracticable. Instead, one standard piece (usually a 1-gram piece) may be used and all other values calculated from this. Of course the unavoidable experi- mental error is thus cumulative in the . larger pieces and the work must be done with extreme care if the corrections are to have any real significance. Calibration. Method of Direct Comparison. Determine the zero point of the unloaded balance. Place the 1-gram piece marked () on the left pan and the one marked () on the right. Adjust the rider, if necessary, to maintain the zero point at its first position. If the rider is not required to restore equilibrium and if the balance arms have been found to be of equal length, then the two pieces are of equal value. If the rider is necessary on the right arm the weight on that side is less in value than the one on the left. If the rider was used on the left arm the reverse is true. Record the result as 1- = i- -f- or n milligram, n being the difference between the values of the two pieces. If the balance arms are unequal this relative value for 1" is to be multiplied bv r' Compare similarly the other pieces of the set, as follows: 68 QUANTITATIVE ANALYSIS Gram Pieces 1- with standard piece, if one is available ! with 1- or 1- 2 with 1- + 1- 5 with 2 + 1- + 1- + 1- 10- with 5 + 2 + 1- + 1" + I"' 10- with 10" 20 with 10- + 10" 50 with 20 + 10- + 10" + 5 + 2 + 1- + ! + 1- Milligram Pieces 500 with 200 + 100- + 100" + 50 + 20 + 10- + 10" + 5 + rider at 5 200 with 100- + 100" 100- with 50 + 20 + 10- + 10" + 5 + rider at 5 50 with 20 -f- 10- + 10- + 5 +rider at 5 20 with 10- +10" 10- with 10" 10- with 5 + rider at 5 5 with rider at 5 Also compare all milligram pieces plus the rider at 5 with 1- or with the standard piece if the latter has been used. If a standardized piece has been used calculate all of the values for the other gram pieces from this; otherwise select the one of the gram pieces of the set which appears to be most nearly normal with respect to the other pieces and call this 1.0000 gram, calculating values for the rest of the gram pieces upon this arbitrary assumption. In the case of the milligram pieces calculate provisional values for each piece, starting with the arbitrary assumption that the smallest piece of the set is correct. (The smallest piece is usually a 5- or 10- milligram piece. Smaller pieces are not needed because the adjustment of the rider provides smaller subdivisions.) Add these provisional values. If the total is not that indicated by the comparison of the collective milligram pieces with the chosen standard then the provi- sional value for each piece is to be multiplied by the factor: gum This will give the new value for the piece, based upon the standard finally adopted. The new sum will equal the true sum unless the dropping of decimals beyond the fourth place has impaired the accuracy of the calculations. GRAVIMETRIC ANALYSIS 69 The ratio, should be calculated as far as six decimal places. This means that considerable work will be involved in multiplying all of the provisional values by this ratio. It is much simpler (and sufficiently accurate in most cases) to use a somewhat different method for changing the provisional values to the true values, thus : Subtract the provisional sum from the true sum, applying the correct sign to the difference. This gives the total correction to be made. Apply 0.5 of this correction to the 500-milligram piece, 0.2 to the 200- milligram piece, 0.1 to each of the 100-milligram pieces, and so on. Method of Substitution. In calibrating by this method a second set of weights should be provided to serve as counterpoise pieces. This may be a low priced set of ordinary weights or a worn out and discarded set of analytical weights, since the accuracy of the calibration does not depend in any manner upon the accuracy of adjustment of the counter- poise. Also it is unnecessary to determine the zero point of the unloaded balance, the only requirement being that the adjustment of the rider shall be made so as to bring the balance into equilibrium about any given scale division as zero point and to maintain this zero point through- out the experiment. This point may conveniently be the zero of the scale. Place the 1-gram piece marked () on the right pan and a 1-gram piece of the counterpoise set on the left. Adjust the rider so as to make the pointer swing about the true zero of the scale and note the rider position. Remove the piece from the right pan and put in its place the I-gram piece marked (). Readjust the rider, if necessary, to restore the equilibrium of the balance. If the rider position is the same as before the pieces () and () have the same value. If the rider was moved to the right in the second experiment the piece () is lighter than the piece () ; if the rider was moved to the left the reverse is true. In either of the latter cases the amount of rider shift is a measure of the numerical difference between the values of the two pieces. Continue this process of comparison of pieces, using the combinations listed in the directions for calibration by direct comparison. In each case the pieces to be compared are placed successively on the right pan, a counterpoise piece of the same denomination being placed on the left pan. Calculate the true or relative values for all of the pieces of the set according to the method already described. Reagents. One of the most vexatious problems with which the analyst has to deal is that of obtaining reagents that are sufficiently pure to suit his purpose. Methods of manufacture 70 QUANTITATIVE ANALYSIS are constantly being improved and better chemicals are now available than in the past, but even at this time the reagent that is assumed to be pure often contains small quantities of impurities which interfere with the accuracy of analytical proc- esses. Attempts have been made by manufacturers to indicate on the label the degree of purity. Thus "c. p." for " chemically pure," signifies a reagent containing no impurity in a quantity that could be detected by chemical tests. "Com." for " com- mercial" means a crude unpurified chemical, "medicinal" sufficiently pure for medicinal purposes, "U. S. P." purity as specified by the United States Pharmacopoeia, etc. Zinc might be labeled "arsenic free" to indicate that it could be used without a blank test for a determination of arsenic by Marsh's method, or "iron free" so that it could be used for reducing solutions in iron analysis without a blank test. If these labels ever did have any real value they early lost it. "c. p." has been made a cover for a multitude of shortcomings in packages of grossly impure reagents. "Medicinal" has meant little more than that the manufacturer hoped that the substance so marked might be sold to the unsuspecting for medicinal purposes. "Silver free" lead (for assaying silver ores) is often lead from which the manufacturer has removed a certain fraction (or none at all) of the silver originally contained in it. Analyzed Chemicals. On account of the carelessness evident in preparing and labeling reagents chemists have come to practi- cally disregard all such indications of purported purity and to rely upon one or both of two sources of information regarding the purity of reagents. These sources are the reputation that the manufacturer bears for producing reliable chemicals and the chemist's own personal test of the chemicals themselves. Many manufacturers have now entirely discarded the abbreviation "c. p." and publish on the label a supposed analysis of the sub- stance contained in the package. "Analyzed chemicals" have thus become popular, but the inexperienced chemist will make a great mistake if he forms the too hasty conclusion that the analysis is always correct. It is often very far from being correct. The passage of pure food and drugs acts in this and many foreign countries has resulted in great improvement in the matter of labeling chemicals that are to be used for medicinal GRAVIMETRIC ANALYSIS 71 purposes, since the label constitutes a legal guarantee as to the contents of the package. When these acts are extended to include the reagents used for scientific purposes the chemist will have a better opportunity for purchasing chemicals of the degree of purity of which he can feel reasonably assured. At present the only safe plan is to make blank tests for such impurities as will interfere in the analysis to be performed. Action upon Glass. While it will be readily conceded that no substance can be made absolutely pure yet certain reagents can only with difficulty be made even approximately pure. Examples are the strong bases, such as sodium hydroxide, potas- sium hydroxide, ammonium hydroxide, barium hydroxide, etc., which readily attack and dissolve glass so that they are always contaminated with silica. On this account their solutions are seldom kept as stock reagents in the laboratory but are made from the solids as required excepting, of course, ammonium hydrox- ide which is a solution of a gas. In such cases the chemist will simply require that interfering substances shall be absent. Basic solutions often contain precipitated matter. The glass bottle is first attacked, the solution accumulating alkali silicates. These are later hydrolyzed, causing the precipitation of hydrated silica. The rule must never be forgotten that solutions are to be filtered just before using for analytical purposes unless they are already quite clear and free from sediment. Chemical Glassware. Glassware that is to be used for analytical work must possess certain properties that are not found in ordinary glass, (a) It must have a very slight solubility in water and in solutions of acids, bases and salts. (6) It must have a low coefficient of expansion and be well annealed, in order to withstand sudden changes in temperature as well as mechanical shock, (c) Its chemical composition must be adapted to the work at hand so that traces dissolving shall not affect analytical results. For example, a lead glass should not be used for solutions in which small amounts of lead are to be determined. Prior to the beginning of the world war manufacturing of chemical " resistance" glassware had not been developed to any great extent in America and most of such material was imported. Jena glass, which is probably the best known of imported ware, is essentially a borosilicate of sodium, zinc and aluminium. 72 QUANTITATIVE ANALYSIS The United States Bureau of Standards has made an extensive investigation 1 of the qualities of two kinds of imported resistance glassware (Jena and Kavalier) and of five American glasses, most of the latter having been developed within the past four years, although one or two American glasses have had an excellent reputation for a long time. The qualities tested were (a) chemical composition, (6) coefficient of expansion, (c) internal stresses, (d) resistance to sudden changes of temperature, (e) resistance to mechanical shock and (/) solubility in water, ammonium hydroxide, mixed solutions of ammonium sulphide and ammonium chloride, solutions of sodium phosphate, sodium and potassium carbonates, and sodium and potassium hydroxides. As a result of this investigation it may be stated that " all of the American made wares tested are superior to Kavalier and equal or superior to Jena ware for general chemical laboratory use." In the following pages Pyrex glass will often be specified but it should be understood that any available resistance glass may usually be substituted. Distilled Water. Natural waters always contain dissolved matter which unfits them for use in analytical work. Besides such natural mineral matter and dissolved gases, water will always dissolve certain quantities of the container when allowed to stand. In order to remove dissolved solids the water is distilled and recondensed. Various forms of stills are in use. In any form of such apparatus the vessel in which the water is boiled should be so far separated from the condensing worm that it is impossible for any spray to enter the latter. The boiler itself may be of any material, but the condensing worm should be of pure tin, silver, or platinum because hot water dissolves most other metals and also glass. The cost of platinum of course precludes its use in any but small stills that are to be used for preparing water for exact investigations, such as are carried out in physical chemical work. Pure tin is the metal generally used for the purpose. Distillation does not free water from dissolved gases and for work in which carbon dioxide, oxygen, nitrogen, or ammonia will interfere it is necessary to boil the water immediately before 'Bur. Stand. Tech. Paper 107 (1908). in : GRAVIMETRIC ANALYSIS 73 using. Boiling should not be unduly prolonged, since the water thus becomes recontaminated with the material of the contain- ing vessel. Transfer of Liquids. The operations involving pouring rea- gents from bottles, pouring liquids into a filter or pouring from one vessel to another are often so clumsily performed as to cause a loss of part of the liquid through splashing or running down the outside of the pouring vessel, thus vitiating the results of the analysis or at least producing a very disagreeable sort of uncleanness of the apparatus. When pouring from a bottle the stopper should never be laid on the desk but is held between the fingers of the right hand. The bottle is then grasped in ch a way as to bring the label under the hand and then a glass d is held in a vertical position against the mouth of the bottle. The latter is tilted to pour out the required amount of liquid, the FIG. 32. Form of label for laboratory reagent bottle. glass rod being kept in position until the bottle is returned to an upright position. No liquid should run down the outside of the bottle in this case. If a drop should escape the label will not be marred because the method of holding the bottle brings the label on the upper side while pouring. When pouring from a beaker the stirring rod is used in the same way for preventing splashing. Records. In no other part of the analysis is system more important than in that of records. Results are many times rendered worthless by uncertainty regarding the meaning of the experimental figures or regarding the pieces to which recorded weights belong. -All analyses are run at least in duplicate, some in triplicate. If at any stage in the work the beakers, crucibles, burettes, or other pieces become interchanged or if the recorded weights, volumes, temperatures, or other data are not properly 74 QUANTITATIVE ANALYSIS labeled and are applied to the wrong pieces, then the calculated results of the analysis are, of course, incorrect. At the very outset the duplicate pieces should be numbered (I and II unless other marks are preferred) in every case where the mark is not objectionable. It is not advisable to mark by labels or pencil SAMPLE OF_ -MARKED GRAVIMETRIC ANALYSIS AMOUNT OF SAMPLE TAKEN WEIGHT OF PERCENT or AMOUNT OF SAMPLE TAKEN WEIGHT OP PERCENT OF FOUND FOUND I I II II AMOUNT OF SAMPLE WEIGHT OF PERCENT AMOUNT OF SAMPLE WEIGHT OF PERCENT FOUND AKEN FOUND 1 I II ir AVERAGES SUBSTANCE DETERMINED PERCENT 8CB8TANOE DETERMINED PERCENT FIG. 33. A convenient blank for reporting the results of a gravimetric analysis. any article that is to be weighed because the mark itself often causes changes in weight through rubbing off or absorption of moisture. For articles that are not to be exactly weighed a small label or pencil such as is used for marking glass and porce- lain may be used. The latter has a soft core composed of pig- GRAVIMETRIC ANALYSIS 75 ment and grease or paraffin so that it will stick to glass. The better grades of glassware are now furnished with a spot rough- ened by sand blasting, this making it possible to use a common graphite pencil or a pen for marking without a label. Even in the case of articles that are not to be marked they can be easily identified if the analyst follows the plan of always keeping No. I on the left and No. II on the right when precipitating, filtering, igniting or allowing crucibles to stand in desiccators Reagent bottles should be marked by a label bearing the name of the student, that of the reagent, the concentration of the solu- tion and the desk number, as shown in figure 32. Many systems of note-book records have been used with greater or less success. If an ordinary blank note book is used it is necessary to so indicate each measured weight or volume that there is no possibility of uncertainty regarding the meaning of the figures. Loose sheets of paper must not, under any circum- stances, be used for records of a permanent character. Loss of such a sheet has frequently caused the loss of days or even weeks of laboratory work. A better device than that of the blank paged note book is found in a systematic record book, with spaces so provided and lettered that mere recording figures is all that is necessary. Such a page for gravimetric analysis is shown in Fig. 33. No indication of the identity of weights or percents is necessary beyond the filling of the blanks as shown. CHAPTER III EXPERIMENTAL GRAVIMETRIC ANALYSIS One of the most apparent differences between the methods of study as applied to qualitative and to quantitative analysis lies in the fact that while in qualitative analysis the metals and acids are grouped according to their susceptibility to the action of cer- tain "group reagents," in quantitative analysis widely different reagents and methods are often used for elements gr acids that are closely allied in most respects. A general review of the conditions that must be fulfilled for accurate gravimetric analysis will serve to show that no such systematic classification as is found in qualitative work is desirable in quantitative pro- cedure, since the reagent and method must always be selected which will give the precipitate which is the most insoluble and the most easily separated and purified and which assumes the most definite form upon the application of heat. For example, calcium, barium and strontium fall in the same periodic as well as the same qualitative group and yet there is no logical reason for studying these metals together in quantitative analysis because barium is most conveniently precipitated as sulphate and calcium as oxalate from water solutions, while strontium is precipitated as sulphate from solutions containing alcohol. Barium sulphate is stable when ignited, and is weighed as such. Calcium oxalate decomposes and is weighed as carbonate or oxide while strontium sulphate is stable and is weighed in this form. Any one of these metals could be precipitated by ammonium car- bonate and the carbonate changed to the oxide by ignition but this would not, in any case, prove to be the most convenient or the most accurate method. In view of these facts it becomes desirable to select, for each element or radical, that method which is, under the circum- stances, the most easily executed and the most accurate and to study these, not in order of qualitative or periodic groups, but 7(3 EXPERIMENTAL GRAVIMETRIC ANALYSIS 77 in that order which will best develop manipulative skill and . accuracy in experimental work. Many methods that were formerly used have become almost obsolete because of the development of better apparatus or more rapid methods. In such cases the older methods will generally receive mention at the proper place and the more modern method will be described more fully. Apparatus. Most of the ordinary apparatus with which the quantitative laboratory is usually stocked is already more or less familiar to the student. Special forms of apparatus will be described in connection with the determinations for which they are to be used. At the beginning of the course all glassware should be thoroughly cleaned and other apparatus should be put in first class order. A principle that should never be forgotten is that both accuracy in results and speed in working are pro- moted by following the practice of cleaning apparatus as soon as its use in a given experiment is finished, so that it will be ready (with perhaps a single rinsing with distilled water) for the next operation that may demand it. The student who works with a desk full of soiled, broken or disorderly apparatus or with spilled chemicals scattered over the tables may as well make up his mind at the outset that he will never be an analyst of any useful sort. After desk apparatus has been invoiced and cleaned, two desiccators like that shown in Fig. 11 are prepared for use. They are first made clean and dry and then a layer, one-half inch thick, of fused, granular calcium chloride is placed in the bottom and a short piece of sodium hydroxide is placed upon this. The former keeps the atmosphere within the desiccator free from moisture while the sodium hydroxide absorbs carbon dioxide and traces of acid vapors that sometimes come from the calcium chloride. The upper part of the desiccator is again wiped with a dry towel to remove calcium chloride dust and a clay or alloy triangle is bent so as to lie freely on the shoulder above the calcium chloride. (Perforated porcelain plates are sometimes used instead of triangles for this purpose.) A thin film of vaseline is rubbed on the ground joint of the cover and the latter is then worked down until it fits well, the surplus vaseline being removed from the edge. The desiccator is now ready for use. Prepare also two wash bottles (one each for hot and cold water) like the one illustrated in Fig. 7, using liter flasks for the purpose. The 78 QUANTITATIVE ANALYSIS neck of the one that is to be used for hot water should be wrapped with cotton cord or sheet cork. A number of stirring rods, 4 to 6 inches long, should be made by fusing and rounding the ends of glass rod or tubing. One or two of these, tipped with pieces of rubber tubing having an end cemented together, are to be used for loosening precipitates from beakers and dishes. These are known as the chemist's "policemen." CALCIUM Calcium may be precipitated from ammoniacal solution as carbonate by alkali carbonates or as oxalate by alkali oxalates. Since ammonium carbonate or oxalate yields volatile ammonium salts as byproducts in the reaction and since traces of these will be expelled upon ignition if not completely washed out of the precipitate, the ammonium salts are always used in preference to those of sodium or potassium. Solubility. The solubility of calcium carbonate in water was determined by Kohlrausch and Rose 1 by conductivity experi- ments and was found to be 0.013 gm (corresponding to 0.0052 gm of calcium) per liter at 18, but the solubility is considerably increased by ammonium salts, such as must be present when ammonium carbonate reacts with a calcium salt in solution. The solubility of calcium oxalate in water was found by Kohlrausch and Rose to be 0.0056 gm of the crystallized salt, CaC 2 O 4 H 2 O corresponding to 0.0015 gm of calcium) per liter. The solubility is considerably reduced by an excess of ammonium oxalate. On account of this difference in solubi ity the oxalate method is preferable. Ammonium chloride should be present in either case to prevent precipitation of traces of magnesium. The reaction involved in the precipitation may be expressed as follows: CaCl 2 + (NH4) 2 C 2 04-CaC 2 04+2NH 4 Cl. Ammonium oxalate does not readily dissolve in water, the saturated solution at containing 2.2 percent of the salt. As the temperature is raised the solubility is increased so that a saturated solution at 20 is easily made by heating in contact with an excess of the salt and allowing to cool. It is not best to keep a stock solution of ammonium oxalate because it readily *Z. physik. Chem., 12, 234 (1893); 44, 197 (1903). EXPERIMENTAL GRAVIMETRIC ANALYSIS 79 undergoes hydrolysis, yielding ammonium hydroxide which attacks the glass, and because there occurs a decomposition in solution as follows: It is preferable to make a small amount of the solution as needed for the determination. A difficulty that is often encountered by the inexperienced analyst is the formation of a precipitate of calcium oxalate which is so finely crystalline that it passes through the pores of the filter, making its complete separation impossible. Refiltration of the portion that has passed through will partially remedy this trouble but when a precipitate is found to be too finely divided for filtration the only satisfactory cure is found in digestion. Certain grades of filter paper are made for filtering fine precipi- tates, their structure being very dense. This renders filtration less rapid than would be the case with papers of ordinary density. The difficulty nearly or quite disappears if the proper conditions are observed during the precipitation. It has been explained (page 20) that too rapid precipitation causes the formation of a large number of small particles rather than a small number of large particles. Two conditions are found to be suitable for the formation of large crystals of calcium oxalate: (1) boiling tem- perature for solutions of calcium salt and ammonium oxalate and (2) slow addition of reagent. These conditions will be elaborated in the directions for the determination. Calcium may also be determined by precipitating as sulphate from a solution containing alcohol 1 or volumetrically by pre- cipitating as oxalate and titrating with a standard solution of potassium permanganate. The latter method will be described in the section dealing with volumetric analysis. Calcium precipitated as either carbonate or oxalate may be weighed as carbonate, oxide or sulphate. Calcium oxalate decomposes as follows, on the application of heat: CaC 2 4 ->CaC0 3 +CO, (1) CaC0 3 -*CaO+C0 2 . (2) 1 Stolberg: Z. angew. Chem., 17, 269 (1904). 80 QUANTITATIVE ANALYSIS Reaction (1) begins at quite low temperatures. Reaction (2) begins as low as 500 but requires long heating at this tempera- ture to complete the decomposition. In practice the highest temperature attainable by the blast lamp is applied and heating is continued until no further loss in weight occurs. Calcium oxide is not reduced by hot carbon and the precipitate may be heated without removing from the paper. Many chemists prefer to ignite the oxalate at a low temperature and to weigh as carbonate but this procedure is of doubtful utility, even for an experienced analyst, on account of the difficulty in stopping the decomposition at a point where all the oxalate has disappeared and no oxide has been formed. It is also possible to add a few drops of sulphuric acid to the crucible containing the calcium oxalate and to weigh the resulting calcium sulphate, after gentle ignition to expel the oxalic acid and the excess of s\ilphuric acid. The most important source of error in this procedure comes from the loss by spattering, a loss which even the greatest care can scarcely prevent. There is also danger of decomposing cal- cium sulphate by strong heating: CaS0 4 ->CaO+S0 3 . It is much preferable to heat strongly the precipitate converting it quantitatively into calcium oxide. The completion of the conversion can be judged by the constancy in the weight of the substance upon further heating. Calcium oxide readily absorbs moisture and carbon dioxide when exposed to the air and the resulting change of weight will become appreciable if the process of weighing is unduly prolonged. If the weight of the crucible and oxide is approximately known most of the weights may be placed on the balance pan before the crucible is removed from the desiccator and the remainder of the process completed in a short time. It is a good plan to keep a small piece of potassium hydroxide in the desiccator in which calcium oxide is to be pre- served. This lessens the absorption of carbon dioxide by keeping the atmosphere free from that gas. The converse of this method may be used for the determina- tion of the oxalate radical, precipitation being made by a soluble calcium salt in basic solution. Volumetric methods are, how- ever, preferable. EXPERIMENTAL GRAVIMETRIC ANALYSIS 81 Determination. Fill a clean dry weighing bottle with the calcium compound to be analyzed. Provide two clean beakers of Pyrex or other resistance glass, having a capacity of 250 cc, and mark them I and II. If the substance is of such nature that it is altered in any way by free exposure to air the sample to be used must be weighed by difference as follows: Place the bottle on the balance pan, using for this purpose a pair of crucible tongs, having short pieces of rubber tubing drawn over the tips, and carefully weigh. Record this weight in the data book at the top of the space marked for sample I, reading the weights as directed on page 62. Carefully remove the stopper, holding over beaker marked I, and pour what is judged to be between 0.2 gm and 0.5 gm into the beaker. Replace the stopper, taking great care that any falling par- ticles drop into the beaker and are not lost, then reweigh the bottle and contents. Record this weight under the first, also at the top of the space for sample II. Remove a second portion to beaker II and reweigh the bottle, recording under the preceding weight. Subtracting the less weights from the greater gives the weights of sample used. If the substance is known to be unaffected by contact with air it may be poured into one of the counterpoised glasses and weighed directly, being then brushed into the beaker by the small pencil brush of camel's hair. After having weighed the two samples for analysis determine, by qualitative tests on another portion of the substance, whether it is solu- ble in water and, if not, whether in dilute hydrochloric acid. If soluble in water dissolve in about 100 cc of distilled water, add 5 cc of 10 percent ammonium chloride solution and treat each sample as directed below. If insoluble in water but soluble in hydrochloric acid first moisten each sample with water then cover the beakers with glasses and carefully add about 20 cc of dilute acid. After effervescence has ceased rinse down the cover glass and the sides of the beaker with water from the wash bottle, dilute to about 100 cc and gently boil for one minute to expel dissolved carbon dioxide. From this point the procedure is the same as for water soluble salts. Prepare ammonium oxalate solution by heating to boiling 5 gm of powdered ammonium oxalate and 100 cc of water in a Pyrex beaker. Part of the salt will crystallize when cooled, leaving a saturated solution. Add dilute ammonium hydroxide (filtered unless already quite clear) to the solution of calcium salt until the liquid smells very distinctly of ammonia. In determining this point the ammonia that is already in the air above the liquid must be blown away before testing the odor. Heat to boiling and add, drop by drop from a pipette, 20 cc of the recently prepared ammonium oxalate solution, stirring vigorously during the addition. If this is carefully done the precipitate should settle readily, 82 QUANTITATIVE ANALYSIS leaving a clear liquid above. When this is the case add a few drops more of ammonium oxalate solution, observing whether any precipitate forms. If so, more reagent must be added in the same manner as at first. When precipitation has been shown to be complete the liquid is digested at a temperature somewhat below the boiling-point over a low flame or on the steam bath for one-half hour or until the super- natant liquid is quite clear, when it is ready for filtration. The odor of ammonia should still be easily perceptible at this stage. Prepare two filters of extracted paper, marking the funnels I and II. Carefully decant each solution, while hot, upon the proper filter, allowing to run through into clean beakers. Observe the directions given on page 73 for proper method of pouring from beakers. Before completing the filtration a few drops of ammonium oxalate solution should be added to the clean filtrate that has already run through. If a precipitate forms, the filter must be well washed; the filtrate and washings returned to the beaker in which precipitation was made and more reagent added in the same manner as before, until precipitation is complete. When no precipitate is produced in the filtrate, complete the filtration, washing the precipitate into the filter by a stream from the hot water bottle, rubbing the beakers with a glass rod tipped with rubber tubing (a "policeman "). Wash the precipitate on the filter until a small amount of the last washings fails to give more than a faint precipitate with silver nitrate and a drop of nitric acid showing that chlorides have been removed. Allow the paper to drain then remove from the funnel, fold as directed on page 34 and place in a previously ignited and weighed porcelain or platinum crucible. The cover should have been weighed with the cruci- ble because the closed crucible will hinder absorption of moisture and carbon dioxide while weighing. The crucible is carefully heated by the burner until the moisture has been expelled and smoking begins. The cover may now be removed and placed on a clean tile while the crucible is heated more strongly until the paper is completely charred. The crucible is now placed on its side, the cover adjusted and complete oxidation of the carbon of the paper is accomplished as described on page 36. The crucible is then placed in an upright position, is covered and subjected to the hottest flame available from the blast lamp. This ignition is continued for 30 minutes, when the crucible is placed in the proper desiccator, allowed to cool to the temperature of the room and quickly weighed. It is ignited for 10 minutes longer, cooled and reweighed. If there is a decrease in weight of more than 0.0003 gm the crucible is reheated for 10 minutes and weighed, the process of heating and weighing being continued until the weight remains constant within the limit given above. The preliminary weighings EXPERIMENTAL GRAVIMETRIC ANALYSIS 83 should be recorded upon the back of the sheet preceding the one used for the final record, the final weight being recorded in the proper blank space. Calculate the percent of calcium in the sample, using the factor already calculated and recorded on page 9, and using a table of loga- rithms for the arithmetical work. Do not discard the ignited product until after the report has been accepted. Errors may have been made which can be corrected if this has been preserved. This is a principle that should be observed, when possible, in all analytical work. SILVER Silver might be gravimetrically determined as chloride, bromide or iodide. The solubilities of these salts in water, shown in the following table, were determined by Kohlrausch and Rose, 1 Milligrams per liter, soluble at 18 Salt Silver equivalent to salt Silver chloride 0017 0013 Silver bromide 0.0004 0.00023 Silver iodide. . 0001 00005 From the comparative solubilities one might conclude that silver chloride is the least desirable form in which to weigh silver. The bromide and iodide, however, are much more sensitive to the action of light, being more readily decomposed into sub- halides with liberation of free halogen. The stabilities of these salts are in the same relation to each other as are those of halides of other metals and of hydrogen. On this account the gravimetric determination of silver is invariably made by weighing silver chloride, using hydrochloric acid as the precipi- tating reagent. Conversely the determination of chloranion is made by using a soluble silver salt as the reagent. A small excess of either a soluble silver salt or a soluble chloride greatly diminishes the solubility of silver chloride as explained in the section dealing with the principles of precipitation. If more than a very slight excess of a metal chloride is present the solu- 1 Z. physik. Chem., 12, 234 (1893). 84 QUANTITATIVE ANALYSIS bility of silver chloride is increased, because of the formation of soluble double salts. On this account hydrochloric acid is used as the precipitant for silver. Silver chloride shows a well defined tendency toward the for- mation of a hydrosol in cold water and when this is formed the solubility follows no definite rule. The sol can be flocculated by boiling with dilute acids or other electrolytes. The precipitate of silver chloride is affected by light as are the other silver halides, it being reduced to a subchloride, Ag 2 Cl ; with liberation of free chlorine. The darkening of silver chloride under the influence of strong light is due to the appearance of the subchloride which is bluish black in color. While some decomposition undoubtedly occurs in daylight of any intensity, if the precipitation is performed in the darker parts of the room ordinary diffused light will not appreciably affect the weight of the precipitate in a short time. The precipitate cannot be ignited in contact with filter paper on account of the ease with which it is reduced to metallic silver. Use can be made of any of the devices for dealing with such pre- cipitates, as mentioned in the general discussion of the ignition of precipitates; we may here follow either the method of removing the precipitate from the filter paper or the method involving the use of the Gooch crucible, both methods being described. Whenever silver chloride is heated, care must be taken to prevent a rise of temperature above the point of fusion, which is 451, since it is sensibly volatile at high temperatures. By making the proper changes in procedure the halogens may be determined, silver nitrate being used as the precipitating reagent. These determinations are discussed later (page 87). Determination. Fill a stoppered weighing bottle with the powdered silver salt and weigh out two samples of about 0.3 gm each for analy- sis, placing in 250-cc beakers, which may be of ordinary hard glass. The weighing may be done upon counterpoised glasses or from the weigh- ing bottle, by following the directions given for the determination of calcium. In this, as in all other cases, care must be exercised to avoid spilling any of the substance upon the balance pan, as this invariably results in corrosion of the pan. Dissolve the weighed sample in about 100 cc of water, and heat nearly to boiling. Add, with stirring, 5 cc of dilute hydrochloric acid EXPERIMENTAL GRAVIMETRIC ANALYSIS 85 and digest on the steam bath until the precipitate is completely floccu- lated and the supernatant liquid is quite clear. Test the clear liquid with more hydrochloric acid as soon as is practicable, to determine whether precipitation is complete. When the precipitate settles com- pletely proceed by one of the following methods. (a) Filtration on Paper. Filter on extracted paper, with or without slight suction, and wash with hot water containing 1 cc of dilute nitric acid in each 100 cc of water, the acid being used in order to prevent the silver chloride from returning to the condition of a hydrosol and thus passing through the filter. Wash until free from chlorides, testing the washings with silver nitrate. Allow the precipitate to drain, then remove the paper, fold over the top and sides (see page 34), place on a watch glass and dry in an oven at 100. When the precipitate and paper are completely dried, place a piece of black glazed paper on the desk, unfold the filter paper and carefully detach as much as possible of the precipitate, using a spatula for this purpose and allowing the precipitate to fall upon the central portion of the glazed paper. While it is desirable to leave as little as possible of the precipitate upon the filter paper it is also essential that no paper fiber be removed with the main portion of the precipitate since this portion is not to be treated to reconvert reduced silver into silver chloride. With the small camel's hair brush the precipitate is now brushed into a pile, loosening any particles that may have been caught by the brush, and is covered with a watch glass. An ignited and weighed crucible is placed upon one corner of the glazed paper. The filter paper is refolded in the same manner as before, is rolled into a tight roll and a stiff plati- num wire is coiled around it in a manner such that the roll can be held over the crucible by means of the wire. Being held in this position it is touched with the oxidizing flame of the gas burner until the paper ignites. The gas flame is to be used only often enough to keep the paper ignited and the outer oxidizing portion of the flame is always to be used for this purpose. The paper is thus burned, the ash falling into the crucible, where the combustion is completed at a low temperature. Some silver has been reduced even with these precautions. To change this again into silver chloride the ash is moistened with a drop or two of concentrated nitric acid and, after a few minutes, a drop of concentrated hydrochloric acid is added. The reduced silver is first dissolved by the nitric acid, forming silver nitrate, and this is changed to silver chloride by the hydrochloric acid. Evaporate the acids by placing the crucible on a water bath, then carefully brush into the crucible the main portion of the precipitate. Heat gently over the burner until the precipitate shows the first appear- ance of fusion where it is in contact with the sides of the crucible. In 86 QUANTITATIVE ANALYSIS case the precipitate has been unduly reduced by light or if it becomes reduced when heated, on account of cellulose derived from the paper, it should be moistened by nitric acid and hydrochloric acid, as directed above, and warmed, when it will usually become white, after which the stronger heating to the fusion point may be performed. Place the crucible in the desiccator and weigh as soon as cool. Calculate the per- cent of silver in the salt, using the factor for silver in silver chloride as already calculated in the table of factors on page 9. (6) Filtration on a Gooch Crucible. A device which is similar to that shown on page 24 is used for applying the suction. Place a porcelain Gooch crucible in the holder, apply the suction and pour into the crucible a suspension of purified and shredded asbestos until a mat about 1 mm thick is obtained on the bottom of the crucible. Asbestos to be used for this purpose should have been prepared by digest- ing for one hour with concentrated hydrochloric acid to dissolve any acid-soluble material, then washing with distilled water until free from chlorides. The desirable thickness of the mat in the crucible will depend somewhat upon the character of the asbestos fiber; if the latter has a fine texture a closer felt will result, with a consequent increase in the efficiency of the filter. Draw all surplus water from the filter by means of the pump, then rinse once with redistilled alcohol. This is for the purpose of promoting rapid drying. Remove the crucible from the holder, wipe the outside then dry in an oven at 105 to 110 for 30 minutes. Cool in a desiccator and weigh. Heat again in the oven for 30 minutes, cool and weigh. This weight will usually be the same as the first but if there is a decrease of more than 0.3 to 0.5 mg the heating and weighing must be continued. When the weight of the crucible has become constant replace the crucible in the holder and again apply suction. Carefully filter the solution from which the silver has been precipitated, finally transferring the entire precipitate to the filter. Wash and test the washings as directed for the method of filtering on paper. Finally rinse once with alcohol, dry to constant weight at 105 to 110 and calculate the per- cent of silver in the sample. This method is usually to be preferred to the one first described. The chief source of error is in the loss of asbestos during filtration and washing. This may be prevented by proper preparation of the asbestos suspension before using, the fine material being removed by sedimenta- tion and decantation. Even when this has been done it is necessary to keep the suction applied at all times during filtration and washing, the asbestos thus being held down in the felt. EXPERIMENTAL GRAVIMETRIC ANALYSIS 87 CHLORIDES (BROMIDES AND IODIDES) The method for the determination of the halogen of halides is the converse of. the one just described, silver nitrate being used as the precipitating reagent. Nitric acid must be present to prevent the precipitation of salts of silver, other than the halide. Determination. From a stoppered bottle weigh into 250-cc beakers duplicate samples of about 0.2 gm of the chloride, bromide or iodide. Dissolve in about 100 cc of distilled water and add 1 cc of dilute nitric acid. Both distilled water and nitric acid must be tested and found free from chlorides. Heat the solution to near boiling and add, slowly from a pipette, a 5 percent solution of silver nitrate until no further precipitate is produced. Digest on the steam bath until the precipitate flocculates readily, leaving a clear solution. Test the clear liquid with another drop of silver nitrate solution to insure complete precipitation. For the nitration either of the methods described for the determination of silver may be used but the Gooch crucible is recommended, especially for the determination of halogens other than chlorine. Filter the solu- tion and wash free from silver with chloride-free distilled water contain- ing nitric acid, testing the washings at the last with a drop of dilute hydrochloric acid. The preparation of the filter and the treatment of the precipitate on the filter are to be exactly as described above for the determination of silver. Calculate the percent of halogen in the sample. The precipitate must be protected from the action of light, especially in the case of silver bromide or iodide, these substances being much more sensitive to light than the chloride. ALUMINIUM The gravimetric determination of aluminium, as well as of iron, chromium, nickel, cobalt and copper, may be accomplished by precipitating them as hydroxides, igniting and weighing these as oxides. For reasons that will presently be discussed all of these metals excepting aluminium are now usually determined by volumetric or electrolytic processes. Aluminium is precipitated as hydroxide by any basic solution, whether it be that of a pure base or of a hydrolyzed alkali salt of a weak acid, such as sodium carbonate or ammonium sul- phide. Aluminium hydroxide readily forms hydrosols which are flocculated by the addition of electrolytes, which must for this purpose be inorganic salts. Since the flocculated hydroxide is also of a colloidal nature (hydrogel) it manifests the phenome- 88 QUANTITATIVE ANALYSIS non of adsorption to a marked degree and the inorganic salts are consequently washed out with considerable difficulty. For this reason, as well as for other and more important ones, am- monium hydroxide is chosen as the precipitant because the by- products of the reaction will thereby be ammonium salts and re- maining traces will be volatilized when the precipitate is heated. Solubility in Bases. Aluminium hydroxide, besides being soluble as a hydrosol, also dissolves in solutions of bases. It thus happens that if an excess of the basic precipitant is in- advertently added, part or all of the precipitate returns to the solution, the amount dissolved depending upon the excess and ionization of precipitant. The strong bases, as sodium or potassium hydroxide, dissolve aluminium hydroxide more readily than the weaker ones and this furnishes a second ^reason for the use of the weaker base, ammonium hydroxide, as the precipitating reagent. In case an excess of this has been added it is possible to remove it by boiling. The solvent action of bases has been explained upon the sup- posed ability of aluminium hydroxide to exist in both acid and basic form. When precipitation is taking place, the solution, besides holding more or less of the hydrosol, is saturated also with the substance in the condition of molecular aluminium hydroxide in equilibrium with two sets of ions. This equilib- rium within the solution might be expressed simply thus: +30H +H 2 AK> 3 (If the substance is an acid it must ionize in three stages: A1(OH) 3 -+H+H 2 A1O 3 , H 2 A10 3 ->H+HA10 3 , HA10 3 -+H+A10 3 . Also Al(OH) 3 -*H+A~io 2 +H 2 0. vSince it must necessarily be very weakly ionized the ion H 2 A10 3 must predominate, but the equilibrium represented above can be but an approximate representation of the real conditions.) EXPERIMENTAL GRAVIMETRIC ANALYSIS 89 The addition of either a strong base or a strong acid to such a system would cause aluminium hydroxide to dissolve if the resulting salt were soluble. The effect of the strong acid upon the acid form would be that of suppression of the (already small) ionization. Its effect upon the basic form would be interaction to form a salt: A1(OH) 3 +3HC1-A1C1 3 +3H 2 0. The disturbance of equilibrium resulting from the disappearance of hydroxyl ions (due to the formation of water) would cause more aluminium hydroxide to ionize in this manner and, since nonionized aluminium hydroxide is also in equilibrium with the undissolved portion, more would go into solution. The effect of a base would be quite similar to that of an acid, although by a different process and through the formation of a different salt. The basic ionization of the aluminium hydroxide would be suppressed while the added base would react with the acid form: H3A103+NaOH-NaH 2 A10 3 +H 2 0, or HA10 2 +NaOH-*NaA10 2 +H 2 0. Here again a salt is formed, although the aluminium appears in the anion. The disturbance of equilibrium has the same ultimate result as before, namely, that the solid substance passes into solution. Adsorption. It has already been stated that the washing of the gelatinous precipitate is more or less difficult on account of the adsorption of dissolved salts. It is impossible to avoid the presence of such salts when making separations or when pre- cipitating aluminium from such compounds as the alums. (Ammonium salts are always present.) There is also danger that, in the case of prolonged washing by water to remove alkali salts or other salts, some of the hydrogel may return to the condition of the hydrosol. In order to prevent this the customary device of having present an ammonium salt in the wash water is used. Some of this necessarily remains with the precipitate at the last. If this salt is ammonium chloride some of the aluminium will be lost by volatilization, the chloride being, as with most other metals, more volatile than the salts of other 90 QUANTITATIVE ANALYSIS acids. The chloride will be formed during ignition by interaction of the aluminium oxide or hydroxide and the ammonium chloride : A1(OH)3+3NH 4 C1->A1C1 3 +3NH3+3H 2 0. Ammonium nitrate should therefore be used in the wash water. Aluminium nitrate, if formed, is decomposed at high temperatures into aluminium oxide and oxides of nitrogen. If the precipitate of aluminium hydroxide is filtered and washed under diminished pressure, care should be exercised that the liquid is not, at any time before the completion of the washing process, drawn out so nearly completely as that the precipitate should harden and crack. In such a case the wash water that is subsequently used will run through the cracks instead of through the body of the precipitate and complete washing will therefore be accomplished only after the use of much water. If it becomes necessary to allow the precipitate to remain in the funnel from one day to the next and before the washing is completed, the precipitate may be kept moist by plugging the stem of the funnel, covering the precipitate with water, and placing a watch glass over the top. By making suitable changes in the procedure, aluminium chloride might be made a reagent for the quantitative determina- tion of hydroxyl. Volumetric methods are always used instead. Determination. Fill a weighing bottle with the powdered sample of an aluminium salt. Choose the method to be used in weighing according to the nature of the substance and weigh two samples of about 1 gm each into Pyrex beakers. Dissolve in 100 cc of water and add dilute, recently filtered ammonium hydroxide, stirring until the liquid is distinctly basic, as shown by a bit of litmus paper thrown into the beaker. Boil until the precipitate is coagulated and until the odor of ammonia above the solution is but faint. Boiling after the odor has disappeared may cause some of the precipitate to return to the solution : A1(OH) 3 +3NH 4 C1-*A1C13+3NH3+3H 2 O. Allow the precipitate to settle and then filter through paper, using a filter pump attached to a bell jar or filter flask and placing a supporting cone of hardened paper or platinum in the funnel (page 23). Wash with hot distilled water containing 1 percent of ammonium nitrate until the washings are free from chlorides, shown by adding a drop of nitric acid and a few drops of silver nitrate solution to a cubic centimeter of the washings caught in a test tube; also from sulphates, as shown by adding a EXPERIMENTAL GRAVIMETRIC ANALYSIS 91 drop of dilute hydrochloric acid and a few drops of barium chloride solu- tion to another portion of the washings. Suck the precipitate as nearly dry as possible and transfer the paper and precipitate to a porcelain or platinum crucible which has been ignited and weighed, folding the paper in the manner already learned. Heat very gently in the covered crucible until the moisture is volatil- ized, then raise the temperature and burn the paper, inclining the cru- cible and placing the cover as in the case of the ignition of the paper containing calcium oxalate. When all of the carbon has been burned, cover the crucible and heat over the blast lamp for 30 minutes. Cool in the desiccator and weigh. Heat again for 10 minutes, cool and weigh. If necessary repeat this process until the weight is constant. Calculate the percent of aluminium in the salt. Aluminium oxide absorbs water from the air, reforming the hydroxide with a corresponding gain in weight. On this account the crucible and oxide should be weighed rapidly. Copper, cobalt and nickel cannot be quantitatively precipi- tated by ammonium hydroxide because of the formation of soluble complex ammonium salts. Sodium hydroxide or potas- sium hydroxide is used as the reagent. Adsorption of the reagent by the precipitate causes a large error and volumetric or electrolytic methods are preferable. BARIDM Barium may be precipitated as sulphate, carbonate or chro- mate. The sulphate and chromate are weighed as such, while the carbonate is ignited to form the oxide, in which form it is weighed. The solubilities are as follows (determined by Kohlrausch and Rose 1 ). Milligrams per liter soluble at 18 Salt Barium equivalent to salt BaCO 3 . 22 15 3 BaSO 4 2 6 1.53 BaCrO 4 3.8 2.06 The sulphate is seen to be the most suitable compound for the separation of barium from solution. The precipitating re- agent may be sulphuric acid or a soluble alkali sulphate. Since the latter produces by the reaction alkali salts that must be *Z. physik. Chem., 12, 234 (1893). 92 QUANTITATIVE ANALYSIS washed from the precipitate, while the former produces volatile acids, sulphuric acid is generally used for the purpose: Ba(NO 3 )2+Na 2 SO 4 -+BaSO4+2NaNO3. Ba(NOs)2+H 2 SO 4 ->BaSO4+2HNO3. Barium sulphate easily precipitates in the form of fine crystals. If precipitation takes place rapidly and from a somewhat con- centrated solution the crystals may be so small as to pass through the filter. Remedies similar to those applied to calcium oxalate may be used also with barium sulphate. These are use of a dense paper for the filter, precipitating slowly from a hot solu- tion and digestion of the precipitate in the mother liquor at a temperature near the boiling point. Sometimes the filter paper is treated before using with a hot, concentrated solution of ammonium chloride which softens and swells the cellulose fibers, making a less permeable filter. Such treatment is of doubtful utility since the ammonium chloride must later be washed out of the paper or cause some volatilization of barium chloride when the precipitate is ignited. No trouble will be experienced if the precipitation is accomplished under proper conditions. The converse of this method may be used for the determina- tion of the sulphate radical and for sulphur of any compound that may be changed to a sulphate. Barium chloride is then the precipitating reagent and the solution is made slightly acid by adding hydrochloric acid. The latter is necessary to prevent the precipitation of barium salts of certain other acids whose salts might be present. Examples of such salts are carbonates, oxalates, phosphates and borates, the barium salts of all of these being insoluble in water but soluble in hydrochloric acid. However, barium sulphate itself is appreciably soluble in hydrochloric acid as is shown in the following table: 1 Solubility of barium sulphate, milligrams per liter Hydrochloric acid 'Barium sulphate Barium equivalent to barium sulphate 1.82 3.65 7.27 18.23 6.7 8.9 10.9 8.6 3.9 5.2 6.4 5.1 ^Banthisch: J. pr. Chem., 29, 54 (1884), EXPERIMENTAL GRAVIMETRIC ANALYSIS 93 Therefore, if hydrochloric acid is present in any considerable quantity in the solution it must be nearly neutralized before precipitating the barium sulphate, not only because of its solvent action above shown but also because of its tendency to increase the occlusion of other salts by barium sulphate. 1 Such occlusion readily takes place if iron salts are present. Barium chloride itself is also readily occluded by barium sulphate. In the latter case the nature of the resulting error depends upon whether barium or sulphuric acid is being determined. If the former, the result is a negative error because part of the barium is weighed in combination with chlorine, whose equivalent weight is less than that of the sulphate radical. If the sulphate radical itself is being determined the error of occlusion is positive because any barium chloride that may be carried down is simply a part of the precipitating reagent, contaminating the pre- cipitate of barium sulphate. In order to avoid either error the concentration of hydrochloric acid should be made as small as will serve to hold such salts as the carbonate, oxalate, etc., in solution and the precipitation of the sulphate must be ac- complished by slow addition of the reagent to the hot solution, stirring vigorously meanwhile. Change of Weight of Barium Sulphate. Considerable care must be exercised in burning the paper upon which barium sulphate has been filtered and in subsequent ignition of the precipitate to expel traces of moisture. If the temperature is allowed to rise to too high a point barium sulphate will gradu- ally decompose, yielding sulphur trioxide and losing weight thereby : BaS0 4 -BaO+S0 3 . - On this account the blast lamp should never be used for heating the precipitate and the temperature is not allowed to rise above that of dull redness. On the other hand, errors may occur through partial reduction of barium sulphate by carbon monoxide or organic gases resulting from heating of the filter paper. Barium sulphide is thus produced and again the material loses weight: BaS0 4 +4CO-BaS+4C0 2 . ' 1 Richards: Z. anorg. Chem., 8, 413 (1895). 94 QUANTITATIVE ANALYSIS In order to avoid this reduction the temperature should be held at as low a point as will serve to accomplish the combustion of the paper and a plentiful supply of air must be maintained by inclining the crucible and cover, as directed on page 36. Even with these precautions some reduction may occur but if heating is continued for a few minutes after the carbon has disappeared, reoxidation will take place: BaS+20 2 ->BaS0 4 . If it should be suspected that either or both of the errors just discussed has occurred in any given analysis a correction may be made by adding a drop of dilute sulphuric acid to the pre- cipitate after the first weighing, then gently reheating to expel the excess of acid and water, and reweighing. A gajn in weight is taken as evidence that sulphide or oxide of barium was present in the first case. The second weight is then the correct one. This addition of acid, with subsequent heating, also serves to correct any error that may have occurred in the determination of barium, through the occlusion of barium chloride by the precipitating barium sulphate. It will be recalled that such occlusion occasions a negative error in the determination of barium, but a positive one in the determination of the sulphuric acid radical. Then in the first case sulphuric acid converts occluded barium chloride into barium sulphate and gives a precipitate of correct composition. In the second case barium chloride is an occluded impurity in the precipitate and its con- version to sulphate merely serves to increase the error. There- fore, when barium chloride is used as the precipitating reagent for sulphuric acid it is highly important that the precipitation should be carried out very slowly by adding the reagent drop-wise and stirring vigorously. This method serves not only to minimize occlusion of the reagent but also to prevent the formation of a very finely divided precipitate. Determination. Weigh about 0.2 gm of a barium salt into each of two beakers, dissolve in water or the least possible quantity of hydro- chloric acid, dilute to 100 cc and heat to boiling. Add, drop by drop, with vigorous stirring, 2 cc of 25 percent sulphuric acid. Allow the precipitate to settle somewhat and test the supernatant liquid, as usual, to determine whether precipitation is complete. Digest on the steam EXPERIMENTAL GRAVIMETRIC ANALYSIS 95 bath for 15 minutes or longer, until the precipitate settles readily. Filter without the use of a pump, on an extracted paper and wash several times with hot, distilled water, testing the washings for sulphates. Remove the paper from the funnel, fold and place in a weighed cru- cible. Incline the crucible as usual for burning the paper and heat at moderate temperature until white. Cover the crucible and heat barely to dull redness for 15 minutes, cool and weigh. Since no decomposition of the precipitate takes place when it is heated at this temperature there should be no change in weight after the first few minutes of heating unless washing has not been thorough, leaving salts that slowly volatilize. Calculate the percent of barium in the barium salt, using the factor for barium in barium sulphate. SULPHATES If a solution of barium chloride is used as the precipitating reagent the sulphate radical may be determined by essentially the same process as that just described for barium. A small concentration of hydrochloric acid must be maintained as other insoluble salts of barium might be formed in a neutral or basic solution. Phosphate, carbonate and oxalate may be mentioned as common examples of such salts. Determination. Weigh duplicate samples of about 0.25 gm of the sulphate into breakers and dissolve in 75 cc of distilled water. Add 1 cc of dilute hydrochloric acid, heat to boiling and add, drop-wise and with constant stirring, a 5 percent solution of barium chloride until the sulphate is completely precipitated. Digest on the steam bath until the precipitate settles and the solution clears, then filter and wash with hot distilled water, testing finally with dilute sulphuric acid to insure removal of barium chloride. Heat in a weighed crucible as directed above for the determination of barium. From the weight of sample and of barium sulphate calculate the percent of the sulphate radical in the sample. In some cases it is desirable to calculate the percent of sulphur or of sulphur trioxide. This may be done by use of the proper factor. Free sulphuric acid may be determined by the same process. However, since the reaction of this with barium chloride produces free hydrochloric acid it is unnecessary to add any of the latter. 96 QUANTITATIVE ANALYSIS STRONTIUM Strontium is best determined as sulphate, precipitating from a solution containing alcohol and an excess of dilute sulphuric acid. Its solubility in water at 18 is 114 mg per liter. 1 The solubility is considerably diminished by a small excess of sul- phuric acid and in 50 percent aqueous alcohol the solubility is very slight, although no definite figures are now available. Determination. Weigh the proper quantity of strontium salt to produce 0.2 to 0.3 gm of strontium sulphate. If soluble in water dis- solve in 50 cc of water and add to the solution 60 cc of alcohol. If insoluble in water dissolve in hydrochloric acid, evaporate in a cas- serole to expel excess of acid and dilute the solution to 50 cc, then add 60 cc of alcohol. Add, slowly and with stirring, dilute sulphuric acid until precipitation is complete. An excess of about 5 Cc is desirable. Stir for some time then allow to stand for 12 hours. Filter and wash twice with 50 percent alcohol containing a few drops of dilute sulphuric acid, then with 50 percent alcohol until the washings fail to give a test for sulphates. Ignite in a weighed crucible at as low a temperature as possible until white. Weigh the strontium sulphate and calculate the percent of strontium in the sample. The separation of barium, strontium and calcium is accom- plished by converting all of the metals into nitrates, evaporating to dryness and taking up with a mixture of alcohol and ether. Calcium nitrate dissolves and the calcium is precipitated as oxalate after evaporating the alcohol and ether and dissolving the residue in water. The barium and strontium nitrates are dissolved in water, barium is precipitated as chromate and stron- tium as sulphate as described above. POTASSIUM AND SODIUM Potassium may be separated from sodium and determined as perchlorate or as chlorplatinate. It may also be precipitated as potassium sodium cobaltinitrite and this determined by a volumetric process or dissolved and the potassium later deter- mined by the other gravimetric methods. It is often stated that it can also be determined by weighing as sulphate or as chloride. Inasmuch as the latter two methods involve no separa- 1 Kohlrausch: Z. physik. Chem., 60, 355 (1905). EXPERIMENTAL GRAVIMETRIC ANALYSIS 97 tion by precipitation and filtration but simply conversion of the potassium into a form other than the one in which it formerly existed and since any other metals that might be present would also be converted into sulphates or chlorides and weighed with the potassium, the value of these methods is not apparent. Of the first two methods the perchlorate method has the advantage of cheapness, while the chlorplatinate method is more convenient and probably more accurate. So far as is indicated by present experience the chlorplatinate method is the most reliable and, at the same time, the most convenient of all of the known methods for the potassium de- termination. The cobaltinitrite method has not yet been im- proved to the point of a satisfactory quantitative method, although useful in qualitative analysis. The perchlorate method has also suffered the drawback of questionable reliability, as well as that of requiring a reagent which is more or less dangerous to prepare and handle. The one serious obstacle to the continued use of the chlorplatinate method is the very high cost of the reagent itself. The great scarcity and the ex- traordinary rise in the price of platinum since the beginning of the world war has made it increasingly desirable, if not absolutely essential, that some method that does not involve this metal shall be used. In the industrial laboratory a considerable portion of the used platinum is systematically recovered but the recovery involves considerable work and expense, while a certain fraction of the metal is lost in each recovery operation. The chlorplatinate method has been varied in matters of detail but essentially it consists in precipitating potassium chlorplati- nate from an alcoholic solution by adding chlorplatinic acid. Compounds of sodium arid magnesium are not so precipitated and potassium is separated from these by filtration but other metals must be absent because of the small solubility of most chlorplatinates. Potassium chlorplatinate may be either weighed as such or ignited in a current of hydrogen : K 2 PtCl 6 +2H 2 -2KCl+Pt+4HCl. The potassium chloride is washed out of the reduced platinum 7 98 QUANTITATIVE ANALYSIS which is then weighed. In practice potassium chlorplatinate is usually weighed without ignition. Washing. The separation from alcoholic solution can be accomplished only under certain conditions, owing to the small solubilities of certain other compounds that may be present. Sodium chlorplatinate is easily soluble in alcohol but ammonium chlorplatinate is soluble to a very slight extent. Ammonium compounds must therefore be volatilized by heating, before the addition of chlorplatinic acid. Sodium chloride and sulphate are nearly insoluble in alcohol, and must be changed into more soluble substances before a separation of sodium and potassium can be accomplished. The following table will serve to show the nature of the questions that must here be met. There is con- siderable disagreement between the results as obtained by dif- ferent investigators but these figures may be regarded as at least approximately correct. Percent alcohol in water Solubility, grams salt per liter of solvent Sodium sulphate 1 0.7 127.0 19.4 26.0 72.0 <0.001 Sodium chloride 80 about 5 . Ammonium chlorplatinate 2 . . . . 55 0.150 76 0.067 95 0.0037 Potassium chlorplatinate 3 7.742 10 3.72 20 2.18 30 1.34 40 0.76 50 0.491 60 0.265 70 0.128 80 0.085 90 0.025 100 0.009 l de Bruyn: Z. physik. Chem., 32, 63 (1900). 2 Fresenius: Z. anal. Chem., 36, 322 (1897). 3 Archibald, Wilcox and Buckley: J. Am. Chem. Soc., 30, 747 (1908). EXPERIMENTAL GRAVIMETRIC ANALYSIS 99 The Lindo Method. The method of Lindo 1 consisted in obtain- ing a solution containing chlorides of no other metals than sodium and potassium, adding sufficient chlorplatinic acid to convert all of the chlorides into chlorplatinates, evaporating nearly to dryness and adding strong alcohol. Sodium chlorplatinate dissolves arid is separated from the potassium chlorplatinate by filtration. The potassium salt is washed with alcohol and weighed or treated as already described. Lindo also showed that the solubility of very fine crystals of potassium chlor- platinate is greater than that of larger crystals (cf. page 21). Using this method, it was necessary that no sulphate should be present because of the very slight solubility of sodium sul- phate in alcohol. The discussion of the laws of precipitation (page 15 et seq.) will make it clear that even if enough chlor- platinic acid were present to combine with all of the sodium present to form sodium chlorplatinate, the less soluble sodium sulphate would still be precipitated. Even when all of the metals are present as chlorides there is without doubt some contamina- tion of the precipitated potassium chlorplatinate by sodium chloride. It was early observed 2 that high results would be obtained by this method for determining potassium if the factor for potassium in potassium chlorplatinate were calculated by the use of the atomic weight of platinum as determined by Seubert 3 (194.8) or even 195.2, which is that given in the table of atomic weights for 1918, while a factor calculated from the atomic weight 197.2, which had previously been accepted, gave correct results. This may be partly due 4 to the fact that the pre- cipitated potassium chlorplatinate contains also some compounds with composition as represented by such a formula as H 2 PtCl 5 OH or H 2 PtCl 4 O. It is undoubtedly also partly due to the presence of some sodium chloride in the precipitate. On this account the method has been modified by using 80 percent alcohol instead of absolute alcohol. Reference to the solubility table above will show that alcohol of this concentration will dissolve potassium chlorplatinate to a greater extent than does absolute 'Chem. News, 44, 77, 86, 97 and 129 (1881). 2 Dittmar and Me Arthur: J. Soc. Chem. Ind., 6, 799 (1887). 3 Ann., 207, 1 (1881). 4 Dittmar and Me Arthur; loc. cit. 100 QUANTITATIVE ANALYSIS alcohol and this negative error seems practically to balance the positive error discussed. Gladding' s Modification. The fact that volatile acids, organic compounds, ammonium salts, etc., can be easily volatilized by heating makes it desirable to obtain the sodium and potassium in a form in which they can be heated to redness without danger of loss. Both chlorides are sensibly volatile at such temperatures while the sulphates are not but, as already stated, sodium sulphate must not be allowed to form because of its small solubility in alcohol. To meet this difficulty Gladding suggested 1 a further modification of the Lindo method in which the sodium sulphate was to be washed out, after the removal of the excess of chlor- platinic acid, by a water solution of ammonium chloride. The solubility of sodium sulphate in water is considerably increased by the presence of ammonium chloride. This follows the general law that the addition of an electrolyte which does not contain an ion in common with the first electrolyte increases its solubility. To avoid loss of potassium chlorplatinate the washing liquid is previously saturated with the pure salt. In using such a solution it is important that no great change in temperature shall occur in the solution after it is withdrawn from the bottle and before it is used for washing the precipitate. This is because the solution is kept saturated by an excess of potassium chlorplatinate in the bottle. If the temperature should rise the solution which was saturated in the bottle now becomes undersaturated and will dissolve some of the precipitate on the filter. On the other hand, if the funnel is at a much lower temperature than the reagent bottle (due to working in a colder part of the room) it will cool the solution and cause a deposition of potassium chlorplatinate upon the precipitate already present. This modified method, known as the Lindo-Gladding method, is now used quite generally, particularly for the determination of potassium (and indirectly of sodium) in industrial products, minerals, etc. Decomposition. During the evaporation of sodium, potassium and ammonium salts with sulphuric acid the first product is, of course, a mixture of the acid sulphates. After the evaporation 1 U. S. Dept. of Agr., Chem. Bull., 7, 38. EXPERIMENTAL GRAVIMETRIC ANALYSIS 101 of the excess of acid and upon further heating, the pyrosulphates of sodium and potassium are formed: 2KHSO 4 -^K 2 S 2 7 +H 2 0; 2NaHSO 4 ->Na 2 S 2 O 7 +H 2 0. These reactions begin at about 350. At a temperature of dull redness the normal sulphates begin to form: K 2 S 2 7 -*K 2 SO 4 +S0 3 ; Na 2 S 2 7 -Na 2 SO 4 +SO 3 . This decomposition requires heating for some time. It is best to test the completion of decomposition by repeating the heating until the weight becomes constant. Ammonium acid sulphate is decomposed and volatilized as ammonia, water and sulphur trioxide: NH 4 HS0 4 -^NH 3 +H 2 0+S0 3 . If it is desired to accomplish the removal of ammonium salts and organic matter by ignition but to avoid the use of the am- monium chloride washing solution the sulphates, first obtained by evaporating with sulphuric acid and igniting, are dissolved and precipitated by barium chloride which precipitates barium sulphate and leaves sodium chloride and potassium chloride in solution. The excess of barium is then precipitated by sulphuric acid. There is no method known for the direct determination of sodium if we exclude the weighing as sulphate or chloride, methods of very limited usefulness. This is because no sodium compound has sufficiently small solubility to make possible its separation from the corresponding salt of potassium. Sodium is usually determined by weighing it with potassium in the form of sulphate or chloride, determining potassium and calculating sodium by difference. It should be noted that such a method throws all of the errors of the potassium determination upon that of sodium, in addition to any errors that may have occurred in the weighing of the combined chlorides or sulphates. Ammonium. Chlorplatinic acid is also used as a reagent for the quantitative determination of the ammonium radical but potassium must be absent. On account of the difficulty ex- 102 QUANTITATIVE ANALYSIS perienced in the removal of potassium from ammonium the latter is more conveniently determined by volumetric methods. Platinum. The converse of the Lindo method for potassium is used for the determination of platinum. Either ammonium chloride or potassium chloride may be used as the reagent but the former is generally used because of its greater solubility in alcohol, which makes the removal of the excess of reagent more easy. Determination by the Lindo-Gladding Method. (To be performed in an atmosphere which is free from ammonia). Use portions of about 0.3 gm of a sample containing salts of potassium and sodium and weigh into small weighed evaporating dishes. Dissolve in a small amount of hot water, add 0.5 cc of concentrated sulphuric acid, evaporate to dry- ness under the hood, using care to avoid spattering, and ignite at bright redness until no more white fumes are evolved and the residue is white. The steam bath should not be used for the evaporation on account of appreciable solubility of porcelain in steam, and consequent loss in weight. Cool, weigh and ignite again to constant weight. Record the weight of sulphates of sodium and potassium. Dissolve the resulting potassium sulphate (mixed with sodium sulphate) in 50 cc of hot water and then add chlorplatinic acid, using about 1 cc more than the theoret- ical amount, calculated upon the assumption that the original salt was potassium chloride. Evaporate on the steam bath to a thick paste but not to dryness, cool and add 50 cc of 80 percent alcohol, stir up the solid matter and allow to stand, covered, for 30 minutes. If the liquid is not visibly colored too little reagent has been used. In this case new samples should be taken and the quantity of chlor- platinic acid increased. Filter and wash the precipitate thoroughly with 80 percent alcohol, washing several times after the washings pass through colorless. The wash bottle should be provided with ground- glass joints so that no rubber will come into contact with the alcohol. Remove the nitrate and washings, pouring these into the bottle provided for platinum waste residues, and wash the precipitate with five or six portions of 10 cc each of 10 percent ammonium chloride solution which is saturated with potassium chlorplatinate. Wash again, thoroughly, with 80 percent alcohol, using particular care in washing the upper part of the paper free from ammonium chloride. Wash until only a faint turbidity is produced by the addition of a drop of silver nitrate to the last washings. Drain most of the alcohol from the paper, slip the latter out of the funnel and dry in the oven at 100. Place a weighed porcelain crucible EXPERIMENTAL GRAVIMETRIC ANALYSIS 103 upon a piece of glazed paper, remove most of the precipitate to the crucible, brushing up any particles that may have fallen upon the glazed paper, and then replace the paper in the funnel. Place the crucible under the funnel and dissolve the remainder of the precipitate in the smallest amount of nearly boiling water, allowing the solution to run into the crucible. Evaporate to dryness on the steam bath, carefully wipe ' the outside of the crucible with a clean towel and dry for 30 minutes at 100. Weigh and calculate the percent of potassium in the salt ana- lyzed. Calculate also the weight of potassium sulphate, subtract this weight from that of the mixed sulphates, and from the remainder calculate the percent of sodium. Optional Method, Using a Gooch Crucible. Proceed as above until ready to filter out the potassium chlorplatinate. Prepare two Gooch niters as directed on page 86, paying attention to the precautions sug- gested, and using strong suction in forming the asbestos felt. Riiise the crucible with alcohol, remove, wipe the outside and dry at 100 to 105 for 30 minutes or until the weight is constant. Weigh and replace in the holder. Before the suction pump is again turned on moisten the asbestos with one or two drops of water. Start the pump and filter and wash the precipitate exactly as above directed. Remove the crucible, dry in the oven and weigh. Calculate potassium and sodium as before. In the foregoing exercise the procedure is based upon the as- sumption that sodium or ammonium salts or both may be pres- ent. The latter are volatilized by heating with sulphuric acid. The former are removed by the ammonium chloride solution. Recovery of Platinum from Waste and Preparation of Chlor- platinic Acid. The recovery and purification of platinum from miscellaneous filtrates and other waste solutions is a matter of increasing importance on account of the extraordinary increase in the price of platinum within recent years. The following method, described by Delong, 1 has been found to serve well for this purpose. Place the solutions in an evaporating dish having a capacity of 2 liters for each 100 gm of platinum and evaporate until most of the water has been expelled. Make basic with sodium hydroxide solution and add, stirring, sodium formate, either solid or in concentrated solution. A quantity of sodium formate equal to about half the weight of platinum will be required. If foaming occurs add more sodium hydroxide. Heat on the steam bath for 1 hour, stirring occasionally, then acidify 1 Chem. Weekblad, 10, 833 (1914). 104 QUANTITATIVE ANALYSIS with hydrochloric acid, 25 percent solution, stirring during the addition of acid. Filter off the precipitated platinum on a soft paper, using suction. Wash twice with hot 2 percent hydrochloric acid, then with hot water until free from acid. Separate the platinum from the paper, dry, ignite and weigh. Pour over the platinum in a porcelain dish five times its weight of 25 percent hydrochloric acid, heat on the steam bath and add slowly 50 percent nitric acid until no more gas is evolved. About 1 cc of nitric acid will be required for each gram of platinum. After the platinum is in solution add 10 cc of 25 percent hydrochloric acid and evaporate to small volume and repeat this process twice. This reduces and eliminates nitric acid. Dilute with water and evapo- rate two or three times to expel hydrochloric acid. Finally dilute, cool and filter on a soft filter whose approximate w r eight is known. If the filtrate is not perfectly clear refilter. Wash the paper free from platinum stain and if any appreciable residue remains, dry and weigh it on the filter. Correct the weight of platinum for this weight of car- bon, etc., then make the solution to the desired concentration. For potassium determinations the solution should contain 0.1 gm of platinum in 1 cc. A method for the recovery of platinum from scrap by elec- trolysis is described by Weber. 1 Chlorplatinic acid prepared by this method is quite free from traces of nitric acid. The Perchlorate Method. The perchlorate method is based upon the fact that potassium perchlorate is almost insoluble in 97 percent ethyl alcohol, while sodium perchlorate dissolves with greater ease. It involves the use of an aqueous solution of perchloric acid, the preparation of which is somewhat trouble- some and dangerous. The solubility of potassium perchlorate in alcohol of various concentrations is as follows: 2 Concentration of alcohol, percent by weight Grams, potassium perchlorate in one liter of solvent Grams, potassium equivalent to potassium perchlorate 97.2 95.8 90.0 0.156 0.20 0.36 0.044 0.06 0.10 1 J. Am. Chem. Soc., 30, 29 (1908). 2 Wenze: Z. angew. Chem., 6, 691 (1891). EXPERIMENTAL GRAVIMETRIC ANALYSIS 105 The solubility is considerably diminished by an excess of per- chloric acid. Sodium perchlorate dissolves easily in alcohol although no definite data are on record. The perchlorate method has been improved by the sub- stitution of a 20 percent solution of perchloric acid for the pure acid, as formerly used. This solution keeps well and involves little or no danger of accident through handling. As in the chlorplatinate method it is necessary to remove ammonium salts. This may be done by gently heating the chlorides or, more safely, by evaporating with sulphuric acid and heating the sulphates rather more strongly. If the latter method is followed it becomes necessary to reconvert the salts to chlorides before precipitating potassium perchlorate because of the limited solubility of sodium sulphate in 97 percent alcohol, which is used for washing the precipitate. The analytical method for materials containing salts of only sodium, potassium and ammonium is described below. Determination. Weigh 1 gm of the sample in which sodium and potassium are to be determined, brushing into a porcelain or platinum dish. Treat with sulphuric acid and evaporate and heat to expel ammonium salts and excess of acid, using the 'procedure as described for the chlorplatinate method. Dissolve the weighed sulphates of sodium and potassium in 50 cc of hot water, add one drop of concen- trated hydrochloric acid, heat nearly to boiling and then add, dropwise and with constant stirring, a 5 percent solution of barium chloride until all sulphate is precipitated. This operation should be performed very carefully in order to have the least possible excess of barium chloride at the end. Digest for a short time over a small flame or on the steam bath, then filter and wash the precipitate and paper well with hot water. Evaporate the nitrate and washings to about 25 cc and add 10 cc of 20 percent perchloric acid solution. Evaporate over a steam bath in a hood until the solution becomes viscous, cool and dissolve the residue in a small amount of hot water. Again add 5 cc of perchloric acid solution and evaporate over the steam bath until the solution evolves dense white fumes of perchloric acid. Cool to room tem- perature and add 25 cc of a solution made by mixing 1 cc of 20 percent perchloric acid with 100 cc of 98 percent alcohol (making practically 97 percent alcohol). If the insoluble potassium perchlorate is caked it should be broken with a stirring rod so that no soluble salts will escape the action of the alcohol. 106 QUANTITATIVE ANALYSIS During the process of evaporation of the various solutions a Gooch filter should be prepared, the asbestos felt being washed with the perchloric acid-alcohol mixture. The filter is dried for 1 hour at 120 to 130, cooled and weighed. Filter the solution on this prepared filter, removing every trace of the precipitate from the beaker by means of a policeman and the prepared washing solution, and wash four or five times with this solution. Dry for 1 hour at 120 to 130, cool and weigh. From the weight of potassium perchlorate thus obtained calculate the percent of potassium in the sample." Also from this weight calcu- late the weight of potassium sulphate which is equivalent to it, subtract this weight from the combined weights of sodium sulphate and potassium sulphate and from the remaining sodium sulphate calculate the percent of sodium in the sample. MAGNESIUM The determination of magnesium is usually made by pre- cipitating from a basic solution as di magnesium ammonium orthophosphate. This is ignited and weighed as magnesium pyrophosphate. The reactions may be expressed thus : MgCl 2 +NH 4 OH+Na 2 HP0 4 -^MgNH 4 P04+2NaCl-|-H 2 0, 2MgNH 4 P04-Mg 2 P 2 07+2NH 3 +H 2 0. No other metals than those of the alkali group may be present, as the phosphates of practically all others are insoluble in am- monium hydroxide. Any soluble phosphate may be used as the precipitating reagent but the ones most used in practice are disodium orthophosphate and sodium ammonium acid ortho- phosphate (microcosmic salt). Solubility. The following tabular statement from the work of Ebermayer 1 shows the solubility of crystallized dimagnesium ammonium orthophosphate in mixtures of ammonium hydroxide and water. Percent by volume of ammonium hydroxide of sp. gr. 0.96 Grams of MgNH 4 PO 4 .6H 2 O per liter of solvent at 15 Equivalent grams of Mg per liter of solvent 25 50 75 0.074 0.027 0.023 0.019 0.0072 0.0026 0.0022 0.0018 J. prakt. Chem., 60, 41 (1853). EXPERIMENTAL GRAVIMETRIC ANALYSIS 107 This statement of the solubility of the precipitate in solutions containing various concentrations of ammonium hydroxide would lead to the conclusion that precipitation from the more concentrated solutions of ammonium hydroxide would result m greater accuracy because of the small solubility of magnesium ammonium orthophosphate. From this standpoint alone the conclusion would be correct. It happens, however, that the basicity of the solution, as well as the presence of other salts, has an important influence upon the composition of the precipitate. The decrease of solubility with increasing concentrations of ammonium hydroxide and also of ammonium salts is to be expected as a consequence of the mass law since these substances + increase the concentration of the ion NEU, a constituent of the precipitate. Substances other than dimagnesium am- monium orthophosphate are precipitated to some extent under the following conditions: Effect of High Basicity. If the solution is strongly basic when the reagent is added there is formed some trimagnesium orthophosphate, Mgs(PO4) 2 , and the quantity of this substance is increased by slow addition of the reagent. This is not decom- posed upon heating and the ignited precipitate is therefore not all magnesium pyrophosphate. This fact makes it undesirable that too much ammonium hydroxide should be present, even though the solubility of the precipitate is lessened thereby. The solubility of dimagnesium ammonium orthophosphate is much less than that given by Ebermayer, according to the work of Bube 1 who states that the saturated solution in pure water contains about 0.00014 gm in 1000 cc. It is also stated that in such a solution the solubility product of trimagnesium orthophosphate is far exceeded and that the solubility of di- magnesium ammonium orthophosphate is increased by large concentrations of ammonium ions. This would probably account for the increased precipitation of trimagnesium ortho- phosphate in solutions made strongly basic by ammonium hydrox- ide, the magnesium ammonium salt changing into magnesium phosphate and ammonium phosphate, the magnesium salt precipitating : 3MgNH 4 P0 4 -Mg 3 (P04) 2 + (NH 4 ) 3 P0 4 . 1 Z. anal. Chem., 49, 525 (1910). 108 QUANTITATIVE ANALYSIS Effect of Ammonium Salts. -If the solution contains excessive quantities of ammonium salts, whether the precipitation takes place from a strongly basic or weakly basic solution the dimag- nesium ammonium orthophosphate will contain certain quantities of monomagnesium ammonium orthophosphate, Mg(NH 4 ) 4 (P04) 2 . This substance, when strongly heated, passes into magnesium metaphosphate, Mg(POs)2, a substance which can be converted into magnesium pyrophosphate only after prolonged heating at high temperatures (2Mg(PO 3 )2 Mg2P 2 07+P2O 5 ). / Ammonium salts should thus be nearly or entirely absent, with the exception of a certain amount of ammonium chloride, which must be present to prevent the precipitation of magnesium hydroxide. If they have accumulated in the solution as a result of the use of ammonium hydroxide in the separation of other metals, they should be removed before precipitation, by (a) evaporating to dryness and heating strongly or (6) evaporating to small volume and heating with concentrated nitric acid or (c) performing a double precipitation, dissolving the first impure precipitate in hydrochloric acid and reprecipitating. Method (a) or (c) is to be preferred. Temperature and Rate of Precipitation. Most chemists prefer to precipitate magnesium from a cold solution although Gibbs 1 recommends a boiling solution. Whether the cold or hot solution is used one of two procedures may be followed in order to conform to the principles outlined above. The entire amount of disodium phosphate solution may be added at once to an acid solution and then dilute ammonium hydroxide slowly added until the solution is basic. After standing a short time most of the precipitate will form and the remaining magnesium can be precipitated by the addition of concentrated ammonium hydroxide. Instead of following this method the solution may be made neutral or faintly basic and disodium phosphate added slowly, thus precipitating nearly all of the substance, when concentrated ammonium hydroxide may be added as before. The second method is recommended. Decomposition upon Heating. The reason for the difference in the mode of decomposition of the precipitate containing more of the ammonium radical from that of the dimagnesium am- 1 Am. J. Sci., [3] 6, 114. EXPERIMENTAL GRAVIMETRIC ANALYSIS 109 rnonium salt is apparent when the properties of the three phos- phoric acids and of the salts are examined. Phosphorus pent- oxide, by combining with different proportions of water, gives rise to three different acids : P 2 5 + H 2 >2HP0 3 , metaphosphoric acid, P 2 5 +2H 2 O >H 4 P 2 O 7 , pyrophosphoric acid. P 2 5 -f-3H 2 0-2H 3 P0 4 , orthophosphoric acid. Either metaphosphoric or pyrophosphoric acid will be trans- formed into the one containing more water if allowed to stand in solution. Also the acids may be changed in the opposite sense by heating. At about 213 orthophosphoric acid loses water and yields pyrophosphoric acid. 2H 3 P0 4 ^H 4 P 2 7 +H 2 0. At about 400 pyrophosphoric acid loses one molecule of water and yields metaphosphoric acid. When heated to higher temperatures the remaining molecule of water is lost and phosphorus pentoxide remains 2HP0 3 -P 2 5 +H 2 0. Phosphorus pentoxide is thus seen to be the final product of any of the three acids when the acid is heated to a high temperature and this is because a volatile substance (water) is produced by heating. Just as the acids are compounds of phosphorus pent- oxide and water, so the salts may be regarded as compounds of phosphorus pentoxide and metallic oxide (which is analogous to hydrogen oxide). Consequently the extent to which the salts may be decomposed by heating will be conditioned by the nature of the metallic oxide or, in other words, by its degree of volatility. Thus the normal phosphates of sodium, potassium, magnesium, calcium, etc., are not decomposable at all, except at extremely high temperatures where phosphorus pentoxide begins to be volatile, while the acid phosphates of these metals are decomposable to whatever extent is denoted by the propor- tion of water that may be formed. Ammonium salts are con- verted completely into phosphorus pentoxide because, instead of the hypothetical metallic oxide, (NH 4 ) 2 0, there are formed 110 QUANTITATIVE ANALYSIS ammonia and water and both of these substances are volatile. Orthophosphoric and pyrophosphoric acids are polybasic and a considerable variety of salts may be prepared, containing varying amounts of metals, ammonium and hydrogen, so that they may be regarded as containing varying amounts of metallic oxide, ammonia, water and phosphorus pentoxide. Examples. The composition and decomposition of the three phosphoric acids and typical examples of their salts are shown in the following statement: Composition | Decomposition by Heat Acids 2HP0 8 H 4 P 2 O 7 =0= 2H 2 O.P 2 S 2H,P0 4 =0=3H 2 O.P 2 O 6 2HPO 3 ->P 2 H 4 P 2 7 -P 2 Os 2HsPO 4 ->P 2 O5+ 3H 2 O Normal Potassium Salts 2KPO S OK K 4 P 2 7 0= 2K 2 O.P 2 O 6 2K,P0 4 03K 2 O.P 2 Os Not decomposed except by slow loss of P 2 O at high temperatures Potassium Acid Salts No acid metaphosphate possible K 2 H 2 P 2 O 7 =c= K 2 O.H 2 O.P 2 O B 2K 2 HPO 4 =^ 2K 2 O.H 2 O.P 2 O S 2KH 2 PO 4 O K 2 O.2H 2 O.P 2 O B K 2 H 2 P 2 7 ->2KPO 3 + H 2 O 2K 2 HP0 4 ->K 4 P 2 O 7 + H 2 O 2KH 2 PO 4 ->2KP0 3 + 2H 2 O Potassium Ammonium Salts No double metaphosphate possible K 2 (NH 4 ) 2 P 2 7 K 2 0.(NH 4 ) 2 O.P 2 O 8 2K 2 NH 4 PO 4 =C= 2K 2 O.(NH 4 ) 2 O.P 2 O 6 2K(NH 4 ) 2 PO 4 OrKz 2K 2 NH 4 PO 4 -K 4 P 2 O 7 4-2NH 3 + H 2 O 2K(NH 4 -) 2 PO 4 -> 2KPO 3 + 4NH j -f- 2H 2 O Monomagnesium ammonium orthophosphate, Mg(NH 4 )4(PO 4 )2, is analogous to monopotassium ammonium orthophosphate, K(NH 4 ) 2 P04, as may be seen from the structural formulae: x NH 4 4 -,T0 4 and N 4 / M g 4 NH 4 / Its decomposition can therefore proceed as far as magnesium metaphosphate : Mg(NH4)4(P0 4 ) 2 -^Mg(P0 3 )2+4NH 3 +2H 2 0. EXPERIMENTAL GRAVIMETRIC ANALYSIS 111 while dimagnesium ammonium orthophosphate can decompose only as far as the pyrophosphate : 2MgNH 4 P04-^Mg 2 P 2 O 7 +2NH 3 4-H 2 O. This makes it necessary that such conditions shall be maintained as will make possible the formation of but one double salt, in order that the composition of the ignited precipitate may be definite and constant. Rate of Crystallization. The complete precipitation of dimagnesium ammonium orthophosphate takes place only after standing for some time. Formerly it was considered neces- sary to allow 24 hours for the action to proceed. It is now gener- ally considered that from 2 to 3 hours is sufficient for the pre- cipitation of all but a minute amount, negligible under ordinary circumstances. The usual method of testing with excess of reagent, to determine whether precipitation is complete, is rendered useless because of the slow crystallization of the pre- cipitate unless several hours are allowed for the possible pre- cipitation of small amounts. The crystalline precipitate may be readily filtered and washed by a dilute solution of ammonium hydroxide or ammonium nitrate. The precipitate is appreciably soluble in distilled water. An application of the laws of solubility, discussed under the head of " Precipitation" (page 15) would lead to the conclusion that any one of the three classes of soluble compounds: phosphates, ammonium salts or magnesium salts, will lessen the solubility of ammonium magnesium phosphate + ++ since the latter dissociates into the three ions NH 4 , Mg and PO^ The addition of magnesium salts to the washing fluid is clearly out of the question if magnesium is to be determined. Phosphates must themselves be removed by washing because only the ammonium phosphates are entirely volatile and these only with some difficulty. Either ammonium hydroxide or ammonium nitrate is suitable for the purpose, excess of either being driven off during drying and ignition of the precipitate. Ignition. Considerable difficulty is often experienced in obtaining pure, white magnesium pyrophosphate by igniting the magnesium ammonium orthophosphate. This is usually due to imperfect washing, sodium phosphate being left in the pre- 112 QUANTITATIVE ANALYSIS cipitate. Upon heating, traces of the salt cause partial fusion, particles of carbon being enclosed and oxidation made difficult. Thorough washing followed by long heating at high temperatures is the only remedy. A similar method may also be used for the determination of arsenic acid, the precipitate of magnesium ammonium arsenate being heated at a moderate temperature until it forms magnesium pyroarsenate : Determination. Weigh into Pyrex beakers portions of about 0.3 gm of a magnesium salt. If the salt is soluble in water dissolve in about 100 cc of distilled water and add a drop of concentrated hydrochloric acid. If not soluble in water dissolve in hydrochloric acid (1 part of concentrated acid to 1 part of water), warming if necessary. Cool and drop in a very small piece of litmus paper and then add, slowly and with stirring, dilute ammonium hydroxide until the solution is faintly basic. Now add from a pipette, slowly and with stirring, 15 cc of a clear 10 percent solution of disodium orthophosphate. Allow to stand for 15 minutes until a considerable part of the precipitate has appeared, then add concentrated ammonium hydroxide solution (sp. gr. 0.90), in such quantity that the solution shall finally contain ammonium hydrox- ide equivalent to one-ninth of its total volume. Cover and allow to stand for three hours or stir continuously for 30 minutes. Filter the precipitate on a filter of extracted paper, in a weighed platinum Gooch crucible or in an ignited and weighed alundum crucible and wash until free from chlorides with a solution containing 2 percent of ammonia or 5 percent of ammonium nitrate, finally testing the washings with silver nitrate after acidifying with nitric acid. If a Gooch crucible has been used place the cap on the bottom and heat over the burner until dry, then over the blast lamp for 20 minutes. An alundum crucible is treated similarly. If a paper filter was used remove the paper from the funnel and, if sufficient precipitate is present to make its removal from the paper feasible, dry and remove most of the precipitate to a sheet of glazed paper, refold the paper and place in a weighed porcelain or platinum crucible. Incline the crucible with the cover leaned against it and heat gently over the burner until the paper is completely burned and the precipitate is nearly white. After the precipitate is white or gray the main portion is added and the cruci- ble is heated for 20 minutes over the blast lamp, cooled in the desiccator and weighed. From the weight of magnesium pyrophosphate calculate that of magnesium and the percent of magnesium in the original sample. EXPERIMENTAL GRAVIMETRIC ANALYSIS 113 PHOSPHATES For the precipitation of the phosphate radical as magnesium ammonium phosphate it is necessary that no metal that can_ form an insoluble phosphate shall be present. In case this condition is not fulfilled a preliminary separation of the phosphate radical is made by precipitating ammonium phosphomolybdate from a solution containing free nitric acid. This operation is described on page 453 and following, in the discussion of the determination of phosphorus in steel. A sample containing no metals other than those of the alkali group does not require this treatment. This simpler determination will now be described. Determination. Prepare a solution of "magnesia mixture" as follows: Dissolve 55 gm of crystallized magnesium chloride and 140 gm of ammonium chloride in water, add 130 cc of ammonium hydroxide (specific gravity 0.90) and dilute to 1000 cc. If this solution is kept in stock for any considerable time it will acquire a flocculent precipitate of hydrated silica, derived from solution of the glass by the base. The solution must be clear when used. This condition may be insured by filtering the solution or by preparing only enough of the reagent to last a short time. Weigh duplicate samples of 0.2 to 0.4 gm of the phosphate into beakers of resistance glass, dissolve and dilute to 75 cc. Drop in a very small bit of litmus paper and if a basic reaction is not shown add dilute ammonium hydroxide until the paper becomes blue. Add 10 cc of a 10 percent solution of ammonium chloride, mix and then add, very slowly, "magnesia mixture" sufficient in quantity to precipitate all of the phosphate. As the precipitate does not form rapidly in a barely basic solution it is not always easy to determine when enough of the reagent has been added. It is then best to use what is thought to be a good excess and to rely upon testing the filtrate which is obtained later. The rest of the procedure is exactly the same as in the determination of magnesium, with the single exception that "magnesia mixture" instead of sodium phosphate solution is used in testing for the completion of precipitation. From the weight of magnesium pyrophosphate finally obtained calculate the percent of the phosphate radical, of phosphorus pentoxide or of phosphorus, the report depending upon the nature of the sample examined. 114 QUANTITATIVE ANALYSIS MANGANESE The method of Gibbs for manganese depends upon the same chemical principles as are involved in the determination of magnesium. A soluble orthophosphate is added to the solution of the manganese salt and the solution is then made basic with ammonium hydroxide. Dimanganese ammonium orthophos- phate is precipitated and this, when ignited, gives manganese pyrophosphate. 2MnNH 4 P04->Mn 2 P 2 7 +2NH 3 +H 2 O. If the manganese is already in its lowest state of oxidation, precipitation is accomplished without further change. If it is in the form of a manganate or permanganate or of manganese dioxide it is first reduced by sulphurous acid: Mn0 2 +H 2 S0 3 -MnSO4+H 2 O. Determination. Weigh enough of the sample to contain about 0.1 gm of manganese. If the sample is a permanganate or manganese di- oxide dissolve in 50 cc of a saturated solution of sulphurous acid, con- taining also 1 percent of hydrochloric acid, filter if necessary and boil to expel the excess of sulphur dioxide. If the sample is a soluble manganese salt omit the treatment with sulphurous acid. In either case proceed as follows : Add to the solution in a Pyrex beaker, 3 percent more than the quan- tity of 10 percent disodium phosphate solution calculated to be neces- sary for complete precipitation of the manganese. Heat to boiling and add dilute ammonium hydroxide solution, drop by drop with constant stirring, until a precipitate begins to form. Boil and stir until this pre- cipitate becomes crystalline, then add another drop of ammonium hy- droxide and stir and boil until the additional amorphous precipitate becomes crystalline. Continue this process until further addition of ammonium hydroxide produces no precipitate. All of the precipitate should now be in the crystalline condition. Add 0.5 cc excess of ammo- nium hydroxide, then cool the solution by placing the beaker in ice-water. Filter and wash with a clear, slightly basic, 10 percent solution of am- monium nitrate or a 2 percent solution of ammonium hydroxide until free from chlorides, then ignite with the same precautions as were ob- served in the ignition of magnesium ammonium phosphate. From the weight of manganese pyrophosphate obtained calculate the percent of manganese in the sample. EXPERIMENTAL GRAVIMETRIC ANALYSIS 115 CHLORINE, BROMINE AND IODINE The members of the halogen group may occur in different forms, requiring different methods of procedure. This occur- rence may be as free halogens, as oxyacids or salts, as hydracids or salts, or as organic compounds. The gravimetric determina- tion of the negative radical of the halogen hydracids is invariably made by precipitating and weighing the silver salt. The solubilities of the latter, as well as the principles involved in the precipitation, washing and ignition, were discussed under the description of the determination of silver. The procedure is also similar to that involved in the determination of silver, the silver salt (silver nitrate) being, in this case, the precipitant while the chloride, bromide or iodide is the substance being investigated. Separation of Chlorine and Iodine. If chlorides and iodides occur together the iodine may be precipitated as palladious iodide by a solution of palladious chloride, PdCl 2 . In another portion the total halogen may be precipitated by silver nitrate and weighed as a mixture of silver iodide and chloride. The proper weight of silver iodide, as calculated from the weight of pal- ladious iodide found, is subtracted and chlorine calculated from the remainder. Indirect Method. The chlorine and iodine may be determined indirectly by precipitating by excess of silver nitrate, weighing the mixed chloride and iodide, then converting the silver iodide into chloride by heating in an atmosphere of chlorine and reweighing. If x = weight of chlorine, a = weight of silver chloride and iodide, b = weight after conversion of silver iodide into chloride and y = weight of iodine, then -^-^ - x = weight of silver chloride oO.4o 234.80 and 12Q92 y = we ^t of silver iodide, therefore: 143.34 . 234.80 ' ^ ^-^ 35.46 126. 92 143.34 143.34 , . Subtracting (2) from (1), ~ y = a-b. Then 2/ = 1.3877(a-6), (3) z = 0.63506-0.3877a. (4) 116 QUANTITATIVE ANALYSIS Problems In a manner similar to that shown above for chlorine and iodine, derive the following formulae: 2. x =2.7855 b -1.7004 a, ?/=2.7004a-3.53806, where x = weight of bromine, y= weight of iodine, a = weight of silver bromide and iodide and 6= weight of silver chloride after chlorination. 3. x = 1.0451 6-0.7976 a, y- 1.7974 (0-6), where x = weight of chlorine, y= weight of bromine, a = weight of silver chloride and bromide and b= weight of silver chloride after chlorination. Determination of Two Halogens in Mixed Halides, Indirect Method- Use about 0.5 gm of sample. Dissolve in 75 cc of water, add 0.5 cc of dilute nitric acid and then, drop by drop and with constant stirring, a slight excess of 5 percent silver nitrate solution. 10 to 30 cc will be sufficient. Digest at near the boiling temperature until the precipi- tate settles readily, leaving a clear supernatant solution. Test for completion of the precipitation then filter in a prepared and weighed Gooch crucible and wash free from excess of silver nitrate. Dry at 105 to constant weight. Ignite a small porcelain boat and cool in a desiccator. Place this on a sheet of black glazed paper and carefully remove the asbestos filter and every particle of the silver halides from the filter, placing these in the boat. Allow this to remain in the desiccator for 15 minutes, then weigh. From this weight subtract that of the silver halides, already found. This gives as the remainder the weight of the boat plus asbestos. Prepare a chlorinating apparatus according to Fig. 34. A is a distilling flask of 100 cc capacity, fitted with a funnel tube which reaches to the bottom of the flask. The latter is connected at the side with a glass combustion tube, B, of 12 to 15 mm internal diameter and 40 cm length. Corks are used in the ends of the combustion tube. Place the entire apparatus under a hood. Insert the boat containing the silver halides and asbestos into the middle of the tube. Place 10 gm of potassium permanganate in the flask and then pour into the tube 5 cc of concentrated hydrochloric acid. Warm, if necessary, to start the reaction and when the tube is filled with chlorine carefully heat directly under the boat, using a wing burner. Add more acid to the flask as it may become necessary, in order to maintain a slow evolution of chlorine, and heat the boat for 15 minutes to a temperature just under the fusing point of the silver chloride. Finally raise the temperature EXPERIMENTAL GRAVIMETRIC ANALYSIS 117 until fusion barely begins, then remove the flame. Disconnect the tube from the flask, attach a tube of calcium chloride to one end and slowly draw air through until the tube is quite cool. Remove the boat from the tube and test the odor to determine whether all free chlorine has been removed. Dry for 30 minutes at 105, cool in the desiccator and weigh. From this weight subtract that of the boat plus asbestos. The remainder is the weight of silver chloride. From this and the weight of the mixed silver halides and of the sample calculate the percent of each halogen in the sample, using one of the formulas derived above. The entire experiment should be conducted away from bright light. B FIG. 34. Apparatus for chlorination of mixed halides of silver. Chlorine and iodine may also be separated by the method of graded oxidation, to be described in connection with the separa- tion of the three halogens. Separation of Bromine and Iodine. Mixtures of bromides and iodides may be analyzed by methods similar to those de- scribed above. Palladious bromide is sufficiently soluble to make possible the precipitation of palladious iodide in the pres- ence of the bromide. Also either the indirect analysis or the method of graded oxidation may be used. Separation of Chlorine and Bromine. For mixtures of chlorides and bromides either the method of graded oxidation or that of indirect analysis may be used. For the latter silver 118 QUANTITATIVE ANALYSIS chloride and bromide are weighed together and the bromide is then converted into silver chloride and re weighed. Separation of Chlorine, Bromine, and Iodine, by Graded Oxi- dation. In the discussion of the decomposition voltages of electrolytes (page 141) it is shown that any electrically neutral element that is capable of ion formation will, when placed in contact with a solution containing its ions, generate a definite potential difference whose magnitude depends upon the solution tension of the element and the concentration of its ions already in solution. Two such systems will generate a definite electro- motive force if external connection is made between the non- ionized elements and if the two solutions are brought into contact. This electromotive force is always in the direction that would cause a current to flow externally from the element having the less solution tension (if its ions are positive, or the greater if its ions are negative) to the one having the greater solution tension. When a metal passes into solution and forms positive ions it is thereby oxidized. When an element capable of negative ion formation passes into solution and forms ions it is thereby reduced. Conversely, when metallic ions are converted into mas- sive, uncharged metal they are reduced and when non-metallic or negative ions are discharged they are thereby oxidized. Ac- cording to this view oxidation consists in the addition of positive charges or the removal of negative ones, while reduction is the addition of negative charges or the removal of positive ones. ++ +++ Thus the change : Fe >Fe (change of the ferrous ion to the ferric ion) is oxidation, while the reverse is reduction and MnO4 > MnO 4 (change of the permanganate ion to the manganate ion) is reduction, while the reverse is oxidation. It follows from these statements that the metal having the greater solution tension is the stronger reducing agent while the non-metallic element having the greater solution tension is the better oxidizing agent. It thus becomes possible to compare the activities of two oxidizing or reducing agents by measuring the magnitude and direction of the electromotive force produced by combining two systems made up of these agents in contact with solutions containing the respective products of reduction or oxidation. EXPERIMENTAL GRAVIMETRIC ANALYSIS 119 If the oxidizing or reducing agents are solids the electrodes are composed of these solids. If gases the electrode is of some material that will superficially dissolve the gases, while if the agents are solutions they are merely brought into contact with electrodes made of indifferent metals, such as platinum. Thus silver as a reducing agent would itself be made to form the electrode material, in contact with an ionized silver salt. Oxygen as an oxidizing agent would be caused to bubble into the solution in contact with an electrode of platinum which is coated with platinum black, in which oxygen dissolves, and this would be immersed in a solution containing hydroxyl ions. Potassium permanganate would be used in simple contact with a platinum electrode and the solution would also contain the positive ++ manganese ion, Mn, the product of reduction of the ion Mn0 4 . In the last case the force, solution tension, is replaced by the tendency of the permanganate anion, already in solution, to become reduced. Oxidation Potential. All of the elementary halogens are oxidizing agents because they exhibit a tendency toward negative ion formation: C1-C1, Br-Br, etc. Conversely the change of halogen ions into neutral elements is an oxidation of these ions. In order to bring about this oxidation it is necessary to apply another oxidizing agent whose " oxida- tion potential, " F (potential of the nonionized electrode minus that of the electrolyte) is greater. This is analogous to the decomposition of a halide by means of the current, which is an oxidizing agent for the negative ion and a reducing agent for the positive ion. Selective Oxidation. If an oxidizing agent can be found, having an oxidation potential greater than that of one of the halogens and less than that of another this agent may be used for separating the two halogens by oxidation of the salt (or acid) of one, removing the liberated element by distillation and leaving the other, which was incapable of being oxidized by this agent. This is analogous to the electrolytic separation of metals by grading the electromotive force which is applied to two electrodes 120 QUANTITATIVE ANALYSIS in the solution. The measurement of oxidation potentials should therefore furnish valuable information for assisting in the selection of oxidizing agents suitable for graded oxidation of the anions of the halogen hydracids. Bancroft has shown 1 that the following differences exist between the oxidation po- tentials of chlorine, bromine and iodine in solutions of salts of their respective hydracids: Chlorine in potassium chloride bromine in potassium bro- mide =0.241 volt. Bromine in potassium bromide iodine in potassium iodide = 0.535 volt. These differences vary somewhat if the concentrations are altered. As examples of oxidizing agents which will serve for the iodine anion without thereby oxidizing bromine or ohlorine anions, may be mentioned monopotassium arsenate and nitrous acid. These were suggested by Gooch. 2 The oxidation potential of nitrous acid (or potassium nitrite and sulphuric acid), according to the measurements of Bancroft, is 0.249 volt higher than that of iodine in potassium iodide, and 0.285 volt lower than that of bromine in potassium bromide. For the selective oxidation of the bromine anion in presence of the chlorine anion the following substances have been used: potassium permanganate in acid solution, lead peroxide in acid solution, potassium dichromate in acid solution, ammonium persulphate in neutral solution and potassium iodate in acid solution. With the exception of the last named oxidizing agent all of these substances possess oxidation potentials higher than that of chlorine in potassium chloride and could therefore be made to serve for a quantitative separation of bromine and chlorine only by carefully regulating the concentrations of oxi- dizing and reducing agents and by stopping the distillation at exactly the correct time. Attempts have been made to regulate the speed of oxidation by acidifying with substances that can furnish only a small concentration of hydrogen ions, since it is only in presence of hydrogen ions that the reactions can proceed. Weakly acid substances that have been used for this purpose 1 Z. physik. Chem., 10, 387 (1892). 2 Chem. News, 61, 235 (1890). EXPERIMENTAL GRAVIMETRIC- ANALYSIS 121 are acetic acid, potassium acid sulphate, ferrous sulphate and aluminium sulphate. The last two are weakly acid through hydrolysis. Potassium permanganate is the oxidizing agent used by Jan- nasch and Aschoff 1 and the oxidation of hydrochloric acid is prevented by acidifying with an acid no stronger than acetic acid and by employing a large dilution. The oxidation potential of potassium permanganate with sulphuric acid is 0.097 volt higher than that of chlorine with potassium chloride. The reduction of potassium permanganate is really a reduction of manganese itself, being a change of heptavalent into bivalent manganese. The complete equation is 5Br+4H 2 0. The ionic change involving manganese is Mn0 4 -4in-j-2O 2 . The electrical change of manganese itself is not oxidation, as would appear from the last equation, but reduction, because the univalent anion, MnO 4 , is composed of one atom of heptavalent positive manganese and four atoms of bivalent negative oxygen, BO that the change of manganese is really V+V + + Mn KMn. If the substance being analyzed is known to be a pure mixture of only two of the halides, one of the halogens may be liberated and removed by distillation without subsequent absorption, the other being determined in the residual solution. If it is not a pure mixture or if it contains salts of three halogens it is necessary to absorb at least one of these and make a direct determination of it in the absorbing solutions. Bugarszky 2 used potassium iodate and sulphuric acid for separating bromine and chlorine, distilling the bromine without absorption. The oxidation potential of acidified potassium iodate was found by Bancroft to be 0.064 volt higher than that 1 Z. anorg. Chem., 1, 144 and 245 (1892); 5, 8 (1894). 2 Ibid., 10, 387 (1895). 122 QUANTITATIVE ANALYSIS of bromine and 0.178 volt lower than that of chlorine. The reaction is as follows: KI03+5KBr+3H 2 S0 4 ->3K 2 SO 4 +5Br+I+3H 2 O. Both bromine and iodine are liberated and distilled and account of this must be taken if the free halogens are absorbed and subse- quently determined. Chlorine is determined in the residual solution, after reducing the excess of iodic acid to hydriodic acid by means of sulphurous acid, then oxidizing by nitrous acid and distilling. Andrews 1 modified the method of Bugarszky by substituting nitric acid for sulphuric acid and by reducing the excess of iodic acid by means of phosphorous acid. His method was not tested, however, except for the determination of chlorides in crude bromides and of chlorine in crude bromine In both cases the chlorine was present in relatively small quantities (less than 10 percent) and it was not adapted to the determination of both bromine and chlorine. For the determination of chlorine, bromine and iodine by direct means ; the method of Jannasch and Aschoff is probably the best of all methods yet proposed, even though permanganic acid is not an ideal oxidizing agent for the separation of chlorine and bromine. In this method the solution of mixed chlorides, bromides and iodides is first acidified with sulphuric acid and potassium nitrite is added: KI+KNO 2 +2H S0 4 -+2KHSO 4 +NO+H 2 O+L The liberated iodine is distilled, absorbed, and subsequently determined. The sulphuric acid is then neutralized by sodium hydroxide, acetic acid and potassium permanganate are added and the bromine is distilled, absorbed and determined. In the residual solution the excess of permanganate is reduced and the chlorine is determined gravimetrically. The absorbent which best serves for iodine and bromine is a solution containing sodium hydroxide and hydrogen peroxide. Bromine and iodine react with sodium hydroxide to form sodium bromide and sodium hypobromite in the one case and sodium iodide and sodium hypoiodite in the other: 2NaOH+2Br-NaBrO+NaBr+H 2 0, 2NaOH+2L-NaIO-fNaI-fH 2 0. 1 J. Am. Ghem. Soc., 29, 275 (1907). EXPERIMENTAL GRAVIMETRIC ANALYSIS 123 If the solutions are allowed to stand for some time bromates and iodates are formed: 3NaBrO-NaBrO 3 -f2NaBr, 3NaIO->NaIO,-f2NaI. The last change does not take place if hydrogen peroxide is present and the solution is kept cold, the hypobromite and hypoiodite being reduced as fast as formed: NaBrO+H 2 O 2 - NaIO+H 2 2 - >NaBr+H 2 O+O 2 , >NaI+H 2 0+0 2 . In the resulting solutions bromine and iodine may be determined as the silver salts in the usual manner after acidifying with sulphuric acid. " ^ V \ I I i FIG. 35. Apparatus for the separation of the halogens. If the absorbing solution has been allowed to become warm some iodate or bromate will be formed. In this case acidification will cause the liberation of free halogen which will escape precipi- tation. If silver nitrate is added before acidifying, any iodate or bromate will remain in solution as the silver salt. This inter- ference of oxysalts may also be prevented by the addition of a sulphite before the addition of acid. The resulting sulphurous acid then reduces the iodate or bromate to iodide or bromide. During the distillation of bromine and iodine it is essential that contact with cork or rubber be avoided, since the halogens are thereby reduced and absorbed. Ground-glass stoppers are necessary in all parts of the apparatus where such contact would 124 QUANTITATIVE ANALYSIS occur and, where rubber connections are used, the glass tubes inside must be pushed together so as to expose as little of the rubber tubing as is possible. All reagents must be tested and found free from the halogens. Determination. Weigh about 1 gm of the mixture of halides, placing the sample in a round bottomed, glass stoppered distilling flask, having a capacity of 1000 cc, and having an inlet tube sealed into the side of the neck and reaching to the bottom of the flask. Connect the appa- ratus as shown in Fig. 35. A is a vessel in which steam may be generated, B is the distilling flask, C and D are bubble tubes having a capacity of 150 cc. The tube a should reach to the bottom of the steam generator and should extend about 18 inches above. This tube provides an inlet for air, in case there is any tendency toward drawing liquid back from B. Each of the absorption tubes C and D contains 50 cc of 5 percent sodium hydroxide and 50 cc of hydrogen peroxide. The union between B and C and between C and D should be made by bringing the glass tubes quite together inside the rubber connections. Dissolve the weighed sample in about 600 cc of water, add 5 cc of 25 percent sulphuric acid and 2 gm of sodium nitrite Heat the solution nearly to boiling and pass steam through the flask for twenty minutes after the solution is colorless. During this time the tubes C and D must be kept cool by immersion in ice water. When all of the iodine has been distilled the boiling is interrupted, the absorption tubes are disconnected and their contents washed into a 300 cc beaker. The tubes are then returned to the apparatus and are refilled with sodium hydroxide and hydrogen peroxide as before. The solution in the flask is barely neutralized with sodium hydroxide solution and evaporated to a volume of about 500 cc. 1.5 gm of potassium permanganate and 60 cc of 33 percent acetic acid are added and the bromine thus liberated is distilled and absorbed in hydrogen peroxide and sodium hydroxide in the cooled tubes. When steam has been passed through the solution for some time after the latter has become colorless the distillation is again stopped and the contents of the tubes washed into another beaker. The solutions now containing the iodine and bromine are boiled until the excess of hydrogen peroxide is completely decomposed. 0.5 gm of sodium sulphite is added and then dilute sulphuric acid until the solu- tion is slightly acid in character. If any color appears at this point it is due to the presence of iodine or bromine produced by iodate or bromate, showing that insufficient sodium sulphite has been added. In this case 0.5 gm more is at once added to reduce the free halogen. EXPERIMENTAL GRAVIMETRIC ANALYSIS 125 When the solutions are acid and colorless a 5 percent solution of silver nitrate is added, drop by drop from a pipette, stirring vigorously until no further precipitation occurs. The liquid is digested at near the boil- ing temperature until the precipitate settles readily, after which it is filtered on a Gooch crucible, as directed on page 86, and the precipitates are washed free from silver nitrate, testing the washings with dilute hydrochloric acid. The crucibles are finally washed once with alcohol to promote rapid drying and are then dried at 110 for one-half hour or until the weight is constant. The percent of iodine and of bromine is calculated. The solution in the large distilling flask is boiled with alcohol to reduce the excess of potassium permanganate and is then poured into a beaker or evaporating dish and evaporated to a volume of not more than 150cc. 5 cc of dilute nitric acid is added and the chlorine is precipitated and weighed exactly as directed in the case of iodine and bromine. Halogen Oxyacids. The oxyacids of the halogens (or their salts) may be reduced to the hydracids by warming with hydrogen peroxide, after which the separation and determination may be accomplished as above directed. Free Halogens existing in solution may be converted into oxysalts by treatment with alkali bases, 'after which their separa- tion and determination may be carried out by methods already discussed. Their determination is more conveniently made by volumetric methods which will be discussed later. Chlorine in gaseous mixtures is also determined by absorption followed by a volumetric process. Organic Halogen Compounds. Compounds of the halogens with organic residues cannot be analyzed by the usual methods because such compounds do not, as a rule, ionize to form the anions of the halogen acids. The compound must be decomposed in such a manner as to leave the halogen in the form of an in- organic compound of one of the well-defined acids. In such cases either the lime method or the method of Carius may be used. In the lime method the material is mixed with pulverized lime, free from halogens, and is placed in a hard glass tube, closed at one end. Lime is placed in the open end of the tube, which is then heated in a combustion furnace. The organic compound is decomposed and the halogen unites with the calcium oxide to form calcium halide, from the acid solution of which the halogen may be precipitated by silver nitrate. If more than one halogen 126 QUANTITATIVE ANALYSIS is present the separation may be made, after the heating is finished, by methods already outlined. In the Carius 1 method the material is heated in a closed tube in contact with fuming nitric acid and silver nitrate. The organic compound is oxidized and the free halogen thus produced is converted into the hydracid. The halogen hydracid at once reacts with silver nitrate and the silver halide is later weighed. The method is not well adapted to separation of the halogens, since a mixture of silver salts is obtained in the tube. Determination. Carius tubes of hard glass may be obtained with one end already closed. The tube should be approximately 50 cm long and 2 cm in diameter. If such a tube is not at hand a good grade of combustion tubing may be used. One end is closed as follows: The tube is carefully heated at a point about 10 cm from one end by rotating in the flame of the blast lamp. When the glass has softened the tube is quickly drawn out, until half closed. It is allowed to cool and is then removed from the flame and cut at the narrow part. The nearly closed end is then fused together until a well-rounded end is produced. This must be annealed with great care or disastrous breaks will occur later. Having prepared a tube that is clean and dry, another small tube about 4 cm long is closed at one end to serve as a weighing tube. About 0.2 gm of the organic material is weighed into the latter. Into the Carius tube is carefully placed about 1.5 gm of powdered silver nitrate and 2 cc of fuming nitric acid, free from halogens. The acid is intro- duced through a funnel with a long stem which reaches at least half way to the bottom of the tube, thus keeping the upper half dry. The weigl ing tube containing the substance to be analyzed is inserted into the end of the Carius tube, the latter being placed in a slanting position. Mixing of the contents of the weighing tube with the acid should not occur until after the Carius tube is sealed. The latter is now heated about 10 cm from the open end, the tube is drawn out while in the flame and the walls are sealed together. A more or less blunt point should be left here as shown in Fig. 36. Since a high pressure will be generated within the tube when heating begins it is necessary to place the tube inside an iron tube having caps screwed over the ends. The glass tubes frequently break on account of high pressure. The iron tube is now placed in a suitable furnace in which it may be gradually and uniformly heated. The temperature and time necessary for heating will vary with the nature of the substance under examination. Most organic compounds will be completely *Z. anal. Chem., 1, 240 (1861); 4, 451 (1864); 10, 103 (1871). EXPERIMENTAL GRAVIMETRIC ANALYSIS 127 decomposed by heating for three hours at 300, while many aliphatic campounds will require a temperature no higher than 150. After the decomposition is completed, as shown by the disappearance of carbon the furnace is allowed to cool, the iron pipe containing the tube is carefully removed, the cap unscrewed and the glass tube taken out. The latter is wrapped in a towel, to minimize the danger due to possible explosions, and the point of the tube, where it was last sealed off, is held in a flame until softened. The internal pressure causes the glass to blow out and the gas escapes, after which the tube may be handled without risk of injury. A scratch is made near the blown-out end, but FIG. 36. Sealed end of Carius tube. on the wide part, and this end is broken off by touching the scratch with a hot glass rod. The contents of the tube are rinsed into a beaker, diluted with water and filtered; the precipitate is washed and weighed by the ordinary process, using either a paper filter or a Gooch crucible. From the weight of silver halide found the percent of halogen in the organic compound is calculated. If the fuming nitric acid contains halogens, blank determinations must be made and corrections applied. CARBONIC ACID AND CARBON DIOXIDE The following cases are to be considered: Carbon dioxide in gaseous mixtures, solutions of carbonic acid and salts of car- bonic acid. Carbon Dioxide in Gaseous Mixtures (air, chimney gases, etc.) This- determination is best made by gasometric methods which will be considered in a later section (pages 333 and 341). Carbonic Acid in Solution. The most frequently occurring case is that of underground waters. Such waters, coming from regions of low temperature and high pressure, often contain con- siderable quantities of carbonic acid. When the water reaches the surface, diminished pressure and rise in temperature cause the release of more or less carbon dioxide, so that a determina- tion is always subject to some uncertainty regarding the rela- tion of the original concentration of carbonic acid to that in the water as the analyst receives it. Determinations are also re- 128 QUANTITATIVE ANALYSIS quired of carbonic acid in carbonated drinks. In such cases provision must be made for transferring the solution from the pressure bottle to the apparatus in which the determination is to be made without loss of carbon dioxide. The procedure for the determination of carbonic acid in water is given on page 401. Carbon Dioxide in Carbonates. Determinations of this class are by far the most common in general analytical practic. The FIG. 37. Rohrbeck's appara- tus for determination of carbon dioxide by loss. FIG. 38. Moor's apparatus for determination of carbon dioxide by loss. carbonate is decomposed by means of a stronger acid than car- bonic acid and the carbon dioxide determined in one of three ways: (1) by a determination of loss in weight, (2) by measuring the gas disengaged, or (3) by weighing this gas after absorbing by reagents in a suitable apparatus. Determination by Loss. Many forms of apparatus may be obtained for the determination of carbonic acid by loss. Three EXPERIMENTAL GRAVIMETRIC ANALYSIS 129 of these are shown in Figs. 37, 38, and 39. Any such apparatus must include means for drying incoming air and outgoing gas. It must also be compact and not too heavy to be weighed on the analytical balance. In using such apparatus the sample is weighed and brushed into the lower generating vessel. Hydro- chloric or sulphuric acid is placed in the upper bulb and the bubble tubes are partly filled with concentrated sulphuric acid. The whole apparatus is accu- rately weighed, after which the cock is carefully opened so that acid drops upon the carbonate, evolving carbon dioxide at a moderate rate. This carbon dioxide passes out through the sulphuric acid in the bubble tube, being freed from moisture by so doing. The apparatus is finally heated and air is drawn through to displace the remaining carbon dioxide. The loss in weight is taken to represent carbon dioxide. This is not accurately the case unless the air that is drawn through the apparatus is first dried. The determination by means of such apparatus is quickly made but is sub- ject to a rather large error on account of the large weight of the apparatus, because of the large surface and largely because of the difficulty encountered in the drying and purification of the outgoing gases unless unduly large quantities of sulphuric acid are used, as well as an absorbent for acid vapors. Determination by Absorption. The direct determination by a somewhat more elaborate apparatus is to be preferred if accuracy is an object. In such a method the purification of the carbon dioxide is rendered complete by elaborating that part in which the purification is accomplished, providing better contact of the gases with drying agents and acid absorbents. Instead of weighing the entire apparatus before and after expulsion of FIG. 39. Schrotter's ap- paratus for determination of carbon dioxide by loss. 130 QUANTITATIVE ANALYSIS carbon dioxide from the carbonate the carbon dioxide is ab- sorbed in a weighed amount of potassium hydroxide, which is again weighed after the absorption. Many variations in the apparatus have been employed but the apparatus here described embodies the essential features of most of these. In Fig. 40, A is a generating flask into which the weighed sample of carbonate is placed. B is a dropping funnel having a capacity of 50 cc, and having the lower end drawn out to a point and turned upward. This part should extend to the bottom of the flask. At the top of the dropping funnel a drying tube C is connected by means of a rubber stopper and a bent glass tube. FIG'. 40. Assembled apparatus for determination of carbon dioxide by absorption. The drying tube is filled with soda lime for the absorption of carbon dioxide from the air that is later to be drawn through. Following the generating flask is a short condenser D and then U-tubes E, F and G. The first U-tube is omitted if sulphuric acid is to be used for decomposing the carbonate, or is filled with an ab- sorbent for hydrochloric acid vapors if this acid is used. The U-tubes F and G are filled with granular calcium chloride which absorbs moisture from the gas mixture. Following these is the apparatus H in which potassium hydroxide is placed for the ab- sorption of carbon dioxide. This apparatus also carries a small EXPERIMENTAL GRAVIMETRIC ANALYSIS 131 tube filled with calcium chloride to prevent the removal of mois- ture from the apparatus, which would occur if the dry entering gases were allowed to leave the apparatus saturated with mois- ture. To provide a means for drawing air through the whole apparatus the aspirator J is placed at the end of the series, while to prevent moisture from diffusing backward into the absorption apparatus the calcium chloride tube / is interposed. Choice of Acid. The choice of acid to be used in decomposing the carbonate will depend upon the nature of the latter. Sul- phuric acid is to be preferred where it can be used, because it is non-volatile and thus needs no absorbent in the purifying apparatus. If, however, the carbonate is one of a metal which forms a sulphate of small solubility (e.g., calcium carbonate or barium carbonate) sulphuric acid soon coats the particles with insoluble sulphate which hinders the decomposition of the interior of the particles. Decomposition is slow and uncertain and for this reason hydrochloric acid is used instead of sulphuric acid. A preliminary test should be made to ascertain whether sulphuric acid forms a complete solution of the carbonate to be analyzed. Absorbent for Hydrochloric Acid. If hydrochloric acid must be used a suitable absorbent is placed in the U-tube E, following the generating flask. Absorbents which serve best for this purpose are silver sulphate and anhydrous copper sulphate. For such a purpose the copper sulphate is prepared by first dropping red hot pieces of pumice stone into a concentrated solution of copper sulphate, removing the pumice stone, allowing to drain and then drying at 200. A supply of the dry pieces is kept in a desiccator and fresh pieces placed in the U-tubes for each deter- mination. A more satisfactory absorbent is a saturated solution of silver sulphate in concentrated sulphuric acid. This may be absorbed by pieces of pumice and used in the same manner as copper sulphate. When sulphuric acid is used in this manner it must not be allowed to come into contact with corks, cotton or any other organic matter. The evidence of such contact is blackening. The result is the formation of both carbon dioxide and sulphur dioxide. Both oxides are absorbed in the potas- sium hydroxide and give rise to errors in the determination. A U-tube with glass stoppers should be used. Silver nitrate 132 QUANTITATIVE ANALYSIS cannot be used because its reaction with hydrochloric acid pro- duces nitric acid, which is nearly as volatile as hydrochloric acid, and also chlorine. Soda Lime. The soda lime which is used for the removal of carbon dioxide from the entering air should be fresh and in the form of lumps. A powdered condition is evidence of having been air-slaked, in which case it is unfit for use since it is already saturated with carbon dioxide. Calcium Chloride. The calcium chloride used for the absorp- tion of moisture should be the granular form which has been fused. Fusion is necessary in order to produce an anhydrous material. This fusion always produces a certain amount of calcium oxide which, if allowed to remain as such, will absorb a certain quantity of carbon dioxide as well as of water. It is best to treat the material directly in the bottle by passing dried carbon dioxide through for several hours, then displacing the carbon dioxide by drawing through dried air. In filling U-tubes only lumps should be used. The tube is filled to just below the side branches and then a loose plug of cotton or glass wool is placed on top in each side to prevent drawing out of any grains of powder that may subsequently be produced. For sealing the tubes a cork is pressed in until it begins to fit closely. It is then cut off even with the top and the smaller part is pressed into the tube about one-eighth inch farther. The shallow cup thus formed is poured full of melted paraffin or sealing wax. When this solidifies an air-tight seal should result, unless bubbles have formed in the sealing material. In the latter case a flame may be lightly touched to the surface of the solid paraffin or wax, which will cause the bubbles to break. Phosphorus Pentoxide. Passing gases are dried to a greater extent by phosphorus pentoxide than by calcium chloride. The former absorbs moisture so rapidly as to make the charging of tubes difficult unless the humidity of the atmosphere is low. The oxide is usually obtained as a fine powder, formed by sub- limation. As this combines with water it forms a sticky mass of phosphoric acid, which soon clogs the tube unless some device is employed to prevent this. The best method for charging drying tubes with phosphorus pentoxide is to arrange a ribbon of glass wool, over which the oxide is sifted. The glass wool is EXPERIMENTAL GRAVIMETRIC ANALYSIS 133 then quickly folded into a narrow strip which is placed in the tube. Choice of Drying Agent. It might seem at first sight that practically perfect drying of carbon dioxide before absorption would be necessary for accurate determinations and that phos- phorus pentoxide would therefore be the ideal drying agent. An inspection of the conditions will show that this is not the case. Of course the increase in the weight of the absorption bulbs must accurately represent the weight of carbon dioxide absorbed. But in order that this may be the case it is only necessary that the unabsorbed air shall pass out of the bulbs with the same degree of hydration that it possesses when entering. Therefore any fairly good dehydrating agent will serve provided that the same agent that is used preceding the bulb in the train is also used in the tube that is attached to the bulbs for drying outgoing gases. Either calcium chloride or phosphorus pentoxide, but not both, may be used in the same train. Similar reasoning will apply to the use of concentrated sulphuric acid with other drying agents. Absorbent for Carbon Dioxide. Potassium hydroxide is generally used for the absorption of carbon dioxide, a solution 33 percent by weight being commonly employed. In practice it is found that absorption becomes so slow as to be uncertain before the point of complete saturation is reached. Tke prac- tical limit is reached when 0.10 gm of carbon dioxide has been absorbed by each cubic centimeter of potassium hydroxide solution. To determine the amount of gas that can be absorbed by the solution in the apparatus the latter is first filled with water to the height at which the liquid is to stand. This is emptied out and measured. The number of cubic centimeters times 0.1 gm is the weight of carbon dioxide which can be absorbed before the solution becomes inefficient. By adding together the weights of gas absorbed in successive experiments it is easy to determine when the bulbs need refilling. The bulbs in which the absorption is to take place furnish the greatest source of error to be encountered in this method. Inaccuracies are due to the large weight, the large surface and the possibility of moisture being carried out by the outgoing air. The effect of the comparatively large weight of the bulbs and their contents 134 QUANTITATIVE ANALYSIS is to decrease the sensibility of the balance. The surface gives rise to a possible error because of the variable amount of moisture which is always dissolved in the surface of the glass. This error may be considerable if the two weights (before and after the absorption) are observed under different atmospheric conditions of humidity. For this reason it is necessary that the two readings of weights shall be made on the same day and as near to each other in point of time as possible. The danger of loss of moisture from the potassium hydroxide solution to the dry air which enters is magnified by the necessary limit which the already large weight of the bulbs places upon the tube which carries the calcium chloride for drying the outgoing air. For this reason many analysts prefer to separate this drying tube from the bulbs, using a small U-tube or even two such tubes, and weighing the apparatus in the two or three parts. Sometimes there is placed in the first half of the U-tube so used, or in the first U-tube if two are used, solid potassium hydroxide to insure complete absorption of carbon dioxide. While this procedure may make more certain the complete drying of the air and thus prevent a loss of weight from this cause, an added uncertainty is introduced due to the accumulation of the errors of four weighings. It is possible to insure complete detention of the moisture by passing the gas at a regular, specified rate, not exceeding a maximum found by experience. Determination. Procure the following parts for assembling: 1 dropping funnel, 50 cc, with 1-hole rubber stopper, 1 short, wide flask, 75 cc, such as is used for fat extractions, with 2-hole rubber stopper, 1 condenser with body not more than 6 inches long, 3 U-tubes with corks to fit, 1 U-tube with glass stoppers, 1 straight drying tube with 1-hole rubber stopper, 1 set "potash bulbs" of some approved form, 1 aspirator bottle, tubulated near bottom, with 1-hole rubber stoppers to fit, 1 piece glass tubing, about 2 feet X | inch, for supporting apparatus, 2 clamps, 2 pinch cocks, 1 small screw clamp (Hoffman screw), 2 retort stands, EXPERIMENTAL GRAVIMETRIC ANALYSIS 135 Glass and rubber tubing for connections. Fill and connect the apparatus in the manner previously described. Measure the capacity of the absorption bulbs by drawing in distilled water, then blowing out and measuring the water. This volume will be used as a basis for the calculation of absorbing power as already directed. When filling the absorption bulbs with potassium hydroxide solution the latter should not be warmer than the air of the room. The bulbs are detached from the apparatus and the solution is drawn in through a tube attached at a, suction being applied at 6. The solution should about half fill the bulb c when air is bubbling through. The ground- glass joint between the drying tube 6 and the bulbs should be lightly coated with vaseline and the tube then twisted on until it fits closely enough that there will be no danger of loosening during the course of an experiment. Any surplus vaseline is removed from the outside of the joint. Place the bulbs in position, close the cock of the dropping funnel and open the pinch cock at e to allow water to flow from the aspirator. Bubbles of air will at first pass through the bulbs but this action will finally cease unless there is a leak in the apparatus, in which case it must be found and closed It is important that all glass tubes be brought entirely together inside the rubber connections since rubber is slightly permeable to gases. After the apparatus has been shown to be free from leaks the pinch cock at/ is closed, the cock of the separatory funnel slowly opened and, after equilibrium is established, the clamp k is so adjusted that when clamp / is opened air will pass through the bulbs at a rate not greater than 3 bubbles per second. Clamp k is not thereafter changed. This provides against too rapid flow of gas under any conditions. Clamp / is now closed, the bulbs are removed, the inlet and outlet tubes are closed by short rubber tubes containing glass plugs and the bulbs are wiped clean and placed in the balance case. A short glass tube is inserted to bridge the gap made by removing the bulbs. The bulbs should be allowed to stand for 15 minutes before weighing. In the meantime about 1 gm of the carbonate is weighed and brushed into the generating flask and a small amount of water is added to moisten the sample. The stopcock of the funnel B and the clamp e are now opened and 500 cc of air is drawn through the apparatus, measured by the outflowing water from the aspirator. This frees the apparatus from carbon dioxide. After the absorption bulbs have stood for 15 minutes the tubes carrying the plugs are removed and the bulbs are weighed. The plugs are then replaced and left so until the bulbs can be connected in the apparatus. 50 cc of dilute sulphuric acid or hydrochloric acid is placed in the dropping funnel, a test having previously been made to 136 QUANTITATIVE ANALYSIS determine whether sulphuric acid will form a clear solution with the carbonate. If such a solution is not produced, of course hydrochloric acid must be used and silver sulphate and pumice must be placed in tube E. Reconnect the apparatus and open all cocks except the stop-cock in the dropping funnel, leaving the clamp k set for the proper rate of gas flow, as previously determined. Slowly open the cock of the dropping funnel, allowing acid to drop just fast enough to evolve carbon dioxide at the prescribed rate. The constant attention of the operator is necessary at this point, for by causing too rapid evolution of gas some moisture may escape absorption in the small tube of the absorption bulbs and the experiment be rendered worthless. The acid should be allowed to run in until about 1 cc is left above the stopcock, this acting as a seal during the subsequent boiling. After the decomposition of the carbonate is complete the solution in the flask is slowly heated until it boils, always with due regard to the rate at which the gas is made to flow through the absorption bulbs. The boiling is continued for one minute, when the flame is withdrawn, the cock of the dropping funnel being opened at the same time to allow air to enter so that no back suction occurs, due to the cooling effect. Air is now drawn through the apparatus until 1000 cc of water has flowed from the aspira- tor. This amount of air should be sufficient to sweep all of the carbon dioxide into the absorption bulbs. The clamp / is now closed, and the absorption bulbs are removed, plugged and placed in the balance case. After 15 minutes they are weighed without the plugs, the increase in weight being the weight of carbon dioxide. From this and the weight of sample the percent of carbonic anhydride (combined carbon dioxide) is calculated. For the duplicate or any subsequent determination the generating flask and the dropping funnel are washed absolutely free from acid, so that no decomposition of the next carbonate sample may occur before the bulbs are in place. The first U-tube should also be emptied and recharged with absorbent, if such is to be used for the next determination. If a large number of determinations is to be made with the same apparatus much time will be saved by providing two decomposition flasks and two absorption bulbs. While one determination is being made another sample may be weighed into the duplicate flask and the second absorption bulb may be weighed. The next determination may then be started while the first bulbs are standing in the balance case, pre- liminary to the final weighing. It is also necessary to determine when the various absorbents have become so saturated as to be inefficient for further work. Soda lime in the tube C is good until the lumps have fallen into a powder. Silver sulphate in the pumice of tube E may become inefficient through absorption of hydrochloric acid or through the EXPERIMENTAL GRAVIMETRIC ANALYSIS 137 accumulation of water in the tube. The solubility of silver sulphate in water is much less than in concentrated sulphuric acid. If the acid solution becomes diluted the silver salt crystallizes and will not there- after readily absorb hydrochloric acid. As the silver sulphate becomes saturated with hydrochloric acid it darkens, on account of the action of light. When the darkening effect has proceeded as far as the middle of the tube the material should be replaced. Calcium chloride must be replaced when it becomes visibly moist for the first third of any absorbing tube. A method has already been given for the determination of the amount of carbon dioxide which can be absorbed by the solution in the bulbs. CHAPTER IV ELECTRO-ANALYSIS We have here to deal with a class of work that, while also gravimetric in most cases, is sufficiently different from what has already been considered to be treated as a separate division. In all of the preceding exercises the element or radical to be deter- mined was precipitated from a solution by chemical reactions produced by other substances which were added for the purpose. In the cases now to be considered the precipitation will be brought about by electrical action, the passage of a current through the solution causing the deposition of a metal upon a cathode in such a form that it can be weighed, or the accomplishment of some change which makes possible the determination of a sub- stance not a metal. The electrolysis of silver sulphate will serve as an example. When a solution of this salt is electrolyzed at platinum electrodes the metal is plated on the cathode and sulphuric acid is produced at the anode, thus: 2Ag+S0 4 -*2Ag+S0 4 , 2SO4+2H 2 0->2H 2 SO4+O 2 . The silver can then be weighed and the sulphuric acid deter- mined volumetrically. While the electrolysis of a simple salt is frequently a tolerably simple and well understood process, the practical accomplish- ment of such a process for the purpose of a quantitative analysis is usually possible only when a certain set of conditions is main- tained. The principal reasons for failure to attain accuracy are three: (1) Deposition may not occur upon passage of a current. (2) The deposit may be contaminated by other products of electrolysis. (3) The deposit may not have the proper physical character, so that it will not adhere to the electrode but crumbles off during the electrolysis or during the process of washing. We have thus to consider the nature of salt to be used, solvents, tem- 138 ELECTRO-ANAL YSIS 139 perature, electrolytic pressure (voltage), current density and nature and kind of electrode. Nature of Electrolyte. Electrolytic methods are more fre- quently applied to the determination of metals than of non- metals, although methods have lately been perfected for the determination of the latter. If the metal alone is to be deter- mined it will usually be possible to obtain it in the form of what- ever salt gives the best results. Certain anions must be excluded in specific cases, either because they yield substances that at- tack the anode or because the acids that are produced by their electrical discharge cause the metal to deposit in an undesirable physical form. As an example of corrosive action upon the anode it is sufficient to mention here the formation of nascent chlorine at a platinum anode when a chloride is electrolyzed. With regard to the effect of the acid that accumulates in the solution as electrolysis proceeds it may be stated that there is little known, at present, of the reasons for the effect of acids, bases and other substances that may be in the solution, upon the nature of the deposit. Experiment shows, however, that such substances often exert a very important influence upon the physical character of a deposited metal and they are often added for this reason, although they may be objectionable for other reasons. A solu- tion of copper sulphate, if electrolyzed without the addition of another substance, usually gives a dark red or brown deposit of finely divided copper which is liable to powder and be lost during washing. If a small amount of sulphuric acid is first added the deposit is improved, while nitric acid causes a still better deposit of bright red, firm and adherent metal. For this reason nitric acid is usually added although it gives rise to more or less danger of resolution when the cathode copper is being washed. On the other hand, a silver salt is best electrolyzed in the absence of nitric acid. If silver nitrate is electrolyzed from water solution with or without the addition of nitric acid (the latter is formed by the electrolysis) the plate of silver on the cathode is so decidedly crystalline that it is very easily detached. The addition of potassium cyanide in quantity sufficient to redissolve the pre- cipitate of silver cyanide first formed gives a solution from which silver will deposit as a white firm plate. The solution in potas- sium cyanide has a comparatively high electrical resistance 140 QUANTITATIVE ANALYSIS so that more energy is consumed in the accomplishment of its decomposition, nevertheless potassium cyanide is generally added. Other examples of similar effects will appear in the exercises. It is desirable to note that little is known of the cause of such effects, also to guard against a very common misconception regarding the purpose of adding other electrolytes to solutions that are to be electrolyzed. It is frequently stated that such substances are added in order to increase the conductivity of the solution. If such substances could increase the ionization of the salt that is to be electrolyzed, or in any manner diminish the fric- tional resistance to the passage of the ions, such an effect would be desirable. It is evident, however, that the addition of a foreign electrolyte can usually increase the conductivity only by itself acting as a carrier of current, in which case it has accomplished no desirable effect since the prime object is not to use a large current but to make the minimum current do the maximum work in discharging an ion already in solution. Solvent. Very little work has been done in any solvents other than water. The use of organic solvents may, in some cases, prove advantageous in producing good deposits where other con- ditions fail to do so. Temperature. Conductivity of solutions usually increases with rise in temperature. This is not due to increased ionization (ionization usually decreases with rise in temperature) but to re- duced viscosity and consequent reduction in frictional resistance to ionic migration. If the complete electro-decomposition of a substance requires considerable time it is not convenient to heat to any definite elevated temperature. In most cases, therefore, the temperature of the solution is not raised above that of the laboratory except at the beginning. Decomposition Voltage. For every electrolyte in solution there is a definite minimum voltage, below which no decomposi- tion will take place. If but one electrolyte is present and the voltage lies below this minimum, a continuous current cannot flow. The minimum voltage necessary to produce a continuous flow of current is called the "decomposition voltage" for the sub- stance in question. If salts of more than one metal are present in solution, the deposit on the cathode will consist of any metals, ELECTRO-ANALYSIS 141 the decomposition voltage of whose salts has been exceeded. If there is sufficient difference in the values of decomposition voltage for the different salts, separation may be made. It is only necessary to adjust the voltage so that it shall exceed the decomposition voltage of the metal that is most easily discharged. After this metal has been completely removed from solution and weighed the voltage is raised until it exceeds the decomposition voltage of the metal next in order and so on. While this consti- tutes the general procedure for electrolytic separations it is necessary to make certain changes in the nature of the solution after each metal is removed in turn as will be understood from the discussion of the solutions to be employed. In order to understand the origin of the decomposition voltage it will be necessary to briefly consider the underlying principles of electrolysis. If a metal is placed in contact with a solution of one of its salts it will be found that a difference in potential exists between the metal and solution. This difference may be either positive or negative, i.e., the metal may be at a higher or lower potential than that of the solution. The difference may be zero in certain cases but such cases are special. The conception of Helmholtz 1 regarding the cause of this potential difference may be thus stated: Whenever two dissimilar substances are in contact a potential difference is established because of the passage of one into the other. In the case of metals and their salt solutions, one of two things may happen : either some metal atoms pass into solution and become charged ions or some ions are discharged by the mass of metal and themselves become elementary In the first case the solution assumes a higher potential than the metal because positive charges have been transferred from metal to solution. In the second case the solution is at a potential lower than that of the metal because positive electricity has passe.d from solution to metal. The direction of the change is determined by the relative magnitude of two opposing forces. The metal shows a tendency to pass into solution in the ionic condition in obedience to a force called by Nernst 2 " electrolytic solution tension." This force varies with different elements, but is constant for a given element and 1 Wied. Ann., 7, 337 (1879). Z. physik, Chem., 4, 129 (1889), 142 QUANTITATIVE ANALYSIS may be relatively large or small. When positive ions have been thrown into the solution the potential difference thus established gives rise to an attraction of an electrostatic nature between the positively charged solution and the negatively charged metal. This may be represented diagrammatically as in Fig. 41. Double Layer. Helmholtz considered that a " double layer " was thus formed, composed of positive and negative charges, and that the components of this layer were very close together. The attraction existing between the components of the double layer increases as more ions are formed and finally reaches equilib- rium with the solution pressure. If ions of the metal in question Fio. 41. Diagram illustrating the "double layer." were already in the solution, then osmotic pressure would oppose the entrance of more ions into the solution and would thus act in conjunction with the electrostatic attraction, so that equilibrium would be reached with a smaller potential difference. Evidently, then, the potential difference between the electrode and elec- trolyte, in the case of a given metal and its salt solution, will be numerically greatest when the initial ion concentration is least, and least when the latter is greatest. If the solution pressure is small and the ion concentration large, equilibrium may be reached only by the actual deposition of ions upon the metal. In this case the potential difference will be positive. If the different elements are compared, with regard to the potential difference established between them and their ion solu- ELECTRO-ANALYSIS 143 tions, the ion concentration being the same in all cases, a series of different values is obtained. It is to be noted that if an element takes a negative charge upon becoming ionized the potential difference is reversed in sign. In the following table 1 several of the elements are given with values for the potential difference. These differences are measured by a method that need not be here discussed, involving the use of an arbitrary standard, so that only the relative values are important. These values are for solutions which contain 1 gm-ion per liter. Element Potential difference electrode-electrolyte Element Potential difference electrode-electrolyte Potassium 3 20 Hydrogen 000 Sodium 2 82 Arsenic. . . <+0 293 Barium -2 82 Copper . + 329 Strontium -2.77 Bismuth . . . < + 391 Calcium Magnesium -2.56 2 54 Antimony <+0.466 + 750 Aluminium -1.276 Silver + 771 Manganese Zinc -1.075 770 Palladium Platinum <+0.789 < + 863 CadnrmiTn 420 Gold < + l 079 Iron 340 + 1 96 Thallium -0.322 Chlorine + 1 417 Cobalt 232 + 993 Nickel... . 228 + 520 Tin < 192 + 1 119 Lead -0.148 In the electrolysis of a salt solution both positive and negative ions are discharged at the electrodes, both electrodes become coated with the products of 'decomposition (polarized), and both then become essentially electrodes of the respective elements, no matter what the original substance might have been. Hence such a system as that just discussed may be considered to exist at each electrode. Since, after electrolytic decomposition has begun, electrolysis consists of electrical discharge of ions it is evident that it must act in opposition to solution pressure and in conjunction with osmotic pressure. That is, in order to produce continuous electrolysis a voltage must be applied to the electrodes that is at least as great as the algebraic difference between the single potential differences normally established at the cathode and anode. This difference constitutes the theoretical "decom- 1 Wilsmore: Z. physik. Chem., 35, 291 (1900). 144 QUANTITATIVE ANALYSIS position voltage" of a given salt solution, and no appreciable electrolysis can take place as a result of the application of a lower voltage. The attempt to calculate the decomposition voltage of a solution from the single potential differences, ex- perimentally determined, does not always give results that agree with those found by experiment. This is because the ion con- centration is not necessarily equal to 1 gm-ion per liter and it changes as electrolysis proceeds. Also where gases are evolved at the anode (a condition generally noticed) the phenomenon of " overvoltage " exerts a very important influence upon the practical decomposition voltage. Importance of this Principle. Agreement is not at all satis- factory in the case of salts of oxyacids. This is partly because oxysalts are not normal with respect to the oxygen or hydroxyl ion, also because oxygen shows overvoltage to a marked degree, giving a much larger decomposition voltage than would be calculated. The matter that is here important is not the possibility of calculating decomposition voltages from known single potential differences but the recognition of the reasons for the fact that a minimum decomposition voltage must exist for every compound, under definite conditions, and that if this value can be determined a method for electrolytic separations is available. We shall under- stand that " decomposition voltage" refers to the fall in potential, measured across the electrodes, below which electrolysis cannot take place continuously. So far as we now know there is no upper limit to the voltage that should be used, excepting that set by the current density that should be employed, unless separations are to be made. Because the variable factors which influence the practical de- composition voltage (such as overvoltage of the anion, varying temperature and added electrolytes) and the consequent diffi- culty that is experienced in its calculation, Sand 1 proposed to measure merely the cathode potential difference and to make metal separations by properly grading this potential. This is, no doubt, the ideal method. For making such measurements, however, more elaborate apparatus is required and much care must be exercised. For most purposes such refinement is entirely unnecessary because the practical decomposition voltage is * J. Chem. Soc., 91, 374 (1906). ELECTRO-ANALYSIS 145 usually fairly accurately known as the result of experiments with a given salt, electrolyzed under specified conditions. Current Density. The relation between the amount of current flowing through a solution of an electrolyte and the amount of substance decomposed is stated in the law of Faraday: 1 (a) For any given electrolyte the amount of decomposition is directly proportional to the amount of current; (6) the amounts of different substances decomposed by the same current are proportional to the combining weights of the substances. According to the first part of this law the rate of decomposition and electro-deposition in any experiment will depend upon the current strength, except- ing the part played by other electrolytes that may be present. Being thus able to limit and control the rate of deposition the question arises as to whether there is any suitable current strength, above or below which good results will not be attained. If it were not possible for any current to pass through a solution ex- cept that carried by the ions of the salt which we desire to de- compose there would probably be no definite limit to the prac- tical current strength to be employed. The ions encounter a large frictional resistance in their migration toward the elec- trodes. In order to overcome this resistance the pressure (volt- age) is raised, often considerably above the decomposition voltage of the metal salt being analyzed, in order to hasten the action. If no other cation is present there is no upper limit to the prac- tical voltage. There is, however, another positive ion that is present in all aqueous solutions and particularly when acids are present. The ion referred to is that of hydrogen. If the pres- sure is raised above its discharge potential it can discharge at the cathode and will do so unless the current strength, concen- tration of hydrogen ion and concentration of metal ion are so related that the current can easily be carried from solution to cathode by the metal ion without the necessity for discharge of the hydrogen ion. This relation will evidently be such that there is a relatively small current, large metal ion concentration and small hydrogen ion concentration. The objection to the deposition of hydrogen on the cathode is based upon the fact that minute bubbles of gas prevent the proper coherence of the deposited metal. Apparently, then, the upper limit of current 1 Pogg Ann., 33, 301 and 481 (1834). 10 146 QUANTITATIVE ANALYSIS will be fixed by the point at which noticeable evolution of hy- drogen (or other gas) occurs. This limit cannot well be calcu- lated but is determined by experiment. It should be noted that the value to be measured is not that of total current flowing across the solution but is that of the current flowing into unit area of the electrode where the desired deposition is taking place, usually the cathode. This gives rise to the term " cur- rent density/' abbreviated to CD. In stating the conditions to be observed in electro-depositions the current density may be more specifically defined as the current in amperes flowing into each 100 sq cm of cathode surface. This is denoted by CZ> 10 o. Evidently total amperes ^ ~ square decimeters cathode surface The proper current density to be employed in a given case is called the " normal density" for that experiment. This is indicated by NDiQQ. The normal density is fixed by the conditions already discussed. There is considerable variation, however, based also upon the form of the electrodes. This will be taken up in the next paragraph. Nature of Electrodes. Electrodes must possess certain proper- ties in order to be suitable for use in quantitative analysis. The electrode material must be insoluble in the solution of electro- lyte, with or without current action. In the process of metal plating as used in the arts it is customary to make the anode of the metal being plated, so that the rate of deposition at the cath- ode is equal to the rate of solution at the anode, the mean con- centration of the metal in the solution remaining constant. This is obviously out of the question in quantitative analysis, where the total metal in solution, and no more, is to be deposited. The material most used for electrodes is platinum. The con- tinued advance in the cost of platinum has led to a search for less expensive materials. However, the combination of high electrical conductivity and low solubility is a rare one. Other metals could be used for cathodes because the current action prevents their resolution, but when the deposited metal is to be removed after the process is finished the solvent used will gener- ally dissolve some of the electrode also. So long as the cost of platinum is sufficiently low to make it possible to provide an ELECTRO-ANALYSIS 147 adequate supply of platinum electrodes it is doubtful whether any other material will supplant it to any great extent. Gooch and Burdick 1 have perfected a method for making electrodes, by which a very small amount of platinum is spread over a relatively large surface of glass. A mixture of glycerine and chlorplatinic acid is spread over the glass surface, which is then heated. The glycerine is evaporated and the chlor- platinic acid is decomposed, elementary platinum fusing into the glass surface. Mercury is used as a cathode for a certain class of work and this will be discussed in a later paragraph. The electrodes must also be chemically unaltered by the passage of a current or else altered in a definite manner. The first condition is more often realized but there are cases where one electrode is altered in a definite manner, as a silver anode, used for the determination of chlorine, bromine or iodine anion becomes coated with chloride, bromide or iodide of silver. The electrode that is to be weighed and is to receive the deposit (usually the cathode) should present the maximum surface for the minimum weight of electrode material Since a practical limit is placed upon the current density the duration of the process of deposition will be inversely proportional to the total electrode surface exposed. The weight of the electrode must not be too large for accurate work and these considerations naturally lead us to consider the form of material where the ratio of surface to weight will be as large as possible. In general, any piece of platinum, small enough to be weighed, may be used as a cathode for metal deposition. Any chemist who has a dish or crucible may make at least an occasional analysis by the use of such an article as cathode. The dish is simpler because the solution may be placed directly in it and a coil of platinum wire used as an anode. The ratio of surface to weight is not large in this case, especially as only one surface is effective. Moreover if there is any sediment in the solution this may be partly caught by the depositing metal and weighed along with the latter. The cathode dish designed by Classen is quite thin and presents a larger surface. A dish of this kind weighing about 40 gm has a capacity of 250 cc and presents an inner surface of about 150 1 Z. anorg. Chem., 78, 213 (1912). 148 QUANTITATIVE ANALYSIS sq cm to the solution. A crucible may be used as a cathode by connecting as in Fig. 42. A rubber stopper is used to help support the crucible, a metal rod passing through and connecting with the negative of the current source. A small platinum wire serves to complete the connection of crucible with rod. This form of cathode also possesses a small relative surface. Other forms of cathodes are open cylinders and cones of foil, and gauze cylinders and plates, made from gauze of small mesh and fine wire. These forms are shown in the illustrations (Figs. 43 and 44). Of all of these forms the gauze electrode is most efficient, not only because the relative surface is greatly increased by constructing of fine wire but also because practically all parts of the surface are equally effective. The latter condition does not obtain for foil electrodes of any form, the surface farthest from the anode being in a relatively weak electrolytic field. Gauze elec- trodes also permit better mixing of the solu- tion and very much higher current densities may be used. Special forms of electrodes for rapid rotation will be discussed later. Other Apparatus. The necessary apparatus for electro- analysis will include, besides the electrodes, a generator of direct current, variable resistance, voltmeter and ammeter. The source of current may be a dynamo, any of the forms of primary cells, secondary or storage cells or thermoelements. The thermoele- ment is not a practical source of current, being both inefficient and unreliable. The direct current from a dynamo may be used and is better than primary cells, the latter being trouble- some in the matter of maintenance. The chief objection to the dynamo current lies in the fluctuations usually resulting from a variable load on the line from the generator. The best and most satisfactory current producer for this class of work is the secondary or storage element. Any of the various forms of accumulators will prove satisfactory, the lead-lead peroxide cell being the best known. The great merit of the secondary cell FIG. 42. Crucible cathode. ELECTRO-ANALYSIS 149 is its constancy and reliability. The E. M. F. of the lead cell is about 2 volts and the necessary voltage for the work may be obtained by connecting several cells in series. Any rheostat will do for this work, provided that the range in resistance is properly related to the other factors entering into the determination of current strength. In the absence FIG. 43. Platinum foil cathodes. FIG. 44. Platinum gauze cathode. of such a rheostat carbon lamps may be used for the current control, if the line voltage is high. The resistance of a 16 c.p. carbon lamp is about 220 ohms and by arranging several in parallel a fairly satisfactory regulation of current may be provided. Voltmeters and ammeters should have the scales graduated with a range as limited as is consistent with the current conditions 150 QUANTITATIVE ANALYSIS to be employed, so that each subdivision may represent a small fraction of a unit. A satisfactory plan is to have double scale instruments, the range of one scale being ten times that of the other. The necessary connections for the apparatus used for electro- analysis are shown in figure 45. B represents the source of current. In series with this are connected A, the ammeter for measuring the current strength, R, a rheostat for varying the resistance in the circuit and thus providing current control, c and a, cathode and anode. The voltmeter V, for measuring the pressure, is given a shunt connection across the electrodes. FIQ. 45. Diagram of connections for electro-deposition of metals. The actual current flowing through V is very small because of the high resistance possessed by the winding of the voltmeter. On this account the indications of the ammeter are practically correct for the current flowing through the electrolyte between the electrodes, although not absolutely so. The following is a description of the apparatus now in use in the Purdue laboratory. Purdue University Laboratory for Electro-Analysis. Current is furnished by storage cells of the lead type, each having a capacity of 48 amp-hours and a maximum charge and discharge rate of 6 amp. They are placed in a closed battery case which is outside the room for electro-analysis. The cells are provided with sand trays and insulators, and also with glass covers which ELECTRO-ANALYSIS 151 almost entirely prevent the annoyance due to acid spray during charging, and the case ventilator makes charging absolutely inoffensive. The interior of the case is protected against the attack of acid by asphaltum paint. Wires from the cells run "into the special laboratory where they are connected with the distributing switchboard. This is FIG. 46. Distributing switchboard of the Purdue University Laboratory for electro-analysis. a 28 X 72-inch board of black oiled slate, providing switches and plug receptacles for the control of all current which is used for any purpose within this room. The cells are con- nected in series groups of three each, the outside terminals of such group having double receptacles. This arrangement en- 152 QUANTITATIVE ANALYSIS ables the operator to connect his cells with any number of groups in multiple, thus giving greater latitude in the selection of voltage and current strength than is possible with the usual series connections. The 110-volt charging current enters the board through a switch which can be connected by plug connectors with any cell or combination of cells, and a slide-wire rheostat on the back of the board makes it possible to charge any number of cells at a time. The cells are protected during charging by an under-load circuit-breaker, and also during both charge and discharge by a fuse panel which is placed on the back of the board. An ammeter with a range of 20 amperes and a double scale voltmeter, with ranges of 150 volts and 15 volts, are pro- vided for proper control of the charging process. ,The 15-volt scale is used for testing the voltage of the cells when they are not in use. A switch on the distributing board controls the 110-volt alternating current which is used for the lights and motors. Finally, on this board are the terminals for all of the desks, so that any operator may connect with his desk any of the cells not then in use, and in almost any combination. It will be seen that the connections on this board make the different cells and desks practically independent of each other; for instance, a part of the cells may be charging while the remainder may be dis- tributed to the various desks as wanted. At each working desk is a 24 X 36-inch slate panel which carries all of the apparatus that will be needed by the analyst, making each board an independent working unit. The volt- meter and ammeter on each board are double-scale instruments with ranges of 2 volts and 20 volts and amperes, respectively. The multiplier for the voltmeter is controlled by a small knife switch, and the shunts for the ammeter are joined to plug receptacles. The current to each desk panel is controlled by a slide-wire taper rheostat. These rheostats sre wound to give a total re- sistance of 130 ohms, in 254 steps. The carrying capacity of the first step is 1.3 amperes and that of the last step 25 amperes, on continuous work. For working with rotating electrodes, 1/30 h.p. alternating- ELECTRO-ANALYSIS 153 current, series motors of the commutator type are mounted on the board and are controlled by a switch and 5-step rheostat. Induction motors are not used, because it is desirable to make variation in speed possible. These motors have a maximum speed of 2200 r.p.m. on 110 volts and are provided with three pulleys of different sizes. FIG. 47. A single desk panel and rotator, Purdue University Laboratory. In many laboratories it is the practice to mount the motor directly on the electrolyzing stand, thus avoiding all belting and making possible direct connection with the rotating electrode. On the other hand, the application of a small belt is a simple 154 QUANTITATIVE ANALYSIS operation and not only very much decreases the vibration of the stand, and consequently the danger of dust particles fall- ing into the bath, but also removes the motor from the region of the bath, which in many cases contains acids and which is frequently hot. Corrosion of the motor parts is in this way largely prevented. The stand for holding the electrodes is of iron, is quite heavy, to prevent vibration, and is fitted with rubber feet. Every portion of the base and vertical rod is heavily enameled and the FIG. 48. Group of five desk panels, Purdue University Laboratory. electrode supports are clamped to the rod by means of heavy thumb screws, but in such a manner that the screw does not come into contact with the rod, so that the enamel is not injured by the grip. Each clamp is insulated from the rod by a fiber bushing and each carries a binding post. There is no glass about the stand, perfect insulation of the electrode clamp being secured by the fiber bushings. The supporting ring for a dish cathode has three brass screw contacts which are ad- justable for dishes of different sizes. Finally, for stationary electrodes, simpler clamps are provided to take the place of the ELECTRO-ANAL YSIS 155 rotator. The rotator, which carries three pulleys of different sizes, is a vertical shaft, the lower end of which carries a universal chuck, electrical contact being insured by a brass brush. SAMPLE: OF_ MARKED ELECTRO -ANALYSIS SOJISTANOE DETERMINED EXPEBUJEST 2iUMBSB I II J II AMOUNT or SAMPLE TAKE* FRACTION UBSD QUANTITY or OTHER ELECTROLYTES PRESENT VOLTS TOTAL AMPERES CATHODE SURFACE *%, TEMPERATURE RANGE TOTAL TiME'or ELECTROLYSIS DESCUIEIIOM ojf ELECTRODES AND BEBtD,:iIi ROX&TED TOEUHTO* JtfEictJFocMf) CHAOTER or DEPOSIT PERCENT. AVERAGE FIG. 49. Blank for reporting results of electro-analysis. Records. Systematic records should be kept of all of the data obtained during the experiment. A satisfactory blank for this purpose is shown in figure 49. 156 QUANTITATIVE ANALYSIS COPPER If a copper salt is to be used a nitrate or sulphate is best suited. Other salts of volatile acids may be converted into the sulphate by evaporating with sulphuric acid, stopping the evaporation before any decomposition into copper oxide occurs. Copper deposits in a coherent form from solutions containing sulphuric acid, nitric acid, oxalic acid and ammonium oxalate, potassium cyanide, phosphoric acid, formic acid or ammonium hydroxide. Of all of these, nitric acid produces the best results and probably sulphuric acid is next best. Chlorides should not be present. If metallic copper is to be analyzed it may be dissolved in nitric acid and the undesired excess of acid removed by evaporation. Determination of Copper in Soluble Salts. Use enough sample to yield 0.25 to 0.50 gm of copper. Dissolve in such a manner that 200 cc of solution will contain about 2 cc concentrated nitric acid. If the sample contains chlorides it must be treated with the least possible excess of sulphuric acid and heated until fumes of sulphur trioxide appear. It is sometimes desirable to make a larger quantity of solution, as 250 cc, and to use an aliquot part for each determination. In this case the acid may be added to the solution as used. The electrolysis may be begun and finished at the temperature of the room, but it will be hastened by warming the solution to about 70 at the beginning. Connect the weighed electrodes and add enough water to cover the cathode, which must extend entirely to the bottom of the beaker, place split cover glasses on the beaker and electrolyze with a pressure not below 1.7 volts and not above 2.0 volts unless other metals are known to be absent. In the latter case there is no upper limit to the voltage that may be used, except that fixed by the current density desired to give good deposits. The current density that may be used will depend upon the kind of electrodes. For foil cones or cylinders or for dishes, NDioo = about 0.1 amp. For gauze electrodes NDi OQ may sometimes be as high as 5 amp. In any case the analyst must use his judgment, watching the deposited metal to discover its character. The copper should appear as a bright red metal with no spots of brown and no tendency to crumble off the cathode. When the disappearance of color indicates that the metal is deposited remove a few drops to a white test plate by means of a pipette and test for traces of copper by adding a drop of concentrated ammonium hydroxide or of potassium ferrocyanide solution. The for- mer is preferable for the first test because if copper is found the solution can be neutralized with nitric acid and returned to the beaker. If no copper is indicated make another test, using four or five drops taken from the bottom of the beaker, and applying the ferrocyanide test. ELECTRO-ANALYSIS 157 When all of the metal is found to be deposited arrange a small siphon tube in such a manner that the solution may be drawn from the bottom of the beaker without interrupting the current. As the solution is removed wash down the exposed portion of the cathode removing every trace of acid before the E. M. F. is removed from the system. It is best to add water fast enough to keep the bottoms of both electrodes covered until the acid has become so diluted that no further action is to be feared. Siphon out the remaining liquid and perform the next operations as quickly as possible. Instead of siphoning, another plan is to have the beaker supported on blocks, removing these and lowering the beaker gradually and washing the cathode as it becomes exposed. Lower the beaker and remove the cathode, taking care to avoid touching cathode to anode and thus making a short circuit, and quickly wash with much water. Set aside until the duplicate cathode has been treated in the same way, then wash both cathodes with redistilled alcohol and dry at 100. The alcohol washing may be omitted, but drying is hastened by this means. Weigh and calculate the percent of copper in the sample. Remove the copper from the cathode by dipping into warm dilute nitric acid. Determination of Copper in Brass and Similar Alloys. 1 Dissolve 0.5 gm of the drillings in a mixture of 5 cc each of concentrated nitric acid and water, in a covered casserole, heating finally to expel all brown oxides of nitrogen. Dilute to 75 cc and filter through a paper of close texture. Wash the residue of metastannic acid with hot water, pre- serving the washings with the filtrate. (The precipitate may be ignited in a porcelain crucible to stannic oxide for an approximate determination of tin.) To the filtrate and washings in a casserole add 5 cc of concentrated sulphuric acid. Evaporate under a hood until copious fumes of sul- phuric acid appear. The nitric acid is thus driven off. Cool, add 35 cc of water and boil for 1 minute in order to dissolve all soluble sulphates. Filter on a Gooch crucible and wash with dilute sulphuric acid until a few drops of the washings show no blue color with a slight excess of ammonium hydroxide. (In case the Gooch crucible has been ignited and weighed before filtering the solution, the lead sulphate may be used for a determination of lead. It is washed with 50 percent alcohol to remove sulphuric acid and is then dried and heated, cautiously at first, to dull redness for 15 minutes. After cooling it is weighed as lead sul- phate, from which the percent of lead is calculated.) 1 The complete analysis of brass is described on page 505. The brief outline here given is with reference to the electrolytic determination of copper. 158 QUANTITATIVE ANALYSIS The solution and washings (not including alcohol washings from the lead sulphate) now contain copper, zinc and sulphuric acid. In pres- ence of so much sulphuric acid it is not desirable to add more nitric acid, even though the latter would improve the character of the deposit of copper to be obtained. Instead, add 1 gm of ammonium nitrate, which results in the substitution of nitric for sulphuric acid. Rinse the solu- tion into the beaker in which electrolysis is to take place, arrange the electrodes and add water until the cathode is covered, mixing well. Conduct the electrolysis and subsequent treatment as directed on page 156, keeping the voltage below 2.5 to avoid the deposition of zinc. Determination of Copper in Ores. The ore will always contain more or less insoluble gangue. It may or may not contain metals whose compounds are soluble in aqua regia and whose decomposition voltages are near to that of copper (bismuth, antimony, arsenic, mercury or silver) . In case such metals are absent or are present in quantities so small that they may be disregarded, proceed as follows : Weigh 0.5 gm of the powdered ore and place in a casserole. Add 10 cc of concentrated hydrochloric acid and 5 cc of concentrated nitric acid, cover and boil slowly to aid in dissolving. Digest on the steam bath until action seems to be complete then add 7 cc of concentrated sulphuric acid and boil under a hood until volatile acids are expelled and fumes of sulphuric acid appear. Cool, add 25 cc of water, boil 1 minute and filter to remove lead sulphate and gangue, receiving the filtrate in the beaker in which the solution is to be electrolyzed. Wash well with hot water, preserving the washings with the filtrate. Add 1 gm of dry ammonium nitrate, connect the electrodes in place, add water until the cathode is covered and mix well. Electrolyze as directed on page 156. If any of the metals named above are present in appreciable quantities the procedure is the same as that just described for other ores, until after the filtrate from lead sulphate and gangue is obtained. This filtrate and the washings are received in a casserole. Dilute, if neces- sary, to 75 cc. Cut a strip of sheet aluminium about 2.5 cm wide and 14 cm long, bend into a triangle and place in the casserole. Cover the solution and boil gently until all of the copper is precipitated, leaving the solution colorless or green from ferrous sulphate. (If this condition cannot be obtained it is because nitric acid has not been expelled com- pletely when evaporating with sulphuric acid.) When all copper is precipitated wash down the sides of the casserole with a stream of hydrogen sulphide solution and pour the solution and copper into a filter paper. Filter rapidly, to minimize oxidation and resolution of copper, and wash the aluminium and copper with hydrogen sulphide solution. The latter will cause the precipitation of traces of ELECTRO-ANALYSIS 159 copper that may have redissolved but this precipitation should take place before the solution has passed through the filter. If the filtrate appears brown this indicates a loss of copper and the solution must be refiltered and washed until a clear filtrate is obtained. Place under the filter the beaker which is to be used for the electrolysis, then pour over the aluminium in the casserole a mixture of 4 cc of concentrated nitric acid with the same volume of water. Heat to dissolve all adher- ing copper then pour the acid slowly over the copper in the filter. When all copper is dissolved wash the paper very thoroughly with hot water. Boil the filtrate and washings to expel oxides of nitrogen, then electro- lyze without further addition of acid. SILVER The deposition of silver from solutions containing free acids may be accomplished, but the deposit is usually either spongy or crystalline so that it cannot be washed without loss. When nitric acid is present there is also a deposit of silver peroxide at the anode if high voltage is applied. The addition of potassium cyanide, although materially lowering the concentration of silver cations, is desirable because it entirely prevents the formation of silver peroxide and also yields a coherent deposit of silver at the cathode. The deposit is without lustre and should be white. The deposition of silver peroxide upon the anode, when high voltage is applied to solutions containing nitric acid, is probably due to the existence of the anion of a silver oxyacid, H 2 AgO2 * + 2H+AgO 2 . This is a theoretical derivative of the dihydroxide of silver, Ag(OH) 2 . The decomposition potential of this anion is higher than that of the univalent silver cation and its concentra- tion is always small, yet it will discharge to some extent at the same time that silver is depositing upon the cathode, if high vol- tage and current density are used: Ag0 2 AgO+0. Such a deposit having formed upon the anode it will finally redissolve as the silver becomes more dilute in the solution because the equilibrium is disturbed by the removal of silver cations, Ag, which are dis- charging at a much greater rate on account of their lower decom- 160 QUANTITATIVE ANALYSIS position potential. The discharge and deposition of silver peroxide is entirely prevented by potassium cyanide. When this is added to a solution of silver nitrate a precipitate of silver cyanide is first formed: AgNO 3 +'KCN->AgCN+KN0 3 . An excess of potassium cyanide redissolves this precipitate, forming a salt of a complex anion: AgCN+KCN-KAg(CN) 2 . In this way the equilibrium, + Ag<=AgO 2 is also disturbed by the conversion of silver mto the new anion Ag(CN) 2 , which, presumably, has a high decomposition potential. The double cyanide of potassium and silver cannot all be in the form represented by the formula KAg(CN) 2 as in this case no silver could be discharged at the cathode, but only hydrogen. There must, therefore, be equilibrium between the salt having the composition represented above and the single cyanides: KAg(CN) 2 ^KCN+AgCN, or the ionic equilibrium Ag(CN) 2 ^Ag+2CN. It is to be supposed that the relatively high decomposition potential of the anion Ag(CN) 2 prevents its discharge under ordinary conditions. Determination. Use sufficient sample to give about 0.3 gm of silver. Dissolve this in a small amount of water and add just double the quan- tity of a solution containing 3 to 5 gm of potassium cyanide in 25 cc of water that is necessary to redissolve the precipitated silver cyanide. Fasten the weighed electrodes in place and dilute with water until the cathode is covered. The E.M.F. required will be about 2 volts and NDi QO = 0.04 to 0.10 amp for electrodes not of gauze or as high as 2 amp if gauze electrodes are used. If the current density is too large the potas- sium cyanide will be decomposed around the anode, giving a dark solu- ELECTRO-ANAL YS2S 161 tion of organic matter that will eventually reach the cathode and darken the silver. Such darkening will occur also if the potassium cyanide is impure. A cyanide of high grade is required for this purpose. The deposition will require 30 minutes to 5 hours, depending upon the current density used. When the deposition is thought to be complete a few drops of the solution may be tested for silver by adding dilute nitric acid in sufficient quantity to decompose the potassium cyanide. Hydrocyanic acid is removed by boiling and then ammonia is added to make basic, this being followed by ammonium sulphide. When the deposition has been shown to be complete the solution is removed and the cathode washed, dried and weighed as in the preceding exercise. Remove the silver from the electrode by dipping into dilute nitric acid. Caution. Avoid inhaling the vapors that arise during the progress of electrolysis. Do not use pipettes and do not fill the siphon by suction. Determination of Silver and Copper in an Alloy. (Other metals, with the exception of gold, are assumed to be absent.) Clean the alloy by polishing, followed by wiping with filter paper. Cut into pieces or drill so that about 0.5 gm may be used for each analysis. Dissolve in a covered casserole in a mixture of 5 cc each of concentrated nitric acid and water. Rinse down the cover glass and evaporate the solution nearly to dryness on the steam bath. Do not heat the dry mass as this would cause the decomposition of some of the nitrates. Redissolve the salts in about 10 cc of water. Silver. Place 3 gm of potassium cyanide in the electrolyzing beaker, dissolve in 50 cc of water and rinse into this the solution of copper and silver nitrates, stirring. Insert the electrodes, dilute until the cathode is covered with solution, mix and electrolyze as directed on page 160, keeping the voltage below 1.5. Copper. To the solution from which silver has been deposited, including the washings from the latter, add 5 cc of concentrated nitric acid. This should be done under a hood as the vapors of hydrocyanic acid are extremely poisonous. Evaporate to 75 cc or less, to expel all of the hydrocyanic acid. Determine the copper as directed on page 156. IRON Iron does not deposit well from solutions containing nitrates, chlorides or strong inorganic acids. If the iron salt is derived from such an acid this acid will be formed as electrolysis proceeds and iron will redissolve from the cathode. To prevent its forma- tion a salt of a weak inorganic or organic acid may be added. 11 162 QUANTITATIVE ANALYSIS Examples of salts that are used for this purpose are ammonium oxalate, ammonium tartrate and sodium citrate. There is always a possibility of depositing some carbon from such solutions, and this is least likely to occur when the oxalate is used. The iron or iron salt should be converted into the sulphate before electrolyzing. Determination. Calculate the weight of sample that will be required to give about 0.3 gm of iron. If the sample is an iron salt, soluble in water, dissolve in water, using no more than is necessary. If the sample is iron or steel, dissolve in the proper quantity of dilute sul- phuric acid, avoiding an excess. In either case, pour the solution slowly into a solution of 5 to 10 gm of ammonium oxalate in as little water as possible, stirring until the precipitate of iron oxalate is redissolved. Dilute, after the electrodes are in place, until the solution covers the cathode. The decomposition voltage for iron in such a solution is about 2 andNDioo= 0.1. The deposited iron should be bright. It does not easily redissolve in the solution when the current is interrupted and may be readily washed. The end of the process is tested by the use of potas- sium ferricyanide or potassium thiocyanate. If the latter is used the solution should be previously warmed with two or three drops of nitric acid, since the iron (if any is present) has been reduced to the ferrous con- dition by the current action. After weighing the iron it should be re- moved from the cathode by dissolving in warm dilute sulphuric acid. LEAD When salts of lead in solution are subjected to the action of a current the lead may deposit upon both cathode and anode upon the former as elementary lead and upon the latter as lead peroxide. Upon the cathode the lead is so spongy that it becomes -impossible to wash and dry it properly. This is the familiar action of the lead storage cell during charging, where lead sulphate is electrolyzed, producing spongy lead at the " negative" and lead peroxide at the " positive." If the lead salt solution contains a considerable excess of nitric acid the entire quantity of lead deposits upon the anode as peroxide and this is the only practicable quantitative method for the electroly- sis of lead salts except where the mercury cathode is used. Lead dioxide cannot be completely dehydrated unless heated to a temperature above 200. ELECTRO-ANAL YSIS 1 63 The deposition of lead peroxide upon the anode when nitric acid is present is to be explained exactly as in the case of silver. The solution contains a small concentration of the amphoteric lead perhydroxide, Pb(OH) 4 , which furnishes anions of an oxy- acid of lead, H 4 Pb0 4 . The electrolysis of this acid and dis- charge of its anion produces lead peroxide and oxygen: Pb = 4 -+Pb0 2 +0 2 . As in the case of silver the formation of the peroxide may be prevented by the addition of some substance which will diminish the concentration of lead cations, such as ammonium oxalate. Since the cathodic deposit of lead cannot well be washed and dried without loss or oxidation the anodic deposition of peroxide is assisted by addition of nitric acid. Determination. Weigh enough lead salt (preferably nitrate) to produce about 0.3 gm of lead, dissolve in the proper quantity of water to cover the electrodes and add 20 cc of concentrated nitric acid for each 100 cc of solution. Connect the electrodes so that the one with the largest surface will be he anode, instead of the cathode as is the case in the electrolysis of most other metals. The lead peroxide will not adhere well unless the anode surface has been roughened, as by sand blasting. Warm to about 50 and keep at this temperature until the electrolysis is finished. Use 2.4 volts. NDioo = 1.5 amp. At the end of the operation test the solution for lead by adding a few drops of hydrogen sulphide solution to a small amount of the electrolyte. Care- fully remove the anode, having previously washed it in the usual way. Dry at 200 to 230 until the weight is constant. NICKEL The best solution from which to deposit nickel is that of the sulphate, containing ammonium sulphate and ammonium hy- droxide. If nitric acid is present there is usually some trouble, due to the oxidation of the deposited nickel. Nickel may also be precipitated from solutions containing ammonium oxalate, tartrate or citrate or from solutions containing an excess of potas- sium cyanide. Determination. Separation of Copper, Nickel and Iron. Thoroughly clean and dry a nickel coin, weigh it accurately, place in a casserole and dissolve in nitric acid (sp. gr. 1.2) the casserole being covered while the coin is dissolving. Carefully add 10 cc of concentrated sulhpuric acid 164 QUANTITATIVE ANALYSIS and evaporate over a flame until the characteristic dense, white fumes of sulphuric acid appear. The evaporation should be accomplished while holding the casserole in the hand, giving it a continuous rotary motion to hasten evaporation and prevent spattering. Allow the mate- rial to cool then wash into a 250 cc graduated flask and dilute to the mark. Measure 50 cc of the solution into the vessel in which electrolysis is to be accomplished and deposit the copper in the manner already described. The voltage should not exceed 2; 7, which is nearly the decomposition voltage of nickel. Unusual care should be exercised in washing and saving the washings because other metals are to be determined. Evaporate the solution from which the copper has been removed, until the volume is about 100 cc. Neutralize with ammonium hydrox- ide, boil to flocculate the colloidal ferric hydroxide which is always present and filter off the precipitate, washing the paper 'and precipitate with hot water. Add to the nitrate 7 gm of ammonium sulphate and 20 cc of ammonium hydroxide (sp. gr.' 0.90), and electrolyze. Decomposition voltage is about 2.8 and any voltage above this value may be used, the upper limit being fixed by the nature of the deposit obtained. Record the current density. Dissolve the ferric hydroxide in the filter paper with 1 cc of oxalic acid solution, saturated at about 20. Wash the solution out of the paper with hot water and into a solution of 5 gm of ammonium oxalate in 100 cc of water. Electrolyze as previously directed. Moving Electrodes. In the discussion of decomposition voltage it was noted that if the voltage is unduly increased in or.der to hasten the decomposition, gas evolution prevents the formation of a dense deposit of metal. Migration of the ions is comparatively slow and current is transferred, not only across the solution, but also from the solution to the cathode, by hydro- gen ions. If the migration of the metal ions is aided by stirring the solution large currents may sometimes be carried without the deposition of enough hydrogen on the cathode to injure the metal deposit. Stirring may be accomplished by any one of five different methods: (1) Heating to produce convection currents, (2) use of stirring apparatus not connected with the electrodes, (3) rotation of the anode, (4) rotation of the cathode, (5) electromagnetic action. Convection currents are of limited usefulness because they are not sufficiently rapid. They will, ELECTRO-ANAL YSIS 165 however, materially shorten the time of electrolysis. Mechanical stirring, whether by. the second, third or fourth method, has practically the same use. Which method of these three is to be chosen will be decided chiefly by matters of convenience. If a stirrer of glass or other non-conducting material is to be used it will require room for its movement and, since it is as easy to rotate one of the electrodes, the stirrer is generally made one of these. Either electrode may be rotated with success. The anode is generally the one chosen for this purpose because it has not the large surface of the cathode and is therefore more easily manipulated. Fig. 50 shows one of the forms of anodes that may be rotated rapidly without becoming bent or distorted. The cathode is most frequently a dish. In order to provide a dish of large surface and capacity Classen devised a very thin platinum dish. The speed of rotation of the anode should be as high as may be attained without danger of throwing the solution out of the vessel. 500 to 1000 r.p.m. may be used. Recently a comparatively slow rotation (150 r.p.m.) of the ordinary spiral anode inside a gauze cathode has been used, with a con- siderable degree of success. Indeed, it is ques- tionable whether this is not more practicable than the rapid rotation because the gauze electrodes may be used without danger and the apparatus needs no watching after starting. The time necessary for complete deposition of the metal may be made about one-fifth of that required when stationary electrodes are used. In the electromagnetic stirring apparatus de- FlG 50 _ ( vised by Frary 1 the solution is placed within an Anode suitable electromagnetic field, generated by a current passing through a solenoid surrounding the electrolyte. The moving ions constitute a conductor in which the current moves radially while the electromagnetic field is vertical. In another form the apparatus is changed using a vertical field and vertical 1 J. Am. Chem. Soc., 29, 1592 (1907). 166 QUANTITATIVE ANALYSIS current lines. In either case the mutual action of the fields causes rotation of the solution within. The Mercury Cathode. By making the cathode of mercury instead of platinum two important gains are made. The large expenditure for electrodes is largely eliminated because of the relative cheapness of mercury, also there is no longer any ques- tion as to the satisfactory nature of the deposit of metal, since the latter amalgamates with the mercury instead of forming a surface deposit. The one obstacle to a nearly universal use of this form of electrode lies in the small surface that may be used, FIG. 51. (a) Mercury cathode cell with (b) drying apparatus. the large specific gravity of mercury prohibiting the use of more than a few cubic centimeters. The most satisfactory form of apparatus for this purpose is a glass cup having a small platinum wire fused into the bottom for the cathode connection (Fig. 51). The upper limit to the normal density is approximately fixed by the undesirable heating effect of large current densities. The anode should be rotated. In the following exercises are given the changes necessary to adapt the foregoing exercises to the use of rotating electrodes and the mercury cathode. Determination of Copper by Use of the Rotating Anode. Set up the apparatus and add to the solution 1 cc of dilute sulphuric acid ELECTRO-ANALYSIS 167 instead of nitric acid. Use rapid rotation and note the current density that can be employed. For slow rotation use the gauze cathode and spiral anode. Use the same solution as with stationary electrodes but increase the current density as much as possible, noting the character of the deposit in deter- mining the practicable maximum. Determination of Copper by Use of the Mercury Cathode and Rotating Anode. Use the small amount of solution made necessary by the size of the cup. Thoroughly clean the cup and place in it pure mercury until the whole apparatus weighs not more than 60 gm. Wash with distilled water, then with redistilled alcohol and finally dry by passing dried air through the cup. The drying tube on the inlet of the drying apparatus must be so arranged that the air is well filtered by cotton and no calcium chloride can enter the cell. The drying will cool the mercury and this must be allowed to warm to the temperature of the laboratory before weighing. Connect the platinum wire with the nega- tive binding post and place the anode in position. Pour in the solution but do not add acid of any kind. Rotate the anode as rapidly as may be done without loss of solution by spattering. Limit the current density only by the tendency of the solution to boil. After the electroly- sis is completed stop the rotation of the anode, reduce the voltage (but not below 2 volts), lower the anode until it almost touches the mercury and siphon and wash until no acid remains. Finally rinse the cell with redistilled alcohol and dry at the temperature of the laboratory. The copper amalgam may be used as a cathode in more experiments but it should be replaced by pure mercury as soon as it shows any tend- ency to form a scale of undissolved copper. Determination of Silver by Use of the Rotating Anode. Use the same solution as with stationary electrodes. The apparatus and general manipulation are the same as with copper. Either rapid or slow rotation of the anode will prove to be satisfactory. Determination of Silver by Use of the Mercury Cathode and Rotating Anode. Dissolve the silver salt in the proper amount of water and do not add potassium cyanide. Proceed as with copper. The deposit of silver peroxide that usually appears upon the anode at first should later redissolve so that all of the silver will finally amalgamate at the cathode. Determination of Iron by Use of the Rotating Anode. The iron should be in the form of sulphate. Note the current conditions required to produce a good deposit. Determination of Iron by Use of the Mercury Cathode and Rotating Anode. Use the same solution as in the preceding exercise and similar current conditions. CHAPTER V VOLUMETRIC ANALYSIS The gravimetric process involves the conversion of a given constituent of a substance into a compound of known composi- tion by the addition of an excess of a precipitating reagent. The volumetric process consists in the addition of a reagent of accurately known concentration (a ''standard solution") until a definite reaction with the substance is exactly completed. In the first case the compound produced is weighed while in the second case a solution of one of the substances reacting is measured by volume, the weight of the reacting substance being thus obtained indirectly. In the gravimetric process the constituent to be determined is actually weighed in a new compound and its weight calculated from the known composition of that compound. In the volumetric process the constituent to be determined is cal- culated from its known reacting ratio to the indirectly observed weight of reagent. Since a measurement of volume is much more quickly and easily made than is the case with the necessary filtration, washing, drying, ignition, cooling and weighing of a gravimetric determination, it follows that the volumetric method frequently results in a great saving of time. This is, however, not always the case and the method to be chosen will be that which, all circumstances considered, can be carried out most easily, quickly and accurately. No generalization can, at pres- ent, be made regarding this choice. The choice itself will not be difficult or uncertain after some experience is gained in general quantitative analysis. Apparatus. It may be observed at the beginning that the balance is practically always concerned even in the volumetric process. The concentration of the standard solution must be determined and this determination is generally gravimetric. This fact might seem to remove the time-saving element from the new class of methods. This is not so because one gravi- 168 VOLUMETRIC ANALYSIS 169 metric determination suffices for a large number of volumetric determinations if a sufficient quantity of standard solution is made. The volumetric process will involve many very accu- rate measurements of volume and consequently several forms of graduated apparatus. These will be described with some detail. Flasks. For measuring relatively large quantities (50 cc to 2000 cc) of liquids in one portion the graduated flask may be employed. This is the form of apparatus that possesses the least relative surface and conse- quently causes the least trouble in washing, draining, etc. The one reading is made at a mark on the neck. The neck must be suffi- ciently small to permit a reading with a slight percentage error but must be large enough to permit filling and emptying without trouble. These are practical limits, fixed as the result of experience. The requirements of practice also demand a neck of uniform bore and of some margin above and below the mark. Volumetric flasks may be graduated to con- tain a stated amount or to deliver this amount when emptied. No container can be made to deliver all of the liquid contained in it if the liquid is one that wets glass, a condition that obtains with water and solutions in water. If the flask is to be graduated to deliver a stated amount the mark must be placed higher than if it indicates the same amount contained in the flask. If a flask, upon emptying, could be made to drain uniformly it could be accurately calibrated for deliverance. On account of the necessary form of the flask -this is impossible and for accurate work the flask is always calibrated for containing the amount indicated by the inscription upon it. Pipettes. Pipettes are filled by suction and allowed to de- liver the liquid by the action of gravity. Common forms of pi- pettes are shown in the illustrations. The pipette which is gradu- ated to deliver one fixed and stated amount is known as a " transfer pipette.'' On account of the smallness of the bore at No.16 contains 200 C.C. II, 20C. FIG. 52. Volumetric flask. 170 QUANTITATIVE ANALYSIS the point where the mark is placed the pipette may be made to liquids with a relatively high degree of accuracy, in their use often render the readings highly These are chiefly due to inconstancy in the length of the period of draining. The pipette must be quite Aam^ especially with regard to the slight film of oily period of time must be allowed for draining in afl cases where a given instrument is used and, in the case of pipettes, a uniform procedure must be followed re- garding the removal of the last drop which is retained in the point of the pipette. The approved practice is to touch the point of the instrument against the into which the liquid flows but not* to blow out the drop that is still retained. Burettes. The burette finds a more taJUaBu w use than any other form of volumetric apparatus. The flask is used in lrihj standard solution often in making solutions of the substance being analyzed flQmritimflgs in subdividing thggg SPJntinmR. Theburette is used in each determination MM? it must be so graduated as to measure any quantity of liquid be- tween the extreme limits of its graduations. Its con- struction, calibration and use therefore require excep- tional care. Its bore must be uniform, its graduations sharp, distinct and correctly placed. It freely and uniformly and it should be provided with a cock of proper construction to permit easy control of the outflow. In reading burettes considerable jmiiM*tjiiiPs result ifUMii TMj*?>Ji?t-T' "jTiiy is be the part of the surface of the liquid which is observed is the lowest point of the meniscus this point is in the center of the bore. If, there- fore, the eye knot in the same horizontal plane with this pointer^ the burette is not in a vertical position the fine upon the exterior of the burette, apparently marking the position of the meniscus, does not ityuacnt the correct reading. This is made evident from Fig. 34 in which the errors are purposely exaggerated. In otder to increase the accuracy of reading and to prevent parallax, VOLUMETRIC ANALYSIS 171 various devices are used. In the Schellbach burette, Fig. 55, a background of white glass bears a stripe of blue. The meniscus appears against this as a point. This improvement is of doubtful value except in rooms where the light is not good, because it does not prevent parallax. The use of floats, Fig. 56, sometimes renders readings more easily made, especially when the liquid is so dark in color as to be nearly opaque. The mark on the float is brought so near to the side of the burette that parallax is also largely prevented. Trouble due to sticking of the float is suf- ficient cause for dispensing with its use whenever possible. The FIG. 54. Effect of Parallax. Fro. 55. Meniscus as seen in SdMflbaea boratt*. best construction of burettes yet devised is that specified by the U. S. Bureau of Standards. 1 Upon burettes made according to these specifications the marks for whole cubic centimeters extend entirely around the burette while those for subdivisions extend half way around. This arrangement absolutely obviates the troubles of parallax and makes quite sharp readings possible. Certain devices are sometimes used for promoting rapid fin- ing of the burette. Fig 58 illustrates some of these. To set up such a burette and keep it in working order requires a certain amount of attention and such devices are of value chiefly in works laboratories where large numbers of routine determinations are to be made by means of the same standard * Bur. Stand., Che. No. 9. 172 QUANTITATIVE ANALYSIS solution. The burette with a plain glass cock is most serviceable for ordinary use. All burettes should be covered by a cap when in use. This excludes dust and lessens evaporation of the solution. Units of Volume. For the requirements of volumetric analysis the same accuracy may be obtained without regard to the par- ticular unit of volume adopted, provided that all of the different pieces are calibrated upon the basis of the same unit. The liter is defined by the International Bureau of Weights () /*^\ an d Measures to be the volume occupied at ) ^ I W 4 C. by water having a mass of 1 kg. This is almost exactly 1000 cc and for all practical purposes may be regarded as such. (The milliliter is 1.000029 cc.) It is now customary to use this true liter as the standard, calibrat- ing apparatus upon this basis, the apparatus to be used at the average room temperature. In America the working temperature is usually taken to be 20. Other temperatures are used as standard working temperatures, particu- larly abroad where 17.5, 15.5 or 15 is used as a calibration and working temperature. When the true liter is made the basis of cali- bration and higher temperatures than 4 are used for the experi- mental part of the calibration corrections must be made for the difference in density of water used for calibrating and, if the water is weighed, also for the buoyant effect of air. In order to avoid making such corrections Mohr 1 suggested a different unit, the "Mohr liter," which is defined to be the volume of 1000 gm of water weighed in air at a standard pressure of 760 mm of mercury and at a temperature of 17.5. Absolute or Relative Capacities. In the discussion of the calibration of weights (page 66) it was stated that true gram values were not required, since accurately known relative values of the various pieces are sufficient for analytical values, analytical results being always stated in some sort of ratio of the con- stituent determined to the material analyzed. For a similar 1 Lehrbuch, Chemisch-analytischen Titriermethoden, 6th ed. (Rev. by A. Classen), 42. FIG. 56. Burette floats. VOLUMETRIC ANALYSIS 173 reason the various pieces of volumetric apparatus may be graduated upon any desired basis, true liter or otherwise, pro- vided only that the different pieces shall have correctly indicated relative capacities. However, it is very desirable that all pieces that are to be used in a given laboratory shall be interchangeable FIG. 57. Form of burette approved by the Bureau of Standards. FIG. 58. Burette with automatic fill- ing and overflow devices. and the adoption of a common standard for all workers in a laboratory is a practical necessity. In such a case calibration to a basis of the true liter, using weights that are calibrated in true gram values, is the logical procedure. Tolerance. Certain experimental errors in the graduation of volumetric apparatus may be regarded as reasonable errors, on account of which the apparatus should not be rejected. This 174 QUANTITATIVE ANALYSIS does not mean that corrections should not be made after the results of calibration are known. Maximum permissible errors in graduation are known as " tolerances" and if the tolerance is exceeded in a given case the piece should not be used without regraduation. The following quotation from the bulletin of the U. S. Bureau of Standards already referred to gives the requirements of that bureau for volumetric flasks, burettes and pipettes to be accepted for testing. These requirements are recognized as, at once, rigid and scientific. Apparatus used, even by the student beginning the study of volumetric analysis, should, whenever possible, conform to these requirements. GENERAL SPECIFICATIONS " (a) Units of Capacity. The liter, defined as the volume occupied by a quantity of pure water at 4 C. having a mass of 1 kg, and the one- thousandth part of the liter, called the milliliter or cubic centimeter, are employed as units of capacity. " (b) Standard Temperature. 20 C. is regarded by the bureau as the standard temperature for glass volumetric apparatus. " (c) Material and Annealing. The material should be of best quality of glass, transparent and free from striae, which adequately resists chem- ical action, and has small thermal hysteresis. All apparatus should be thoroughly annealed at 400 C. for 24 hours and allowed to cool slowly before being graduated. "(d) Design and Workmanship. The cross section must be circular and the shape must permit of complete emptying and drainage. "Instruments having a base or foot must stand solidly on a level sur- face, and the base must be of such size that the instruments will stand on a plane inclined at 15. Stoppers and stopcocks must be so ground as to work easily and prevent leakage. "The parts on which graduations are placed must be cylindrical for at least 1 cm on each side of every mark, but elsewhere may be enlarged to secure the desired capacities in convenient lengths. "The graduations should be of uniform width, continuous and finely but distinctly etched, and must be perpendicular to the axis of the appa- ratus. All graduations must extend at least halfway around, and on subdivided apparatus every tenth mark, and on undivided apparatus all marks must extend completely around the circumference. "The space between two adjacent marks must not be less than 1 mm. The spacing of marks on completely subdivided apparatus must show VOLUMETRIC ANALYSIS 175 no evident irregularities, and sufficient divisions must be numbered to readily indicate the intended capacity of any interval. Apparatus which is manifestly fragile or otherwise defective in construction will not be accepted. " (e) Inscriptions. Every instrument must bear in permanent legible characters the capacity in liters or cubic centimeters, the temperature in Centigrade degrees at which it is to be used, the method of use, i.e., whether to contain or to deliver, and on instruments which deliver through an outflow nozzle the time required to empty the total nominal capacity with unrestricted outflow must be likewise indicated. "Every instrument should bear the name or trade-mark of the maker. Every instrument must bear a permanent identification number, and detachable parts, such as stoppers, stopcocks, etc., belonging thereto, must bear the same number. SPECIAL REQUIREMENTS "(a) Flasks. At the capacity mark or marks on a flask the inside diameter should be within the following limits : Capacity of flask (in cc) up to and including 2000 1 000 500 250 200 100 50 ?5 Maximum diameter (in mm) Minimum diameter (in mm) 25 18 20 14 18 12 15 10 13 9 12 8 10 6 8 6 "The neck of a flask must not be contracted above the graduation mark. "The capacity mark on any flask must not be nearer the end of the cylindrical portion of the neck than specified below: Capacity Distance from upper end, cm Distance from lower end, cm 100 cc or less. 3 1 More than 100 cc 6 2 "Flasks of 1 liter or more but not less may be graduated both to contain and to deliver, provided the intention of the different marks is clearly indicated. "(6) Transfer Pipettes. Pipettes for delivering a single volume are designated "transfer" pipettes. "The suction tube of each transfer pipette must be at least 16 cm long, and the delivery tube must not be less than 3 cm nor more than 25 cm long. "The inside diameter of any transfer pipette at the capacity mark must not be less than 2 mm and must not exceed the following limits : 176 QUANTITATIVE ANALYSIS Capacity of pipettes (in cc) up to and including .... Diameter (in mm).. . 25 4 50 5 200 6 "The outside diameter of the suction and delivery tubes of transfer pipettes exclusive of the tip must not be less than 5 mm. "The capacity mark on transfer pipettes must not be more than 6 cm from the bulb. "The outlet of any transfer pipette must be of such size that the free outflow shall last not more than one minute and not less than the following for the respective sizes : Capacity (hi cc) up to and including Outflow time (in seconds) 5 15 10 20 50 30 100 40 200 50 " (c) Burettes and Measuring Pipettes. Only those emptying through a nozzle permanently attached at the bottom are accepted for test. "So-called "Shellbach" burettes that is, those having a milk-glass background with a colored center line will not be accepted for test. "The distance between the extreme graduations must not exceed 65 cm on burettes nor 35 cm on measuring pipettes. "The rate of outflow of burettes and measuring pipettes must be restricted by the size of the tip and for any graduated interval the time of free outflow must not be more than three minutes nor less than the following for the respective lengths: Length graduated, cm Time of outflow, sec Length graduated, cm Time of outflow, sec 70 160 35 60 65 140 30 50 60 120 25 40 55 105 20 35 50 90 15 30 45 80 40 70 "The upper end of any measuring pipette must be not less than 10 cm from the uppermost mark and the lower end not less than 4 cm from the lowest mark. "On a burette the highest graduation mark should not be less than 5 cm nor more than 15 cm from the upper end of the burette. "(d) Burette and Pipette Tips. Burette and pipette tips should be made with a gradual taper of from 2 cm to 3 cm, the taper at the extreme end being slight. "A sudden contraction at the orifice is not permitted and the tip must be well finished. VOLUMETRIC ANALYSIS 177 "In order to facilitate the removal of drops and to avoid splashing when the instrument is vertical, the tip should be bent slightly. "The approved form of tips for burettes, measuring pipettes, and transfer pipettes is shown in Fig. 59. "Special Rules for Manipulation. These rules indicate the essential points in the manipulation of volumetric apparatus which must be observed in order that the conditions necessary to obtain accurate measurements may be reproduced. "(a) Test Liquid. Apparatus will be tested with water and the capacity determined will, therefore, be the volume of water contained or delivered by an instrument at its standard temperature. FIG. 59. Tips of burette and pipettes, approved by the Bureau of Standards. "(6) Method of Reading. In all apparatus where the volume is limited by a meniscus the reading or setting is made on the lowest point of the meniscus. In order that the lowest point may be observed it is necessary to place a shade of some dark material immediately below the meniscus, which renders the profile of the meniscus dark and clearly visible against a light background. A convenient device for this purpose is a collar-shaped section of thick black rubber tubing, cut open at one side and of such size as to clasp the tube firmly. " (c) Cleanliness of Apparatus. Apparatus must be sufficiently clean to permit uniform wetting of the surface. "(d) Flasks and Cylinders. In filling flasks and cylinders the entire interior of the vessel will be wetted, but allowed a sufficient time to 12 178 QUANTITATIVE ANALYSIS drain before reading. Before completely filling to the capacity mark flasks should be well shaken to completely mix the contents. "Flasks and cylinders when used to deliver should be emptied by gradually inclining them until when the continuous stream has ceased they are nearly vertical. After half a minute in this position the mouth is brought in contact with the wet surface of the receiving vessel to remove the adhering drop. "(e) Pipettes and Burettes. In filling pipettes and burettes excess liquid adhering to the tip should be removed when completing the filling. "In emptying pipettes and burettes they should be held in a vertical position, and after the continuous unrestricted outflow ceases the tip should be touched with the wet surface of the receiving vessel to com- plete the emptying. "Stopcocks, when used, should be completely open during emptying. "Burettes should be filled nearly to the top, and the setting to the zero mark made by slowly emptying. " While under normal usage the measurements ordinarily are from the zero mark, other initial points may be used on burettes of standard form without serious error. "Tolerances. (a) Flasks Capacity (in cc) less than and including Limit of error, cc If to contain | If to deliver 25 0.03 0.05 50 0.05 0..10 100 0.08 0.15 200 0.10 0.20 300 0.12 0.25 500 0.15 0.30 1,000 0.30 0.50 2,000 0.50 1.00 (6) Transfer pipettes Capacity (in cc) less than and including Limit of error, cc 2 5 10 30 50 100 200 0.006 0.01 0.02 0.03 0.05 0.08 0.10 VOLUMETRIC ANALYSIS 179 (c) Burettes and measuring pipettes Capacity (in cc) of total graduated portion less than and including Limit of error (in cc) of total or partial capacity Burettes | Measuring pipettes 2 0.01 5 0.01 0.02 10 0.02 0.03 30 0.03 0.05 50 0.05 0.08 100 0.10 0.15 " Further, the error of the indicated capacity of any ten consecutive subdivisions must not exceed one-fourth the capacity of the smallest subdivision." Calibration. For accurate work the apparatus as supplied by the makers should never be regarded as correctly graduated until it has been tested (calibrated) by the user. Some manu- facturers use great care in graduating, especially with such ap- paratus as must pass inspection by one of the national standardiz- ing bureaus. Others are less careful and pieces are often found to have large errors in their graduation. Two general methods are in use for calibrating instruments for capacity. In the first the quantity of pure water or other liquid of known density which would exactly occupy the desired volume at the stated temperature is measured by weighing. The position of the meniscus is then compared with the mark upon the apparatus. If the latter was previously unmarked the position of the menis- cus is then marked. In the second method the capacity of the instrument is determined by allowing water or another liquid to flow into it from a previously standardized piece and the capacities are compared. In the first method temperatures must be accurately noted and corrections made for any departure from the standard temperature, also for air displacement. In the second method such corrections have already been made when the standard piece was calibrated and we have merely a comparison to make of two instruments of capacity. The second method is, therefore, shorter in point of time. Errors may occur if proper attention is not given to certain details of manipulation. Calibration by Weighing. Water is the most conveniently used liquid for this purpose. Since water solutions are generally 180 QUANTITATIVE ANALYSIS to be used in volumetric analysis water possesses a second advantage in that the form of meniscus is most nearly that of the solutions later to be measured. The problem in calibrating, in the case of flasks or other pieces having but one or two marks and therefore easily remarked, is to determine the correct posi- tion of the mark when the piece contains the rated quantity of liquid. For this reason, in laboratories where all of the apparatus purchased is regularly calibrated it is best and cheapest to purchase flasks unmarked, though conforming to certain specifi- cations as to dimensions and shape. If the true liter is to be taken as a basis it will first be necessary to determine the apparent weight of this volume of water in air, correcting also for the expansion due to the difference in tempera- ture between that at which the apparatus is calibrated and used and the temperature upon the basis of which the liter is defined (4). From the table on page 181 it is seen that the density of water at 20 working temperature is 0.99823. One liter of water would therefore weigh 998.23 gm in a vacuum. In air both water and weights are apparently lighter than in a vacuum. If the density of these were equal the effect of air upon them would be equal and no error would be introduced. Since the density of the weights is greater and their volume less, the buoyant effect is greater upon the water and the apparent weight of the water in air is less than the true weight. One liter of air, at 760 mm pressure and at 20 and having a humidity of 50 percent, weighs 1.19 gm. This is the buoyant effect upon the water. Analytical weights, whether plated or not, are usually constructed of brass. The density of brass may be taken as 8.4. The weight of air displaced by 998.23 gm of brass weights is then 0.99823 X 1.19 -4- " = 0.14gm. This is the buoyant effect upon the weights. The difference 1.19 0.14 = 1.05 gm is the apparent loss in weight of the liter of water when weighed in air. 998.23-1.05 = 997.18. There- fore one liter of water at 20 and in air apparently weighs 997.18 gm. The weight of fractions of a liter will be calculated from this. VOLUMETRIC ANALYSIS 181 EXPANSION OF WATER ACCORDING TO P. CHAPPUIS' Density of pure water free from air, by tenths of degrees from to 40 and under standard pressure Standard degrees Tenths of degrees Mean differ- ences 1 2 3 4 5 | 6 7 | 8 | 9 0.999 8681 8747 8812 8875 8936 8996 9053 9109 9163 9216 + 59 1 9267 9315 9363 9408 9452 9494 9534 9573 9610 9645 + 41 2 9679 9711 9741 9769 9796 9821 9844 9866 9887 9905 + 24 3 9922 9937 9951 9962 9973 9981 9988 9994 9998 0000 + 8 4 1.000 0000 *9999 9996 *9992 *9986 *9979 9970 *9960 *9947 *9934 - 8 5 0.999 9919 9902 9884 9864 9842 9819 9795 9769 9742 9713 - 24 6 9682 9650 9617 9582 9545 9507 9468 9427 9385 9341 - 39 7 9296 9249 9201 9151 9100 9048 8994 8938 8881 8823 - 53 8 8764 8703 8641 8577 8512 8445 8377 8308 8237 8165 - 67 9 8091 8017 7940 7863 7784 7704 7622 7539 7455 7369 - 81 10 7282 7194 7105 7014 6921 6826 6729 6632 6533 6432 - 95 11 6331 6228 6124 6020 5913 5805 5696 5586 5474 5362 -108 12 5248 5132 5016 4898 4780 4660 4538 4415 4291 4166 -121 13 4040 3912 3784 3654 3523 3391 3257 3122 2986 2850 -133 11 2712 2572 2431 2289 2147 2003 1858 1711 1564 1416 -145 13 1266 1114 0962 0809 0655 0499 0343 0185 0026 *9865 -156 16 0.998 9705 9542 9378 9214 9048 8881 8713 8544 8373 8202 -168 17 8029 7856 7681 7505 7328 7150 6971 6791 6610 6427 -178 13 6244 6058 5873 5686 5498 5309 5119 4927 4735 4541 -190 19 4347 4152 3955 3757 3558 3358 3158 2955 2752 2549 -200 20 2343 2137 1930 1722 1511 1301 1090 0878 0663 0449 -211 21 0233 0016 *9799 *9580 *9359 *9139 *8917 *8694 *8470 8245 -221 22 0.997 8019 7792 7564 7335 7104 6873 6641 64C8 6173 5938 -232 23 5702 5466 5227 4988 4747 4506 4264 4021 3777 3531 -242 24 3286 3039 2790 2541 2291 2040 1788 1535 1280 1026 -252 25 0770 0513 0255 *9997 *9736 *9476 *9214 *8951 *8688 *8423 -261 26 0.996 8158 7892 7624 7356 7087 6817 6545 6273 6000 5726 -271 27 5451 5176 4898 4620 4342 4062 3782 3500 3218 2935 -280 28 2652 2366 2080 1793 1505 1217 0928 0637 0346 0053 -289 29 0.995 9761 9466 9171 8876 8579 8282 7983 7684 7383 7083 -298 30 6780 6478 6174 5869 5564 5258 4950 4642 4334 4024 -307 31 3714 3401 3089 2776 2462 2147 1832 1515 1198 0880 -315 32 0561 0241 *9920 *9599 *9276 *8954 8630 *8304 *7979 *7653 -324 33 0.994 7325 6997 6668 6338 6007 5676 5345 5011 4678 4343 -332 34 4007 3671 3335 2997 2659 2318 1978 1638 1296 0953 -340 35 0610 0267 *9922 *9576 *9230 *8883 *8534 *8186 7837 *7486 -347 36 0.993 7136 6784 6432 6078 5725 5369 5014 4658 4301 3943 -355 37 3585 3226 2866 2505 2144 1782 1419 1055 0691 0326 -362 38 0.992 9960 9593 9227 8859 8490 8120 7751 7380 7008 6636 -370 39 6263 5890 5516 5140 4765 4389 4011 3634 3255 2876 -377 40 2497 2116 1734 1352 0971 0587 0203 98X8 9433 *9t)47 -384 41 0.991 8661 1 Travaux et Memoires, Bur. Intern. Poids Mesures, 13 (1907.) 182 QUANTITATIVE ANALYSIS Calibration at Other Than the Standard Temperature. Carelessness with regard to the temperature of the water used in calibrating may lead to serious errors. The apparent weight of one liter of water at 25 is 996.04 gm and at 15 it is 998.05 gm, instead of 997.18 gm. This gives a mean variation of 0.20 gm for each degree of variation from the standard tem- perature of 20, within the limits of 15 and 25. This error is partly compensated by the change in the actual capacity of the glass apparatus, due to contraction or expansion with change in temperature. The average coefficient of cubical expansion of the glass used for volumetric apparatus is 0.000025 for each degree of change in temperature. The correction is, of course, similar in sign to the temperature variation from 20, while the correction to be applied to the required weight of water is opposite in sign to the temperature variation. That is, if the temperature is above 20 a smaller weight of water should be taken for a true liter; but a glass vessel is actually larger and should have, from this standpoint alone, a greater weight of water to indicate a mark which can be used at 20. For this reason the net correction represents the difference between the two. The mean correction of 0.20 gm in the apparent weight of a liter of water is not sufficiently accurate for a range of more than a degree or two variation. The following table gives the apparent weight of a liter of water for various temperatures between 15 Temperature, degrees Apparent weight of 1 1 of water in air, grams Weight to be taken for calibration 15 998.05 997.93 16 997.90 997.80 17 997.74 997.66 18 997.56 997.51 19 997.38 997.36 20 997.18 997.18 21 996.97 996.99 22 996.76 996.81 23 996.53 996.61 24 996.29 996.39 25 996.04 996.16 VOLUMETRIC ANALYSIS 183 FIQ. 60. Calibrating device of the Bureau of Standards. 184 QUANTITATIVE ANALYSIS and 25, also the corrected weight of water to be taken for cali- brating flasks that are to be used at 20. The latter weight includes the correction for expansion and contraction of glass. This table is to be used only in case it becomes difficult to main- tain the laboratory temperature at 20. Calibration by Standardized Bulbs. Any bulb or tube that is to be used as a comparison standard for calibrating by the second 50 FIG. 61. Morse-Blalock calibrating bulbs. method must be of such a form that it will drain well upon emptying. It must also have the graduated portion small enough to make possible accurate readings at this part. The apparatus must be quite rigid so that varying pressure may have only an inappreciable effect upon the volume. The apparatus shown in Fig. 60 is that used by the Bureau of Standards. VOLUMETRIC ANALYSIS 185 In Fig. 61 are shown the standard bulbs devised by Morse and Blalock. 1 The three pieces shown provide a means for calibrating vessels of all of the various capacities common to the analytical laboratory, if proper combinations are made. In calibrating these bulbs it is necessary to determine the capacity from the single mark to the first stem division, also the capacity of the stem for the smallest subdivision. If the FIG. 62. Morse-Blalock bulb arranged for calibrating flasks. water used is kept at the standard working temperature no correction for this factor need be introduced and the value 997.18 gm, previously deduced as the apparent weight of a liter of water, may be used without change. The bulbs are supported in such a manner that they may be readily filled from a reservoir of distilled water at 20. The water from the bulb is carefully weighed and its volume calculated. That from the entire graduated portion of the stem is then weighed in a smaller vessel and the calculated volume is divided by the total 1 Am. Chem. J., 16, 479 (1894). 186 QUANTITATIVE ANALYSIS number of stem divisions and the result recorded as the value of one division. In order to use the bulbs for the calibration of vessels that are to contain a specified volume of liquid, the vessel to be calibrated must first be cleaned as hereafter directed. The proper bulb is then placed in such a position that it may drain directly into the instrument being calibrated and the latter is marked at the meniscus. Instruments to be calibrated to deliver, such as burettes, are better calibrated by weighing the water delivered. If, however, it is desired to use the standard bulbs for this purpose the burette is so connected that it may empty into the bulb from below. The details of manipulation will be made clear in the exercises that follow. Cleaning Solution. Prepare a cleaning solution by dissolving 5 gm of powdered commercial sodium dichromate in 500 cc of commercial sulphuric acid. The solution may be kept in a bottle having a wide mouth, such as those in which dry chemicals are purchased. Burettes may be inverted and left standing in the bottle the solution then being drawn up by suction and held in the burette by closing the cock. For cleaning flasks the solution may be allowed to remain in the flask for some time or a small amount may be warmed and the flask rinsed with it. The chromic acid produced by the interaction of sulphuric acid and sodium dichromate oxidizes all organic matter and leaves the glass thoroughly free from it. Exercise: Calibration of the Standard Bulbs. Clean the bulb and set it up in a manner similar to that shown in Fig. 60. Distilled water must be used. Place the bulb so that the graduated stem extends down- ward. A glass stopcock must be used for controlling the outflow, since pinchcocks with rubber connections would involve an uncertain change in volume of the apparatus. The tip of the outflow tube should be contracted to restrict the outflow, according to the size of the bulb, as follows : Size of bulb, cc Time of outflow, sec 50 200 500 30 50 65 Fill the bulb to the upper mark. With the stopcock wide open allow the water to flow into a previously dried and weighed flask until the first division (zero) on the lower stem is reached. After 15 seconds VOLUMETRIC ANALYSIS 187 adjust the water level to coincide with the mark, put the stopper in the flask (a glass stopper is desirable) and weigh. Place a weighing bottle under the delivery tube and allow the stem to drain to the last mark, stopper the bottle and weigh. If the water was at 20 divide its weight in grams by 0.99718 and record the' result as cubic centimeters capacity of bulb and of stem. Record also the value of each stem division and the division to be used as the mark for the rated capacity of the bulb. If the temperature was not 20 determine from the table on page 182 the weight of water that should be used at the observed temperature for calibrating bulbs to be used at 20. In the preceding exercise as in those that follow the bulb and reservoir may also be set up as in Fig. 62. The chief objection to this method of assembling lies in the fact that the water entering the bulbs must, for reasons already explained, pass through a glass stopcock, which is necessarily lubricated with some kind of grease. The result is that no matter how well the bulb may have been previously cleaned it acquires, at the first filling, a film of oil that absolutely prevents the proper draining at the next stage in the experiment. If the first arrange- ment is used and the bulb is filled from above a rubber connec- tion and pinchcock may be used and this annoyance avoided. Exercise : Calibration of Flasks by the Standard Bulbs. With the proper bulb in the position used in the preceding exercise, and with the dried flask to be calibrated placed under the delivery tube, allow water to flow into the flask until the proper mark on the lower stem is reached, exactly following the directions as to time of outflow as given in the pre- ceding exercise. Avoid handling either bulb or flask at the parts con- taining the water as the temperature is thereby raised. It has already been explained that after the bulbs have been standardized they may be used for further calibrations without regard to the temperature of the water provided only that the temperature does not change during the progress of the experiment. To mark the flask cut a strip of gummed label, long enough to reach around the neck and about 1/4 inch wide. Carefully paste this with the original straight edge at the level of the meniscus, where the mark is to be made. Melt a small quantity of paraffin and brush a thin layer over the label and over a space of about 1 inch on either side of it. Using the point of a knife or of a sharpened piece of wood trace the straight edge of the label around the neck of the flask, making a mark sufficiently wide to be easily visible. The label here merely serves as a guide, making a 188 QUANTITATIVE ANALYSIS regular line possible. Using a small feather as a brush apply a few drops of hydrofluoric acid .and allow this to remain on the flask for two or three minutes, after which the acid may be washed off and the paraffin removed by warming. In case the flask already has a graduation and the calibration shows this mark to be incorrectly placed it is desirable to indicate the new mark by making a small, well-defined arrow with the point resting exactly upon the new mark. The operator's initials may be placed beside the arrrow and if this is done carefully, no interference will result. If the flask contains no inscription etch the side like that shown in Fig. 52, page 169. Exercise: Calibration of Flasks by Weighing. This is, in many respects, the most satisfactory method of calibration although more time is required. Have the flask clean and quite dry. Place on a balance of capacity sufficiently great to carry the filled flask. Counter- poise, then add weights to the right pan at the rate of 997.18 gm for each liter. Remove the flask from the balance and fill with recently boiled distilled water at 20, nearly to the point where it is thought that the mark will be placed. Remove drops from the inside of the neck, above the level of the water, using a roll of filter paper. Replace the flask upon the balance pan, then carefully drop in water from a pipette until the balance is in equilibrium. Mark as directed in the preceding exercise. In both of these exercises the flasks have been calibrated to contain the rated quantity and this is the only way in which flasks will be used in this work. Exercise : Calibration of Burettes by Weighing. The marking of a burette is too complex to be easily changed and the calibration will therefore consist of finding what, if any, corrections must be applied to the existing graduations. First inspect the burette to determine whether it conforms to specifi- cations, especially with respect to outflow time. Clean the burette with cleaning solution and distilled water. Fill with distilled water at 20. Weigh accurately a 25 cc weighing bottle to the third decimal then measure 5 cc of water into it from the burette, and reweigh. Add another 5 cc and weigh, continuing until the bottle is full. Empty the bottle, reweigh and continue the process until the water from the entire graduated portion of the burette has been weighed. Repeat the process in order to have a check upon the work. Calculate the true capacity of each of the ten portions, using the weight 0.99718 gm for 1 cc of VOLUMETRIC ANALYSIS 189 water. Record as follows, the capacities in the last two columns being recorded only as far as the second decimal place. Mark. Weight of water, each interval. True capacity, each interval. True total capacity, zero to e of interval. BBtj Construct a curve showing the true reading at all points. In case any marked irregularity is observed at any part of the burette so that corrections taken from the curve would be inaccurate, recalibrate this portion, using 1 cc at a time. Exercise: Calibration of Burettes by the Standard Bulbs. Set up the apparatus as in Fig. 63. The reservoir must be higher than the top of the burette and this, in turn, must be placed so that the lowest graduation is higher than the bulbs. The tubing leading from the reser- voir to the burette may be of well-cleaned rubber. That between the stopcock a and the burette and bulbs respectively must be of glass, the necessary connections being of heavy rubber tubing with the glass tubes pushed together until they touch inside the rubber. With the three-way cock b closed, open the cock a and fill the burette with water. -Close a and open b so that the 2 cc and 3 cc bulbs may fill, then drain the burette to the zero mark and the bulbs to that mark on the stem of the 2 cc bulb which represents exactly 2 cc. This leaves the bulbs moistened as they will be throughout the experiment. Leave the burette cock open. Turn the cock b and measure 5 cc of water from the burette into the small bulbs. Observe the position of the meniscus upon the stem of the 3 cc bulb and calculate the true capacity of the first portion of the burette, using the values for the stem divisions as determined in the calibration of the bulbs. Repeat this process for the other nine portions of the burette and record as follows: Mark. Bulb reading. True capacity, each interval. True total capacity, zero to end of interval. For a more nearly complete calibration the burette may be calibrated 2 cc at a time, using the 2 cc bulb alone or 1 cc or any fraction at a time using the standard tube, Fig. 64. Such. calibration is necessary if the bore of the burette is found to be very irregular. Calibration of Pipettes. Pipettes which are graduated in small subdivisions from zero to full capacity (" measuring pi- pettes") are calibrated in the same manner as are burettes. Transfer pipettes are best calibrated by the method of weighing. Exercise : Calibration of Transfer Pipettes. Determine whether the time of outflow conforms to the requirements as set forth on page 176. 190 QUANTITATIVE ANALYSIS If not alter the tip of the pipette before calibrating. Provide a weigh- ing bottle having a capacity of 10 cc, also a larger one having a capacity equal to that of the pipette. Cut a strip of paper about 2 mm wide FIG. 63. Morse-Blalock bulb set up for calibrating burettes. and 5 cm long and carefully rule this in divisions of centimeters, mark- ing from to 5, and subdivisions of millimeters, using fine lines. Deter- mine the approximate location of the capacity mark on the pipette by a rough experiment, unless the pipette is already marked. Paste VOLUMETRIC ANALYSIS 191 the paper strip on the stem of the pipette with the division 2.5 at the supposed place for the capacity mark and with the zero toward the point of the pipette. Having cleaned the pipette with chromic acid solution it is drawn full of distilled water which is at a temper- ature of 20, and the water is allowed to flow out until the zero mark is exactly reached. The pipette must be held in a vertical position and the eye must be in the same horizontal plane as is the meniscus. The pipette tip is now touched against the side of the beaker to remove the last drop. The finger is then removed from the top of the pipette and the water is allowed to flow, at full speed, into the larger weighing bottle, which has already been weighed. The tip is immediately touched to the side of the weighing bottle to remove the hanging drop. The weighing bottle is ther stoppered and weighed. Cal- culate the volume of the water from the observed weight and record this as the capacity of the pipette to the zero mark. Using the small weighing bottle determine in a similar manner the capacity of the pipette stem between and 5. Divide this capacity by 50 in order to obtain the value of the smaller sub- divisions. From the capacities so determined calculate the number of stem divisions to be added to the zero in order to obtain the rated capacity of the pipette. Mark the point so determined, using the method directed for marking flasks. CALCULATION OF THE RESULTS OF VOLUMETRIC ANALYSIS Although the first exercises in volumetric analysis will necessarily have to do with the making of solutions and with their standardiza- tion and adjustment to desired concentration, it will be simpler to deal first with the calcula- tion of the results of the analysis. When making the determi- nation of silver by the gravimetric method, a definite amount of the silver compound was weighed, dissolved in water and a slight but somewhat indefinite excess of hydrochloric acid was FIG. 64. Morse Blalock tube. 192 QUANTITATIVE ANALYSIS added, thus precipitating all of the silver as silver chloride, the following reaction taking place: AgNOs+HCl-^AgCl+HNOt. The silver chloride, representing the entire amount of silver present in the compound of unknown composition, was then filtered, washed, dried, and weighed and from this observed weight and the known weight relations between silver chloride and silver the percent of the latter was calculated. The formula AgCl expresses the fact that for each 143.34 parts, by weight, of silver chloride, there was involved 35.46 parts of chlorine and 107.88 parts of silver. In other words, in any given weight of , . . , 35.46 , .:. . , ^ . , . . , 107.88 . silver chloride, 1/to OA of this weight is chlorine and 1/)0 , is ' 143.34 143.34 silver. If the weight of silver chloride found in the analysis 107 88 is multiplied by the fraction 1 AQ QA and the result divided by 14O.OT: the weight of sample taken, the quotient will be, when multiplied by 100, the percent of silver in the sample. If W = the weight of silver chloride found, and S = the weight of sample taken, this would be expressed shortly as follows: 107.88 W X 100 A p -- = percent of silver in sample. (I) Instead of adding the hydrochloric acid in slight but indefinite excess to the solution of the silver salt, one might add exactly the amount required to complete the reaction, but no more. This would involve the use of some method for determining when the reaction is exactly completed (the "end point "), such as noting when another drop of hydrochloric acid solution fails to produce any further precipitation of silver chloride. Suppose, also, that the concentration of the hydrochloric acid solution, in grams per cubic centimeter, were very accurately known. We should then have the following data: (a) Weight of silver salt taken, (6) Volume of hydrochloric acid required to react with silver, (c) Concentration of hydrochloric acid. Just as the formula for silver chloride expresses the weight relations between silver, chlorine and silver chloride, so the VOLUMETRIC ANALYSIS 193 equation for the reaction between silver salt and hydrochloric acid expresses the weight relations between all of the elements and compounds involved. We are here particularly concerned with the relations between silver and hydrochloric acid, and we note that for every 107.88 parts, by weight, of silver, we require 36.468 parts of hydrochloric acid for complete precipitation of the silver as silver chloride. Conversely, if the reaction has been exactly completed, for every 36.468 parts of hydrochloric acid used there will have been present 107.88 parts of silver. The weight of pure hydrochloric acid used is found by multiplying the number of cubic centimeters by the concentration in grams per cubic centimeter, i.e., VC = Wt. HC1 used, where F = cc of acid solution used and C f = gm hydrochloric 107 88 acid in 1 cc. If this weight is multiplied by the fraction QA ob . 4o the result will be the weight of silver in the sample. Expressed briefly: 107.887CX100 36.46S percent silver. (II) This is the most general expression for the calculation of the results of a volumetric analysis. A comparison of expressions (I) and (II) will show that the volumetric calculation differs from the gravimetric calculation in two respects only: (1) A weight of a substance is obtained indirectly by measuring the volume of a solution of known concentration, instead of directly by weighing the substance. (2) The substance whose weight is desired is one which reacts with the substance being determined instead of one which is produced by this substance. The gravi- metric factor for the ratio of the weight of one substance to the weight of another substance which contains the first becomes the volumetric factor for the ratio of the weight of one substance to the chemically equivalent weight of another substance which does not contain the first. The substance which is a visible indication of the end point of the reaction is called the " indicator." Indicators will be discussed at length in a later section. The solution whose con- is 194 QUANTITATIVE ANALYSIS centration is accurately known and of which we measure the volume required, is called a " standard solution" because it is actually a standard by which the quantity of the substance under investigation is measured. The process of running in the stand- ard solution until the end point is reached is called "titration." The examples given below will serve to illustrate the principles above outlined: 1. 0.5436 gm of a silver salt was dissolved and titrated by a standard solution of hydrochloric acid, 1 cc of which contained 0.00304 gm of the pure acid. 27.2 cc of the standard was used. Required, the percent of silver in the salt. 27.2 X 0.00304 = gm of pure acid used; 107 88 Sfi 46 ^ m ^ hydrochloric acid = gm of silver present. 107.88X27.2X0.00304X100 Therefore 36.46X0.5436 = 45 - = percent Sl1 ' ver in the original sample of silver salt. Use of Aliquot Parts. It often happens that, in order to elimi- nate the error due to the lack of uniformity of a sample being analyzed, or for reasons of convenience, a larger quantity than is necessary for the titration is weighed and this dissolved in a definite quantity of solvent and an aliquot part taken for the titration. In such a case the final calculation must include an expression of this fact. Thus in Example (1) instead of weighing 0.5436 gm of the silver salt, suppose that 10.8720 gm was weighed, dissolved, and the solution diluted to 1000 cc, 50 cc being then titrated. The statement would then be 27.2X0.00304X107.88X20X100 10.8720X36.46 = percent Sllver m 2. In a sample of undried, but otherwise pure, sodium hydrox- ide, the percent of the base was to be determined. For this purpose 5.5310 gm of the sample was weighed and dissolved and diluted to 250 cc. Portions of 25 cc each were measured and titrated by a standard solution of hydrochloric acid, the average amount of acid solution required for complete neutrali- zation being 45.1 cc; 1 cc of the standard acid contained 0.00960 gm of the pure acid. Required, the percent of sodium hydroxide. " The solution of the problem is as follows: VOLUMETRIC ANALYSIS 195 45. 1X0. 0096 = gm of hydrochloric acid used; ' ' . Xgm of hydrochloric acid = gm of sodium hydroxide in 25 cc of solution. 250 -7^-Xgm of sodium hydroxide = gm in 5.5310 gm of sample. AO The condensed expression is 45.1X0.0096X40.008X250X100 "36.46X25X5. 5310 = 85. 89 = percent sodium hydroxide. The " equivalent weight" of a substance is the number of weight-units chemically equivalent to eight weight-units of oxygen. This definition is sufficiently broad to apply to any system of weights, although there are very few cases in scientific work where "grams" might not be substituted for "weight-units," since the metric system is quite universally accepted and used in scientific work. The result of this substitution is the "gram- equivalent." In the effort to determine what is the equivalent weight of a substitute it is always necessary to inspect the equation for the reaction that occurs in that particular case. In the reaction HC1 + NaOH-^NaCl + H 2 0, it is easily seen that the equivalent weights for all of the elements, radicals, ions or compounds are the atomic, radical, ionic or molecular weights, respectively. In the reaction: H 2 SO 4 +BaCl 2 -BaS0 4 +2HCl, the equivalent weights of the compounds are seen to be one- half of their molecular weights, with the exception of hydro- chloric acid, whose equivalent weight is its molecular weight. This will be more easily understood if we first determine the ''hydrogen equivalent" of each substance, this being the number of atoms of hydrogen chemically equivalent to one molecule, atom, radical, etc., of the substance under consideration, as denoted by the equation for the reaction that has taken place. The hydrogen equivalent of sulphuric acid is 2 because it reacts by substituting another element for two hydrogen atoms. That of barium chloride is 2, because one atom of a bivalent element gives place 196 QUANTITATIVE ANALYSIS to hydrogen or because 2 atoms of a univalent element give place to one radical which is bivalent. That of barium sulphate is 2 and that of hydrochloric acid is 1, for similar reasons. Since one atom of oxygen is chemically equivalent to two atoms of hydrogen, it follows that any body that is equivalent to 1.008 weight units of hydrogen will be equivalent to 8 weight units of oxygen and therefore the equivalent weight is the molecular (atomic, etc.] weight divided by the hydrogen equivalent. Problems Find the equivalent weights of the substances whose formulas are in bold face in the following equations : 4. 2HCl+Na 2 C0 3 2NaCl+H 2 CO 3 . 6. HCl+NaHCOs >NaCl+H 2 CO 3 . 6. HCl+Na 2 CO 8 >NaHCO,+NaCl. 7. (NH 4 ) 2 C 2 04 + CaCl 2 -CaC 2 O 4 +2NH 4 CL 8. (NH 4 ) 2 C 2 O 4 +HC1-NH 4 C1+NH4HC 2 O4. 9. (NH 4 )2C 2 04+2HC1->2NH 4 CH-H 2 C 2 O4. 10. FeCli+2AgNO,->Fe(NO,)+2AgCl. 11. 2FeCl 2 +Cl 2 2FeCl 3 . 12. H 2 SO 4 +Zn ZnSO 4 +H 2 . 13. H 2 S0 4 +2H->SO 2 +2H 2 O. 14. CuCl 2 +2AgNO 3 -Cu(NO 3 ) 2 +2AgCl. 16. 2CuCl 2 +Fe >2CuCl+FeCl 2 . Calculation of the Weight of One Substance, Chemically Equiv- alent to a Stated Weight of Another. In the solution of nearly all problems of quantitative chemistry there is involved a calcu- lation of the weight of one substance chemically equivalent to a given weight of another. In the examples just considered this is a calculation of the weight of silver or of sodium hydroxide equivalent to the weight of hydrochloric acid that is contained in the quantity of standard solution used. In example (1) this 1 07 88 was expressed as on 4^ FX 0.00304 and in example (2) as 36 46 ^X- QQ96 - I* i s eas i lv seen tnat both of these cal- culations involve the multiplication of the weight of hydro- chloric acid by a ratio of equivalent weights. From this fol- lows the rule that to find the weight of one substance chemically VOLUMETRIC ANALYSIS 197 equivalent to a stated weight of another, multiply the stated weight by the fraction: equivalent weight of substance calculated equivalent weight of substance given This is a simple and very useful rule and its application will obviate the use of the more cumbersome rule of proportions. Applied to gravimetric analysis the fraction given above is the factor. Problems 16. What weight of oxygen is equivalent to 0.3460 gm of hydrogen, direct oxidation to water being understood? 17. What weight of carbon dioxide is equivalent to 0.5693 gm of carbon, direct oxidation being understood? 18. Calculate the weight of ferric chloride and of iron equivalent to 0.5243 gm of chlorine, the following reaction taking place: 2Fe+3Cl 2 -2FeCl 3 . 19. Calculate the weight of potassium hydroxide equivalent to 1.7521 gm acetic acid, assuming complete neutralization. 20. In problems (4) to (15) calculate the weights of the substances whose formulas are in bold face, equivalent to 3 gm of the substances reacting with them. 21. A solution of hydrochloric acid of specific gravity 1.05 contains 10 percent by weight of the pure acid. What volume of the solution is required to precipitate the silver from 0.75 gm of silver sulphate? 22. 0.4321 gm of impure potassium sulphide was oxidized to potassium sulphate and precipitated by barium chloride. 0.8035 gm of barium sulphate was produced. What was the percent of sulphur in the sample? 23. What weight of tartaric acid is equivalent to 3.52 gm of sodium hydroxide, the following reaction taking place? Na 2 C4H4O6 + 2H 2 O. Use of a Standard Solution for the Titration of but One Sub- stance. When a standard solution is to be used for the titra- tion of but one substance the calculations will all involve the constants representing (1) equivalent weight of the active sub- stance in the standard solution, (2) equivalent weight of the substance to be determined and (3) concentration of the standard solution. If the standard solution of example (1) is to be used for 107.88X0.00304 tne determination ot silver, the expression : - ~a A& 198 QUANTITATIVE ANALYSIS contains quantities that are constants for all such determinations. These constants should then be combined in a single constant: 0.00899. From what was said in the preceding paragraph this is seen to be the weight of silver equivalent to 1 cc of this par- ticular standard solution of hydrochloric acid. All such calcu- lations would therefore be made by the expression 7X0.00899X100 ; p = percent silver (III) o This is obviously a very simple calculation and such simplifica- tion is possible and should be made whenever a given standard solution is to be used for a considerable number of determinations of a single substance. Burette Reading a Direct Percentage Reading. If some care were exercised in adjusting the weight of silver salt used for analysis where statement (III) is to enter the calculation, so that exactly 0.8990 gm of sample were used, the expression would become VX0.00899X100 0.8990 " = percent Sllver whence, V = percent silver. In this case the volume of standard solution used is the percent of silver in the sample. From this follows the rule: To make the burette reading a percentage reading first calculate the weight of the titrated substance that is equivalent to 1 cc of the standard solution, then use 100 times this weight of sample. Problems 24. What weight of soda ash must be used for analysis in order that 1 cc of the hydrochloric acid solution containing 0.0031 gm shall be equivalent to 1 percent of sodium carbonate, assuming complete decomposition? 26. A standard solution of sulphuric acid contains 40.2 gm in 1000 cc. What weight of potassium hydroxide must be taken so that each cubic centimeter of the standard acid required shall indicate 0.1 percent of potassium hydroxide? 26. A standard solution of barium hydroxide contains 20.35 gm in 1000 cc. What weight of vinegar is necessary in order that 1 cc of barium hydroxide solution shall indicate 0.1 percent of acetic acid in the vinegar? No System. In the examples given above there was no definite basis for the choice of the concentration of the standard solution, VOLUMETRIC ANALYSIS 199 all that was required being an accurate knowledge of the existing concentration. Thus, in the first example the standard hydro- chloric acid contained 0.00304 gm in 1 cc, while in the second it contained 0.00960 gm in 1 cc. There is no connectio , ap- parent or real, between these concentrations; they were chosen, at least to a certain extent, at random, upon the assumption that the determination of the concentration (standardization) was carried out with the greatest possible accuracy but that in making the solution no particular care was exercised. The method of calculating analyses made by means of such standard solutions would in all cases be analogous to these examples and any substance can be determined by means of such solu- tions, provided that the reaction involved is definite, complete and well understood, and that the end point can be determined accurately. Normal System. The calculations of volumetric analysis may be considerably shortened by the proper adjustment of the concentration of the standard solution. The reaction between hydrochloric acid and silver nitrate is expressed by the equation: HC1 + AgNO 3 -HNO 3 + AgCl, and expression (II) was deduced for the calculation of the per- cent of silver in a salt that had been titrated by a standard solution of hydrochloric acid. This expression was 107.887CX100 36.46S -= P^cent silver. By using the proper indicator in each case we might have completed such reactions as the following for the titration of the substances indicated : HC1 + NaOH-^NaCl + H 2 0, (a) HC1 + KOH->KC1 + H 2 O, (b) HC1 + NH 4 OH-+NH 4 C1 + H 2 0, (c) 2HCl+Na 2 CO 3 ^2NaCl+H 2 0+CO 2 , (d) HCl+NaHCO 3 -*Naa+H 2 O+C0 2 , (e) 2HCl+Ba(OH) 2 -*BaCl 2 +2H 2 0, (f) 2HCl+CaC0 3 ->CaCl 2 H-H 2 0+C0 2 . (g) 200 QUANTITATIVE ANALYSIS Many other substances may also be titrated by this same standard solution and in each case the expression for the per- cent of the substance to be calculated would be the same as (II) with the exception that, for the equivalent weight (combining weight) of silver (107.88) we should substitute the equivalent weight of the substance to be calculated; we should then have: 40.0087CX100 = percent NaOH (a') 36 46 56.1087CX100 percent KOH (b') 35.05 FCX100 = percent NH 4 OH (c') 36 17.03 7(7 -X 100 36.46S = Percent NH 3 53 7CX100 ,, - 36 46 ^ -- = percent Na 2 CO 3 (d') 84.0087CX100 , ,. 36 46 s -- = percent NaHCO 3 (e') 85.6937CX100 36 46 s -- 50.0457CX100 36 , ff . = percent Ba(OH) 2 f ,, = percent CaC0 3 (gO In all of these expressions for the percent of the various substances as titrated by a single standard solution, the only difference lies in the equivalent weight of the substance. The volume of standard required will depend, among other things, upon the purity of the sample and, since this is unknown, the volume required cannot be predicted. The concentration of the standard is under control and may be arbitrarily fixed at any desired figure. The equivalent weights concerned are constants, in any given case, and the weight of sample may be made whatever is desired. If the standard solution is made of such strength that the tumber of grams contained in 1000 cc will be represented by the equivalent weight (in the case of hydrochloric acid 36.46 gm) the concentration in grams per cubic centimeter will then be VOLUMETRIC ANALYSIS 201 Q 0.03646, and the fraction .> which is involved in all of the expressions, will become =0.001, so that we shall then have 0.04008FX100 = percent NaOH (ai) 0.0561087X100 ^ = percent KOH (bi) 0.035057X100 = percent NH 4 OH (ci) and so on. The standard solution of hydrochloric acid thus made, con- taining 1 gram-equivalent of the active substance in 1000 cc of solution, is a solution of general application and the calcula- tion of the results of analyses of various substances is simplified by this choice of concentration. Such a solution is called a " normal solution" which will be defined as a solution containing 1 gram-equivalent of the active substance in 1000 cc. From the foregoing discussion the following deductions may be made. 1. 1 cc of any normal solution is equivalent to one-thousandth of one gram-equivalent ( = one milHgram-equivalent) of any substance. This is because 1 cc of any normal solution contains one-thousandth of one gram-equivalent of the active substance.. 2. If the milligram-equivalent of the substance titrated in a given determination is multiplied by the number of cubic centi- meters of a normal solution used for the titration, the result is the weight of the former in the sample. This follows as a result of (1). 3. 1 cc of any normal solution is equivalent to 1 cc of any other normal solution. This also follows as a result of (1). 4. The relative volumes of various standard solutions equiva- lent to each other are inversely as the respective normalities of these solutions. Thus 6 cc of a fifth-normal solution is equivalent to 26 cc of a tenth-normal solution. These principles are very important and their intelligent ap- plication will serve to shorten many of the calculations of volumetric analysis. 202 QUANTITATIVE ANALYSIS It frequently happens that the normal solution is too concen- trated or too dilute for convenient use in a given analysis. In this case the advantage of the normal solution may be retained by making the concentration of the solution to be some simple multiple of the concentration of the normal solution, such as 2> 3, V, y~, -rpr, .TT^-T, etc. This factor must then be introduced O JLU OU 1UU into the calculations involving the solution. Solutions made oi normal or a simple multiple of normal strength are said to be made in the "normal system" and are, for the sake of brevity, N N N designated as N, 2N, , , , etc. Problems 27. 1 cc of normal acid is equivalent to what weight of ammonium car- bonate, assuming complete decomposition? N 28. 1.1256 gm of a silver alloy is dissolved and titrated by ^ potassium thiocyanate solution according to the following equation : KCNS + AgNO 3 ->KNO 3 + AgCNS. 35.2 cc of standard solution is required. What is the percent of silver in the alloy? N 29. 0.5 gm of limestone was dissolved in 50 cc of -= acid. The unused o N excess of acid was titrated by 16.2 cc of JK base. What was the percent of calcium carbonate in the limestone? What percent of calcium? N 30. 0.4 gm of soda ash was titrated by 20.9 cc of y acid. What was the percent of sodium carbonate in the sample? 31. 0.5' gm of an ammonium salt was decomposed by sodium hydroxide N and the resulting ammonia distilled into 50 cc of -^ acid solution. The N unused excess of acid was titrated by 29.3 cc of ^Q base. What percent of ammonia in the salt? Decimal System. Instead of using the normal system, a further simplification may be made by adjusting the standard until each cubic centimeter shall be equivalent, not, as in the normal system, to a decimal fraction of a gram-equivalent of the VOLUMETRIC ANALYSIS 203 substance to be titrated but to a decimal or simple fraction of a gram of the substance. For example, a solution of hydro- chloric acid would be made with each cubic centimeter equiva- lent to 0.0100 gm, 0.0010 gm, 0.0050 gm, etc., of silver. This results in a very much simplified calculation and still more time is saved if the weight of sample used bears a definite and simple relation to the equivalence of the standard. Such solutions as these are frequently made for technical work in industrial laboratories, where large quantities of standard solutions are often required for the titration of a single con- stituent of a large number of samples. Mention may be made of the use of potassium permanganate or potassium dichromate solutions for the titration of iron in ores, sodium thiosulphate solutions for the determination of the available chlorine in bleach- ing powder, potassium ferrocyanide solutions for the determina- tion of zinc and hydrochloric acid solutions for the determina- tion of hardness of water. The method of calculation of the necessary concentration of a solution to be made in the decimal system is the reverse of the method for calculating the equivalence of a solution of given concentration. Example: What must be the concentration of a solution of potassium hydroxide in order that each cubic centimeter shall be equivalent to 0.001 gm of sulphuric acid? The reaction involved is: 2KOH+H 2 S0 4 -K 2 S04+2H 2 0. The equivalent weight of potassium hydroxide is 56.108 and that of sulphuric acid is 49.02. Each cubic centimeter must con- rn - /-kQ tain * no X 0.001 gm of potassium hydroxide. This is *.\}.\jZi 0.00114 gm. Problems 32. Calculate the concentration ( ) of standard solutions of hydro- \ cc / chloric acid such that 1 cc=c=the following weights of other substances: 0.002 gm of silver; 0.005 gm of silver chloride; 0.010 gm of potassium hydroxide; 0.005 gm of sodium hydroxide; 0.002 gm of sodium; 0.002 gm of ammonia. 204 QUANTITATIVE ANALYSIS 33. Calculate the concentration of a nitric acid solution such that 1 cc=c= the following weights of substances: 0.040 gm of potassium hydroxide; 0.005 gm of calcium carbonate; 0.001 gm of nitrogen as ammonia. 34. What is the concentration of a potassium hydroxide solution of which 1 ceo 0.010 gm of potassium acid tartrate? Choice of System. Summarizing, it has been shown that volumetric analysis may be carried out by the use of standard solutions made in "no system," in the "normal system" or in the "decimal system," and that for any of these systems a defi- nite, precalculated weight of sample may be taken so that the burette reading in cubic centimeters will indicate directly the percent or simple fractions of percent of the constituent being determined. Which of these systems shall be selected in prac- tical work will be determined by the circumstances. If but a few titrations are to be made with a given standard solution the time saved in simplified calculations will not justify the expendi- ture of time required for adjusting the concentration to the nor- mal or decimal system. If many titrations are to be made, one of the latter two systems will be used. The normal system is most useful for standard acids and bases because their application is more general and a solution so made will give simplified cal- culations for the titration of many other substances. There are many standard solutions which are not to be used so generally but which are made for the titration of but one substance. In such instances the decimal system will always be used. Temperature Correction for Standard Solutions. It is often difficult to control the temperature of the laboratory within close limits and errors may thereby be introduced into volumetric determinations, due to changes in the density of standard solu- tions and in the capacity of measuring instruments when the temperature varies from 20. The following table, adapted from tables published by the Bureau of Standards, 1 indicates the corrections for water and for two concentrations of most of the common standard solutions of acids and bases. Approxi- mately these corrections will apply also to most other solutions of similar concentrations. 1 Bur. Stand., Circ. 19, table 33, VOLUMETRIC ANALYSIS 205 Correction, cc per observed liter to give Temperature, volume at 20 degrees N N Water 77: Solutions -^ Solutions 10 2 15 +0.8 +0.8 + 1.0 16 +0.6 +0.7 +0.8 17 +0.5 +0.5 +0.6 18 +0.3 +0.4 +0.4 19 +0.2 +0.2 +0.2 21 -0.2 -0.2 -0.2 22 -0.4 -0.4 -0.5 23 -0.6 -0.6 -0.7 24 -0.8 -0.9 -1.0 25 -1.0 -1.1 -1.3 It will be seen that a variation of 2, either way, from 20 involves an error of 0.04 percent in measurements of tenth- normal solutions, or 0.04 to 0.05 percent for half-normal solu- tions. This may be ignored for much of the routine work of the industrial laboratory but for more exact work the corrections should be applied. The use of standard acids and bases provides a means for the quantitative determination of practically any acid or base and of many salts. This is an extremely useful department of work, in view of the fact that no gravimetric method will serve to de- termine the essential constituent of acids and bases, the ionizable hydrogen and hydroxyl. For example, from potassium hy- droxide potassium may be determined as chlorplatinate or perchlorate, but this gives no information concerning the per- cent of potassium hydroxide since potassium from any salt present is also precipitated and weighed. By using the proper indicator salts of strong bases with weak acids or of strong acids with weak bases may also be titrated. Thus sodium carbonate may be titrated by standard hydrochloric or sulphuric . acid if methyl orange is used as indicator. Adjustment to Exact Concentration. In most of the exercises of the following pages the student is directed to adjust his standard solutions to the exact stated concentrations. If a 206 QUANTITATIVE ANALYSIS solution is desired to be tenth-normal, or perhaps of such con- centration that 1 cc is equivalent to 0.1 mg of phosphorus or 0.5 mg of tin, etc., it is first made to this approximate concentration. It is then standardized and finally adjusted to the required strength, with or without an additional standardization to con- firm the accuracy of the dilution. The only exceptions to this rule are in cases of solutions of unstable substances which change on standing. Of course this process of adjustment is, in some cases, a some- what tedious procedure and there is a too common custom of omitting the final adjustment, using a correction factor in the calculations of analyses made by means of this solution, this factor having been found by the first standardization. Thus, N if the first standardization showed a solution to be 1.0359-^, o all calculations of titrations would be made as though the solu- tion were fifth-normal, except that the factor 1.0359 would enter. This figure is known as a "normality factor." To the inexperienced analyst it may seem that this is, after all, the best method of dealing with the question that the use of such a factor requires much less work than is involved in the process of exact dilution and restandardization. But here again the question must be resolved with respect to the way in which the solution is to be used, as was done in the matter of choice of system. The work involved in adjusting the standard solution is balanced against the labor that is saved by simplified calculations. But the latter quantity is, of course, to be mul- tiplied by the number of determinations that will be made before the solution deteriorates and requires restandardization, or before the supply is exhausted. Obviously, this means that if any considerable use is to be made of a given standard solution it should be adjusted to the exact desired concentration. It may also be remarked that if adjustment is not to be made there is no logic in trying to work to either the normal or the decimal system. The method described on page 197 is far simpler in this case. CHAPTER VI COLOR CHANGE OF INDICATORS In a broad sense the word "indicator" applies to all substances which, by undergoing any visible change, indicate the end point of reactions. When the indicators are inorganic the reactions are usually definite and well understood. The indicators used in acidimetry and alkalimetry are organic and the direct cause of color change is, even now, not thoroughly understood. Many of the organic dyes show, in acids, a color different from that in bases. The color change is generally reversible an indefinite number of times. The molecular structure of the dye is often very complex and it is not easy to follow the changes in structure. Simple lonization Theory. Most or all of the indicators of this class are known to possess, in certain conditions at least, the prop- erties of acids or bases. The acid or basic nature is usually weakly emphasized. From this Ostwald deduced a theory as to the cause of color change. 1 According to this theory these dyes are, when uncombined with other acids or bases, weak electrolytes and largely in the molecular state. If a base is added to a weakly acid indicator, the salt is formed and this is highly ionized, ac- cording to the general rule. The molecule possesses one color (or is colorless), while the anion shows a different color. The result of the addition of a base is therefore a color change. If another acid is now added to the ionized salt the weak acid is reformed, the molecule reappears and the color change is reversed. The added acid has a further effect upon the indicator acid in suppress- ing the already small ionization. Similar reasoning would apply to basic indicators. Phenolphthalein is in the presence of acids, a derivative of phthalic anhydride and phenol having the following constitution : CO = (C 6 H 4 OH) 2 . 1 Scientific Foundations of Analytical Chemistry, 118. 207 208 QUANTITATIVE ANALYSIS According to the theory of Ostwald this is a very weak acid, giving a small concentration of ions thus: HPh 5 H + Ph, the symbol Ph representing the negative radical of the compound. The ionization constant is very small and equilibrium occurs with an inappreciable concentration of the anion. Upon the addition of a base the ionized acid is neutralized, equilibrium is disturbed and the ionized salt is produced, hence the color of the anion (red) appears. Methyl orange is known to have, under certain conditions, the structure This is an acid, the red molecular form predominating in acid solutions, and the yellow anion appearing in basic solutions. Theory of Chromophors. This explanation is not sufficient in itself for several reasons. The silver salt of phenolphthalein is intensely purple, even when dry, and the dry salt cannot be highly ionized. Ethers of tetrabromphenolphthalein have been prepared; 1 these are non-ionizable but colored. The monoethyl ether is xCO 2 C 2 H 5 r-C 6 H 2 Br 2 OH Litmus is known as both blue and red in the dry state, when it must be chiefly molecular, no matter what the color may be. Also the studies of recent years upon the constitution of organic dyes have shown that in many cases a change of molecular structure takes place upon the addition of an acid or a base. Phenolphthalein is known to have the structure shown above but in basic solutions there is a salt of a carboxyl acid which is a quinone derivative. The phenol derivative is then in equilibrium 1 Nietzki and Burckhardt: Ber., 30, 175 (1897). COLOR CHANGE OF INDICATORS 209 with the quinone derivative and this equilibrium is disturbed in one direction or the other by the addition of an acid or a base. CO COO C 6 H 4 / +H (C 6 H 4 OH) 2 C = C 6 H 4 = \C 6 H 4 OH If an acid is added the first (colorless) molecular form is produced because suppression of ionization results from the increase in concentration of hydrogen ions. If a base is added the ionized form is neutralized, forming ionized salt and water, and thus the new structure predominates. When methyl orange changes from the sulphonic acid to one of its salts a change of structure also takes place. The structure peculiar to the non-ionized body (present when an acid is added) is not that of an azo compound but one containing the quinone ring. There is then equilibrium between the two forms: (CH 3 )2N-C 6 H4-N = N-C e H 4 S0 3 H^ Yellow, predominates in basic solution. (CH 3 ) 2 N = C 6 H 4 = N-NH-C 6 H 4 SO a ! ___ . _ j Red, predominates in acid solution. The acid, by suppressing the ionization of the first form, causes the second, a lactonic form, to predominate. A base, by forming salt and water from the sulphonic acid, causes the first form to predominate. With the acid the quinone ring gives a red color. With the base the azo group gives a yellow color. The quinone ring, =C 6 H 4 = , is one of a class of groups known as "chromo- phors" because, wherever they appear in any compound they give rise to color. Other well characterized chromophors are the azo group N = N , the nitro group, N0 2 , and the dicarbonyl group, CO CO - 1 Classification of Organic Indicators. Ostwald classified the indicators according to their supposed dissociation constants into three groups: (a) Very weak bases and relatively strong acids. (b) Moderately strong acids and bases. (c) Very weak acids and relatively strong bases. 1 Vide Hantzsch: Ber., 32, 575 (1899), and Stieglitz: J. Am. Chem. Soc. f 26, 1112 (1903). 14 210 QUANTITATIVE ANALYSIS Since he explained the color change upon the basis of salt formation it would necessarily follow that the relative sensitive- ness would vary in the three classes. In class (a) the indicators would be highly sensitive to bases but not easily affected by acids, except by very strong ones. The indicators of class (b) would be moderately sensitive to both acids and bases, while in class (c) they would be highly sensitive to acids and to none but strong bases. While some of these indicators are here called " relatively strong" acids or bases, it must be remembered that, compared with the strongest electrolytes, all are weakly ionized and all lie in the class of weak electrolytes. While the theory of color change by ionization must be regarded as based upon incorrect assumptions, the above classi- fication is still a convenient one since the same relative sensi- tiveness would follow from the application of any of our theories. Phenolphthalein may be taken as an illustration. According to the simple ionization theory there is to be considered merely the following system in equilibrium : HPh^H+Ph, where Ph is understood to mean the negative radical C 6 H 4 (CO) 2 C 6 H 4 OH.C 6 H 4 0. Phenolphthalein falls in the class of very weak acids and it is consequently chiefly molecular unless a strong base be present, a weak base forming an easily hydrolyzed salt. Even weak acids can decompose the salt, therefore phenolphthalein will be easily affected by acids but will not be highly sensitive to bases. According to the view that color is due to the existence of chromophors the equation CO COO No <=* C e H 4 / +H = (C 6 H 4 OH)2 C CeH 4 O \C 6 H 4 OH is an expression of equilibrium between the non-ionized colorless form and the ionized form containing a chromophor. Here again a strong base will be required to produce the ion containing COLOR CHANGE OF INDICATORS 211 the chromophor, while a weak acid will reform the colorless molecule and for the same reasons that are given above. The classification according to ionization loses much of its significance when it is remembered that each indicator exists in at least two forms. On this account we shall rather lay stress upon the sensitiveness of the indicator toward acids and bases and shall so classify a few of the more commonly used indicators. Class I Highly sensitive to acids, less sensitive to bases Class II Moderately sensitive to acids and bases Class III Highly sensitive to bases, less sensitive to acids Phenolphthalein Rosolic acid Litmus p-nitrophenol Lacmoid Methyl orange. Ethyl orange. Cochineal. Erythrosine. Methyl red. This classification shows that an indicator which is highly sensitive to acids is, in a corresponding degree, weakly sensitive to bases. For this reason the only generally serviceable standard acid or base is a highly ionized one. Such a standard may be used for the titration of either weak or strong electrolytes, the selection of indicator being made with reference to the substance titrated rather than to the standard. Many of the indicators undergo a color change upon the addi- tion of certain salts, as well as of acids or bases. In all such cases the salt is one derived from an acid and a base of unequal degree of ionization. The result of the partial hydrolysis of such a salt is the production of an excess of ions of either hydrogen or hydroxyl, according to whether the acid or the base is the more strongly ionized. An indicator of the proper sensibility will be affected exactly as though the solution were that of an acid or a base. Thus sodium carbonate in solution yields, by partial hydrolysis, sodium hydroxide and bicarbonate. The difference in degree of ionization of these two electrolytes is so great that all indicators show the color that is ordinarily exhibited in basic solutions. On the other hand the hydrolysis of ferric chloride yields a strong acid and a weak base and here again the difference in ionization is sufficiently large to give an "acid reaction'' with 212 QUANTITATIVE ANALYSIS most indicators. The exact point at which the indicator will change color depends upon the relative strength of acid and base and also upon the sensibility of the indicator itself. If a solution of sodium carbonate containing methyl orange is titrated by hydrochloric acid the color change does not occur until the sodium carbonate is completely decomposed, according to the following equation : Na 2 C0 3 +2HC1^2NaCl+H 2 C0 3 . This is because methyl orange is practically insensible to such a weakly ionized acid as carbonic acid and a slight excess of hy- drochloric acid is necessary in order to affect the indicator. If phenolphthalein, an indicator of high sensibility to acids, is used instead of methyl orange the color change occurs when one-half of the reaction represented above is completed: Na 2 C0 3 +HCl^NaHC0 3 +NaCl. In other words sodium bicarbonate, yielding upon hydrolysis two equivalents of carbonic acid for one of sodium hydroxide, is a neutral body to an indicator that is easily affected by even weak acids and only with difficulty by bases. Orthophosphoric acid may be taken as a final example. The molecule of this sub- stance contains three atoms of ionizable hydrogen. Two of these atoms are ionized to but a small extent. If the acid is titrated by a solution of a strong base the point at which the color change occurs (" end-point ") will depend upon the indicator used. If the indicator is litmus the color changes from red to blue gradually instead of suddenly and this change comes after one-third and before two-thirds of the hydrogen is neutralized. In the pres- ence of methyl orange the color changes at the completion of the reaction: NaOH+H 3 P04->NaH 2 PO 4 +H 2 O. In the presence of phenolphthalein the end point is somewhat in- definite but occurs at the neutralization of two-thirds of the acid : 2NaOH + H 3 P0 4 -Na 2 HP0 4 + 2H 2 0. DESCRIPTION OF INDICATORS Following is a brief discussion of the preparation and properties of the indicators named in the table on page 211. COLOR CHANGE OF INDICATORS 213 Phenolphthalein. The chemical nature and changes of phenol- phthalein have already been discussed. The compound is pre- pared by heating together 5 parts of phthalic anhydride, 10 parts of phenol and 4 parts of concentrated sulphuric acid for several hours at a temperature between 120 and 130. The mass is then boiled with water, filtered and the residue dissolved in dilute sodium hydroxide solution. The solution is filtered and neutralized with hydrochloric acid. Phenolphthalein precipi- tates and is purified by recrystallization from alcohol. The pure substance is a yellowish-white crystalline powder, practically insoluble in water but soluble in alcohol. For use in volumetric analysis a solution of 5 gm in 1000 cc of 50 percent alcohol is used. Rosolic Acid (Corallin). Commercial rosolic acid is a mix- ture of aurine, CigHuOs, oxyaurine, Ci 9 Hi 6 O 6 , methylaurine, CsoHieOs and pseudorosolic acid, CaoHieO^ Each of these sub- stances contains the quinone ring. It is prepared by heating together phenol, sulphuric acid and oxalic acid. The changes occurring as acids or bases are added are not thoroughly under- stood. The solution as used in the laboratory is a 1 percent solution in 60 percent alcohol. The indicator is red with bases and yellow with acids. It is highly sensitive to acids and can be used for the titration of weak acids. Litmus. Litmus is obtained by the action of ammonium hydroxide and potassium hydroxide upon certain species of plants, followed by fermentation. The essential constituent of the indicator is azolitmin, C 7 H 7 N04, of unknown constitution. It is colored red by acids and blue by bases. Its sensitiveness toward both acids and bases is only moderate and this fact makes it a very valuable indicator for general qualitative purposes but of little use for quantitative analysis. A 10 percent solution in water is used in the laboratory. p-Nitrophenol. This indicator is prepared by the action of /N0 2 (4) nitric acid upon phenol. Its formula,C 6 H 4 <( X)H (1), is indicated by its name. It is yellow with bases and colorless with acids. It is applicable to the same uses as is litmus. The indicator solution is a 0.02 percent solution in water. It should not be kept in a closely stoppered bottle. 214 QUANTITATIVE ANALYSIS Methyl Orange. The constitution and chemical properties have already been discussed. The substance as obtained in commerce is usually the sodium salt. This is a yellow powder, soluble in water. A solution containing 0.5 gm in 1000 cc is used. In presence of ammonium salts the indicator is not very sensitive. Also the color is destroyed by iron or aluminium salts. Ethyl Orange. This substance is analogous in constitution to methyl orange, it being the diethyl ester of amidoazobenzene- sulphonic acid. Its properties and uses are the same as those of methyl orange. Cochineal. Cochineal is the dried female insect Coccus cacti Linne". The essential coloring matter is carminic acid, CiiHi 2 7 whose constitution is not definitely established. The solution for use- as an indicator is made by digesting 3 gm of the dried, unpowdered insects with 250 cc of 75 percent alcohol until the coloring matter is extracted, then decanting. The bottles containing the solution should not be closely stoppered. The indicator is violet with bases and red with acids. Its sensitive- ness is not diminished by ammonium salts. Lacmoid. Lacmoid is prepared by heating together resorcin, sodium nitrite and water. It is a deep blue dye of unknown constitution; the molecular composition is probably represented by the formula Ci 2 H 9 N04. It is soluble in alcohol and less so in water. In solution it is colored blue by bases and red by acids. Its sensitiveness toward both acids and bases is moderate but it finds an application in the titration of carbonates in boiling solution, carbonic acid being decomposed as fast as it is formed. Litmus may be used in the same way but is not so sensitive in hot solutions. The indicator solution should contain 2 gm of purified lacmoid in 1000 cc of 50 percent alcohol. Erythrosine (lodeosine). This dye is tetraiodofluorescein, a synthetic derivative of fluorescein, and it has the constitution represented by the formula CO C 6 HI 2 OH >0 6 HI 2 OH. COLOR CHANGE OF INDICATORS 215 The substance is therefore a phthalein. It is almost insoluble in water but is soluble in hot alcohol and ether. A solution of 0.5 gm of the sodium salt in 1000 cc of recently boiled water is used as indicator. It is colored red by bases and light yellow by acids. It is highly sensitive to bases and is therefore appli- cable to the titration of the alkaloids. Its sensitiveness to acids is correspondingly small so that it may be used for the titration of carbonates of the alkali and alkaline earth metals without boiling to expel carbonic acid. On account of its limited solu- bility titrations are made by adding 10 to 20 cc of the solution in ether to the titrating solution, shaking after each addition of acid. Methyl Red. This dye is p-dimethylaminoazobenzene-o- carboxylic acid: The indicator solution is prepared by dissolving 1 gm of the solid in 100 cc of 95 percent alcohol. The solution is pale yellow in basic solutions and violet red with acids. It belongs to Class III and is especially good for the titration of ammonium hydroxide and the alkaloids, all being weak bases. It cannot be used if much carbonic acid is present, hence is useless for the titration of carbonates. All indicator solutions must be adjusted to exact neutrality before using. CHAPTER VII STANDARDIZATION Thus far we have dealt with only the calculation of the results of analyses, assuming that the standard solution was ready for use in the experiment. The determination of the concentration of the standard solution is called " standardization." The details of the experimental work will be treated later and will be mentioned here only so far as they may serve to illustrate the methods of calculation. Standardization may be accomplished by one or more of four methods : Direct Weighing. The active substance of the solution is accurately weighed and dissolved to make a definite volume of solution. This method is applicable to only those substances that can be obtained in a pure state or in a state of uniform and accurately known composition. Most of such substances are crystallized salts or acids, or soluble gases. Weighing a Substance Produced by a Measured Volume of the Solution. Sulphuric acid solution may be standardized by precipitating a measured volume by adding an excess of barium chloride. From the weight of barium sulphate found the weight of sulphuric acid may be calculated by the method given on page 196. Similarly hydrochloric acid solution may be standardized by precipitating as silver chloride. Measuring the Volume of Solution Required to React with a Known Weight of a Substance of Known Purity. An acid may be allowed to react with a pure carbonate and the required volume noted. Sodium thiosulphate solution may likewise be titrated against a weighed quantity of iodine (or indirectly) against a weighed quantity of arsenic trioxide. Titration Against Another Solution Which has Already Been Standardized. This is a very common expedient. 216 STANDARDIZATION 217 Primary Standards. It will be noticed that in each of these cases there is some substance of known composition that is measured or weighed and the solution is somehow compared with this for standardization. This- substance of known com- position is called the " primary standard," whether it be the substance dissolved in the solution, something produced by the solution or something reacting with the solution. The following examples will illustrate the methods of calcula- tion in each of the cases discussed. (1) The method of calculation for the first method of stand- ardization is self-evident. The normality is equal to the ratio of the number of grams dissolved in 1000 cc to the number of grams in 1000 cc of a normal solution. That is, . _ grams per 1000 cc ~ equivalent weight (2) A solution of hydrochloric acid was standardized by precipitating the chlorine from 40 cc, as silver chloride. The weight of silver chloride found was 0.6327 gm. Required, the normality of the solution. 0.6327 1 cc acid solution =0= gm silver chloride. 1 cc normal acid solution^ 0.1433 gm silver chloride. Therefore normality = '^ 27 -* . 1433 = = 0. 1 107 N. To make the solution decinormal 1000 cc would be diluted to 1107 cc. (3) A similar solution was standardized by titration of pure sodium carbonate in presence of methyl orange, the following reaction being completed: Na 2 CO 3 +2HCl->2NaCl+H 2 CO3. It was found that 32.2 cc acid o 0.1638 gm of the primary standard, sodium carbonate. Required the normality. 0.1638 ' 1 cc acldo ~32~2~ gm sodium carbonate and 1 cc normal acido 0.053 gm sodium carbonate. Therefore normality =^^53 =0.9598 N. 218 QUANTITATIVE ANALYSIS (4) Another acid solution was standardized by titration against a measured volume of standard potassium hydroxide solution in presence of methyl orange according to the equation: HC1+KOH-^KC1+H 2 0. 1 cc of the primary standard contained 0.00468 gm of potas- sium hydroxide. It was found from the titration that 50 cc of potassium hydroxide solution 43. 5 cc of hydrochloric acid solution. The weight of potassium hydroxide in 50 cc of solution = 50X0.00468 gm. Since this weight was equivalent to 43.5 cc of acid, the potassium hydroxide equivalent to 1 cc acid = 50X0.00468 43.5 50X0.00468 gm. The normality of the hydrochloric acid solu- 0.095 N. 43.5X0.0561 In case the primary standard is a solution already standardized in the normal system the normalities of the solutions are inversely as the respective volumes that are equivalent to each other. N (5) 30.0 cc of ^r sodium thiosulphate solution is found by titration to be equivalent to 29.8 cc of iodine solution. The normality of the latter is required. If solutions are to be standardized in the decimal system the calculations involve nothing more than finding the weight of the substance in terms of which the standardization is to be ex- pressed, equivalent to 1 cc of the solution which is being stand- ardized, always using as the starting point the known weight of the primary standard and following the method explained on page 196. In many cases the standardization is to be expressed in terms of the primary standard itself. For example, iodine solution is to be standardized against pure arsenic trioxide and expressed in terms of the same substance. Here we have the very simple method of weighing a suitable amount of arsenic trioxide, then dissolving and titrating by the iodine solution. Then gm As 2 Q 3 1 cc iodine solution =o - I _ Bolution ' STANDARDIZATION 219 Other familiar examples of this class of methods are the standardization of permanganate solutions against elementary iron or antimony for obtaining the weights of these elements equivalent to one cubic centimeter of the solution. The following example will serve to illustrate the first case just discussed: (6) A solution of potassium permanganate was standardized against sodium oxalate as follows: 2.5340 gm of sodium oxalate was dissolved and the solution was diluted to 1000 cc. 25 cc portions were titrated and gave an average of 24.25 cc of potas- sium permanganate solution equivalent to the oxalate solution used. Required the weight of iron and of calcium equivalent to 1 cc of permanganate solution. 25 cc of the oxalate solution contained 0.025X2.5340 gm 0.025X2.5340 and 1 cc of permanganate solution is equivalent to . _ gm of sodium oxalate. This weight, multiplied by the ratio of the equivalent weight of iron or of calcium to that of sodium oxalate, will give the weights of these substances that are equivalent to 1 cc of the standard solution. Then 0.025X2.5340X55.88 Ice solution^ 24.25X67.005 =0 - 00218 S m Fe or 0.025X2.5340X20.035 24.25X67.QQT~ =0-00078 gm Ca. Problems 35. 30.0 cc of sulphuric acid solution yields 0.3625 gm of barium sul- phate. Calculate the normality and the dilution necessary to make the solution tenth-normal. 36. 44.6 cc of silver nitrate solution yields 1.2870 gm of silver chloride. Calculate the normality and the dilution necessary to make the solution fifth-normal. 37. 39.7 cc of barium chloride solution yields 2.5346 gm of barium chromate. Calculate the normality and dilute to make the solution half-normal. 38. An acid solution is standardized by titrating pure sodium carbonate, using methyl orange as indicator. 45.1 cc=c= 2.4065 gm of sodium car- bonate. Calculate the normality and dilution necessary to make the solu- tion exactly normal. 220 QUANTITATIVE ANALYSIS 39. 50 cc oi nitric acid solution is added to 0.4530 gm of pure calcium carbonate. The unused excess of acid is titrated by a solution of a base and 6.15 cc of the latter is required. The base is then titrated against the acid in order to compare their concentrations and 21.3 cc of acid is found to be equivalent to 19.2 cc of base. Calculate the dilution necessary to make each solution fifth-normal. 40. A sulphuric acid solution is standardized by titrating a sample of potassium bicarbonate which contains 98.45 percent of the pure compound. 35 cc acid =0= 0.0391 gm of the sample. Calculate the normality and the dilution necessary to make the solution hundredth-normal. 41. 32.9 cc of potassium hydroxide solution exactly neutralizes 0.3118 gm of pure potassium acid tartrate, KHX^H^Oe- Calculate the dilution necessary to make the solution twentieth-normal. 42. 45.9 cc of sodium hydroxide solution was added to a solution con- taining 0.25 gm of crystallized oxalic acid, H 2 C 2 O4.2H 2 O, the indicator being phenolphthalein. The excess of base was titrated by 1.3 cc of acid. 4.9 cc of the acid is found to be equivalent to 50 cc of base. Calculate the normality of each solution. 43. A solution of barium hydroxide was standardized by titration against succinic acid, H 2 C4H4C>4, in presence of phenolphthalein. 20.9 cc of barium hydroxide solution neutralized a solution of 1.22 gm of succinic acid. Cal- culate the normality. 44. 38.1 cc of a sodium bicarbonate solution exactly neutralizes 36.7 cc of a tenth-normal solution of hydrochloric acid. What is the normality of the first solution? CHAPTER VIII EXPERIMENTAL VOLUMETRIC ANALYSIS Standard Acids. It has already been shown that the most serviceable acids or bases for standard solutions are those that belong to the class of strong electrolytes. For standard acids this practically limits one to the use of hydrochloric, sulphuric or nitric acid. On account of the ease with which it may be re- duced, nitric acid is not to be recommended for standard solu- tions of general application. Of the first two acids named, hy- drochloric acid is usually to be preferred because it is monobasic and cannot form acid salts. Materials for Standardization. For standardizing an acid use may be made of any method which involves a definite re- action with a pure substance or which produces a precipitate or gas that may be weighed or measured. This makes either volumetric or gravimetric methods available. Since most strong bases are hygroscopic and also combine readily with carbon di- oxide, purification and weighing are difficult and these substances are unsuitable for use as primary standards for standardizing acid solutions. We have left the alkali and alkaline earth car- bonates for this purpose. Calcium carbonate and sodium car- bonate are suitable and both substances may be obtained nearly pure. Precipitated calcium carbonate may be used but it is seldom free from other salts and an analysis must be made be- fore it is used for standardizing. The form most often used is the natural crystallized calcite known as Iceland spar, which is often nearly 100 percent calcium carbonate, although its purity should not be assumed without analysis. The chief disadvan- tage in the use of any form of calcium carbonate for standardizing acids lies in the fact that it is insoluble in water and an excess of acid must be used in order to hasten the process of solution. In such a case a direct titration cannot be made but the excess of acid must be determined by titration with a solution of a 221 222 QUANTITATIVE ANALYSIS base, whose concentration as compared with the acid must be known. Several of these difficulties may be obviated by the use of sodium carbonate, which is soluble in water. As this substance is obtained in commerce it contains variable quantities of water and salts incident to the process of manufacture, such as sodium sulphate and sodium chloride. This is largely due to the com- paratively large solubility of the salt and the consequent diffi- culty in purifying it by crystallization. At 20 a saturated solution contains sodium carbonate to the extent of 17.7 per- cent of its weight. The pure salt is best obtained from the bicarbonate by heating. Sodium bicarbonate dissolves to the extent of 8.8 percent of the weight of solution at 20. It may therefore be more readily purified by fractional crystallization, especially if the purest obtainable commercial salt is used for the purpose. When heated sodium bicarbonate decomposes according to the equation 2NaHCO 3 -^Na 2 C0 3 +H 2 O+CO 2 . The dissociation tension of carbon dioxide from sodium bicar- bonate is as follows: 1 Temperature 55 60 70 80 90 100 Tension mm. of mercury 19 25 43 70 125 310 The tension of carbon dioxide in the atmosphere is less than 1 mm and, in consequence, sodium bicarbonate will slowly de- compose at temperatures below 55. At 100 the decomposition is fairly rapid and the bicarbonate is completely changed into normal carbonate by heating for a short time at about 300. At still higher temperatures the normal carbonate will yield some sodium oxide and carbon dioxide. Gravimetric Standardization. Acids may also be standardized gravimetrically in case insoluble salts can be produced. Such a method will apply to hydrochloric or sulphuric acid but not to nitric acid, since no insoluble nitrate is known. A point fre- quently overlooked is that this method is really a standardiza- tion with respect to the negative radical and is an acid standardi- Lescoeur: Ann. chim. phys. [6] 28, 423 (1892). EXPERIMENTAL VOLUMETRIC ANALYSIS 223 zation only in case no salts of the acid are present. Even in the purest commercial acids ammonium salts are often present and the weighing of silver chloride or of barium sulphate will thus not give a basis for the correct calculation of acid strength. Standardization by Direct Weighing. It is possible to weigh the active substances directly in the exact amount necessary for a solution of desired strength only in case the substance is available in pure form. This is not the case with most of the inorganic acids 1 and with comparatively few salts. It then becomes necessary to calculate the approximate quantity re- quired to make a solution somewhat stronger than that desired, to standardize the solution so made and dilute exactly to the required strength. Such dilution may be accomplished with accuracy in case the water to be added may be measured in a burette, i.e., if less than about 10 cc is required. The final volume obtained by dilution is the sum of the initial volume and that of the added water only in case no expansion or contrac- tion occurs upon mixing. This is practically the case if the solu- tion is already dilute and the relative amount of water to be added is small. Dilution may then be accomplished by measuring a specified amount of solution in a flask, running in the calculated amount of water from a burette and mixing directly in the gradu- ated flask. The neck of the flask must be capable of easily holding the required water above the graduation. This fact, together with considerations of volume changes already men- tioned, places a practical limit upon the amount of dilution that may be accurately made in one process. If more than 10 cc of water must be added to 1000 cc of solution it is necessary to dilute to nearly the required amount, restandardize and redilute to the exact value required. For example, if it is found N that a solution is 1.3462 X TT: and it is necessary to dilute the solution to make it exactly tenth-normal, each 1000 cc must be diluted to a volume of 1346.2 cc. The addition of 346.2 cc of water could not be accurately made because there is no gradu- ated vessel capable of accurately measuring this quantity. Such an addition might also cause an appreciable volume change. 1 Vide, Moody: J. Chem. Soc., 73, 658 (1898), and Acree: Am. Chem. J., 36, 117 (1906). 224 QUANTITATIVE ANALYSIS The correct procedure would be to add first to each 1000 cc of solution about 335 cc of water, measured in a graduated cylinder, restandardize and then carefully complete the dilution in the manner already explained. Exercise: Preparation of Pure Sodium Carbonate Use the besi grade of sodium bicarbonate that is obtainable. Make qualitative tests for sulphates, chlorides and potassium, using an approximately 5 percent solution of the material for these tests. If impurities are found, purtfy by recrystallization as follows: Make a saturated solution of sodium bicarbonate by warming the purest obtainable salt with distilled water. Decant from any undissolved matter remaining and evaporate the solution in a large porcelain or platinum dish, at a temperature not higher than 40, until crystals begin to separate. Allow to cool and stand, uncovered but in a place which is protected from dust, until about 25 gm of crystals have formed, pour off the solution, wash the crystals once with cold water and press between filter paper. Dry at 100, powder and heat in a platinum crucible at a temperature between 270 and 300 until the weight is constant. The product should be pure sodium carbonate but a test for sulphates and chlorides should be made. Preserve in a closely stoppered weighing bottle. Exercise: Preparation of Tenth-normal Hydrochloric Acid. Deter- mine the specific gravity (if not already known) and from this the per- cent of hydrochloric acid in the concentrated solution found in the laboratory. From the data so obtained calculate the weight or volume necessary to make 2500 cc of tenth-normal solution. Measure 2 percent more than this amount into a 1000 cc graduated flask and fill to the mark with water. Empty into a bottle having a capacity of about 2500 cc and add 1500 cc more of water. Stopper and mix thoroughly by shaking. Since the solution has been warmed by the reaction between acid and water it should be allowed to stand until the temperature of the room is attained before standardizing. Calculate the weight of sodium carbonate necessary to make 250 cc of a tenth-normal solution. Weigh this quantity of the prepared pure material on counterpoised glasses, then brush into a beaker. Dissolve the weighed carbonate in distilled water and carefully rinse into a 250 cc graduated flask. Fill to the mark and mix thoroughly. Imperfect mixing is often found to be the source of discrepancies in titrations with the acid solution. The solution will not remain constant in concentra- tion and should not be kept for more than one day. Fill a burette with the solution and another with the acid solution. Before making the titrations practice reading the color change as follows: Place 100 cc of distilled water in a beaker and add a drop of methyl orange and 0.5 cc EXPERIMENTAL VOLUMETRIC ANALYSIS 225 of sodium carbonate solution. Drop in the acid solution until the last drop changes the tint from yellow to pink. Now drop in sodium car- bonate solution until the yellow color reappears. Repeat the alternate additions of carbonate and acid until the color change can be observed when but one drop of either solution is added. It will aid in the next process if this solution is preserved and another prepared, the two showing the two colors of methyl orange. These may be set aside for comparison. Measure 40 cc of the carbonate solution into another beaker or Erlenmeyer flask, dilute to about 100 cc and titrate with the acid solution in presence of a drop of methyl orange. Calculate the normality of the solution, also the volume of water to be added to each 1000 cc to make exactly tenth-normal. If water to be added is more than 10 cc add nearly the required amount to each liter of the acid, mix, and restandardize. If the quantity to be added is less than 10 cc the acid is diluted as follows: Fill a dry 1000 cc graduated flask to the mark with the acid solution. This flask should be capable of holding the required amount of water above the mark. From a burette add the calculated quantity of water directly to the solution in the flask and mix thoroughly. Pour into a dry bottle and make a second liter of diluted solution in the same manner, having first rinsed and dried the graduated flask. Check the accuracy of the dilution by restandardization. Record upon the label of the bottle your name, the came of the standard solution, its normality and the date of standardization, thus : Hydrochloric Acid N/10 Feb. 6, 1913 FIG. 65. Form of label for standard solution. SODA ASH The commercial grade of sodium carbonate known as "soda ash" contains, in addition to sodium carbonate, considerable water, some potassium carbonate and varying quantities of other sodium and potassium salts, such as sulphates and chlorides. A complete analysis would include the determination of all radicals 15 226 QUANTITATIVE ANALYSIS but on account of the fact that soda ash is used in many indus- tries because of its basic properties its valuation generally in- cludes a determination of basicity and of water with other impurities. The first determination may be made by direct titration by a standard .acid in the presence of methyl orange, while the last may be determined directly or taken by difference, in which case the percent of water includes all other impurities. In view of the fact that the basicity of the alkali carbonates toward methyl orange is due to the hydrolysis of a salt of a strong base and a weak acid, it is obvious that any salt derived similarly will likewise be basic and that therefore the titration by acid is really a method for determining the radical of salts of weak acids and is not a basis for the calculation of any particular salt, such as sodium carbonate. Thus, for example, if a mixture of sodium carbonate, potassium carbonate, sodium bicarbonale, sodium silicate, and sodium borate were being titrated these salts would all be decomposed by the standard acid before an end point with methyl orange would be reached. In the absence of a more extended analysis it would be impossible to calculate the percent of any one of these radicals or compounds. Since all are basic in solution and since all would serve for most purposes where sodium carbonate is used industrially it is customary to report the percent, arbitrarily calculated, of either sodium carbonate or sodium oxide (regarded as being combined), assuming that all basicity of soda ash is due to sodium carbonate. It should also be noted that sodium carbonate could contain either alkali hydroxides or bicarbonates but not both in the same sample. Ordinarily no attention is given to either. On account of the lack of uniformity of most commercial soda ash it is necessary to select a rather large sample, dissolving in water and measuring an aliquot part for the titration. The directions for sampling on pages 9 to 14 should be carefully followed but exposure to air should not be unduly prolonged. Determination. Fill a 20 cc weighing bottle with soda ash, properly sampled. Assuming that the sample is pure sodium carbonate, calcu- late the approximate weight that should be contained in 500 cc of solution so that 25 cc shall require about 35 cc of tenth-normal acid for its titration. Weigh this quantity of soda ash into a graduated 500 cc flask, dissolve and dilute to the mark with water. Mix thor- EXPERIMENTAL VOLUMETRIC ANALYSIS 227 oughly by inverting the flask and shaking. Measure out 25 cc by means of a calibrated pipette, allowing this portion to run into 200 cc breakers or Erlenmeyer flasks. Add just enough methyl orange to tint the solu- tion and then tirate with the tenth-normal acid. Make at least two more titrations and calculate the percent of sodium carbonate, also of sodium oxide, assuming that the basicity of the substance is entirely due to the sodium oxide combined as carbonate. "Pearl ash" (crude potassium carbonate) may be evaluated in a similar manner. MIXTURES OF CARBONATES AND BASES The use of the two indicators, methyl orange and phenol- phthalein, provides a means for the determination of carbonates and bicarbonates when in mixture, and also of carbonates and soluble bases. Bases and bicarbonates (acid salts) cannot occur in the same mixture. If phenolphthalein is added to a solution containing sodium carbonate and sodium hydroxide and the solution is titrated by a standard acid, the end point is reached when the sodium hydroxide is neutralized and the sodium car- bonate is changed to bicarbonate: NaOH+ HCl-NaCl+ H 2 O, (1) Na 2 CO3+HCl-NaHCO+NaCl. (2) Since the solution has now become colorless, methyl orange may be added and the titration continued until the red tint appears, when the sodium bicarbonate has been completely decomposed: NaHCO 3 + HCl-NaCl+ H 2 CO 3 . (3) Represent by A the cubic centimeters of acid used in completing the first titration and by B that used in the second. #X normality XO. 106 = gm of sodium carbonate, (A B)X normality X 0.040 = gm of sodium hydroxide. It must be noted that here the equivalent weight of sodium carbon- ate is 106 instead of 53, since by equation (3) IHCl^ !NaHCO 3 and by equation (2) lNaHCO 3 -lNa 2 C0 3 . The quantitative conversion of sodium carbonate into sodium bicarbonate by means of an acid can take place only when care is 228 QUANTITATIVE ANALYSIS taken to prevent the escape of carbon dioxide. At the point where the acid enters the solution there is at first complete con- version of a part of the carbonate into the normal sodium salt of the added acid. If carbon dioxide escapes from the solution at this point more acid will be required to produce the first end point than would otherwise be the case. The escape of gas may be prevented by keeping the solution at a temperature near and by adding the standard acid very slowly and while stirring vigorously. 1 It is very difficult to avoid the escape of carbonic acid unless the relative amount of carbonate is small. On this account the method is best suited to bases, in which carbonate occurs as an impurity, rather than to materials in which carbonate is an essential constituent. Determination of Hydroxide and Carbonate in Commercial "Caustic Soda" or "Caustic Potash." Arbitrarily assuming that the sample is pure sodium or potassium hydroxide, calculate the approximate quan- tity necessary to dissolve and dilute to 1000 cc so that 50 cc shall require about 40 cc of tenth-normal acid. From a stoppered weighing bottle weigh this amount of well-mixed sample into a 1000 cc graduated flask. Dissolve in 500 cc of water and cool to 20. Dilute to the mark, mix, measure out 50 cc by means of a pipette, add a drop of phenolphthalein, cool to in ice water and titrate to the disappearance of the pink color. Add a drop of methyl orange and continue the titration to the next color change. Calculate the percent of sodium carbonate and of sodium hydroxide. MIXTURES OF CARBONATES AND BICARBONATES If a mixture of a carbonate and bicarbonate is to be investi- gated the procedure is the same as in the preceding exercise. In this case the phenolphthalein changes color at the completion of the reaction Na 2 CO 3 +HCl->NaHCO 3 +NaCl. When the color change of methyl orange occurs the sodium bi- carbonate so produced, as well as that originally present, has been decomposed. If A is the acid used in the first titration and B that used in the second titration A X normality X 0.1 06 = gm of sodium carbonate, and (BA) X normality X 0.084 = gm of sodium bicarbonate. 1 Kuster: Z. anorg. Chem., 13, 127 (1897). EXPERIMENTAL VOLUMETRIC ANALYSIS 229 These calculations are based upon the same arbitrary assumption regarding the presence of other salts as is noted in the discussion of the valuation of soda ash. Determination of Sodium Carbonate and Bicarbonate in a Mixture. Proceed as in the preceding exercise, except that the calculation is to be made as above indicated. A thermometer must be placed in the solu- tion and the temperature lowered to 0. The standard solution is added very slowly. HARDNESS AND ALKALINITY OF WATER The difference between "hard" and "soft" water lies in the fact that the former contains various inorganic compounds which form insoluble soaps when used with soap for cleansing. The salts of calcium and magnesium are chiefly responsible for this action. When ordinary soap is dissolved in such a water there is at once formed a precipitate of calcium and magnesium salts of the fatty acids. Temporary Hardness. Bicarbonates of the metals named above are quite common in ground waters. When the water is boiled the excess of carbonic acid is expelled and normal car- bonates are formed: Ca(HCO 3 )2-CaCO 3 +H 2 0+CO 2 ; Mg(HCO 3 ) 2 ^MgCO 3 +H 2 O+C0 2 . Because of the very limited solubility of the normal carbonates a precipitate is formed and the hardness of the water is diminished to this extent. Such hardness as is removed in this manner by boiling is known as "temporary hardness" and the water so treated is partly "softened." Permanent Hardness is due to the slight amount of normal carbonates (about 0.013 gm per liter at 20) that remains to saturate the water, and to non-decomposable, soluble salts of calcium, magnesium, iron or other metals that may form in- soluble soaps of fatty acids. The most common of such com- pounds are chlorides, sulphates and nitrates. Alkalinity. The "alkalinity" of a natural water represents its content of carbonate, bicarbonate, borate, silicate, phosphate and hydroxide, or such of these as may be present. It is obvious that hydroxides and bicarbonates cannot exist in the same solu- 230 QUANTITATIVE ANALYSIS tion. Also a water which contains sodium or potassium carbon- ate is not likely to contain calcium, magnesium or iron as bi- carbonates, because reactions like the following would occur: Ca(HCO 3 ) 2 +Na 2 C0 3 ^CaCO 3 +2NaHC0 3 . Methods for the Determination. If none of the other salts mentioned as causing alkalinity were present a simple titration with standard acid in presence of methyl orange or erythrosine would serve as a determination of bicarbonate hardness. Also if one could neglect the fact that carbonates are not entirely precipi- tated when formed by heating bicarbonates, such a determination of bicarbonate hardness could be reported as of temporary hardness. Because neither of these conditions is fulfilled the most accurate method for the determination of temporary hard- ness is by making the titration in presence of methyl orange before and after boiling. Permanent hardness equals non-carbonate hardness plus hard- ness due to a saturated solution of calcium carbonate, mag- nesium carbonate, etc., as already explained. If a water should have no temporary hardness its non-carbonate hardness and permanent hardness would be identical. Non-carbonate hard- ness is best determined by the use of "soda reagent/' a standard solution of equal weights of sodium hydroxide and sodium carbonate. A measured quantity of the water is boiled to decompose bicarbonates and to this is added the standard solu- tion in excess. Reactions like the following take place: CaS0 4 +Na 2 C0 3 -*CaCO 3 +Na 2 S0 4 ; MgCl 2 +Na 2 C0 3 ->MgCO 3 +2NaCl. After filtering to remove insoluble carbonates the solution is titrated by standard acid in presence of methyl orange and the hardness is calculated from the amount of soda reagent that has been used by reactions like those represented above. The use of sodium hydroxide in the standard solution appears to diminish the solubility of the normal carbonates that are formed. Clark's method for the determination of total hardness is based upon the use of a standard soap solution, added until a permanent EXPERIMENTAL VOLUMETRIC ANALYSIS 231 lather is produced. It was formerly extensively used but it is inaccurate and is now little used. Expression of Results. The small percent of dissolved salts usually found makes desirable a method of expressing results which is different from that used in other connections. Instead of percent it is customary to report parts per hundred thousand of water, parts per million or grains per gallon. The Imperial English gallon of water weighs 70,000 grains while the United States gallon weighs in air at 15.5 and 760 mm pressure, 58335+ grains. This gives at least four different systems which have been at various times and in various countries commonly used for the expression of the results of water analysis. Hardness itself has also been expressed in Clark's degrees (grains of calcium carbonate per Imperial gallon), German degrees (parts of calcium oxide per 100,000 parts of water) and French degrees (parts of calcium carbonate per 100,000 parts of water). This has re- sulted in much confusion but there is now a general tendency toward the practice of expression in parts per million or, more exactly, milligrams per liter, although in industrial operations the report is often made as grains per United States gallon. Upon the assumption that one liter of water, at the usual working temperature, weighs 1000 gm, every milligram of dissolved matter will represent one part per million of water. This as- sumption is not quite correct and "milligrams per liter" is a better expression than "parts per million." The following exercises may be performed at this point by students who will not carry out a more extensive analysis of water later. (See page 339.) Determination of Alkalinity. Prepare a fifth-normal solution of sulphuric acid, foPowing the general directions given on page 224 for tenth-normal hydrochloric acid and standardizing against pure sodium carbonate in presence of methyl orange. Dilute 100 cc of this fifth-normal solution to 1000 cc with recently boiled and cooled distilled water. The work of standardization and dilution must be done very carefully in order to avoid the necessity for ^standardization of the last solution. . Potassium acid sulphate may be substituted for sulphuric acid in this determination. As this substance usually contains considerable water due allowance should be made when calculating the required weight for the fifth-normal solution. 232 QUANTITATIVE ANALYSIS Measure 100 cc of the water sample in a volumetric flask and rinse into a porcelain dish or casserole with recently boiled water ; or measure the sample directly into the porcelain dish by means of a pipette. Add 2 drops of methyl orange indicator and titrate with the fiftieth-normal acid solution already prepared. Calculate the alkalinity in the conventional way as milligrams of calcium carbonate per liter. Determination of Temporary Hardness. Determine the alkalinity of the original sample as already directed. Measure a second portion of 200 cc of the sample and rinse into a 500 cc Erlenmeyer flask. Boil gently for 10 minutes, cover and cool to room temperature and immedi- ately rinse into the 200 cc volumetric flask. Dilute to the mark with recently boiled and cooled distilled water and mix thoroughly. It is important to have distilled water which is quite free from carbonic acid for this determination, as otherwise part of the precipitated carbon- ates will be redissolved and later titrated as alkalinity. On this account the wash bottle must not be used in the ordinary manner *by blowing into it. Stopper the flask containing the boiled sample and allow to stand until the precipitate has settled, then remove 100 cc of the clear liquid by means of a pipette and redetermine the alkalinity. The difference (if any) between the two titrations represents the temporary hardness. While temporary hardness is due to bicarbonates of calcium, mag- nesium or iron (and sometimes other metals of the earth or alkaline- earth groups) it is customary to calculate it in terms of calcium carbon- ate, the most common of the products of boiling hard water. The convenience of a fiftieth-normal acid for this use should be noted. Since the equivalent weight of calcium carbonate is almost exactly 50, 1 cc of fiftieth-normal acid is equivalent to 1 mg of calcium carbonate. Report the temporary hardness in milligrams of calcium carbonate per liter. Determination of Non-carbonate Hardness. Prepare an approxi- mately tenth-normal (to methyl orange) solution of soda reagent, using equal weights of sodium carbonate and sodium hydroxide, standardizing against tenth normal hydrochloric acid and using recently boiled water for dilutions during the titration. Fill a dry 200 cc volumetric flask (or one that has been rinsed with the sample) with the sample of water. Rinse this into a 500 cc Erlenmeyer flask of resistance glass, using 200 cc of distilled water. Boil for 15 minutes to expel carbon dioxide then add exactly 25 cc of standard soda reagent from a pipette. Boil for 10 minutes, rinse into the 200 cc volumetric flask and dilute to the mark, using recently boiled and cooled distilled water. Filter through a dry filter and discard the first 50 cc of nitrate. From the filtrate that is subsequently obtained measure EXPERIMENTAL VOLUMETRIC ANALYSIS 233 50 cc portions by means of a pipette into Erlenmeyer flasks and titrate at once with fiftieth-normal sulphuric acid or potassium acid sulphate, using methyl orange or erythrosine. If erythrosine is used, add 1 cc of the indicator solution and 5 cc of neutral chloroform, titrating until the pink color just disappears from the chloroform when violently shaken. Calculate the non-carbonate hardness in the conventional manner by assuming the typical calcium sulphate as the hardness-giving agent and reporting milligrams of this compound per liter. In waters of the "alkali" type, containing carbonates of sodium or potassium, a negative value will be found for non-carbonate hardness, i.e., more standard acid will be required after the treatment above outlined than is equiva- lent to the soda reagent added. Standard Bases. Standard solutions of bases are subject to change in basic concentration and must be frequently restand- ardized. This is because glass is appreciably soluble in bases and the accumulation of alkali silicates in the solution gives an increase in basicity toward all indicators. Bases also absorb carbon dioxide when exposed to the air and this results in a decreased basicity toward indicators of the class of phenol- phthalein. For this reason it is desirable to avoid unnecessary contact with the air after standardization. For the preparation of standard basic solutions free from carbonates one may either use a substance whose carbonate is insoluble or one which contains little carbonate and whose carbonate may be precipitated by the addition of another sub- stance. For the first method barium hydroxide is generally used. This base always contains some barium carbonate but this remains undissolved when the solution is made. If, thereafter, carbon dioxide is absorbed by the solution, an equivalent amount of barium carbonate is precipitated and the solution remains free from carbonate, although it must be restandardized. Sodium hydroxide or potassium hydroxide may be obtained nearly free from carbonates by dissolving in alcohol, decanting or filtering from undissolved carbonate and evaporating the alcohol in an atmosphere that is free from carbon dioxide. Bases so prepared may now be obtained from the manufacturers and should be used for the preparation of standard solutions whenever possible. 234 QUANTITATIVE ANALYSIS For the second method potassium hydroxide or sodium hy- droxide may be used and a slight excess of barium chloride added to the solution, barium carbonate being thereby precipitated. The material used should be the sticks that have been crystallized from alcohol. As sodium and potassium carbonates are only FIG. 66. Burette with three-way stopcock, connected with reagent bottle. slightly soluble in alcohol the bases from this solution are nearly free from carbonates. In either case reabsorption of carbon dioxide is prevented by passing entering air through a soda lime tube, removing the solution through a siphon directly to the burette, as in Fig. 66. Covering the solution with a layer of EXPERIMENTAL VOLUMETRIC ANALYSIS 235 toluene or. of any oily substance is not to be recommended because this results in fouling the burette. It should be understood that these precautions are unneces- sary in most titrations. While it is true that alkali carbonates have not the same basicity toward some of the indicators as have the alkali hydroxides, a proper correction is made by standardiz- ing in presence of the indicator that is to be used in the determina- tions. This is a principle that must always be observed. One must not, for instance, use the standardization in presence of methyl orange as a basis for the calculation of determinations made in presence of phenolphthalein. Selection of Base for Standard Solutions. The proper base for a standard solution will depend upon the nature of the titra- tion to be made. For reasons discussed on page 211 one will generally select a highly ionized substance and from this stand- point sodium hydroxide and potassium hydroxide are about equally good. On account of the relative cheapness of the former it should be given preference, wherever possible. However, in certain cases (see analysis of edible oils, beginning on page 345) a standard solution of a base is used for the saponification of oils. On account of the greater solubility of potassium soaps, potassium hydroxide is best for this purpose. Standardization. The standardization of solutions of bases may be accomplished by indirect methods, titrating against a previously standardized acid solution, or by direct methods, titrating against a weighed substance of known purity. In general the first method is to be recommended because the availa- ble solid acids of uniform purity are limited to the organic acids. This makes necessary the use of phenolphthalein in the standardization, which limits the use of solutions so standardized to determinations that can be made by means of this indicator. Some solid substances that may be used as primary standards are oxalic acid, H 2 C 2 04.2H 2 O, benzoic acid, HC 7 H 5 2 and succinic acid, H 2 C 4 H 4 04. No direct gravimetric method can be used because there is no precipitating reagent for the hydroxyl radical and a determination of the metal would not give the basic strength because of the invariable presence of salts of the same metal. Exercise : Preparation of Tenth-normal Sodium Hydroxide Solution. Select the best grade of sodium hydroxide obtainable, that purified 236 QUANTITATIVE ANALYSIS by alcohol being preferable. Calculate the weight necessary for 2500 cc of tenth-normal solution and weigh out the quantity with 2 percent added to compensate for carbonates, water, and other impurities. Dis- solve in a 2500 cc bottle having a solid rubber stopper. Fill the bottle with recently boiled and cooled distilled water, mix thoroughly and -allow to cool to the temperature of the room. Titrate portions of about 30 cc each against the tenth-normal acid, using methyl orange as indicator. Continue these practice titrations until the results agree closely. Dilute to make the solution exactly tenth-normal, using recently boiled and cooled water. Restandardize against the acid, using methyl orange, then use, in turn, phenolphthalein, methyl red, cochineal and lacmoid, as well as such other indicators as are available. The indi- cators of the first class (page 211) will indicate a weaker base than those of the third class, on account of the presence of small amounts of carbonates in the standard solution. Record the normality of the basic solution according to each indicator and use the proper figure for sub- sequent titrations, according to the indicator there used. Determination of the Concentration of the Laboratory Acids. Use the "dilute" acids, either sulphuric, hydrochloric, nitric or acetic. The sample should not be weighed in the analytical balance and it is better to determine the specific gravity and then take a measured volume. Determine the specific gravity with an accurate hydrometer. From the approximately known percent of acid in the solution calculate the dilu- tion required to make a solution approximately equivalent in concentra- tion to the standard base. Make 500 cc of solution, measuring accu- rately the volumes used. Titrate the finally diluted solution against the standard base, using methyl orange for any of the acids above named except acetic acid. For this use phenolphthalein. Calculate the per- cent of acid, by weight, in the original solution. If tables are at hand compare the results of the experiment with the percent corresponding to the specific gravity found. Determination of the Purity of Citric Acid. Weigh about 3 gm of commercial citric acid, dissolve and dilute to 500 cc. Titrate portions of 30 cc to 40 cc by the standard base, using phenolphthalein. Calculate the percent of the tribasic acid HsCeHsOT.H^O assuming that no other acid is present. VINEGAR Vinegar contains from 3 to 6 percent of acetic acid, in addi- tion to coloring matter, dissolved solids and sometimes unfer- mented sugar or alcohol. Cider vinegar contains also from EXPERIMENTAL VOLUMETRIC ANALYSIS 237 0.08 to 0.16 percent of malic acid. Vinegar is sometimes adul- terated with other added acids, notably sulphuric acid. The complete analysis will include the determination of the sub- stances just mentioned and others that serve to characterize the vinegar with respect to its origin or quality. The determination of total acidity alone is of value in determining the '' strength " of the vinegar. This is practically due to acetic acid alone in pure vinegars other than those made from cider. In cider vinegar the determination of malic acid is also of importance as indicating its origin. Determination of Total Acidity of Vinegar. Weigh a clean, stoppered weighing bottle of at least 25 cc capacity. Add about 25 cc of vinegar, stopper and reweigh with an accuracy of 1 mg. Carefully transfer this sample to a 100 cc volumetric flask, using a stirring rod. Rinse the bottle and rod with recently boiled water and dilute the vinegar and rinsings to the mark on the flask. Mix thoroughly. By means of a pipette measure two or three separate portions to Erlenmeyer flasks, dilute to 50 cc, add a drop of phenolphthalein and titrate with standard base. Calculate the total acidity as percent of acetic acid, using the normality of the base as determined in presence of phenolphthalein. BORIC ACID Boric acid is one of the weakest of all inorganic acids, having a percentage ionization not far above that of hydrocyanic acid. It is therefore not possible to titrate it -by standard bases in ordinary solution because no indicator will give a sudden color change at any point in its neutralization. It can be determined by titrating in presence of glycerine, phenolphthalein being used as indicator. 1 Hydroxylated organic compounds form ester- like compounds with the various boric acids, the result being substances more strongly ionized than boric acid. Tartaric acid forms such substances, with a resultant change of its optical rotatory power. With glycerine and orthoboric acid the com- pound C 3 H 5 (OH).HBO 3 is probably formed because the substance reacts as a monobasic acid. Boric acid combined as salts may be determined by first adding, in presence of methyl orange, enough hydrochloric or sulphuric acid to completely decompose the 1 Thomson: J. Soc. Chem. Ind., 12, 432 (1893). 238 QUANTITATIVE ANALYSIS borate. In case carbonates are present carbonic acid is also produced and this must be expelled by boiling because phenol- phthalein is to be used in the last titration. If the more common biborates are thus decomposed they yield orthoboric acid, methyl orange indicating the end-point at the completion of the reaction : Na 2 B40 7 +5H 2 O+2HCl->2NaCl+4H 3 B03. The addition of glycerine produces the reaction: The monobasic acid is then neutralized in the titration: Commercial borax is essentially crystallized sodium biborate, Na 2 B 4 07.10H 2 0. It loses water rapidly and seldom* corresponds to this formula. Determination. Crush and mix a sample of a borate quickly. Weigh about 10 gm, dissolve and dilute to 500 cc. Titrate measured portions o f 20 cc with tenth-normal acid in presence of methyl orange, first diluting the solution to 50 cc. Now add to other diluted portions of the borate solution the exact amount of acid indicated by the titration but no indi- cator. Boil for a few minutes, then cool. Add a drop of phenolphthal- ein and 30 cc of glycerine. The glycerine must be neutral. Titrate with standard base to the appearance of a pink color. Add 10 cc more of glycerine and if the pink disappears add more base. Continue the addition of glycerine and standard base until a permanent end-point is attained. Calculate the percent of boron trioxide, B 2 03, in the original borate. Reference to the formula for the complex of boric acid and glycerine will show that the equivalent weight of boron trioxide must be one-half its molecular weight. Use of Two Standards. There are cases where a direct titra- tion cannot be made conveniently and where two standard solutions may be used with advantage. If the titrated substance is nearly insoluble in water and dissolves only when the standard solution is added a direct titration is a very tedious operation. This is due to the fact that it would not be permissible to add more standard than is chemically equivalent to the substance with which the standard reacts, so that the active mass of the standard is, necessarily, always small, particularly toward the end of the titration, where it approaches zero. The velocity of solution EXPERIMENTAL VOLUMETRIC ANALYSIS 239 would therefore become extremely low, this also approaching zero toward the end of the titration. For this reason it is much simpler to add an excess of the standard solution, the total quantity used being measured accurately. The active mass of the standard is then relatively large throughout the operation and solution of the sample pro- ceeds fairly rapidly. After solution is complete the unused excess of standard is titrated by means of a second standard solution. This gives sufficient data for the calculation. The volume used of the second standard is calculated to the equivalent volume of the first. This figure is then subtracted from the total volume used of the first standard and the remainder is the volume used by the titrated substance. A familiar example of the use of two standards is in the titra- tion of the acid-neutralizing power of limestone for agricultural purposes. The first standard is hydrochloric acid and the second is either sodium hydroxide or potassium hydroxide. A tenth-normal solution of acid is not active enough for this purpose and a fifth-normal solution should be prepared. The preparation and standardization of fifth-normal solutions will follow the same lines as that of tenth-normal solutions, with which the student is already familiar. "Alkalinity" of Limestone for Agricultural Purposes. Have the sample finely powdered. Weigh exactly 0.5-gm samples on counter- poised glasses and brush into Erlenmeyer flasks. Moisten with water then pipette 75 cc of fifth-normal hydrochloric acid into each flask. After effervescence has become slow, connect with a reflux condenser and heat to boiling. Boil for 1 minute to expel carbon dioxide, then cool and rinse down the condenser, the stopper and the upper part of the flask. Add a drop of methyl red or phenolphthalein and titrate the unused excess of acid with fifth-normal base. (In standardizing the base against the acid, use must be made of the same indicator as is chosen for the determination.) Calculate the percent of calcium carbonate in the limestone. This will give a fictitious value for dolomitic limestones, which contain magnesium carbonate having an equivalent weight of 42, as com- pared with 50 for calcium carbonate. In such cases the calculated percent of calcium carbonate may be more than 100. CHAPTER IX OXIDATION AND REDUCTION Reactions of neutralization constitute a very important group in quantitative chemistry. Of no less importance is that group which is composed of reactions of oxidation and reduction. These do not involve the use of the organic indicator but of some inorganic substance, this being, in some cases, the standard reagent itself. While the quantitative relations Between the reacting substances are here different from those of neutraliza- tion, the same principles will serve for the calculation of the results of the titration. The equivalent weights will be deter- mined by dividing the molecular weights by the hydrogen equivalents but the latter will be determined, not by the valence of the reacting parts but by the change of valence in the reaction, because oxidation always involves a change of valence of the oxidizing and reducing agents. In the reaction: BaCl 2 +H 2 SO 4 ->BaSO4+2HCl the hydrogen equivalent of each radical is plainly represented by its valence, for it is in this measure that it may enter into the exchange of double decomposition. In the reaction: 2KMn04+10FeS0 4 +8H 2 S0 4 -^K 2 SO4+2MnSO4+ 5Fe 2 (SO 4 )3+8H 2 the hydrogen equivalents of potassium permanganate and ferrous sulphate cannot be so represented because something that is different from ordinary double decomposition has taken place. Manganese has left a negative radical and entered a positive one and has thereby changed its valence and has been reduced. The change of valence will represent its power as an oxidizing agent. Iron has taken on more of the radical with which it was already combined, has increased its valence and has been oxidized. It 240 OXIDATION AND REDUCTION 241 is not difficult to determine the change of valence of manganese if its state of oxidation in the two compounds is first determined. Apparent Valence. For the purposes of this inspection the question of the actual valence need not be considered. There has been much difference of opinion regarding the structural formulas that should represent inorganic compounds and the real valence of an element in an oxide is indicated by the structural formula of that oxide. For example, the dioxide of manganese might be represented by the 1) formula Mn^ , in which the valence of manganese is 4, or by O ,0 Mn^ | in which the valence is 2. If the element is regarded as X always being in direct combination with the oxygen the latter may be taken as a measure of the apparent valence and the loss of oxygen or its equivalent in other elements, as the result of a reaction, will be measured by the apparent change of valence. This change will therefore be the hydrogen equivalent of either oxidizing or reducing agent. The apparent valence of an element which is in the negative radical of an oxyacid or of its salt is found by subtracting the total valence of the positive radical from the total valence of the other element of the negative radical and dividing the result by the number of atoms of the element in question. Thus the apparent valence of manganese in potassium permanganate is (4X2) 1 = 7; that of chromium in potassium dichromate is (7X2)-2 _ 6 When the element in question forms a simple radical its valence is easily seen from an inspection of the formula. The valence of chromium in chromium chloride is 3, that of manganese in man- ganous sulphate is 2, that of iron in ferric sulphate is 3, etc. In the reaction involving the reduction of manganese from a permanganate to a manganous salt, the apparent valence changes from 7 to 2 and the hydrogen equivalent of the compound is therefore 5. The equivalent weight of potassium permanganate, containing 16 242 QUANTITATIVE ANALYSIS one atom of manganese in each molecule, is one-fifth of its molecu- lar weight or 31.606. Iron is oxidized in the reaction. Its appar- ent change of valence is from 2 to 3, its hydrogen equivalent is therefore 1 and its equivalent weight is its atomic weight. In the reaction: K 2 Cr 2 O 7 + 6FeCl 2 + 14HC1->2KC1 + 2CrCl 3 + 6FeCl 3 + 7H 2 0, potassium dichromate is the oxidizing agent. Closer examina- tion shows that the element chromium is the real cause of the oxidizing action, since it leaves the negative radical by changing to a base-forming lower oxide. The change of valence (reduction) of chromium is from 6 to 3, its hydrogen equivalent is 3, and that of potassium dichromate, containing two atoms of chromium in the molecule is 6. The equivalent weight of potassium dichro- mate is then one-sixth of its molecular weight. The assignment of apparent valence to the oxidizing and reduc- ing elements is a valuable aid in balancing oxidation and reduc- tion equations. For example the equation last given is to be balanced. The empirical equation is K 2 Cr 2 O7+FeCl 2 +HCl-^KCH-CrC] 3 +FeCl3+H 2 0. First inspect the equation to determine the oxidizing and reducing elements, which are seen to be chromium and iron. Determine the changes in apparent valence and from these their hydrogen equivalents and those of the compounds containing them. Write these above the respective compounds thus: VI I K 2 Cr 2 O 7 +FeCl 2 -f-HCl->KCl+CrCl3+FeCl 3 4-H 2 O. The hydrogen equivalent of the reducing agent is to be the coefficient of the oxidizing agent, and vice versa. These coeffi- cients will not thereafter be changed. From these will follow the coefficients of potassium chloride, chromium chloride and ferric chloride and from all of these will be calculated the coefficient of hydrochloric acid From the latter will follow the coefficient of water. The balanced equation is then: K 2 Cr 2 O 7 + 6FeCl 2 + 14HC1-*2KC1 + 2CrCl 3 + 6FeCl 3 + 7H 2 0. OXIDATION AND REDUCTION 243 Problems 45. Determine the equivalent weights of the oxidizing and reducing agents in the following equations and balance the latter: (a) KMnO 4 +MnSO 4 +KOH^K 2 SO 4 + MnO 2 +H 2 O. (b) K 2 Cr 2 O 7 +SnCl 2 +HC1-KC1 +SnCl 4 +CrCl 3 +H 2 O. (c) H 3 AsO 3 +I 2 +H 2 O-H 3 AsO 4 +HI. (d) HgCl 2 +SnCl 2 -HgCH-SnCl 4 . (e) HgCl 8 +SnCl 2 -Hg-+SnCl4. 46. How much potassium permanganate must be contained in 1000 cc of solution so that 1 cc shall be equivalent to 0.002 gm of iron? To 0.002 gm of manganese? 47. How much potassium dichromate must be contained in 1000 cc of solution so that 1 cc shall be equivalent to 0.005 gm of iron? 48. A solution of potassium permanganate contains 2.83 gm in 1000 cc. What weight of pure ferrous ammonium sulphate, FeSO 4 (NH 4 ) 2 SO 4 .6H 2 O, should be taken for standardization so as to require 30 cc of the permanga- nate solution? 49. A solution of potassium dichromate is of such concentration that 1 cc =0=0.005 gm iron. A solution of potassium permanganate contains 3.26 gm of the salt in 1000 cc of solution. To what volume should 1000 cc of the stronger solution be diluted to make the two equivalent to the same weight of iron? 50. What weight of iodine is required for 1000 cc of a tenth-normal oxidiz- ing solution? 51. What weight of arsenic trioxide is equivalent to 1 cc of tenth-normal iodine solution? Potassium Permanganate. This substance is a very con- venient one for use as a standard oxidizing agent, not only because it is readily and quantitatively reduced by many reducing agents, but because it serves as its own indicator. The color of even a dilute solution is quite intense, while the reduced salt, man- ganous sulphate, forms colorless solutions. The least excess, after the oxidation is finished, is made evident by the color of the unchanged permanganate. Practically, standard potassium permanganate is most often used for the determination of iron, less frequently for the determination of manganese, and occa- sionally for the determination of tin or antimony and indirectly for the determination of phosphorus, calcium and many thero elements. Theoretically it could be used for the direct or indirect determination of any reducing agent or oxidizing agent but other methods are frequently better for substances other than those named 244 QUANTITATIVE ANALYSIS A solution of potassium permanganate, if made by simply dissolving the salt in water, decomposes slowly and must be fre- quently restandardized. Morse 1 has shown that this is due to the presence of manganese dioxide, traces of which may have been contained in the original salt or it may have been formed by reduction of potassium permanganate by dust or organic impuri- ties. The work of Morse shows that if the solution is filtered through asbestos, and is kept in clean, glass-stoppered bottles its concentration will remain constant almost indefinitely. IRON Iron may be readily and quickly titrated by a solution of po- tassium permanganate. For this purpose the standard solution should be made in the decimal system. The normal system is not adapted to this determination because the potassium per- manganate will not often be used for the determination of sub- stances other than iron and the decimal system provides more sim- ple calculations. If the iron compound, as an ore, is taken in such quantity that the weight of sample is a simple multiple of the iron equivalent of the standard, the burette reading is a direct percentage reading. Thus, if 1 cc of standard is equivalent to 0.002 gm of iron and if 2 gm of iron ore is taken for analysis each cubic centimeter of standard represents 0.1 percent of iron. Reduction of Permanganate by Chlorides. One serious disadvantage in the use of potassium permanganate for the titration of iron lies in the fact that if hydrochloric acid is pres- ent, it also is oxidized by the standard, the result being a ficti- tious value for the percent of iron. But most iron ores dissolve best in hydrochloric acid, the solubility in sulphuric acid being very slight. After solution of the ore is accomplished it is neces- sary to remove the hydrochloric acid or to find some method for avoiding its reducing action upon the permanganate. To re- move the acid, the solution of the ore may be evaporated with sulphuric acid. Use of Manganous Sulphate. Instead of removing the hydrochloric acid it is possible to minimize its reducing action by the addition of manganous sulphate to the solution. In 1 Am. Chem. J., 18, 401 (1896); 20, 521 (1898). OXIDATION AND REDUCTION 245 order to understand the remarkable action of manganous salts in preventing the oxidation of chlorides in dilute solutions it is necessary to examine more closely the reaction between potassium permanganate and ferrous chloride. This reaction, occurring in an acid solution, is usually represented as follows: KMn0 4 +5FeCl 2 4-8HCl->KCl+MnCl 2 +5FeCl 3 +4H 2 O. (1) This reaction really takes place in at least two stages: KMnO 4 +4FeCl 2 +8HCl^KCl+MnCl 3 +4FeCl 3 +4H 2 O, (2) MnCl 3 +FeCl 2 -+MnCl 2 +FeCl 3 . (3) Equations (2) and (3) are seen by inspection to indicate the same end products as equation (1) and if no side reactions of any sort took place the ordinary calculations would not be affected by the occurrence of two stages in the reaction. But manganese trichloride is a very unstable compound and if not reduced at once by ferrous salts in the immediate vicinity (a condition that especially maintains near the end of the titration) this compound proceeds to decompose spontaneously: 2MnCl 3 +2MnCl 2 +Cl 2 . (4) Again, if this liberated chlorine could react quantitatively with ferrous chloride: 2FeCl 2 +Cl 2 -+2FeCl 3 (5) there would be no error involved in the titration for equations (2), (4) and (5), in sequence, are still equivalent to equation (1). This condition does not obtain. Reaction (5) is slow and un- certain and one of the reacting bodies (chlorine) is a gas which readily escapes. Therefore, according to Barneby, 1 the function of manganous salts, added in large quantities, is to increase the concentration of the manganous ion, thus obstructing the progress of reaction (4). By this means the spontaneous decomposition of manganese trichloride is prevented and it is reduced only by contact with ferrous chloride, with which it reacts very readily. Observation of End Point. As ferrous chloride becomes oxidized a yellow color appears in the solution, deepening to red as the concentration of the ferric salt increases. Experience 1 J. Am. Chem. Soc., 36, 1429 (1914). 246 QUANTITATIVE ANALYSIS shows that this renders difficult or impossible the observation of the first faint pink of the permanganate anion, which constitutes the indication of the end point of the reaction. There is no corresponding color in the case of sulphates of iron and this goes to show that the color of ferric chloride solutions is not due to the ferric ion but rather to hydrolysis. Partial or complete hydrolysis may yield any or all of the products indicated by these equations. FeCl 3 + H 2 0->FeOHCl a + HC1, FeOHCl 2 + H 2 O->Fe (OH) 2 C1 + HC1, Fe(OH) a Cl+H 8 0->Fe(OH) 8 +HCl. The basic chlorides or ferric hydroxide form colloidal solutions (sols) which are colored. As these reactions are reversible the addition of an excess of hydrochloric acid hinders hydrolysis and lightens the color but it has already been shown that a large excess of this acid is undesirable because of its reducing action upon permanganate. Reinhardt 1 showed that the formation of color can be prevented by the addition of phos- phoric acid to the solution. Evidently, hydrolysis is here prevented by diminishing the concentration of one of the re- acting ions (ferric) through the formation of the phosphate, which has a very small ionization. FeCl 3 +H 3 PO 4 ->FeP04+3HCl. The "Zimmermann-Reinhardt solution" contains manganous sulphate, phosphoric acid and sulphuric acid and it is added to the largely diluted ferrous chloride solution just before titration of the latter by standard permanganate solution. Primary Standards. Potassium permanganate cannot be obtained in a sufficiently high state of purity to make possible direct weighing of the substance for standardization. Primary standard reducing agents of known and uniform purity must be used. Some reducing agents that may be used for this purpose are iron, ferrous ammonium sulphate, oxalic acid and ammonium oxalate. Iron. Pure iron is not obtainable commercially and can be prepared only with great difficulty. Iron wire may be obtained 1 Chem. Zg., 13, 323 (1889). OXIDATION AND REDUCTION 247 having a fairly high degree of purity. A common fallacy has found acceptance by many chemists, to the effect that the so- called " piano wire," purchased for standardizing, is uniformly 99.6 percent or some other stated percent of iron. The purity of such wire varies widely and this material cannot be used as a primary standard for accurate work unless it has been carefully analyzed. Such analysis is best accomplished by electrolysis. The gravimetric determination of iron by precipitation as ferric hydroxide and weighing as ferric oxide is not accurate unless made with extreme care, on account of the difficulty experienced in purification of the precipitate. If iron is to be used for standard- izing solutions, weighed portions of the previously analyzed material are dissolved in dilute sulphuric acid. The necessary limitations as to accuracy render the use of iron as a primary standard decidedly unsatisfactory, especially since it has become possible to obtain iron salts in a high state of purity. Ferrous Ammonium Sulphate. Ferrous ammonium sulphate ("Mohr's salt") may be purified very readily by crystallization and preserved without oxidation, if kept dry. Many analyses conducted upon high-grade samples have shown that the salt may be obtained in a state of practical purity, the percent of iron being almost exactly that calculated from the formula. Oxidation in air does not readily occur, so that the standard sample may be preserved almost indefinitely if kept in a stop- pered bottle. Sodium Oxalate. This is one of the best primary standards now available for standardizing oxidizing solutions. The reaction with potassium permanganate is represented thus: 5Na 2 C 2 O 4 +2KMnO4+8H 2 S04->K 2 SO4+2MnSO4+5Na 2 SO 4 + 10C0 2 +8H 2 The salt may be purified by ordinary methods of recrystallization from water solutions or by the addition of alcohol to rather concentrated water solutions. By the latter method crystal- lization is more rapid and the crystals are consequently finer. At temperatures between 240 and 250 the crystals may be dried completely and without decomposition. 1 The solid -does not contain water of crystallization. The standardization of potas- 1 Bur. Stand., Circ. 40. 248 QUANTITATIVE ANALYSIS slum permanganate solutions is made by titration of weighed portions in presence of sulphuric acid, at temperatures between 80 and 90. Standard, certified samples of sodium oxalate may be obtained from the Bureau of Standards and these may be kept indefinitely, if in a stoppered bottle. Stock, solutions cannot be preserved unchanged. Oxalic Acid. Oxalic acid and ammonium oxalate may be puri- fied by recrystallization to the condition required for primary standards. If previously analyzed they are nearly as satisfac- tory as ferrous ammonium sulphate. The student is cautioned against the practice of assuming the purity of any of his primary standards without an analysis as the basis for the assumption. Reduction of the Iron Solution. Iron exists in the ferric con- dition in most ores or other minerals. In order to reduce the solution of ferric salt either stannous chloride, zinc, sulphurous acid or hydrogen sulphide may be employed. The first two are the only ones now commonly used. Stannous chloride, in solution, possesses the advantage of instantaneous action if added to the hot solution of ferric chloride. If the iron is to be reduced by stannous chloride an addition of this salt to the ore during the process of solution will materially hasten the action. For the final reduction the stannoils chloride solution may be added from a pipette, the disappearance of the red color of basic ferric chloride providing an approximate indica- tion of the end-point. In the analysis of iron ores there is occasionally trouble at this point unless certain precautions have been taken. In the first place, many iron ores contain appreciable quantities of organic matter and this serves to produce a yellow color when the ore is dissolved. As color due to this cause does not disappear when the iron has been reduced it is not possible to determine when the correct amount of stannous chloride has been added. This trouble may be avoided by igniting the weighed sample for a short time in a porcelain crucible, before dissolving. The second cause of irremovable color lies in fusion of insoluble residues in platinum crucibles. The pyrosulphate which is used as a flux dissolves traces of platinum and this, with stannous chloride, forms a yellow solution containing a complex of tin and OXIDATION AND REDUCTION 249 platinum. This interference is avoided by the substitution of porcelain crucibles for those of platinum. After a slight excess of stannous chloride has been added the solution is cooled and a considerable excess of mercuric chloride is added, the unused stannous chloride being thereby oxidized: 2HgCl 2 +SnCl 2 ->SnCl 4 +2HgCl. Mercuric chloride will not oxidize ferrous chloride and hence may be left in the solution. If an insufficient excess of mercuric chloride is used, or if it is added too slowly, free mercury may be produced : HgCl 2 -f-SnCl 2 -*SnCl 4 +Hg. The indication of such action is the appearance of a gray precipi- tate of mercury instead of the characteristic white silky crystals of mercurous chloride. If mercury is so produced the deter- mination is ruined because this mercury will itself reduce some of standard oxidizing solution during the process of titration of the iron. Stannous chloride cannot be used to reduce ferric solutions previous to titration by potassium permanganate unless the interference of chlorides is to be prevented by the addition of manganous sulphate. Instead, pure zinc or zinc of predeter- mined iron content may be used to reduce the iron. For this purpose an approximately weighed quantity of granular zinc or zinc dust may be added directly to the solution and dissolved in it, or the ferric solution may be passed through a redactor. The latter is a tube having the dimensions of a burette, filled with amalgamated zinc and having a dropping funnel fixed in the top. The acidified solution is passed through the tube once or twice and the iron is thereby reduced. Most zinc contains iron or other reducing matter and if it is to be dissolved in the iron solution, as above described, a blank determination should be made to determine the amount of standard that will be required to oxidize the reducing matter of the zinc. On account of the slow reducing action of zinc, stannous chloride is much to be preferred, where conditions will permit its use. Sulphurous acid and hydrogen sulphide reduce iron solutions quickly but the disadvantages involved in their preparation prevent their extensive use for this purpose. 250 QUANTITATIVE ANALYSIS Exercise: Preparation of Standard Potassium Permanganate Solu- tion. Calculate the weight of potassium permanganate required for 2500 cc of solution, each cubic centimeter to be equivalent to 0.005 gm of iron, using for this purpose the method shown on page 196. Weigh 2 percent more than the calculated weight and dissolve in about 1000 cc of distilled water. Allow to stand at least an hour, then filter through a double asbestos filter arranged as in Fig. 67, all vessels hav- ing been previously cleaned by chromic acid mixture. Dilute the solution to 2500 cc and standardize as follows: Calculate the weight of ferrous ammonium sulphate that will be approximately equiva- lent to 35 cc of the permanganate solu- tion. Weigh three such portions into Erlen- meyer flasks of 250 cc capacity.* Dissolve each portion, immediately before titrating, in 50 cc of distilled water and add 10 cc of dilute sulphuric acid. Titrate at once with the potassium permanganate solution, the first appearance of a permanent pink tint being taken as the end-point. If sodium oxalate is used as the primary standard, proceed as follows: Warm the acid solution of oxalate to 90 and titrate, stirring vigor- ously and continuously. The permanganate must not be added more rapidly than 10 to 15 cc per minute and the last 0.5 to 1 cc must be added drop-wise, with particular care to allow the color from each drop to disappear before another drop is added. At the end of the titration the temperature must not be below 60. A thermometer should be kept in the solution during the entire experiment. Calculate the value of the permanganate solution in terms of iron and dilute two liters of the solution to the exact equivalence of 0.005 gm of iron per cubic centimeter. From this iron value calculate also the calcium and manganese values (see pages 252 and 253), and record all upon the label of the bottle and in the record book. Determination of Iron in an Ore. Sample the ore and grind the last selection to pass the 100-mesh sieve. Weigh exactly 0.5 gm of ore on the counterpoised glasses, brushing into each of three porcelain cru- FIG. 67. Apparatus for filtering potassium per- manganate solution. OXIDATION AND REDUCTION 251 cibles. Heat the crucibles without covers for 5 minutes, using the ordinary desk burner, then allow the crucibles to cool, place in casseroles and add to each 25 cc of concentrated hydrochloric acid. If method (b) is to be used for reducing the iron add also at this point 5 cc of 5 percent stannous chloride solution. Cover and warm until solution is complete or until no further action appears to take place. If the residue is not colored, proceed, without nitration, by one of the methods (a) or (b) given below. If the residue is colored it may contain iron. In this case filter on a small paper and wash the paper free from iron solu- tion with hot water. Set the filtrate and washings aside and burn the paper at a low temperature in a porcelain crucible. If the residue is small in amount and apparently contains little silicious matter it may be decomposed by fusing with potassium pyrosulphate. Cool and dissolve the mass in hot water, adding the solution to the former filtrate. Now proceed by one of the methods given below. (a) Add 4 cc of concentrated sulphuric acid to the iron solution and evaporate by holding the casserole over a free flame, keeping in constant motion to hasten evaporation and to prevent bumping. Evaporate until the characteristic white fumes of sulphuric acid appear, this being the point at which all water and hydrochloric acid have been expelled. Cool and dilute to 50 cc, rinsing the solution into a 250-cc Erlenmeyer flask. Add 2 gm of granular zinc, as free as possible from iron, place a funnel with a short stem in the flask and warm until the zinc is dissolved. The iron solution should now be quite colorless or faintly green. Cool and titrate at once with standard potassium permanganate solution. Make a blank determination without the ore, to determine the amount of iron and other reducing matter in the zinc, by dissolving 1 gm of zinc in 25 cc of dilute sulphuric acid, under the same conditions as noted above, and titrating. Express the result of the blank experiment as the number of cubic centimeters of potassium permanganate reduced by 1 gm of zinc. The proper value will then be subtracted from the volume of permanganate used in the iron titration and the percent of iron then calculated. If the zinc is nearly pure it will dissolve very slowly. Solution may be hastened by dropping a coil of platinum wire into the flask, and keep- ing it in contact with the zinc. (b) Concentrate the iron solution, if necessary, to about 50 cc and transfer to a 1000 cc Erlenmeyer flask. While the solution is nearly boil- ing add, drop by drop from a pipette, a 5 percent solution of stannous chloride until the ferric chloride has just been reduced, this being made evident by the disappearance of the red color. Add two drops more of stannous chloride solution then cool quickly by immersing the flask in running water. When cool add, all at once, 25 cc of a 5 percent 252 QUANTITATIVE ANALYSIS solution of mercuric chloride and mix well with the solution. The pre- cipitate should be pure white mercurous chloride without a trace of gray mercury. Dilute to 500 cc and add 50 cc of a solution containing 144 gm of phosphorous pentoxide, 245 gm of sulphuric acid and 67 gm of. crystallized manganous sulphate in each liter of solution. Titrate at once with standard potassium permanganate solution and calculate the percent of iron in the ore. CALCIUM Calcium may be precipitated as oxalate, filtered and washed free from ammonium oxalate, then dissolved in hot, dilute sul- phuric acid and the resulting oxalic acid titrated with potassium permanganate. CaC 2 4 + H 2 S0 4 -+CaS0 4 + H 2 C 2 4 , 10C0 2 +8H 2 0. In order to determine the equivalent weight of oxalic acid it is necessary here, as in other cases, to note the change which it undergoes as it becomes oxidized. The products of oxidation are water and carbon dioxide. By using the method explained on page 241 it will be found that the apparent change of valence of carbon is 1. Since each molecule of oxalic acid contains two atoms of carbon the hydrogen equivalent of oxalic acid as a reducing agent is 2 and its equivalent weight is one-half its mo- lecular weight. When oxalic acid reacts as an acid its hydrogen equivalent is also 2, although there is no necessary connection between the two cases. Since one molecule of oxalic acid is formed by the decomposition of one molecule of calcium oxalate, containing one atom of calcium, the equivalent weight of calcium is also one-half its atomic weight. The calcium equivalent of the potassium permanganate solution may be calculated from the iron equivalent by the expression: eq. wt. of calcium ,. * i s - Xwt. iron = wt. of calcium. eq. wt. of iron Problems 62. 1 cc of a solution of potassium permanganate is equivalent to 0.002 gm of iron. Calculate the weight of calcium, calcium oxalate and oxalic acid equivalent to 1 cc. OXIDATION AND REDUCTION 253 63. What weight of potassium dichromate is equivalent to 4.75 gm of potassium permanganate, both being used for the oxidation of iron? 54. What weight of crystallized oxalic acid, H 2 C2O 4 .2H 2 O, is equivalent to 3.56 gm of crystallized ferrous ammonium sulphate, Fe(NH 4 )2(SO4)j.6H2O? 65. What is the normality of a solution of potassium permanganate, 1 cc of which is equivalent to 0.005 gm of iron? 56. 30 cc of potassium permanganate solution oxidizes 0.1905 gm of crystallized oxalic acid. How dilute 1000 cc of the solution to make exactly tenth-normal? Determination. Use samples of not more than 0.2 gm of the calcium compound. Dissolve, precipitate, filter and wash the calcium oxalate according to the method already learned in an earlier exercise (page 81). Pierce the point of the filter paper and wash the precipitate into a beaker with the least possible quantity of hot water. Thoroughly wash the paper in the funnel with successive portions of 5 cc of hot dilute sulphuric acid until any remaining precipitate shall have been dissolved. Again wash the paper with hot water, then warm (but not boil) the acid and precipitate in the beaker until the precipitate is dissolved. Titrate with standard potassium permanganate, inserting a thermometer and keeping the solution at 80 to 90 (refer to page 250). Calculate the percent of calcium in the sample. MANGANESE In acid solutions potassium permanganate is always reduced to the form of manganous salts, manganese being thereby reduced to its lowest state of oxidation, corresponding to the monoxide. In basic solutions the reduction goes only so far as to produce manganese dioxide. If the reducing agent is a manganous salt it is also oxidized to manganese dioxide : 2KMn0 4 +3MnS04+2H 2 0->K 2 SO 4 +5MnO 2 +2H 2 SO4. (a) This reaction may be made the basis of a determination of manganese by titration with potassium permanganate. 1 The manganese ore is dissolved in hydrochloric acid, manganous chloride being formed. The chloride is converted into sulphate and the hydrochloric acid removed by evaporation with sul- phuric acid. When the titration is made the solution must be only feebly basic. If a strong base is present a reduction of 1 Volhard: Chem. News, 40, 207 (1879). 254 QUANTITATIVE ANALYSIS potassium permanganate to potassium manganate occurs, the reducing agent being manganese dioxide or manganese hydroxide : 2KMnO 4 +4KOH+Mn0 2 ^3K 2 Mn0 4 +2H 2 0, (b) 4KMn0 4 +6KOH+Mn(OH) 2 -^5K 2 MnO 4 +4H 2 O. (c) These undesirable reactions may be prevented by having present a considerable excess of a weak base, such as is produced by shak- ing an excess of zinc oxide with water, this giving a suspension of zinc oxide in a saturated solution of zinc hydroxide. The latter is so dilute and so weakly ionized that the formation of potassium manganate does not take place. It does provide, however, suffi- cient base to neutralize the sulphuric acid produced by reaction (a), because the excess of zinc oxide keeps the solution saturated with zinc hydroxide. Manganese dioxide possesses, to a slight extent, acM-forming properties, since it is able to produce a class of salts that are theoretical derivatives of manganous acid, H 2 MnOs( = H 2 O.Mn0 2 ) . This acid is not known in the free state but certain manganites, as those of calcium and zinc, CaMnO 3 and ZnMnOs, are produced when manganese dioxide is formed in presence of soluble calcium or zinc compounds. If no such metal is present at the moment of oxidation of manganous salts to manganese dioxide, manganous manganite, MnMnOs, is precipitated, thus removing a certain amount of unoxidized manganese from the solution. An error would thereby be introduced but this is prevented by having zinc hydroxide present. The saturated solution of manganese dioxide can have but a small concentration of manganous acid and this is at once precipitated as zinc manganite. The addition of zinc oxide therefore serves a double purpose. It maintains a feebly basic solution throughout the titration and also prevents the precipitation of manganous manganite. Problems 67. From the equation for the reaction between potassium permanganate and manganous sulphate, calculate the equivalent weight of potassium permanganate and of manganese in manganese sulphate. 68. What weight of potassium permanganate must be contained in 1000 cc of a solution, 1 cc of which will oxidize 0.002 gm of manganese? 69. If a solution of potassium permanganate is fifth-normal with respect OXIDATION AND REDUCTION 255 to iron in acid solution what is its normality with respect to manganese in basic solution? 60. 1 cc of a solution of potassium permanganate is equivalent to 0.010 gm of iron. What weight of manganese will be oxidized by 1 cc? Determination. Calculate the approximate weight of pyrolusite or other manganese compound that is necessary to reduce about 40 cc of the standard potassium permanganate solution already made, arbitrarily assuming a state of purity for the sample and also that one-fifth of the sample is finally to be titrated. Dry the sample to constant weight at 120 and weigh the calculated quantity of dried material, placing in a casserole. Dissolve by warming with concentrated hydrochloric acid. When solution is complete or when no further action is apparent filter and wash the residue and paper, preserving the filtrate and washings. If the residue contains any dark material, burn the paper in a platinum crucible and fuse with 1 to 2 gm of sodium carbonate. If manganese is present the fusion will be colored green by sodium manganate. Dis- solve the fused mass in hot water and add to the main solution. Remove hydrochloric acid by evaporating with sulphuric acid (about 2 cc) to the appearance of fumes of sulphur trioxide. Redissolve, adding a little nitric acid if necessary, wash into a 250 cc volumetric flask, dilute to the mark and mix. Treat 50 cc portions as follows: Measure into an Erlenmeyer flask of 1000 cc capacity and neutralize by the addition of zinc oxide suspended in water, shaking and continuing the addition until any iron is precipitated as ferric hydroxide and a sufficient excess of zinc oxide is present to maintain a milky appearance throughout the subse- quent titration. Dilute to about 300 cc, heat nearly to boiling and titrate, adding the potassium permanganate solution 5 cc at a time until a permanent color is produced. Treat a second 50 cc portion of the manganese solution in a similar manner but adding, all at once, 5 cc less of standard solution than was added to the first portion then adding 1 cc at a time. To a third portion add the entire quantity used in the second, less 1 cc, then complete the titration by adding the standard solution 0.1 cc at a time. Treat the fourth portion in the same manner as the third. From the last two titrations calculate the percent of manganese in the ore. AVAILABLE OXYGEN Manganese dioxide is used not only as an ore of manganese but also as an oxidizing agent in various laboratory processes and in a commercial way, as, for example, in the production of chlorine from hydrochloric acid. In such cases the percent of manganese 256 ' QUANTITATIVE ANALYSIS is not as important as is that of oxygen available for oxidation processes. In most cases the available oxygen may be calculated with sufficient accuracy for commercial requirements from the percent of manganese. This can be accurately done only in case no other manganese compound and no other peroxide is present. Generally no other peroxide is present, although manganese fre- quently occurs in pyrolusite in small quantities as other com- pounds than the peroxide. If an accurate determination of availa- ble oxygen is required it may be made by reducing a weighed sample of the manganese dioxide by a measured amount of a reducing agent, titrating the excess of the latter by standard potassium permanganate. The reducing agent may be any of those already discussed in connection with standardization of potassium per- manganate. Ferrous ammonium sulphate or oxalic k acid is to be preferred. The reaction between manganese dioxide and these reducing agents in presence of sulphuric acid is represented by the following equations, which should be balanced by the student as an exercise in calculation of hydrogen equivalents. Determine also what fraction of the total oxygen of manganese dioxide is "available." Mn0 2 + FeS0 4 + H 2 S0 4 -+MnSO 4 + Fe 2 (S0 4 ) 3 + H 2 0, Mn0 2 +H 2 C 2 4 +H 2 S0 4 -+MnS0 4 +C0 2 +H 2 0. Another method for determining the available oxygen is described on page 265. Determination. Dry the sample of either pyrolusite or commercial manganese dioxide to constant weight at 120. Calculate the weight that would be equivalent to approximately 40 cc of the standard potas- sium permanganate already made, arbitrarily assuming that the sample is pure manganese dioxide. Weigh samples of the calculated weight into 250 cc Erlenmeyer flasks. Calculate the weight of crystallized ferrous ammonium sulphate or oxalic acid that would reduce approxi- mately 50 cc of the standard solution of potassium permanganate and add this quantity to each flask. Calculate the approximate volume of dilute or concentrated sulphuric acid necessary to enable the reactions to proceed and add three times this volume to each flask. Add 50 cc of water, warm to 70 and titrate immediately with standard potassium permanganate solution. Calculate the percent of available oxygen in the sample. OXIDATION AND REDUCTION 257 Potassium permanganate solution is also useful for the titra- tion of hydroferro cyanic acid and hydroferricyanic acid. Ferro- cyanides are oxidized in acid solution : 10H4Fe(CN) 6 +2KMnO4+3H 2 S0 4 ->10H 3 Fe(CN) 6 +K 2 S04 +2MnSO 4 +8H 2 0. Ferri cyanides may be reduced in basic solution by ferrous sulphate and then titrated by potassium permanganate. The reaction is as follows: K 3 Fe(CN) 6 +FeSO4+3KOH->K4Fe(CN)6+Fe(OH)3+K2S0 4 . Potassium Bichromate. The equation for the reaction of potassium dichromate with ferrous salts is given on page 242. This substance possesses several advantages over potassium per- manganate as a standard oxidizing agent. It is relatively more stable and therefore may be obtained in a state of uniform purity. This makes it possible to standardize solutions by direct weighing when the degree of purity of the salt has been established by analy- sis. The relative stability is the same with solutions and the standard solution can be kept almost indefinitely without chang- ing its concentration. Potassium dichromate may also be used for the titration of iron and other reducing agents in presence of hydrochloric acid or chlorides, without oxidation of the latter taking place. This is a very decided advantage in the determina- tion of iron since it makes possible the use of stannous chloride as a reducing agent without the addition of manganous sulphate and phosphoric acid. There is no indicator that can be added directly to the solution which is being titrated by potassium dichromate and the color of the standard solution is not suffi- ciently intense to be of any use for this purpose. The indicator that is commonly used is potassium ferricyanide, placed in drops on a white porcelain "spot plate." Drops of the solution are removed from time to time by means of a stirring rod and allowed to touch the drops of ferricyanide. So long as ferrous iron is present the blue of ferrous ferricyanide is apparent on the spot plate. When the last trace of iron has been oxidized there is produced on the plate only the light brown ferric ferricyanide. There being nothing in the appearance of the solution of the iron salt to -indicate the approach to the end-point, the titration is 17 258 QUANTITATIVE ANALYSIS necessarily somewhat tedious unless a system is devised for rapid readings. Such a system has been used in connection with the determination of manganese and is indicated in the next exercise and this removes the last objection to the use of potas- sium dichromate for the titration of iron. Problems 61. A solution of potassium permanganate contains 25.38 gm in 1000 cc. What must be the concentration of a potassium dichromate solution in order that it shall have the same oxidizing power toward iron? 62. Balance the following equation and calculate the equivalent weight of tin. K 2 Cr 2 O 7 +SnCl 2 +HC1-KC1 +CrCl 3 +SnCl 4 +H 2 O. Exercise: Preparation of Standard Potassium Dichromate Solu- tion. The solution should be of such concentration that 1 ccis equivalent 0.005 gm of iron. Calculate the weight of potassium dichromate neces- sary for 2000 cc of such a solution. If the salt is known to be pure, weigh exactly the calculated weight and omit further standardization. If it is not pure but its oxidizing power known from previous determina- tions, calculate the weight of impure sample required and use this weight. If nothing is known of the purity use 1 percent more than the weight of pure salt required for 2500 cc of solution and standardize the solution as directed below. In any case dissolve the weighed salt and dilute to the proper volume. In case titration for standardization is to be omitted and direct weighing is to be made the basis for standardization, 2000 cc of the solution should be accurately made and poured into a dry bottle. Standardization, if this should be necessary, is accomplished by titra- tion against ferrous ammonium sulphate or iron wire, the first method being preferable. Write and balance the equation for the oxidation of ferrous sulphate by potassium dichromate in presence of sulphuric acid referring, if necessary, to the equation for the oxidation of the chloride, page 242. Calculate the weight of crystallized ferrous ammonium sulphate necessary to reduce 35 cc of the dichromate solution. Weigh five portions of exactly this weight into 250 cc beakers and dissolve each in 50 cc of recently boiled and cooled water just before titrating. Pre- pare a 0.01 percent solution of potassium ferricyanide and place a drop in each of the depressions of a white porcelain spot plate. Add to the solution of ferrous ammonium sulphate three times the calculated amount of sulphuric acid necessary, as indicated by the equation, and titrate at once, as follows: To the first solution add the dichromate solution 5 cc at a time, stirring well after each addition, and test by removing a drop by means of the stirring rod and touching to a drop of potassium ferri- OXIDATION AND REDUCTION 259 cyanide solution on the spot plate. The end-point is reached when a blue color is no longer produced on the plate, after standing for 2 minutes. Dust or reducing gases will interfere by reducing traces of ferric chloride. Titrate the second solution by adding 5 cc less than tEe amount of dichromate solution used in the first, then adding 1 cc at a time. Titrate the third solution by adding 1 cc less than the total used in the second, then adding 0. 1 cc at a time. Titrate the fourth and fifth solutions in the same manner and take the average of the last three titrations for permanent record. Calculate the value of the solution in terms of iron. Dilute to make 1 cc equivalent to 0.005 gm of iron. Instead of weighing five portions of ferrous ammonium sulphate a standard solution may be made by dissolving ten times the required amount, adding the necessary sulphuric acid and diluting to 500 cc. Portions of 50 cc are then measured and titrated. The solution oxidizes upon exposure to air and the method of weighing separate portions and dissolving just before titration is preferable. IRON Determination of Iron. Prepare a sample of iron ore by grinding to pass a 100-mesh sieve. Weigh five portions of exactly 0.5 gm each, using the counterpoised glasses and brushing the ore into porcelain crucibles. Heat the inclined crucibles for 5 minutes over the desk burner, cool, place in casseroles and dissolve in hydrochloric acid, with or without the addition of stannous chloride. Reduce each solution just before titration, following the directions given for dissolving and reduc- ing by method (b) of the permanganate method. Do not add the solu- tion of manganous sulphate and of phosphoric and sulphuric acids. Dilute to 100 cc. The titration is carried out exactly as directed for standardizing potassium dichromate solution. Calculate the percent of iron in the ore. CHROMIUM The most important ore of chromium is a compound of iron and chromium known as "chromite," having a composition corresponding with the formula FeO.Cr 2 O3. Although chro- mium is here in its lowest state of oxidation the substance is thought to be a salt of a hypothetical chromous acid, H 2 Cr 2 04. Chromite cannot be dissolved in acids nor is it possible to decompose it easily by fusion with alkali carbonates. Fusion with sodium peroxide decomposes it, oxidizes the iron to ferric oxide and the 260 QUANTITATIVE ANALYSIS chromium to chromium trioxide, forming then sodium chromate. Upon dissolving in water and filtering, ferric oxide is removed. The addition of acid produces sodium dichromate. This can then be reduced by adding an excess of a standard reducing agent, such as ferrous ammonium sulphate, titrating the excess by stand- ard potassium dichromate or permanganate. The reactions are expressed by the following equations, which should be balanced by the student. FeCr 2 04+Na 2 2 ^Fe 2 O 3 +Na 2 Cr04+Na 2 O, Na 2 CrO4+HCl^Na 2 Cr 2 O 7 +NaClH-H 2 O, Na 2 Cr 2 7 +FeS04+HCl^NaCl+Fe 2 (S04)3+ FeCl 3 +CrCl 3 +H 2 0. Iodine and Sodium Thiosulphate. Iodine and thiosulphates react quantitatively, forming sodium iodide 1 and sodium tetra- thionate : 2Na 2 S 2 03+l2-Na 2 S 4 6 +2NaL This is an oxidation of sodium thiosulphate by iodine, which is itself reduced. The solution may be originally neutral or acid, or alkali bicarbonates may be present. Normal carbonates or hydroxides should not be present since they also combine with iodine : 2NaOH + I 2 -+NaIO + Nal + H 2 O, 2Na 2 CO 3 +l2->NaIO+N3l+CO 2 . The color of dilute solutions of iodine is sufficiently intense to serve as a fairly accurate indicator. Much more accurate re- sults are obtained by the use of starch as an indicator, mere traces of iodine producing a visible blue or rose red color with starch. Because of the fact that iodine is an excellent oxidizing agent for many substances when a bicarbonate is present, and that hydri- odic acid is oxidized by many oxidizing agents when an acid is present, free iodine being liberated, the two standard solutions of iodine and sodium thiosulphate form a most useful pair for volumetric analysis. As an example of their use the reactions of arsenic may be noticed. Arsenious acid or an arsenite is oxi- dized by free iodine thus: H 3 AsO 3 +I 2 +H 2 Orfl3As0 4 +2m, OXIDATION AND REDUCTION 261 This reaction does not take place quantitatively but is reversible. If, however, sodium bicarbonate is present in excess the hydri- odic acid is neutralized as fast as it is formed and the reaction is completed. Standard iodine solution may, in this way, be used for the titration of arsenious acid. On the other hand arsenic acid is reduced by hydriodic acid: H 3 As04+2HI->H 3 AsO3+H 2 0+l2. This is seen to be the reverse of the reaction expressed above and it would follow that it also is incomplete unless one of the products is removed. This may be doiie by adding sodium thiosulphate to remove the iodine, in which case the standard sodium thiosulphate indirectly titrates the arsenic acid. In prac- tice hydriodic acid is not kept as a reagent because of its insta- bility but potassium or sodium iodide and hydrochloric acid are used instead, hydriodic acid being thus made available in the solution. Standardization. By properly purifying iodine standard solutions may be made by direct weighing. Commercial iodine is usually not sufficiently pure for this purpose and must be analyzed if it is to be used in this way. Iodine solutions will not remain constant in oxidizing power because of interaction between iodine and water, and it is usually not advisable to attempt to dilute solutions to a definite concentration because they must be restandardized after short intervals of time. For this reason standardization by direct weighing is not practicable and the iodine need not be purified before dissolving. The solution may then be standardized by titrating against any standard reducing solution. The best substances for this purpose are sodium thio- sulphate and arsenious oxide. Iodine does not dissolve easily in water but is readily soluble in a solution of potassium iodide or sodium iodide. Such a solu- tion probably contains an iodide having the formula KI 3 or Nals. Many organic liquids are good solvents for iodine. Examples are the alcohols and acetic acid. - These will be discussed in the sec- tion dealing with the analysis of fats and oils. When starch and iodine are brought together a deep, indigo-blue color is produced and this serves as a very delicate test for either starch or iodine. The nature of the blue substance has long been the subject of 262 QUANTITATIVE ANALYSIS investigation and discussion. It is probably a solid solution of iodine in starch. Sodium thiosulphate may sometimes be obtained in a suffi- ciently pure condition to allow standardization by direct weigh- ing. It is better to make the solution somewhat more concen- trated than that desired and to standardize and dilute to a definite concentration. For standardization the solution may be directly titrated against standard iodine solution or indirectly against potassium dichromate or a salt of copper. It has already been stated that potassium dichromate may be obtained in a state of uniform purity. If to a standard solution of this salt potassium iodide and hydrochloric acid are added, iodine is liberated as follows: K 2 Cr 2 7 +6KI+14HCl->8KCl+2CrCl3+7H 2 0^3I 2 . The liberated iodine may be titrated by sodium thiosulphate and the latter thus standardized. The solution of potassium dichro- mate used for iron determinations may be used also for this purpose. It was standardized in the decimal system, however, and it will be necessary to calculate its value in the normal system because the solution of sodium thiosulphate is to be used for the determination of several different substances. The following ex- ample will illustrate the method of calculation of standardization. Example. 40 cc of a solution of potassium dichromate liberates iodine equivalent to 22 cc of sodium thiosulphate solution. 1 cc of potassium dichromate solution is equivalent to 0.005 gm of iron. What is the normality of the thiosulphate solution? 1 cc of sodium thiosulphate solution is equivalent to ~~ cc of potas- 40 sium dichromate solution and to ^ X 0.005 gm of iron. A normal solution would be equivalent to 0.05584 gm of iron. Therefore the ... . 40X0.005 A1 . OOAT normality is ^-^-0.1828 N. The standardization against a salt of copper is also an excellent method. This is described on page 268 in connection with the determination of copper. Sodium thiosulphate is quite stable in solution and may be kept for months without appreciable change in concentration if the water contains no trace of acid. Even carbonic acid causes OXIDATION AND REDUCTION 263 decomposition and free sulphur is deposited from the solution, sulphurous acid being formed. This is because thiosulphuric acid is very unstable and rapidly decomposes : Na 2 S 2 3 +H 2 C0 3 ^Na 2 C03+H 2 S 2 03, H 2 S 2 3 -H 2 S0 3 +S. Even a very small amount of carbonic acid is sufficient to start the decomposition by liberating some thiosulphuric acid. As sulphurous acid accumulates it aids the decomposition which is thus progressive: Na 2 S 2 03+H 2 S0 3 -+Na 2 SO 3 +H 2 S 2 03, H 2 S 2 3 -H 2 S0 3 +S. In order to avoid starting this series of reactions the water should be boiled and cooled before making the solution. Problems 63. A solution of potassium dichromate contains 4.95 gm of the salt in 1000 cc. What weight of sodium thiosulphate is equivalent to 1 cc? 64. A solution of potassium dichromate contains 6.235 gm in 1000 cc. 30 cc of this solution is equivalent to 42.9 cc of sodium thiosulphate solution. What is the normality of the latter? 65. What weight of potassium dichromate must be dissolved in 250 cc to make a solution, 25 cc of which is equivalent to 35 cc of sodium thiosulphate solution containing 13.65 gm of the crystallized salt in 1000 cc? 66. 1 cc of potassium dichromate solution is equivalent to 0.005 gm of iron. What is the iodine equivalent? 67. 25 cc of iodine solution is equivalent to 0.125 gm of potassium dichro- mate. To what volume should 1000 cc be diluted to make the solution tenth-normal? 68. 20 cc of potassium dichromate solution oxidizes 0.0240 gm of oxalic acid, H 2 C 2 O4.2H 2 O. 1 cc of the same solution oxidizes the same weight of iron as does 1.2 cc of potassium permanganate solution. What is the nor- mality of the latter solution? Exercise : Preparation of Tenth-normal Sodium Thiosulphate Solu- tion. Calculate the weight of crystallized sodium thiosulphate, Na 2 S 2 03.5H 2 0, required for 2500 cc of tenth-normal solution. Crush the salt and dissolve 2 percent more than this weight in cold, recently boiled water and dilute to 2500. cc. Keep the bottle well stoppered and out of direct sunlight. Make 200 cc of a solution containing 30 gm of potassium iodide. The starch solution is made as follows: Moisten 1 gm of starch with 264 QUANTITATIVE ANALYSIS cold water to make a thick paste. Heat 200 cc of water to boiling and pour it into the starch paste. Boil, with constant stirring, for one minute. The solution does not keep well and should be made each day as required. Standardize the sodium thiosulphate solution against potassium dichromate. If the solution used in iron determinations is at hand, use this, otherwise make 250 cc of exactly tenth-normal solution by weighing the salt, dissolving and diluting to the required volume. Measure 35 cc of either solution into an Erlenmeyer flask, add 40 cc of potassium iodide solution and 10 cc of concentrated hydro- chloric acid. Titrate at once with sodium thiosulphate solution, deferring the addition of starch as long as possible. If starch is added before the iodine is nearly all reduced a precipitate of starch iodide will form, free iodine being, in this way, removed from the possibility of reacting with thiosulphate. A false end point is then obtained. The solution of chromium chloride, formed by reduction of potassium dichromate, is green. The solution has an amber tint as long as much free iodine is present. Upon the addition of starch the solution acquires a blue-green color and the change to pure green at the end point may be difficult to detect at first trial. With a little experience the difficulty will disappear. Make at least three titrations and calculate the normality of the sodium thiosulphate solution. Dilute to make tenth-normal. Make a blank test upon the potassium iodide, omitting the potassium dichromate but adding the hydrochloric acid. If iodine is found, correct the observed volume of sodium thiosulphate before calculating its concentration. OXIDIZING POWER OF PEROXIDES Such peroxides as those of manganese, barium, lead, and hy- drogen readily oxidize hydriodic acid and liberate iodine. The titration of the latter by standard sodium thiosulphate solution constitutes an indirect determination of the oxidizing power, or "available oxygen" of the peroxide. In practice it is sometimes not found convenient to add an iodide and hydrochloric acid directly to the peroxide because the solution is usually colored by impurities dissolving as chlorides. In such a case hydrochloric acid is added to the peroxide and the liberated chlorine is distilled into potassium iodide solution. In the case of manganese perox- ide the reactions may be represented thus: Mn0 2 +4HCl-+MnCl 2 +2H 2 C1 2 +2KI-*2KC1+I 2 . OXIDATION AND REDUCTION Problem 265 69. Calculate the equivalent weight of manganese dioxide and of available oxygen and find the weight of each that is equivalent to 1 cc of tenth-nornral sodium thiosulphate solution. These reactions are analogous to those occurring with other peroxides and the determination of available oxygen in manga- nese dioxide istthe most frequently made of all. The apparatus for carrying out the decomposition arid distillation should have ground glass joints and should not allow contact of iodine or chlorine with organic matter. The modified apparatus of Bun- sen, Fig. 68, may be used. The receiver must be kept cold in order to avoid loss of iodine. FIG. 68. Modified Bunsen's apparatus for the determination of available oxygen. Determination. Dry 2 to 4 gm of either commercial manganese dioxide or pyrolusite at 120 until the weight is constant. The sample already used for the determination of manganese may be used for this determination and the results obtained by the two methods compared. Weigh enough sample to be equivalent to about 35 cc of the standard sodium thiosulphate solution and place in the flask of a Bunsen distil- ling apparatus or of some other suitable type. 1 Place in the receiver 2 gm of potassium or sodium iodide, dissolve this in water and dilute until the bend is just sealed when the apparatus is in the proper position. Immerse the receiver in ice water, then add to the flask containing the manganese dioxide 30 cc of concentrated hydrochloric acid and quickly insert the stopper carrying the delivery tube. Warm the acid gently, distilling the chlorine into the potassium iodide solution. Raise the 266 QUANTITATIVE ANALYSIS temperature gradually until the acid is boiling and boil for five minutes after action is completed. While the burner is still under the flask lower the receiver until the delivery tube is entirely out of it, then remove the burner. Remove the delivery tube from the flask and rinse it inside and outside, the water flowing back to the receiver. Rinse the whole iodine solution into an Erlenmeyer flask and titrate with sodium thio- sulphate solution. Make a blank test on the iodide used, as follows: Weigh out the same amour t as was used in the determination of available oxygen, the weighing being accurate to centigrams. Dissolve in 100 cc of distilled water and add 5 cc of concentrated hydrochloric acid. If a yellow color appears, indicating the presence of free iodine, titrate with sodium thiosulphate, using starch at the end. Deduct the thiosulphate used in the blank from that used in the determination of available oxygen and calculate the percent of available oxygen, also the theoretical percent of manganese and of manganese dioxide. If the sample is the same as that used for the direct determination of manganese and of available oxygen by potassium permanganate an interesting comparison of results of different methods may be made, although the calculation of available oxygen from the percent of man- ganese may not check with the direct determination, for reasons already discussed. Sodium thiosulphate may be used to titrate the iodine produced by the action of almost any oxidizing agent" upon a solution of potassium iodide and hydrochloric acid. Peroxides have already been discussed. Other substances that may be determined are free halogens (chlorine and bromine being allowed to displace iodine from potassium iodide), easily reducible oxyacids and their salts, as the halogen oxyacids, nitrous acid and persulphuric acid, oxysalts of metals that exist in acid radicals, as dichromates, chromates, permanganates and manganates, and salts of metals that possess more than one valence, as iron, copper, mercury and arsenic. While sodium thiosulphate may be used for the determination of almost any oxidizing agent it is not necessarily true that this provides the best method for all such materials. In many cases other methods will be found to give better results or to be more conveniently applied. OXIDATION AND REDUCTION 267 Problem 70. Complete the following equations, balance and determine the equiva- lent weights of each of the oxidizing agents. Br 2 +KI KBrO+KI+HCl KBrQs+KI+HCl-* KC1O+KI+HC1 > K 2 Cr 2 O 7 +KI+HCl > KMnO 4 +KI+HCl-> FeCl 3 +KI- CuCl 2 +KI- COPPER The gravimetric determination of copper may be made by precipitating as cupric hydroxide, heating and weighing as cupric oxide, or by precipitating as cupric sulphide, heating with sulphur and hydrogen and weighing as cuprous sulphide. Both methods are difficult of execution and are subject to considerable errors. Electrolytic methods are more accurate and more easy of accomplishment. Copper may be determined volumetrically by several methods, one of the best being Low's "iodide method." 1 This method depends upon the insolubility of cuprous iodide and the instability of cupric iodide. If to a solution of a cupric salt, containing no highly ionized acid and no other oxidizing agent, potassium iodide is added there is an immediate precipita- tion of cuprous iodide with liberation of iodine : The iodine may be titrated by standard sodium thiosulphate solution and copper calculated. Problems 71. Calculate the equivalent weight of copper and the weight which is equivalent to 1 cc of tenth-normal sodium thiosulphate solution. 72. What weight of a copper ore should be taken for analysis in order that 1 cc of fifth-normal thiosulphate solution should indicate 1 percent of copper in the ore? 1 J. Am. Chem. Soc., 18, 458 (1896); 24, 1082 (1902). See also a comparison of methods by Fernekes and Koch: Ibid., 27, 1224 (1905). 268 QUANTITATIVE ANALYSIS If sodium thiosulphate solution is to be standardized against pure copper, the metal is dissolved in nitric acid, most of the nitrogen oxides are expelled by boiling and any remaining trace of nitrous acid is oxidized by bromine. The excess of nitric acid is then neutralized by ammonium hydroxide, acetic acid and potassium iodide are added and the free iodine titrated at once. If a copper ore or crude copper is to be analyzed all metals whose iodides are insoluble or whose salts will oxidize potassium iodide must first be removed. The addition of metallic alu- minium to the solution containing sulphuric acid will precipitate copper and leave in solution all other metals of higher decomposi- tion potentials as well as those soluble in sulphurip acid, providing that nitric acid be absent. The latter can be removed ; by evapo- rating with sulphuric acid. This treatment also precipitates lead as sulphate, which may be removed by filtration. After the copper is precipitated by aluminium it is removed by filtration, washed, dissolved in nitric acid and determined as in the stand- ardization of sodium thiosulphate. During the process of filtration and washing, copper oxidizes and dissolves to some extent in the sulphuric acid. This would occasion a loss and to prevent this a solution of hydrogen sulphide is used for the wash liquid. Any small amount of copper that might be dissolved is reprecipitated as cupric sulphide. If a brown color appears in the filtrate below this is an indication of incomplete precipitation by the aluminium or of resolution during filtration. In either case the determination is spoiled unless the copper can be recovered by refiltration. Exercise: Standardization of Sodium Thiosulphate Solution. The standard solution already prepared may be used and the copper equiva- lent calculated. In case it is desired to standardize against copper or a copper salt, proceed by one of the following methods: (a) Standardization against Metallic Copper of Known Purity. Weigh sufficient copper to require about 35 cc of sodium thiosulphate solu- tion. Place in a 250 cc flask and dissolve by warming with 5 cc of a mixture of equal volumes of concentrated nitric acid and water. Dilute to 25 cc and boil to expel nitrogen oxides. Add 5 cc of bro- mine water and boil until all excess bromine is removed. Cool and add strong ammonium hydroxide until a clear blue solution is obtained, then OXIDATION AND REDUCTION 269 boil until copper hydroxide begins to precipitate. Acidify with acetic acid and boil, if necessary, to dissolve any precipitated cupric hydroxide. Cool, add 3 gm of potassium or sodium iodide and titrate the liberated iodine with sodium thiosulphate solution. Calculate the copper equiva- lent of the solution. (b) Standardization against a Copper Salt. Weigh the proper amount of cupric sulphate of known purity, dissolve in 25 cc of water, make slightly basic with ammonium hydroxide and from this point proceed as in (a). Determination. Dissolve 0.5 gm of ore in a covered casserole with 10 cc of concentrated hydrochloric acid and 5 cc of concentrated nitric acid, warming if necessary to aid solution. If the sample is an alloy use enough to contain 0.2 to 0.4 gm of copper and dissolve in 10 cc of nitric acid, 1:1 (specific gravity 1.2). Add 7 cc of concentrated sulphuric acid and evaporate until dense white fumes of sulphur trioxide appear. Cool, add 25 cc of water and boil to dissolve the sulphates. Filter to remove lead sulphate and gangue, allowing the filtrate to run into a beaker. Wash the residue and paper and dilute the filtrate and washings to 75 cc. Cut a strip of sheet aluminium about 2.5 cm wide and 14 cm long, bend into a triangle and stand on its edge in the solution. Cover and boil until all copper is precipitated and the solution is colorless or green from ferrous sulphate. If this condition cannot be attained it is because nitric acid was not completely removed when evaporating with sul- phuric acid. When all copper is precipitated, wash down the sides of the beaker with a jet of hydrogen sulphide solution, pour the solution into a filter paper and filter quickly. Transfer the copper to the filter, washing the aluminium with hydro- gen sulphide solution while still in the beaker. Wash thoroughly with hydrogen sulphide solution and then place a clean flask under the filter. Add to the beaker containing the aluminium 6 cc of nitric acid, sp. gr. 1.2. Boil shortly to dissolve adhering copper then pour the acid slowly over the filter to dissolve the copper on the paper. When all copper seems to be dissolved pour over the paper 5 cc of bromine water. Wash beaker and paper thoroughly with hot water then open the paper and wash into the flask any particles of copper that have escaped the action of the acid. Boil until all bromine is removed, add strong ammonium hydroxide until a deep blue is obtained, boil until copper hydroxide begins to precipitate and from this point proceed as in the standardization of sodium thiosulphate solution. Calculate the percent of copper in the ore. 270 QUANTITATIVE ANALYSIS BLEA.CHING POWDER When gaseous chlorine is passed over slaked lime it is absorbed with formation of an unstable compound that is easily made to yield chlorine under certain conditions and the compound pro- vides a convenient means for storing and transporting chlorine to be used for bleaching, disinfecting, etc. This compound, known as " bleaching powder," is a double salt of calcium with hydro- chloric and hypochlorous acids and may be represented by the /Cl formula CaCaC0 3 +H 2 0+Cl 2 . This is due to the fact that when hydrochloric acid and hypochlo- rous acid come together, even in dilute solutions, they act upon each other with the formation of chlorine : HC1+HC10-VE 2 O+C1 2 . Because of the easy decomposition of bleaching powder by car- bonic acid it rapidly deteriorates when exposed to air, chlorine escaping. Loss of efficiency also occurs through loss of oxygen : 2CaCl.ClO->2CaCl2+0 2> and through a decomposition such that calcium chlorate is formed : 6CaCl.ClO->Ca(ClO 3 ) 2 +5CaCl 2 . The decompositions represented by the last two equations result in the formation of chlorine compounds in which the chlorine is not liberated upon acidification. A determination of total chlo- rine would therefore be of little value as an estimate of the useful- ness of bleaching powder. " Available chlorine" is better determined by a volumetric process. For this purpose the acidi- fied solution may be treated with potassium iodide and the liber- ated iodine titrated with standard sodium thiosulphate solution, or the solution may be titrated directly by a standard solution of sodium arsenite. For the last titration the indicator is a paste of starch and potassium iodide used on a porcelain plate or absorbed OXIDATION AND REDUCTION 271 by filter paper and dried. This method of reading the end point is inconvenient and the first method of titration is the better one. If calcium chlorate is present in bleaching powder and a strong acid is used for liberating the chlorine, the chlorate will be de- composed, though but slowly. This is because chloric acid is formed by the reaction of chlorate with added acid and chloric acid is slowly reduced by hydrochloric acid, liberating chlorine: HC10 3 +5HCl->3H 2 0-f3Cl 2 . During the titration the effect of these reactions is seen in an uncertain end point. As sodium thiosulphate is added the blue color of starch iodide disappears and then returns and deepens. As the addition of thiosulphate is continued the blue finally permanently disappears, but this end point does not represent the titration of chlorine really available in bleaching processes because that which comes from calcium chlorate is evolved too slowly to be of much use. This interference with the titration may be almost entirely averted by using a weak acid instead of a strong one for the decomposition of the chlorohypo chlorite. The concentration of chloric acid does not then become suffi- ciently large to cause more than slight oxidation of hydriodic acid. The most suitable acid for the purpose is acetic acid. Determination. If bleaching powder were pure calcium chloro- hypochlorite, CaCl.CIO, it would contain about 56 percent of chlorine. For reasons already discussed the amount of available chlorine is much less than this and in the average commercial product it is not much more than 25 percent. Upon this basis calculate the weight that should be used when 50/1000 of the weighed sample is to be taken for the final titration. Weigh from a closed weighing bottle into a 1000 cc graduated flask. Fill to the mark with water and agitate until the powder is thor- oughly disintegrated and all soluble matter is in solution. Measure 50 cc portions into flasks, add 5 gm of potassium iodide and 25 cc of 10 percent acetic acid to each and titrate with sodium thiosulphate solution. Calculate the percent of available chlorine in the powder. Standard Iodine Solution. Iodine solutions do not maintain a constancy of oxidizing power and standard solutions must be restandardized frequently for accurate work. It has already been explained (page 261) that free iodine is usually dissolved with the aid of an iodide and that a molecular compound with a formula such as NaI 3 is present in such solu- 272 QUANTITATIVE ANALYSIS tions. Two-thirds of the iodine in this compound (better repre- sented as NaI.I 2 ) is so loosely bound that it behaves as free iodine. The formula would indicate a necessary ratio of 150 : 254, sodium iodide to iodine, or 166:254, potassium iodide to iodine. However, in practice it is found necessary to use a much higher ratio (2:1) in order to dissolve the iodine readily and to pre- serve the solution in a fairly stable condition. The twentieth- normal solution is convenient for the following determination. ARSENICAL INSECTICIDES Two of the most important insecticides containing arsenic are London purple and Paris green. The former is a waste product of certain aniline dye industries and contains much dye in addi- tion to a fairly large quantity of arsenic. Paris green is a fairly definite compound of cupric arsenite and cupric acetate, repre- sented by the formula: Cu 3 (As0 3 )2.Cu(C2H 3 02)2- This com- pound is decomposed by boiling with sodium hydroxide, pre- cipitating cuprous oxide and forming sodium arsenate and arsenite in solution. The formation of cuprous oxide is due to the reducing action of sodium arsenite, forming sodium arsenate. If the solu- tion is to be titrated for the determination of total arsenic this arsenate must first be reduced. For this purpose the solution is concentrated, then hydrochloric acid and potassium iodide are added and the resulting free iodine is removed by sodium thiosulphate. Na 3 As04+2HCl+2KteNa3As0 3 +2KCl+H 2 O+I 2 , The exact removal of iodine must be determined without the aid of starch. In strongly acid solutions starch is partly inverted, dextrine being one of the intermediate products and dextrine forms with iodine a deep red color which is not later removed and which interferes in the titration of iodine solution. The first equation above represents a reaction that can be quantitatively reversed at will. The complete reduction of pentavalent arsenic has just been accomplished in acid solution, one of the products (iodine) being removed. If the solution is now made basic, thus removing one of the products (hydro- chloric acid or hydriodic acid) of the reverse reaction and if OXIDATION AND REDUCTION 273 standard iodine solution is added a quantitative oxidation of trivalent arsenic occurs. The addition of a strong base is not permissible because this will combine with iodine: 2KOH+I 2 -KIO+KI+H 2 0. Neither is it possible to use a normal carbonate, for similar rea- sons. Alkali bicarbonates may be present and are used in prac- tice for neutralizing the acid. The method that was formerly given as "official" by the Asso- ciation of Official Agricultural Chemists 1 is based upon the prin- ciples just discussed. Several difficulties which are experienced in carrying out the method have led to a change to a distillation method as the official one. One of the principal difficulties is the formation of a yellow colloidal solution of arsenious iodide when potassium iodide and hydrochloric acid are added to reduce the arsenic solution. This color makes impossible the exact removal of iodine by sodium thiosulphate. If* the analysis is performed carefully, as described below, this difficulty will disappear. Determination of Total Arsenic and of Copper in Paris Green. To 2 gm of Paris green in a 250-cc flask add about 100 cc of a 2-percent solution of sodium hydroxide. Boil until all of the green compound has been decomposed and only red cuprous oxide remains. Cool, filter into a 250-cc volumetric flask, washing the paper well, dilute to the mark and mix well. Reserve the cuprous oxide on the filter for the copper determination. Measure two or three portions of 50 cc of the solution into 250- cc flasks and concentrate by boiling to about half the original volume. Cool to 60, add 10 cc of concentrated hydrochloric acid and 1 gm of potassium iodide. Mix and allow to stand for about ten minutes. From a burette carefully add sodium thiosulphate 'solution until the iodine is all reduced. Starch should not be added but care should be exercised in reaching the end point. Allow to stand for 5 minutes longer and if iodine color reappears carefully add more thiosulphate solution. Immediately add, as rapidly as can be done without loss by effervescence, 15 gm of sodium bicarbonate, free from lumps. Titrate at once with standard iodine solution, deferring the addition of starch until near the end point. Calculate the percent of total arsenic, expressed as arsenious oxide, in the Paris green. The residue of cuprous oxide is treated on the filter with 5 cc of nitric acid, specific gravity 1.2, the solution being caught in a 250 cc flask. l Bur. Chem., Bull. 107. 18 274 QUANTITATIVE ANALYSIS Wash the paper well with hot water and proceed as directed for the standardization of sodium thiosulphate against metallic copper, page 268, beginning with "Dilute to 25 cc and boil ." Calculate the percent of copper in the Paris green. The result may also be expressed as cupric oxide, if desired. If preferred, the solution of cuprous oxide in nitric acid may be diluted and electrolyzed, after boiling to expel nitrous acid. This determination is described on page 156. The present official method 1 for the determination of total arsenic in Paris green is based upon the volatility of arsenic trichloride with steam. The sample is dissolved in concentrated hydrochloric acid in a distilling flask. Cuprous chloride is added to reduce any pentavalent arsenic. The distillate containing the arsenous chloride and hydrochlo- ric acid is absorbed in cold water. The excess of acid is then neutralized with sodium hydroxide, sodium bicarbonate is added in excess and the arsenic is 'titrated with standard iodine solution. The reactions involved have already been discussed. Determination of Total Arsenic. Official Method. Prepare the following standard solutions: (a) Arsenous Add. Dissolve 2 gm, accurately weighed, of pure arsenous oxide in a beaker by boiling with about 200 cc of water and 10 cc of concentrated sulphuric acid. Cool the solution, transfer to a 500 cc volumetric flask, dilute to the mark and mix well. Keep in a stoppered flask or bottle. N (6) Iodine Solution, ^Q. Mix by grinding in a porcelain mortar 6.35 gm of pure iodine with 12.5 gm of potassium iodide. Dissolve in water, filter into a 1000 cc volumetric flask, dilute to the mark and mix well. Standardise against solution (a) as follows: Pipette 50 cc of the arsenous acid solution into a 1000 cc Erlenmeyer flask, add 400 cc of water, then gradually add 10 gm of sodium bicar- bonate. Mix and titrate at once with the iodine solution, adding 5 cc of starch solution when the slow disappearance of the iodine color indi- cates that the end point is nearly reached. Calculate the value of the iodine solution in terms of arsenous oxide, As20a. For the determination of total arsenic in Paris green the apparatus shown in Fig. 69 is necessary. A is a distilling flask having a capacity of 250 cc and fitted with a 50 cc dropping funnel. The capacities of the Erlenmeyer flasks B, C and 1 J. Assoc. Off. Agr. Chem., Vol. II, No. 1, Pt. 2, p. 5. OXIDATION AND REDUCTION 275 D are 500 cc, 1000 cc and 100 cc, respectively. B and C are surrounded by cracked ice and contain 40 cc and 100 cc, respectively, of water. D contains 50 cc of water which serves as a trap. The upright tube of flask B reaches to the bottom of the flask and acts as a safety valve, preventing liquid from drawing back from B when the distillation slackens or stops. Place 5 gm of cuprous chloride in the distilling flask. Calculate the theoretical weight of Paris green that would be equivalent to 250 cc of the standard iodine solution. Weigh this amount and rinse into the distilling flask by means of 100 cc of concentrated hydrochloric acid. Distill until only about 40 cc of liquid remains in the distilling flask, FIG. 69. Apparatus for arsenic distillation. then add 50 cc of concentrated hydrochloric acid through the dropping funnel and distill. Continue this process until 200 cc of distillate has been obtained. Stop the distillation and rinse down the condenser and all connecting tubes into the flasks. Rinse the contents of all three of the receiving flasks into a 1000 cc volumetric flask. Allow the solution to attain the temperature of the room then dilute to the mark and mix well. Measure 100 cc of this solution into a 1000 cc Erlenmeyer flask, add two or three drops of phenolphthalein solution and nearly neu- tralize with a saturated solution of sodium hydroxide, leaving the solu- tion slightly acid. Add 10 gm of sodium bicarbonate and titrate the arsenic with standard iodine solution as directed for the standardization of the solution. Calculate the percent of total arsenic in the sample, expressing as arsenous oxide. CHAPTER X TITRATIONS INVOLVING THE FORMATION OF PRECIPITATES The completion of the reactions of neutralization depends upon the small ionization of one of the products (water). The completion of reactions of oxidation and reduction depends upon the relative potentials of oxidizing and reducing agents. Certain other reactions are made the basis of volumetric determinations, completed because of the formation of a precipitate. In some cases an indicator is added while in others the cessation of pre- cipitation with further addition of standard solution is the indicator. SILVER An example of titration without an added indicator is to be found in Gay-Lussac's 1 method for silver. This method is one of the oldest of those analytical methods that have survived to the present day and, while it is not now extensively used because it is somewhat troublesome in the matter of execution, it is one of the most exact of all known volumetric processes. It depends upon the titration of the solution of a silver salt by a standard solution of sodium chloride. The very small solubility of silver chloride renders the reaction practically complete. The con- verse of this method may be used for the determination of chlorine, bromine, or iodine in soluble halides. Exercise : Preparation of Standard Solutions. Calculate the weight of pure sodium chloride that is equivalent to 5 gm of silver, weigh this quantity, dissolve in distilled water and dilute to 1000 cc in a volu- metric flask. Make a second solution by diluting 100 cc of this solu- tion to 1000 cc. Record the silver equivalent of 1 cc of each of these solutions. Determination. Silver may be determined in any alloy that contains no other metal forming insoluble chlorides but the approximate percent 1 Instruction sur 1' essai des matieres d' argent par la voie humide. Paris, 1832. 276 THE FORMATION OF PRECIPITATES 277 of silver should be known. A silver coin may be used. United States silver coinage contains approximately 90 percent of silver. Weigh enough of the alloy to give 0.5 gm of silver, place in a 250 cc flask having a ground glass stopper and dissolve in 10 cc of a mixture of equal volumes of water and concentrated nitric acid. Both water and acid must be tested and found free from chlorine. Boil to expel oxides of nitrogen, assisting this action by drawing air through the flask by means of a filter pump. Add to the solution in the flask exactly 99 cc of the more concentrated standard salt solution, stopper and shake until the pre- cipitated silver chloride flocculates and settles readily. Add from a second burette the more dilute standard solution, 0.5 cc at a time, allowing the solution to run down the sides of the flask and observing whether turbidity is produced. Shake the flask if more silver chloride is formed and continue the addition of the dilute standard solution until the last 0.5 cc fails to produce a visible precipitate in the clear, superna- tant liquid. Do not use the last 0.5 cc in the calculation. It may sometimes happen that the percent of silver in the alloy is not known with sufficient accuracy and either too much or too little of the more concentrated solution is used. In the first case the first addition of the dilute solution fails to produce a precipitate while in the second case an unduly large quantity of the dilute solution is required to reach the end point. In either case the determination should be begun again, the proper alteration being made in either the weight of sample taken or the volume of concentrated standard solution. From the results of the titration calculate the percent of silver in the alloy. In the determination of silver by the method of Volhard 1 an inorganic indicator is added to the solution. The silver should be in the form of nitrate, a solution of a ferric salt, acidified to suppress hydrolysis, is added and the silver is titrated by a standard solution of potassium thiocyanate or ammonium thio- cyanate. Silver is precipitated as silver thiocyanate: AgNO 3 +KCNS-AgCNS+KN0 3 . When all of the silver is removed from the solution an additional drop of the standard solution of thiocyanate produces the red color of soluble ferric thiocyanate: Fe(N0 3 )3+3KCNS->re(CNS)3+KN0 3 . Mercury thiocyanate is insoluble in dilute nitric acid and mercury must therefore be absent. The color of salts of copper, 1 J. prakt. Chem., [2] 9, 217 (1874). 278 QUANTITATIVE ANALYSIS nickel and cobalt obscures the end point and these metals should be absent although as much as 60 percent of copper may be present. The converse of this method may be used for the determina- tion of the thiocyanate radical. Exercise : Preparation of Solutions. Make a solution of silver nitrate, 1 cc of which contains 0.005 gm of silver. Standardize gravimetric- ally by precipitating and weighing silver chloride, or by Gay-Lussac's volumetric method. Make 1500 cc of a solution of potassium thiocyanate or ammonium thiocyanate by weighing 2 percent more than the calculated quantity of salt required to make 1 cc equivalent to 0.005 gm of silver. Make 100 cc of a solution (saturated without heating) of ferric ammonium sulphate, adding enough nitric acid to remove turbidity and to cause the red color to give place to pale yellow. Standardize the thiocyanate solution as follows: Measure 35 cc of the silver nitrate solution into a beaker or Erlenmeyer flask, dilute to about 75 cc, add 1 cc of ferric ammonium sulphate solution and titrate with the thiocyanate solution until a permanent red tint is obtained. Determination. Weigh not more than 0.25 gm of a silver alloy con- taining no mercury, nickel or cobalt and not more than 60 percent of cop- per and place in a 250 cc flask. Dissolve in 10 cc of a mixture of equal volumes of concentrated nitric acid and water, boiling to expel oxides of nitrogen. Cool, dilute to about 75 cc and titrate exactly as in the standardization of the thiocyanate solution. Calculate the percent of silver in the alloy. HALOGENS AND THE CYANIDE RADICAL Volhard's method also applies to the determination of the halogen hydracids and cyanogen. A measured excess of stand- ard silver nitrate solution is added, precipitating all of the chlorine, bromine, iodine or cyanogen. The excess of silver nitrate is determined by titration by standard thiocyanate solution by the method already described. In the original method the precipitated silver halide was not removed by filtra- tion before titration of the excess of silver. Rosanoff and Hill have shown 1 that the silver chloride reacts with the red soluble ferric thiocyanate, which is produced at the end point, as follows : 3A g Cl+Fe(CNS) 3 -*FeCl 3 +3AgCNS, *J. Am. Chem. Soc., 29, 269 (1907). THE FORMATION OF PRECIPITATES 279 This occurs to an appreciable extent, even though the solu- bility of silver chloride is less than that of silver thiocyanate. Rosanoff and Hill found that as much as 43 percent of ammonium thiocyanate is changed in two minutes by reaction with silver chloride. It is therefore necessary to remove the precipitate by filtration before the final titration. Determination. Use the standard thiocyanate and silver nitrate solutions prepared for the preceding exercise. Weigh enough of a soluble chloride, bromide, iodide or cyanide to be equivalent to about 40 cc of the silver nitrate solution. Dissolve in a small amount of water, acidify with nitric acid and add 50 cc of the standard solution of silver nitrate. Filter and wash thoroughly and titrate the excess of silver nitrate by standard thiocyanate solution. Calculate the percent of halogen or cyanogen in the sample. A method for the direct titration of the halogens by standard silver nitrate solution is described on page 397 in the discussion of water analysis. ZINC Ferrocyanide Method. The ferrocyanide titration of zinc has been practised for a long time, and many modifications 1 of the details of the method have been published. Concerning the accuracy of this method there has been considerable controversy, particularly as it applies to the determination of zinc in ores. However, most of the errors have been traced to methods used in separating zinc from interfering metals, preceding the titra- tion. The modified Waring method 2 is described below. Most of the important zinc ores are soluble in acids, although the aluminates require fusion with potassium pyrosulphate. In the acid solution silica is rendered insoluble by evaporation, as otherwise zinc silicate might precipitate. After evaporation with sulphuric acid to expel nitric acid the solution is boiled with a piece of aluminium, which precipitates lead, copper, cad- mium and bismuth, or such of these as are present, these metals all lying below aluminium in the electrochemical series. Iron is not precipitated but is reduced to the ferrous condition. In 1 J. Ind. Eng. Chem., 4, 468 (1912). 2 J. Am. Chem. Soc., 26, 4 (1904) and 29, 262 (1907). 280 QUANTITATIVE ANALYSIS the clear solution sulphuric acid is neutralized by sodium bicar- bonate and then formic acid is added in slight excess. In pres- ence of this weakly ionized acid zinc is precipitated by hydrogen sulphide. Zinc sulphide, so separated from interfering metals which would form insoluble ferrocyanides, is redissolved in hydrochloric acid and titrated with a standard solution of potassium ferro- cyanide, the following reaction first taking place : 2ZnCl 2 +K 4 Fe(CN) 6 ->4KCl+Zn 2 Fe(CN) 6 . (1) Zinc ferrocyanide, so formed, does not flocculate readily but as more ferrocyanide is added to the hot solution a potassium zinc ferrocyanide is precipitated : 3Zn 2 Fe(CN) 6 +K 4 Fe(CN) 6 -^2K 2 Zn 3 [Fe(CN) 6 ] 2 . (2) Equations (1) and (2) may thus be combined: 3ZnCl 2 -h2K 4 Fe(CN) 6 -^6KCl+K 2 Zn 3 [Fe(CN) 6 ] 2 . (3) As indicator, a solution of uranium acetate or nitrate or of ammonium molybdate is used on an outside test plate. The yellow or brown color that is produced by a slight excess of ferrocyanide with ammonium molybdate is of unknown composi- tion. When uranium salts are , used a brown ferrocyanide of uranium is formed. The ferrocyanide titration of zinc is easily performed and is fairly accurate if the details of the standardization of the solution and of the determination are watched closely. All things con- sidered the gravimetric method, weighing zinc as pyrophosphate, is to be preferred. This method is described on page 507. Determination of Zinc in Ores. Weigh 0.5 gm of powdered ore and brush into a 250-cc casserole. Add 5 cc each of concentrated nitric and hydrochloric acids, cover and heat to decompose the ore. Finally boil to expel all red oxides of nitrogen, then remove the cover and rinse this and the sides of the casserole. Cool and add 5 cc of concentrated sul- phuric acid. Evaporate until sulphur trioxide fumes are freely evolved, the casserole being held in the hand and agitated during the process of evaporation. Cool, add 50 cc of water and warm until only insoluble gangue remains. Bend a strip of heavy aluminium foil, 2.5 cm wide and 14 cm long, into a triangle and place in the solution. Boil for 10 minutes, which should remove all color except a faint green due to ferrous sul- THE FORMATION OF PRECIPITATES 281 phate. Filter through a paper containing a piece of aluminium, into a 500-cc flask containing a rod or strip of the same metal, and wash with hot water. Add a drop of methyl orange and neutralize with a solution of sodium bicarbonate (about 5 percent). Barely restore the pink color by adding a 20-percent solution of formic acid, a drop at a time, then add an excess of 5 drops. Dilute to 200 cc and then add 4 gm of ammonium thiocyanate, unless iron is known to be present in only a small amount. Remove the rod of aluminium and insert into the neck of the flask a rubber stopper, through the single hole of which passes a tube leading to the bottom of the flask. Connect the tube with a Kipp generator for hydrogen sulphide and heat the solution to boiling. With the stopper loosely fitted, pass a stream of gas through the boiling solution until most of the zinc has been precipitated, then push the stopper in so that the pressure of the gas from the generator will aid absorption. When the white zinc sulphide settles readily remove the stopper and rinse. Filter through paper and wash with hot water, but without attempting to remove adhering precipitate from the flask. Finally place the paper and precipitate in this flask and add 10 cc of concentrated hydrochloric acid and 50 cc of water, allowing the acid to act upon any precipitate adhering to the tube. Warm until all of the sulphide in the flask is in solution and boil to remove hydrogen sulphide. Drop in a bit of litmus paper and neutralize the acid with ammonium hydroxide, then add 3 cc excess of concentrated hydrochloric acid. Rinse the solution into a 500-cc beaker, dilute to about 250 cc, heat nearly to boiling and titrate with standard potassium ferro- cyanide, using a drop of nearly saturated uranium nitrate or acetate solution or of a 2-percent ammonium molybdate solution, on a white test plate as indicator. Near the last the solution is stirred and tested after each drop of standard solution is added. When a yellow or brown color finally appears it will be found that two or three of the tests immediately preceding this one will develop a color after standing for a few minutes. When this is the case the burette reading corresponding to the earliest positive test for ferrocyanide is taken as the amount of solution equivalent to the zinc. Calculate the percent of zinc in the ore. The solution of potassium ferrocyanide should be made so that 1 cc is equivalent to 0.005 gm of -zinc. It must be standardized against zinc, zinc oxide or zinc sulphate of known purity, following exactly the same method that has been outlined for the ore, beginning with the dissolving of zinc sulphide. Record the concentration of the solution in terms of zinc equivalent to 1 cc of ferrocyanide solution. PART II ANALYSIS OF INDUSTRIAL PRODUCTS AND RAW MATERIALS In most of the exercises in the preceding portion of this book determinations have been made of single constituents of various substances and interfering substances have usually been either absent or capable of being removed with comparative ease. Standard methods have been employed and attention has been centered upon the chemical principles underlying the method and the proper manipulation. In the pages that follow the student will become acquainted with the application of these and other determinations to the testing and analysis of some materials which are of importance to our industrial life. Such materials are often quite complicated in composition and most varied procedures are necessary in a determination of their industrial value. The chemist will then find it necessary to have at his command all of the chemical principles and methods of analysis that have already been learned and to apply these to an intelli- gent study of the material under examination. He will also be prepared to take up other methods of testing. Some of the tests are purely physical but they are, in industrial practice, applied by the chemist and not by the physicist because the former is usually engaged in the analysis of the same or similar materials. Other analytical determinations are empirical, rather than exact, in their nature but must be made with the same degree of care and attention as the determinations involving definite elements or compounds. 283 CHAPTER XI ROCK ANALYSIS CARBONATE MINERALS The most important and abundant of the carbonate minerals are the calcites and the dolomites. The calcites consist essen- tially of calcium carbonate and the dolomites of double carbon- ates of calcium and magnesium but these compounds seldom or never occur in a pure state in nature. Iceland spar i one of the best-known examples of a nearly pure natural variety of calcium carbonate, yet in many samples of Iceland spar substances other than calcium carbonate occur in appreciable amounts. For pur- poses of geological investigation there is usually required a com- plete analysis with the utmost accuracy that can be attained. For technical purposes this is not the case. The mineral is to be used for a given industrial purpose where the essential constituent is the one of chief importance and where impurities are important only to the extent that they may reduce the percentage of the essential constituent or that they exert an undesirable influence 'in the industrial operation to which the mineral is to be submitted. The particular application of the mineral to the industrial proc- ess will determine which impurities are of considerable and which are of minor importance. Those of minor importance are fre- quently grouped, with no attempt at separation, into certain arbi- trary classes. For example a limestone may be used as a source of quick lime, as a flux in iron smelting, as a paving material, as a building stone, as a raw material for. hydraulic cements, or for any one of a variety of other purposes. All limestone contains more or less of material insoluble in acids, consisting chiefly of various silicates and of quartz. For the first purpose named these substances are important only as they act as diluents of the essential calcium carbonate, unless they occur in relatively large 284 ROCK ANALYSIS 285 quantities. For such a purpose the analysis would be so made as to include all such materials as simply " insoluble" or "silicious matter/' no separation of the components being made. If the limestone were to be used as a flux in the smelting of iron ore, the nature of this insoluble material should be more exactly determined, since it not only reduces the actual percent of cal- cium carbonate but also may contain substances that themselves require a flux or that may even add very objectionable impurities, such as sulphur or phosphorus, to the iron itself. For paving or building material the physical properties of the mineral are of great importance and the chemical analysis might be considerably condensed. As another example of such empiricism in analysis, may be men- tioned the usual report on calcium. This element is usually pre- cipitated as the oxalate. It will readily be understood, however, that if barium or strontium is present and not previously sepa- rated it will also precipitate and will be included in the finally weighed oxide. Unless it is known that barium or strontium is present in more than very small amounts the percent of " cal- cium" alone is made a part of the report for technical purposes, strontium or barium serving the same purpose as does calcium. This, obviously, involves a slight error, not only in the naming of the element but in the percent as well, because the factors for these three metals in their oxides are all different. For exact scientific purposes the separation and determination of all ele- ments or radicals may be necessary while for technical purposes the analysis will be ordered according to the use to which the sub- stance is applied. This is an example of the so-called " proxi- mate" analysis, as distinguished from the " ultimate" analysis. It is important to note that the term "proximate" does not imply carelessness in working or neglect of sources of error. It should not even convey the idea of inexact figures, but merely grouping together of more than one substance to be reported by one generic term, as, for example, "insoluble matter" above. The proximate analysis of coal will include the determination of percents of " volatile combustible matter," "fixed carbon," "ash," and "moisture," yet each one of these terms covers many substances which are all determined together with no attempt at a separation into the ultimate constituents, simply because the 286 QUANTITATIVE ANALYSIS figures so determined serve a useful purpose in fixing a valuation on the coal. It was formerly the custom to report the analysis of acids, bases and salts, not as radicals but as anhydrides. Calcium carbonate would be reported as calcium oxide and carbon dioxide, sulphuric acid as water and sulphur trioxide, etc. This custom has now largely fallen into disuse in most lines of analytical chemistry but has been retained in the analysis of minerals. The ultimate analysis of carbonate minerals is exhaustively and scientifically treated in a bulletin of the U. S. Geological Survey and only reference to this will be made. 1 The exercise to follow will deal with the analysis made with industrial ends in view. This exercise will be the student's introduction to separations in quantitative analysis. Heretofore the work with the solution has terminated with the filtration and the removal of the precipitate. The filtrate could contain nothing but impuri- ties and by-products of the reaction and therefore could be of no further importance to the analyst. In the next and in many later exercises the filtrate must be carefully conserved because it contains substances still to be determined. The quantity of wash liquid must be made as small as possible, not merely to minimize its solvent action upon the precipitate but also because the washings must be added to the chief filtrate and the total bulk must not be excessive for subsequent operations. Even with the exercise of great care in this regard an occasional con- centration of the solution by evaporation is necessary in order to reduce its volume to a workable value. Another point that will here appear for the first time is that many of the elements or radicals that must be separated and determined are present in the mineral in relatively small quanti- ties. The student has been accustomed to a rapid appearance of a considerable quantity of precipitate and if this should not appear when the appropriate reagent is added he is likely to conclude that none of the substance is present and to pass to the next determination. None of the constituents ordinarily present in a given mineral or other complex material should be assumed to be absent. The reagent should be added and sufficient time allowed for the precipitation to become completed, remembering 1 U. S. Geol. Surv., Bull. 700, by W. F. Hillebrand. ROCK ANALYSIS 287 that precipitation starts and proceeds slowly from very dilute solutions. Even when no precipitate is finally visible it is the safest plan to filter, wash and ignite the paper in a weighed cruci- ble, when a small amount of precipitate will often be detected^ when otherwise it would have been weighed with the next precipitate to be produced. Analysis of Carbonate Mineral. Read again the discussion of sam- pling on page 9 and apply this to the preparation of a sample of limestone, dolomite or other carbonate mineral, for analysis. The small sample finally used should weigh about 10 gm and should pass a sieve having 100 meshes in each linear inch. Carbon Dioxide. Determine carbon dioxide exactly as directed on page 134, noting that if dolomite is under investigation solution will proceed rather slowly while the acid is cold. It is obvious that hydro- chloric acid must be used since considerable quantities of calcium are present and the solubility of calcium sulphate is not large. Silica or Insoluble Matter. The residue from the carbon dioxide determination may be used for this determination but it is better to use new samples. Weigh duplicate portions of 0.5 gm each into casseroles. Dissolve in 5 cc of concentrated hydrochloric acid, covering the cas- serole while the mineral is dissolving. Rinse down the cover glass and the sides of the casserole and evaporate to dryness on the steam bath. Heat the dry material at dull redness until the organic matter has been oxidized, leaving the residue white or reddish brown from iron oxide. Cool, add 5 cc of concentrated hydrochloric acid and warm until all soluble matter has passed into solution. Dilute to about 50 cc with hot water, boil and filter into a Pyrex beaker, using particular care in removing all of the residue to the filter paper, since the white casserole makes this process somewhat uncertain. Wash the residue free from chlorides with hot water, collecting the washings and filtrate in the same beaker. The total volume, after filtration and washing, should not be greater than 100 cc. If it is greater than this amount it should be con- centrated by evaporation. Place the paper and residue in a weighed platinum crucible, burn the paper and then ignite for 10 minutes over the blast lamp. Report the percent of silicious matter. If silicious matter amounts to more than 0.5 percent it should be separated into its constituents. In this case add to the residue in the crucible 2 gm of sodium carbonate and fuse over the blast lamp until the silicate is com- pletely decomposed, as shown by the cessation of effervescence. Cool, place the crucible in a casserole containing 50 cc of water and warm until the material is completely dissolved or disintegrated. Carefully add 288 QUANTITATIVE ANALYSIS to the covered casserole concentrated hydrochloric acid until efferves- cence no longer occurs, then remove the crucible and rinse. Evaporate the solution and heat at about 120 for 10 minutes. Add 5 cc of con- centrated hydrochloric acid and warm until soluble matter is dissolved, then dilute with 5 cc of water and filter on an extracted paper. Wash with hot water until chlorides are completely removed, adding the filtrate and washings to the original solution of the mineral. Ignite the residue and paper in a weighed platinum crucible, weigh and report as silica. Iron and Aluminium. If the solution has a volume greater than 100 cc it should be evaporated to concentrate to about this volume. Drop into the solution a very small bit of litmus paper and then add dilute ammonium hydroxide, stirring, until the solution is distinctly basic, avoiding undue excess of ammonium hydroxide. Boil for 5 minutes or until the odor of ammonia is faint. Filter through an extracted paper and wash until free from chlorides, adding the washings to the filtrate. Remove the paper from the funnel, fold and burn in a weighed porcelain crucible. Burn the paper at a low temperature in presence of an excess of air, inclining the crucible to facilitate oxidation. Weigh, and if the amount of the oxides is not greater than 0.5 percent report as aluminium oxide and iron oxide. If more than this amount is present a separation is usually made. In this case add 1 gm of potassium acid sulphate to the crucible containing the oxides of iron and aluminium. Fuse at a relatively low temperature until violent effervescence has ceased then heat to redness until the oxides have been completely dissolved. Cool the crucible and dissolve the mass in hot water. Reduce the iron and titrate with standard potassium dichromate or potassium permanganate according to the methods already learned. Manganese. Add bromine water to the filtrate and washings from iron and aluminium until a yellow color is produced, then boil. If manganese is present it will precipitate as brown manganese dioxide. The quantity is usually small but it must not be disregarded. Filter, wash free from chlorides and ignite. Weigh the oxide MnsC^ and calculate as MnC>2, assuming that the manganese was originally pres- ent in this form. (This is an arbitrary assumption because manganous carbonate is of common occurrence.) Calcium. Acidify the solution with hydrochloric acid and concen- trate to about 100 cc, boiling until all bromine is removed. Add a bit of litmus paper, then ammonium hydroxide until basic. Heat to boiling and add, drop by drop with stirring, 10 cc of a saturated solu- tion of ammonium oxalate, or enough to precipitate all of the calcium. Determine the calcium as directed on page 81 or 253, with the following addition, designed to complete the separation of magnesium: Filter ROCK ANALYSIS 289 the precipitate of calcium oxalate and wash once with hot water but without making any attempt to transfer completely to the filter. Place the beaker under the paper and add to the precipitate on the filter enough concentrated hydrochloric acid to dissolve all calcium oxalate^- 2 cc should be sufficient. Wash the paper thoroughly with hot water, precipitate the calcium once more and determine as already directed. Add the filtrate from the second filtration to that from the first. Calcu- late the percent of calcium oxide in the sample. Magnesium. Determine the magnesium in the filtrate and washings as was done in the case of a magnesium salt. The determination is discussed on page 112. Report the percent of magnesium oxide. Sodium and Potassium. These elements are not often present in more than traces in the carbonate minerals and their determination is not often required for industrial purposes. If such a determination is to be made, a new sample of mineral should be used. Follow the pro- cedure directed under the analysis of silicate minerals, page 294. If the analysis has been made with care and every substance present has been determined the sum of the percents of all of the constituents of the mineral should be 100. This will serve as a check upon the accuracy of the work but the sum will rarely be exactly 100. The omission of the determination of other substances present in small quantity will give rise to a negative error, while imperfect washing and other experimental errors will summate as a positive error, so that the sum of all percents may be either greater or less than 100. The student should be able to work so that the sum of all the errors should not be greater than 1 percent. SILICATE MINERALS Silica, as a constituent of various simple and complex silicates, is distributed widely in the earth's crust. Associated with other minerals or in a nearly pure form silica itself is also to be found. These minerals are only slightly soluble in acids or bases and their analysis requires a preliminary decomposition by some agent which will react at elevated temperatures. When silicon dioxide or a silicate is heated with an alkali carbonate to the point of fusion the corresponding alkali, silicate is produced, carbon diox- ide is evolved and whatever heavy metals may have been origi- nally present as silicates are left in the form of oxides. The alkali silicates are soluble in water (as colloids) and most of the metallic oxides so produced are soluble in hydrochloric acid. The pre- 19 290 QUANTITATIVE ANALYSIS viously insoluble mineral is, by this means, obtained in solution and the ordinary analytical processes will henceforth apply. All of the natural silicates may be regarded as being derived from silicon dioxide, the anhydride of the various silicic acids. These acids are not known in the free state but their existence may be supposed from the composition of the salts. Thus H 2 Si0 3 = H 2 O.Si0 2 , H 4 SiO 4 = 2H 2 O.Si0 2 , H 6 Si 2 7 = 3H 2 0.2Si0 2 , H 4 Si 2 O 6 = 2H 2 O.2SiO 2 , EUSisOs =2H 2 0.3SiO 2 , H 2 Si 2 O c =H 2 O.2Si0 2 . These may be taken as the acids from which various natural silicates are derived. Kaolin, the essential constituent of the various impure clays, is Al 2 Si 2 O7-2H 2 O, a salt of H 6 Si 2 07. The felspars are double silicates derived from the acid H^SisOs- As examples of the felspars may be mentioned orthoclase, KAlSi 3 Os, and albite, NaAlSi 3 8 . Fusion of orthoclase with sodium carbonate causes reactions which may be simply represented thus : 2KAlSi 3 08+6Na 2 C03-^K 2 SiO3+5Na 2 SiO3+2NaA10 2 +6C0 2 . The completion of the reaction is assured by the presence of a considerable excess of sodium carbonate. According to the terms of the mass law the reaction should be completed by simply heating at a sufficiently high temperature to decompose com- pletely any carbonates that may be formed by the decomposition. The mass of silicates resulting from the fusion may be decom- posed by hydrochloric acid but if this is not preceded by disin- tegration and solution of the water soluble parts by hot water the result of such treatment will be to form a protective coating upon lumps of the fusion, thus retarding the action of the acid. Upon addition of hydrochloric acid to the mixture of substances after treatment with water, oxides of earth and alkaline earth metals form soluble chlorides, while the alkali silicates are de- composed with formation of alkali chlorides and silicic acid. The first separation occurs in the removal of the silicic acid which must first be converted into the less soluble silicon dioxide. This conversion is partly, but imperfectly, accomplished by evapora- tion to dryness and heating to about 120. Rehydration readily occurs and the silica partly redissolves because of its marked tend- ency toward the formation of hydrosols. This tendency is diminished by long heating at high temperatures since such treat- ment results in incipient fusion and change into the irreversible ROCK ANALYSIS 291 colloid, silicic acid, the production of the latter being promoted by the presence of strong acids. It is practically impossible to com- pletely separate silica by one evaporation and filtration, a small proportion invariably returning to the solution. By evaporating the filtrate, heating, and again filtering, all but a trace of silica may be removed. The residue of silica is never pure but con- tains small amounts of oxides of iron, aluminium and calcium. In order to correct the error arising from this cause the precipitate is treated by hydrofluoric acid, which converts silica into the gaseous silicon tetrafluoride. After volatilization of this and of the hydrofluoric acid the residue is weighed and its loss reported as silica. After the separation of silica the metals will be determined by the usual methods, such as those used in the analysis of carbon- ate minerals. It will readily be seen that, after the fusion of the silicate with sodium carbonate or potassium carbonate, a deter- mination of the alkali metals in this portion of the sample will be without significance. Other- methods must be employed for decomposing the silicate. Two such methods are in general use. J. L. Smith Method. The method of J. Lawrence Smith 1 depends upon the action of calcium chloride upon silicates at about 800, resulting in the formation of alkali chlorides and sili- cates of calcium and other metals. The sample is intimately mixed with ammonium chloride and precipitated calcium carbon- ate, and gently heated. The reaction that occurs might be repre- sented as follows, assuming orthoclase to be the silicate. +2NH 3 +6C0 2 . This can be only an approximate representation of what has really happened during the heating. Ammonium chloride dis- sociates at about 450 into ammonia and hydrochloric acid: NH 4 C1-NH 3 +HC1. The ammonia escapes while the hydrochloric acid combines with calcium carbonate: CaC0 3 +2HCl->CaCl 2 +H 2 O+CO 2 . x Am. J. Sci., [3] 1, 269 (1871). 292 QUANTITATIVE ANALYSIS That the decomposition of silicates is, in a large measure, due to calcium chloride is undoubtedly true. That calcium carbonate, as such, also plays an important part in the reactions would be inferred from the above interpretation of the reaction, since in this reaction only one-third of the calcium carbonate could form the chloride. The significance of this equation is lessened by the fact that it is a representation of a series of reactions that cannot well be tested. The silicate resulting from the decomposi- tion is probably not calcium metasilicate alone but is much more complex than this. After the decomposition is complete the mass is treated with hot water which dissolves chlorides of sodium, potassium and calcium, as well as those of other metals present. After filtra- tion the most of the calcium, iron and aluminium is precipitated by ammonium carbonate and ammonium oxalate; the solution is then evaporated in a platinum dish and heated to expel am- monium salts. If desired, sulphuric acid may be added to convert sodium, potassium and ammonium salts into sulphates, thus providing less liability of loss of sodium and potassium during the heating. In this case the Gladding modification of the Lindo method must be used or else the sulphates must be converted into chlorides by precipitating with barium chloride. If sul- phuric acid is not added the volatilization of ammonium salts must be conducted with greater care, since the chlorides of sodium and potassium are appreciably volatile at a temperature of bright redness. Read again the discussion of the determination of sodium and potassium on page 96. Hydrofluoric Acid Method. Another method for treating silicates without the addition of sodium or potassium carbonate is that of decomposing by means of hydrofluoric acid. The finely powdered silicate is moistened with concentrated sulphuric acid and then hydrofluoric acid is added. The silica is volatilized, upon warming, as silicon tetrafluoride. After evaporation of the excess of hydrofluoric acid and of sulphuric acid the residue is dissolved in water and the solution analyzed by practically the same procedure as is followed in the Smith method. Analysis of Silicate Mineral. To insure complete decomposition of the silicate it must be ground much more finely than is necessary for most other minerals. After it is pulverized to pass a 100-mesh sieve ROCK ANALYSIS 293 about 3 gm of the sample is ground in an agate mortar until it will a 200-mesh sieve. Moisture. Weigh about 0.5 gm of the silicate into a platinum cru- cible, and dry for one hour at 100 to 105, the loss being calculated as hygroscopic moisture. If combined water is also to be determined this can be done by heating to a temperature above 200 in a combustion tube in an atmosphere of dried carbon dioxide, absorbing the moisture in weighed tubes filled with calcium chloride. Silica. Mix with the sample about 5.0 gm of sodium carbonate, approximately weighed. Place the cover on the crucible but slightly toward one side so the contents may be observed. Heat gently at first in order to avoid violent effervescence and consequent loss by spatter- ing, gradually raising the temperature until the full heat of the burner fails to cause more than slight action. Place the crucible over the blast lamp and heat for 15 minutes after carbon dioxide has ceased to be evolved. Remove the lamp, lift the crucible by means of the tongs, using care to avoid contact of the latter with the fusion, and slowly rotate the crucible in such a manner that the fused mass will be spread over the sides of the crucible as it solidifies. After cooling place the crucible and its contents in a Pyrex beaker ; a platinum dish or a porcelain casserole and cover with hot distilled water. Digest until the entire mass has become disintegrated and re- moved from the crucible. Cover the beaker and add, gradually, from a pipette, concentrated hydrochloric acid until the carbonates are com- pletely decomposed. Remove the crucible and cover by means of a stirring rod, rinsing well. Crucible tongs should not be used for this purpose unless they are tipped with platinum. The solution is now evaporated to dryness. If speed is essential and if a casserole has been used the greater part of the liquid may be evaporated by holding over a free flame, giving a rotary motion to the casserole to prevent spatter- ing or bumping. If other work may be carried on at the same time the solution should be evaporated over the steam bath. When completely dry heat for 15 minutes at about 120, cool and add 5 cc of concentrated hydrochloric acid and 50 cc of water, warm until soluble salts are dis- solved and filter the residue of silica at once, washing until free from chlorides, adding the washings to the filtrate. Evaporate the solution again to dryness, heat at about 120 and repeat the treatment with hydrochloric acid and water, filtering in a different paper. Place both papers with their residues in a platinum crucible, dry and burn the papers in the usual way, then ignite over the blast lamp for 20 minutes, cool and weigh. Small quantities of metal salts will have been retained in the residue. In order to correct for their presence the residue is moistened with one or two drops of sulphuric acid and 294 QUANTITATIVE ANALYSIS about 5 cc of hydrofluoric acid is added. The silicon tetrafluoride being volatilized, the acids are evaporated and the residue, which should be small, is heated over the burner. The loss in weight is taken to repre- sent silica. The residue of oxides is dissolved by warming with a few drops of hydrochloric acid and is added to the filtrate from the silica. Iron, Aluminium, Manganese, Calcium, and Magnesium. Concen- trate the nitrate, if necessary, to about 100 cc then determine iron, aluminium, manganese, calcium and magnesium exactly as directed for the analysis of carbonate minerals, cal- culating each as the oxide. Sodium and Potassium. Grind the sample in an agate mortar to pass a 200-mesh sieve and weigh 0.5-gm samples on counterpoised glasses. Place 0.5 gm of resublime4 ammonium chloride (free from sodium and potas- sium) in the clean agate mortar, add the weighed sample and grind together. Add 3 gm of powdered calcium car- bonate and grind again. Place a gram of powdered calcium carbonate on the bottom of the crucible and brush in the mixture already prepared. Add another gram of calcium carbonate to the mortar, grind to remove the last traces of sample from the mortar and' brush this in on top of the charge. This should not entirely fill the crucible. Adjust the crucible in a hole in a piece of asbestos board in such a manner as that the board comes just to the upper level of the charge. The upper part of the crucible will then act as a condenser for any vapors of alkali chlorides that might form in the lower part of the crucible. Heat the lower part of the crucible gradually until the evolution of ammonia becomes slow, then more strongly with the nearly full flame of two flat burners. This heating should be finally applied to all but the upper end. After 45 minutes of such heating the crucible is allowed to cool and the charge emptied into a platinum dish. The materials will usually be sintered together in a mass that has shrunken away from the crucible, the calcium carbonate in the bottom of the crucible serving to prevent sticking of the mass to the crucible. Actual fusion into a FIG. 70. Smith's crucible for the determination of alkali metals in silicates. ROCK ANALYSIS 295 slag that is not decomposed by water is an indication that the tempera- ture was too high or that more calcium carbonate should be used. Add 75 cc of water to the materials in the dish and warm until the mass is thoroughly disintegrated. Boil for a minute, allow to settle and filter. Add 50 cc more of water to the residue, boil, allow to settle and filter through the same paper, transferring all of the residue to the paper. Wash with hot water until a few drops of the washings show only a faint test for chlorides. To determine whether decomposition was complete in the crucible, dissolve the washed residue in hydrochloric acid. Any gritty residue must be retreated as before, as this indicates insufficient heating. Evaporate the combined filtrates and washings to a volume of 75 cc or less, then add 5 cc of dilute ammonium hydroxide and sufficient ammonium carbonate solution to precipitate all iron, aluminium and calcium present. Digest at a temperature just below the boiling point until the precipitate settles readily, then filter and wash once. Dissolve the precipitate in dilute hydrochloric acid and repeat the precipitation, washing the precipitate with hot water. Evaporate the two filtrates and all washings in a platinum dish. When dry, carefully heat over the free flame of the burner until all ammonium salts are volatilized, but without, at any time, allowing the dish to become red. Cool, dis- solve in not more than 10 cc of hot water, add a few drops of dilute ammonium hydroxide solution and 1 cc of saturated, recently prepared ammonium oxalate solution. Digest until the small amount of calcium oxalate settles, then filter and wash the paper, collecting filtrate and washings in the platinum dish. Add 1 cc of concentrated hydro- chloric acid and evaporate to dryness. Heat carefully to volatilize ammonium chloride. Cool in a desiccator and weigh as sodium chloride and potassium chloride. Dissolve in a few cubic centimeters of hot water and transfer to another dish. Dry, ignite and weigh the first dish with any matter that was not soluble and subtract this weight from the one already observed. The difference is sodium chloride and potassium chloride. Determine the potassium in this mixture as directed on page 102, omitting the washing with ammonium chloride solution, since we are here dealing with chlorides instead of sulphates of sodium and potassium. From the weight of potassium chlorplatinate found, calculate (a) the percent of potassium oxide in the silicate and (b) the weight of potassium chloride in the mixed chlorides. Subtract the latter weight from the total sodium and potassium chlorides and from the remain- ing sodium chloride calculate the percent of sodium oxide in the silicate. The sum of all oxides determined should approximate 100 percent. Since experimental errors accumulate and since small quantities of other 296 QUANTITATIVE ANALYSIS substances than those named will generally remain undetermined, the sum of all will usually be less, rather than more than 100 percent. It is customary with many chemists to report this discrepancy as "undeter- mined" but it will be remembered that such a percent is not merely that of undetermined matter but that it also includes the accumulated errors of all of the other determinations. CHAPTER XII FUELS COAL As might be expected from a knowledge of its origin, coal is made up of a large number of organic and some inorganic com- pounds. The attempt to separate and identify these compounds has long engaged the attention of chemists and geologists but comparatively little progress has been made in this direction. The reason for the lack of success in this attempt is that most chemical methods of analysis involve the breaking up of organic substances and the disappearance of the original forms. Such an analysis would prove extremely useful from the stand- point of geology and would, no doubt, be of service in indus- trial applications. In the absence of adequate methods for this purpose, the examination of coal is made with one or more of three ends in view: (1) To determine the geological origin of the coal; (2) to determine its adaptability to various industrial uses, such as steaming, heating, manufacture of pro- ducer gas or illuminating gas, coke, tars, etc.; or (3) to determine its fuel value in heat units per unit weight of coal. The methods used are classified under the head of proximate analysis or of ultimate analysis. The proximate analysis of materials may be defined as "the determination, not of elements or radicals but of groups of compounds falling within approximate limits of composition and having similar properties." The ultimate analysis is, as the word indicates, a determination of the ele- mentary composition. It is made with greater difficulty, involves more expensive apparatus and requires longer time than the proximate analysis, often gives no more useful informa- tion than is given by the proximate analysis and is, for this reason, less frequently made. Methods for the analysis of coal were standardized by a com- 297 298 QUANTITATIVE ANALYSIS mittee of the American Chemical Society in 1899. 1 A joint com- mittee of the American Chemical Society and the American Society for Testing Materials has since made two reports, 2 revising, in many respects, the methods of the original committee. Sampling. The correct sampling of coal is a very difficult process. In a substance showing such a lack of uniformity in composition it is obvious that the sample must be selected with extreme care if the results of the analysis are to express the average composition. For a thorough discussion of this matter, and especially with regard to the selection of samples from cars, heaps, mines, etc., the student is referred to a paper by Bailey. 3 When the laboratory sample is finally selected it should be sealed in an air-tight container to prevent changes in the moisture content. This container may be a tin or galvanized can with a screw top, sealed with a rubber gasket and adhesive tape, or fruit jars, tightly sealed. In the chemical laboratory this sample must be further pul- verized and divided in order finally to obtain a sample which is small enough in quantity and fine enough to serve for the actual analysis, and which shall also be representative of the original material. When the sample reaches the laboratory it is treated to the usual progressive crushing and quartering until the proper fine sample is obtained. During this process there is a continued loss of moisture so that the fine sample finally obtained has not the same percentage composition as the sample as received. If the coal as received is taken as a basis its composition will not be the same as that of another coal of different moisture content but otherwise identical with the first. The only scientifically correct method is to calculate all other percents to a dry coal basis, reporting the moisture in the sample as received. It is sometimes necessary, however, to base all calculations upon the coal as received. In this case moisture must be determined at once, as well as after crushing and preparing the sample for analysis, as already mentioned. Loss of moisture occurs at all times when the coal is exposed. The sample is ground fine 1 J. Am. Chem. Soc., 20, 281 (1898); 21, 1116 (1899). 2 J. Ind. Eng. Chem., 6, 517 (1913); 9, 100 (1917). 3 J. Ind. Eng. Chem., 1, 161 (1909). FUELS 299 enough to serve for the analysis and a determination of moisture is made upon this sample. The difference between the percent of moisture in the original coarse sample and the fine sample provides a basis for the calculation of the analytical results, to the original coal basis. Proximate Analysis. The proximate analysis of coal as usually carried out includes the determination of moisture, volatile combustible matter, coke, " fixed carbon" and ash. The figures thus obtained give considerable information as to the geological age of the coal, determine its fitness for industrial uses and provide a basis for an approximate calculation of fuel value. Moisture. The accurate determination of moisture in coal is a difficult operation. If the coal is heated to 100 or above this temperature there is danger of loss of volatile constituents other than moisture, whether these were originally present or were formed upon heating. Oxidation also takes place during heating. This may result in either gain or loss in weight, according to whether the greater part of the oxygen is retained in solid com- pounds or is lost as volatile ones. The usual tendency is toward a gain in weight so that this error to some extent compensates the loss first mentioned. Such compensation is, however, not to be depended upon and such a determination is therefore not ideal. A method that removes some of these objections is that of drying at ordinary temperatures over sulphuric acid, under diminished pressure. The partial pressure of oxygen being reduced to a negligible quantity, oxidation is entirely prevented. There is also no tendency toward breaking down of the non- vola- tile organic constituents with the production of simpler volatile ones. The method seems to give low and somewhat variable results, however, due to the low rate at which moisture is lost by the coal toward the last of the drying process. Volatile Combustible Matter, Fixed Carbon and Coke. When coal is subjected to dry distillation out of contact with air variable quantities of volatile products are expelled and a residue of inorganic -matter and non-volatile carbon is left. This residue is the "coke," the carbonaceous portion of the coke being called "fixed carbon." It should be understood that the original coal did not consist of free carbon and volatile organic compounds, but that heating at high temperatures resulted in the formation of 300 QUANTITATIVE ANALYSIS such substances by a decomposition of the non-volatile bitumens composing the coal. Such a decomposition of complex com- pounds into simpler and more volatile compounds is technically known as "cracking." Cracking is a somewhat indefinite proc- ess and the products depend to a large extent upon the tempera- ture and time of heating. The volatile products of the distilla- tion of coal have a very important application in the industries. The non-volatile coke is also an industrial material, being exten- sively used as a fuel for reducing ores, for the manufacture of producer gas and water gas and for many other purposes. The relative quantities of volatile matter, fixed carbon and mineral matter are of importance in the determination of the fitness of various coals for various industrial uses. On account of the variation in results obtained by variation in the manner of heating it becomes difficult or impossible to make an intelligent comparison of different coals unless a standard method is adopted for their examination. The control and measurement of the temperature at which they are heated involves the use of a pyrometer and an easily controlled furnace. It is also difficult entirely to exclude air during the heating and a variable oxidation occurs The student is again reminded that volatile combustible matter, fixed carbon and coke are arbitrary classifications of substances produced by an arbitrary method and that they have little scientific meaning except as they provide a basis for comparisons. Ash. Most of the inorganic matter of coal is left behind as "ash" when the coal is burned but the compounds so remaining are not identical with those of the original coal. Oxidizable substances are oxidized, sometimes to a variable extent depending upon the degree of excess of oxygen in the atmosphere in which the coal is burned. Decomposable compounds are also changed at the furnace temperatures. The first class of changes is illustrated by iron disulphide, the essential compound of iron pyrites. This is oxidized to ferric oxide and sulphur dioxide, the latter escaping with the other volatile products of com- bustion. Carbonates of the earth and alkaline-earth metals and hydrated silicates are examples of compounds which are decomposed by heating. Such carbonates are nearly or com- FUELS 301 pletely converted into oxides and the hydrated silicates (such as clay) are partly deprived of their water of hydration. Because of such changes as these the percent of "ash," as determined by burning a weighed sample of coal and weighing the residue, must not be regarded as indicating directly the percent of inorganic matter or "non-coal" in the original ma- terial, but rather as the percent of non-combustible matter that would be left after the combustion of the coal in air. This in- accuracy of expression affects the report on fixed carbon as well as that of ash, since the former is obtained by subtracting the sum of ash, moisture and volatile combustible matter from 100. In the majority of cases the error is comparatively small and the correction is scarcely worth making. But if coal carries unduly large amounts of pyrite or of carbonates such a correction may be necessary, if a report on fixed carbon is to have any significance. When iron disulphide is burned in air the sulphur is lost as sulphur dioxide and iron is left as ferric oxide: Upon the assumption that total sulphur represents the sulphur of iron pyrite, the loss on burning each molecule of iron disul- phide results from the substitution of three atoms of oxygen for four of sulphur. That is, (4X32) -(3X16) =80. 80 5 Therefore 77^ or ~ of the total sulphur percent should be added l^o o to the ash percent as a correction for use in obtaining the correct percent of fixed carbon. Such carbonates as limestone lose carbon dioxide when strongly heated. It is not by any means certain that the con- version to calcium oxide, etc., is complete at the temperatures employed in burning the coal and if a correction is to be made it is better to add a few drops' of sulphuric acid to the ash, evapo- rating the excess of acid and heating gently before a weighing is made. The ash which has been treated with sulphuric acid now contains sulphates instead of carbonates, and it is heavier on this account. Upon a separate sample of coal a determination 302 QUANTITATIVE ANALYSIS of carbon dioxide of carbonates is made by any of the standard methods (see pages 128 to 137). The gain in weight through sulphation is represented by the difference between the molecular weights of sulphur trioxide and carbon dioxide. That is, 80-44 = 36. 36 9 Therefore ' -r or ^- of the percent of carbon dioxide of car- 44 11 bonates should be subtracted from the percent of sulphated ash to obtain the original carbonated inorganic matter. It should be noted that the term "ash" is a misnomer when applied to the percent corrected as above described. The latter should more properly be called "inorganic matter" or "non- coal." Fusing Point of Ash. The composition of coal ash has an important bearing upon the industrial uses of the coal. This is particularly true with regard to the fusibility of the ash, which is the principal property upon which depends the forma- tion of clinker. If the inorganic matter of coal is of such com- position as to produce an easily fused ash, a slag will form in the furnace before combustion is complete, with the result that more or less coal matter is so glazed and protected from oxida- tion as entirely to prevent its combustion. Thus there is a troublesome clinker formed which clogs the grates and interferes with proper stoking, and also a variable waste of combustible matter. This action is, of course, more pronounced in cases of high furnace temperatures. Clinker will often contain a very high percent of combustible matter, the latter being deter- mined by grinding the solid clinker and burning at temperatures under the fusing point of the ash. In most high grade coals the inorganic matter consists very largely of silicates of aluminium, all more or less refractory. Clay is typical of these. If substances capable of yielding basic oxides upon ignition are present, slag formation is promoted and clinker forms. The most abundant of such base formers are iron pyrite and calcium carbonate. On this account a high sulphur content is usually indicative of ability to form a fusible ash. A similar indication is provided by a red ash, since most ^of such color comes from ferric oxide. The presence of lime- FUELS 303 stone is betrayed by neither high sulphur nor red ash. Con- sequently a coal may contain little sulphur and form a white ash and yet clinker badly, the calcium carbonate (or oxide) readily uniting with refractory silicates to produce complex silicates which fuse or soften at the prevailing furnace temperatures. On account of the objectionable character of clinker a deter- mination of fusibility of ash often becomes highly significant. Of course such a complex mixture as that which composes coal ash can have no very definite fusing point, so that observations of the behavior of the ash at high temperatures must be made according to somewhat arbitrary standards. The ash is ground, mixed with water and (usually) some organic binder, and molded into pyramids similar to the well known "Seger cones," which have long been used for measuring furnace temperatures. The "cone" is really a triangular pyramid having a base perpen- dicular to one of the plane sides. When such a piece is heated to its softening temperature it leans toward the vertical side. The temperature at which the apex curves down to touch the supporting plane is taken as the "fusing point." Fieldner and Hall 1 and Fieldner and Feild, 2 working in the Bureau of Mines, found that the fusing temperature so observed varies somewhat according to the size, shape and inclination of the test piece, the fineness of grinding the ash before molding and the atmosphere in which the pyramid is heated. The effect of the surrounding atmosphere is to cause changes in the composi- tion of the ash, the latter reacting with oxidizing or reducing gases. The nearest imitation of furnace conditions was found in a mixture of equal volumes of hydrogen and water vapor. This keeps the iron of the ash in the ferrous condition, in which form it is usually found in furnace clinker. This determination is described with great detail in the papers already cited. If the results are to have any great value the determination must be made by a carefully standardized method and equipment, including the special furnace there described, or a similar one. Preparation of Sample, (a) When Coal Appears Dry. If the sample is coarser than 4-mesh and larger in amount than 10 pounds quickly 1 J. Ind. Eng. Chem., 7, 399 and 474 (1915). 2 Ibid., 7, 742 and 829 (1915). 304 QUANTITATIVE ANALYSIS crush it with a jaw crusher to pass a 4-raesh sieve and reduce it an a riffle sampler (page 14) to 10 pounds, then crush at once to 20- mesh by passing through rolls or an enclosed grinder and take, without sieving, a 60-gm total moisture sample, immediately after crushing. This sample should be taken with a spatula from various parts of the 20-mesh product and should be placed directly in a rubber-stoppered bottle. Thoroughly mix the main portion of the sample, reduce on the small riffle sampler to about 120 gm and pulverize to 60-mesh by any suitable apparatus without regard to loss of moisture. Mix and divide the 60- mesh sample on the small riffle until it is reduced to 60 gm. Preserve this in a rubber-stoppered bottle. Determine moisture in both the 60-mesh and the 20-mesh samples by the methods given under the head of moisture. (6) When Coal Appears Wet. Spread the sample on weighed pans, weigh and dry for 12 hours at room temperature or in a sp'ecial drying oven through which air circulates freely at 10 to 15 above the room temperature. Reweigh and continue the drying until the loss in weight is not more than 0.1 per cent per hour. Complete the sampling as with dry coal and calculate the percent of moisture by air drying. To find the total moisture in wet coal, as received, compute as follows: Let a = percent of moisture .by air drying, MZQ= percent of moisture in 20-mesh coal. , Then - 17^: -- +a = total moisture as received. 1UU The percents of the various constituents of the coal are de- termined on the 60-mesh coaL To compute these percents to the dry-coal basis or the as received bases, proceed as follows: Let M 6 o = per cent moisture in 60-mesh coal, M r = total percent moisture in coal as received, PW = percent of any constituent in 60-mesh coal, PO = percent of any constituent in dry coal, P r = percent of any constituent in coal as received. Then P. /100-J/A /100-Jf, Pr= 100 / Po= FUELS 305 Proximate Analysis. Procure a sample as directed in the preceding discussion. Moisture. The oven for drying the 20-mesh and 60-mesh samples ' must be so constructed as to provide a uniform temperature in all parts and a minimum of air space. Provision must be made for renewing the air in the oven at the rate of two to four times a minute, with air dried by passing through concentrated sulphuric acid. A convenient form of crucible for the moisture determinations and one which allows the ash determinations to be made on the same sample is a flat bottomed porcelain crucible. A fused silica crucible of similar shape may be used. In either case a well fitted aluminium cover should be provided. Glass capsules, with covers ground on, may also be used but ash determinations will then require different samples. Sixty-mesh Sample. Heat the empty crucible under the conditions under which the coal is to be dried, cover, cool in a desiccator over con- centrated sulphuric acid for 30 minutes and weigh. Place approxi- mately 1 gm of the sample in the crucible, cover and reweigh. Remove the cover and place the crucible in the oven, which is maintained at 104 to 110. Heat for 1 hour then cover the crucible, cool over sul- phuric acid for 30 minutes and weigh. Calculate the loss as moisture. Twenty-mesh Sample. Use 5-gm samples weighed with an accuracy of 2 mg and heat for IK hours. The procedure is otherwise the same as for the 60-mesh sample. Permissible differences in duplicate determinations: Same analyst Different analysts Moisture under 5% 0.2% 0.3% Moisture over 5% 0.3 0.5 Ash. Place the uncovered porcelain crucible containing the dried 60-mesh coal in a cold muffle furnace through which air may be drawn and gradually raise the temperature to between 700 and 750. When combustion appears to be complete remove the crucible, cover, cool in a desiccator, and weigh. Repeat the heating for 15-minute periods until the change in weight is not greater than 1 mg. Calculate the percent of ash. Permissible differences in duplicate determinations: Same analyst Different analysts No carbonates present -. . 0.2% 0.3% Carbonates present 0.3 0.5 Carbonates and pyrite present and more than 12 % ash 0.5 1.0 20 306 QUANTITATIVE ANALYSIS Volatile Combustible Matter. For this determination an electri- cally heated, vertical-tube furnace is desirable, although a muffle furnace, heated by gas or electricity may be used. If the tube furnace is available the vertical tube should be about l^j inches in diameter and 6 inches deep. The junction of a thermo-couple connected with a pyrometer meter should be placed immediately below the bottom of the crucible. The furnace is to be kept covered during a determination. If the determination of volatile combustible matter is not an essential part of the specifications under which the coal is bought a No. 4 Meker burner may be used for the heating. The platinum crucible should have a capacity between 10 and 20 cc, a diameter between 25 and 35 mm and a depth between 30 and 35 mm. The cover must fit closely. Weigh the crucible then add, as nearly as possible without careful adjustment, 1 gm of 60-mesh coal, cover the crucible and reweigh. If a furnace is used for the heating it should be brought to a temperature of 950 (20), and the crucible is placed on a support of platinum or of nickel-chromium in the furnace. If a Meker burner is used, adjust the flame so that its extreme height is 15 cm and support the crucible on a platinum or a nickel-chromium wire triangle so that the bottom is 1 cm above the top of the burner. After the luminous flame above the crucible has disappeared, tap the crucible cover lightly in order to seal more perfectly and thus guard against the entrance of air. Heat for exactly 7 minutes then remove the crucible from the furnace or flame without disturbing the cover. Cool in a desiccator and weigh. Calculate the loss as percent of total volatile matter, including moisture. Subtract the percent of moisture found in the 60-mesh sample and report the percent of volatile combustible matter. Svh-bituminous coal, lignite or peat are given a preliminary gradual heating for 5 minutes to expel part of the large quantity of volatile matter. They are then heated as above directed for 6 minutes. Permissible differences in duplicate determinations: Same analyst Different analysts Bituminous coals 0.5% 1 .0% Lignites 1.0 2.0 Fixed Carbon. Subtract from 100, the sum of moisture, volatile combustible matter and ash, corrected to the basis of either dry coal or as received. The remainder is the percent of fixed carbon on the basis considered. FUELS 307 Ultimate Analysis. The complete ultimate analysis of coal will include the determination of all elements. The com- plete analysis is not often made, the determination of sulphur, carbon, hydrogen and nitrogen being all that is usually required. Sulphur. Sulphur may exist in coal in one or more of four forms: elementary sulphur, inorganic sulphides (principally iron pyrite) inorganic sulphates and organic compounds. The accurate determination of the amount present in the different forms is difficult and not often required. Three methods have found considerable use for determining total sulphur in coal. The powdered sample may be gently heated with sodium carbonate (Atkinson's method 1 ) or with a mixture of sodium carbonate and magnesium oxide (Eschka's method 2 ) ; or it may be fused with sodium peroxide. In the first two cases air is the oxidizing agent while in the last sodium peroxide performs this function. The ultimate result in all three is that the organic portion of the coal is burned while sulphur is oxidized, either dioxide or trioxide being retained by the basic constituents of the added reagents, soluble sulphites or sulphates being produced. In the Eschka method magnesium oxide serves to provide a porous mixture into which air readily penetrates. It is necessary to provide against absorption of sulphur oxides from the surrounding atmosphere. For this reason ordinary illuminating or coal gas cannot be used for direct heating of the mixture unless it is first purified. Alcohol burners are suitable for this purpose but a muffle furnace, heated by electricity, is preferable. When the oxidation of the coal is complete the mixture is boiled with water and bromine, the latter to oxidize to sulphates any sulphites that may have been formed. After filtration the solu- tion is acidified and boiled, whereby bromates or hypobromites are decomposed and the bromine expelled from the solution: NaBrO 3 + HCl-NaCl + HBrO 3 , NaBrO + HCl-^NaCl + HBrO, HBr0 3 +5HBr-3H 2 0+3Br 2 , HBrO + HBr-H 2 -t Br 2 . 'J. Soc. Chem. Ind., 6, 154 (1886). . 'Chera. News, 21, 261 (1870). 308 QUANTITATIVE ANALYSIS It is necessary thus to decompose salts of oxyacids of bromine because of the tendency shown by barium sulphate to occlude these compounds at the moment of precipitation. Nitrogen. Nitrogen may be determined by combustion or by the Kjeldahl, the Gunning or the Kjeldahl-Gunning methods. The principles underlying the latter are discussed on page 513 and following. In the combustion method the coal is mixed with fine cupric oxide and is heated in a tube closed at one end. Oxidation occurs and the gases are passed through more heated cupric oxide to complete the oxidation of carbon monoxide or of gaseous hydrocarbons, then over heated copper, the latter to reduce oxides of nitrogen to elementary nitrogen. The mixture of carbon dioxide, water vapor, sulphur dioxide and nitro- gen is passed into sodium hydroxide solution where all gases except nitrogen are absorbed. The latter is collected in a eudiometer and measured, the weight being then calculated. A special form of eudiometer called the " nitrometer/' is useful for this purpose. This is shown in Fig. 71. In order to force the gases out of the combustion tube a short space in the closed end is filled with sodium bicarbonate. Carbon dioxide is evolved at will by heating this and it causes the expulsion of other gases from the tube. Carbon and Hydrogen. The deter- mination of carbon and hydrogen is made by combustion and absorption of water vapor and carbon dioxide in weighed tubes containing ap- propriate absorbents. The method is the same as that used in other connections for organic substances containing sulphur and nitrogen. For the absorption of water vapor calcium chloride or sulphuric acid may be used, the latter giving more nearly complete absorption while the former 'is more conveniently used. For FIG. 71. Schiff's nitrometer. FUELS 309 carbon dioxide, solid soda lime or a solution of potassium hydroxide is used, any of the standard forms of absorbing tubes or bulbs serving as containers. A combustion tube of glass, silica or porcelain is used for the decomposition of the coal. The relative merits of these materials were discussed in an earlier section (pages 37 and 42). For the present purpose the silica tube is most satisfactory and any suitable combustion furnace may be used. The tube should be 95 cm long and should project 10 to 15 cm beyond the furnace at each end. The combustion is effected by heating the coal in an atmosphere of oxygen. Since this also produces volatile organic compounds and some carbon monoxide, it is necessary to pass the gases through a solid oxidiz- ing agent in order to complete the oxidation. Cupric oxide is used for this purpose in the analysis of most organic substances, but when sulphur is present, as in all coals, sulphur oxides are not completely retained by cupric oxide and are therefore ab- sorbed in bulbs containing potassium hydroxide. In the com- bustion of such materials lead chromate is substituted for part of the cupric oxide. Lead sulphate is formed and this is not decom- posed by heating. A lower temperature is used to avoid fusing the material into the tube. Oxygen for the combustion may be made from manganese dioxide and potassium chlorate and stored in a large "gasometer" or it may be purchased in steel cylinders. Oxygen made by the first method named always contains chlorine oxides and it must be purified by passing through a concentrated solution of potas- sium hydroxide. Most of the commercial oxygen formerly obtainable in compression cylinders contained chlorine oxides, carbon dioxide and hydrogen, and such oxygen must be properly purified before it enters the combustion tube. Since the develop- ment of the manufacture of oxygen from liquid air it is possible to obtain gas having a high degree of purity. It is then only necessary to pass the oxygen through potassium hydroxide or soda lime in order to remove traces of carbon dioxide. In order to be able to control the flow a " gasometer" should be filled from the high-pressure cylinder and this used as the, supply for the combustion or else a special high-pressure control valve should be used on the tank. Oxygen. The direct determination of oxygen is difficult and 310 QUANTITATIVE ANALYSIS is seldom attempted. The percent is sometimes estimated by subtracting the sum of percents of all other elements, water and ash, from 100. This is but a rough approximation but is usually all that is required. (Jltimate Analysis of Coal: Sulphur. Eschka Method. Thoroughly mix two parts of light calcined magnesium oxide and one part of anhy- drous sodium carbonate. These materials should be as free as possible from sulphur. Place 3 gm of the mixture on a sheet of white glazed paper and then add 1 gm of 60-mesh coal, accurately weighed. Mix well, then transfer to a porcelain, silica or platinum dish, 2 inches in diameter and 1 inch deep. Cover with about 1 gm of the Eschka mixture. A flame of illuminating gas cannot be used for heating on account of absorption of sulphur dioxide from the flame by the basic mixture. A gas or electric muffle furnace is best for the purpose but an alcohol, gasoline or natural gas flame may be used. If a flame is applied directly, heat the dish in a slanting position on a triangle over a low flame until most of the volatile matter is driven off, then gradually increase the temperature and heat for 30 minutes or more, stirring occasionally, until all black particles have disappeared. If a muffle furnace is to be used, place the crucible in the cold furnace and gradually raise the temperature so that about 1 hour is required to reach 870 to 925. Maintain this temperature for l}4 hours, then cool in the furnace or in air that is free from gases containing sulphur compounds. After either treatment rinse the material into a 200-cc beaker, add 100 cc of nearly boiling water and digest on the steam bath for 30 min- utes, stirring frequently. Filter and wash the insoluble matter thor- oughly with hot water. The filtrate and washings should total about 250 cc. Add 20 cc of saturated bromine water to the solution and stir, then make slightly acid with dilute hydrochloric acid (shown by the failure of more acid to cause effervescence) and boil until all bromine is removed and the solution is colorless. Add two or three drops of methyl orange or methyl red and neutralize with 10 percent sodium hydroxide solution, then add 1 cc of approximately normal hydrochloric acid, or an equivalent volume of a solution of any other normality. Heat to boiling and add, dropwise and with stirring, 20 cc of 5 percent solution of barium chloride. Digest on the steam bath until the pre- cipitate settles readily and then filter and wash free from chlorides with hot water. Fold the paper, place in a weighed platinum, porcelain, silica or alundum crucible and burn the paper at a low temperature FUELS 311 and with free access of air (see page 93), finally heating to dull red- ness for 10 minutes. Cool and weigh. Calculate the percent of sul- phur in the 60-mesh coal. The residue of magnesium oxide, ash, etc., should be dissolved in hydrochloric acid and tested for sulphur. If any is found this must be determined quantitatively and added to the percent already found. A blank experiment should be performed, using all of the reagents of the regular experiment but omitting the first heating. Any sulphur that is so obtained from the reagents is to be subtracted from that found in the analysis of the coal. Permissible differences in duplicate determinations: Same analyst Different analysts Sulphur less than 2% 0.05% 0.10% Sulphur more than 2% 0.10 0.20 Carbon and Hydrogen. A tube combustion furnace of any of the approved types and about 75 to 80 cm long is necessary. The combus- tion tube may be of hard glass, silica or porcelain. It should be long enough to project for at least 10 cm at each end of the furnace, in order to prevent heating of the rubber stoppers that must be inserted in the ends. The internal diameter of the tube should be 12 to 15 mm. Since coal always contains sulphur, lead chromate must be used in the combustion tube and this is prepared by fusing, cooling and crushing about 100 gm. The largest pieces should be small enough to easily enter the tube. By sifting the crushed material, using a 40-mesh sieve, a finer grade will be obtained and this is used for mixing with the powdered coal. The combustion tube should have well rounded ends. It is filled according to the following directions, assuming that the length of the furnace is 75 cm and that of the tube is 95 cm. Into one end of the tube insert a closely fitting roll of copper gauze, 5 cm long, and push this in until a space of 10 cm is left at the end of the tube. Into the other end pour the coarsely crushed lead chromate until a space 50 cm long is filled. The material should be well settled but not packed in such a way as to obstruct the passage of gases. Insert another roll of copper gauze like the first, to hold the lead chromate in place. Another roll of copper gauze, 10 cm long, is inserted in such a way as to leave a space of 10 cm at the end of the tube, and a space of 5 cm between the two rolls. The latter space is for the boat containing the coal. The ends of the tube are closed by rubber stoppers carrying short glass tubes for connecting with the rest of the apparatus. The method of filling the tube is shown in Fig. 72. Fig. 73 shows the II' 312 QUANTITATIVE ANALYSIS method of assembling the complete apparatus. Entering oxygen and entering air are passed through cylinders A and A', re- spectively, containing a good quality of soda lime. The gases are next passed through U-tubes, B and B', containing fused, granular calcium chloride. These tubes are connected with the combustion tube by means of a three-way stopcock. Gases leaving the combustion tube first pass through two U-tubes C and D (preferably glass stoppered) containing calcium chloride, then through the carbon dioxide absorption bulbs E containing potassium hydroxide solution and calcium chloride. Following these tubes is a guard tube F containing -< calcium chloride, also an aspirator G. For the detailed direc- 8 tions for filling and connecting the various absorption tubes *S and aspirator, refer to the discussion of the determination of carbon dioxide in carbonates. ^ When the combustion tube and all parts of the apparatus are in order start a slow but steady current (three bubbles per J3 second in the bulbs) of air by means of the aspirator, then heat gradually the entire length of the combustion tube. The 3 drying tubes C and D and the potassium hydroxide bulbs g need not be in the train because this preliminary heating is ' for the purpose of thoroughly drying the contents of the .g combustion tube and oxidizing any organic matter with which o the tube might be contaminated. They are therefore re- moved, carefully wiped clean and dry, and are then closed "JJ and placed in the balance case. After they have stood " for 15 minutes, if ready to proceed with the blank test, these pieces are weighed. The temperature of the part of the tube containing lead chromate must not be higher than is indicated by dull redness although other parts may be heated to any temperature under the softening point of the tube. When moisture has been expelled from the tube to the extent that no condensation is noticed on the forward end the calcium chloride tubes C and D and the potassium hydroxide bulbs are weighed and placed in the absorption train. The flow of air is continued for 20 minutes, when the aspirator is stopped and the absorption tubes are again removed, stop- pered, placed in the balance case and weighed after standing for 15 minutes. If there is a gain of more than 0.5 mg in either the weight of the potassium hydroxide bulbs or the combined weights of the two calcium chloride tubes the entire operation must be repeated until there is no greater gain than 0.5 mg. r | FUELS 313 When this is the case, that part of the combustion tube which is at the left of the lead chromate is allowed to cool. Provide a porcelain or platinum boat, about 5 cm long and of the proper width for insertion into the combustion tube. In the bottom of this place a layer of powdered lead chromate 1 mm deep and then weigh into the boat about 0.5 gm of powdered coal which has been properly sampled and dried. Mix with a platinum wire. Remove the rubber stopper at the left end of the tube and quickly remove the roll of copper gauze (now largely oxidized to cupric oxide) by means of a wire hook and insert the boat, pushing the latter in until it touches the roll of cupric oxide which confines the lead chromate. Replace the first roll and the rubber stopper as quickly as possible, start a current of oxygen through the tube and gradually heat the cooled portion of the tnrr-tr-\ C D E F II Li G FIG. 73. Diagram of connections for combustion apparatus. tube. Volatile matter will escape from the coal but this will be com- pletely oxidized by the lead chromate in the forward part of the tube. Backward diffusion of volatile combustible matter will occasion no loss by condensation because the roll of cupric oxide behind the boat will serve to oxidize a small quantity of such gases. When all glowing of the coal has ceased, turn the three-way stopcock so that air is drawn into the tube, gradually lower the temperature of the left end so as to avoid cracking and continue the passage of air until about 1000 cc more of water has run out of the aspirator. Remove the absorption tubes and bulbs, close and allow to stand 15 minutes and then weigh. From the total gain in weight of the two calcium chloride tubes, due to absorption of water vapor, calculate the percent of hydro- gen in the coal. From the weight of carbon dioxide absorbed in the potassium hydroxide bulbs calculate the percent of carbon in the coal. Duplicate determinations should be made. If many samples are to be analyzed much economy of time will result from the use ofltwo boats and two sets of absorption tubes. As each experiment is finished another can be started and the combustion will proceed while the first set of tubes is standing in the balance and being weighed. 314 QUANTITATIVE ANALYSIS Fuel Value. After a decision has been reached as to the class of coal that is most suitable for a given industrial purpose the inquiry that is next in importance concerns the number of heat units that can be obtained from unit weight of the various grades of coal entering into that class. The custom of purchasing coal upon a tonnage basis at a contract price, with nothing more than the variety of coal named, is rapidly being displaced, by large consumers, by the method of purchasing upon a heat unit basis. A contract price is made, based upon a specified number of heat units per pound or ton and any deviation from this fuel value involves a corresponding alteration in the price. In scientific work the fuel value is always calculated as calories per gram or kilogram of fuel, while in industrial work it is gener- ally calculated as " British thermal units" per pouijd of fuel. The calorie is the quantity of heat required to raise 1 gm of water 1 C. in temperature. The British thermal unit (B. T. U.) is the quantity of heat necessary to raise 1 Ib of water 1 F. in tem- perature. The calculation of fuel values in either system in- volves equal weights of water and fuel and the relation -P ~, ;?., - 57- is therefore the relation of the centigrade degree 9 to the Fahrenheit degree. That is, cal per gmX = = B. T. U. per o Ib, and B. T. U. per lbXo = cal per gm. y This may be demonstrated as follows : Let a = temperature equivalent of 1 C.. b = temperature equivalent of 1 F., c = weight equivalent of 1 gm, d = weight equivalent of 1 pound. Then B. T. U. perlbx| = B. T. U. per gm; IB. T. U.= cal, ac therefore B. T. U. per Ib X~ = cal per gm, or B. T. U. per lbXn = cal per gm. y Conversely cal per gmX^B. T. U. per Ib. o FUELS 315 Calculation of Fuel Value from the Ultimate Analysis. The cal- culation from the ultimate analysis is based upon the assumption that the heat of oxidation of a compound is equal to the sum of the heats of oxidation of the elements composing it. The ele- ments of coal that are oxidizable are carbon, hydrogen, and sul- phur. Nitrogen is mostly evolved in the free state and inorganic matter other than sulphides is of little or no value for producing heat. Water is incombustible and absorbs heat in becoming vaporized, thus reducing the available heat energy. Oxygen is not only incombustible but, because of the fact that it is already combined with carbon and hydrogen it reduces the per- cent of these elements still available for combustion and therefore reduces the amount of heat that is available when the fuel is burned. The fuel value of elementary carbon is 8080, that of hydrogen is about 34,500 and that of sulphur is 2162 calories per gram, if it is understood that the products of combustion are cooled to ordinary temperature after combustion. This assump- tion is not realized in practice but it is customary to make the assumption in calculating fuel values. There is, of course, no way of knowing the manner in which the elements are combined in the organic compounds making up the coal substances. The arbitrary assumption is, however, made that all oxygen that is already contained in the dry coal is in combination with hydrogen. One-eighth of the percent of oxygen would then be the percent of hydrogen not available as fuel. The complete statement for calorific value, based upon these assumptions, is given in the following formula of Du Long: 34500 H-JO +8080 C+2162 S q = 1QQ - cal per gm, where H, O, C and S represent percents of hydrogen, oxygen, carbon and sulphur respectively. However this formula is hardly an accurate statement of avail- able heat. Before any compound can be burned to the oxides of the constituent elements the compound must be dissociated into its elements. This change will involve absorption or liberation (usually absorption) of heat energy and the amount of energy 316 QUANTITATIVE ANALYSIS change will depend upon the method of combustion. This is not known for coal and the correction cannot be applied. There is also a fuel value (positive or negative) for the mineral matter contained in the coal, since the ash left upon burning is not the same as the original matter. This heat must also be omitted from the calculation because its quantity is not known. The changes taking place during the conversion of wood into the various forms of coal are represented by the following approxi- mate figures for the composition of the combustible portion, disregarding small quantities of elements other than those des- ignated, as well as ash. Carbon Hydrogen Oxygen Wood 49 6 43 Lignite 70 5 25 Bituminous coal 86 4.5 9.5 Anthracite coal 95 2 3 The gradual loss of volatile matter involves a relatively large loss of oxygen. This fact explains the steady rise in fuel value as the coals progress toward the anthracite, although the decrease in the ratio of hydrogen to carbon remaining would lead one to expect a fall in fuel value. Calculation of Fuel Value from the Proximate Analysis. Many attempts have been made to devise a formula that will serve for calculating fuel value from the results of the proximate analysis. Such a formula must necessarily be purely empirical and must fail in many cases because coals of practically identical proximate composition may vary widely in ultimate composition and constitution. The chief value of all such formulas lies in making possible an approximate valuation of the fuel where neither the ultimate analysis nor calorimetric determinations can be obtained. Following are a few examples of these formulas. The results of such calculations must be used with great caution. Formula of Haas: 156.75 [100- (% ash+% S+%H 2 O)]+40.5X % S = B.T.U. per Ib. FUELS 317 Formula of Goutal: 1 82 C + AM = cal per gm when C = % fixed carbon, M = % volatile combustible matter and A = a coefficient whose value is fixed by the volatile combustible matter as follows: M = 2-15 15-30 30-35 35^0 A =130 100 95 90 Formula of Gmelin: [100- (% H 2 0+% ash)] 80-6CX% H 2 = cal per gm, in which C is a coefficient which varies with the percent of mois- ture as follows: Percent moisture C <3 - 4 3-4.5 + 6 4.5-8.0 + 12 8.5-12.0 + 10 12-20 + 8 20-28 + 6 >28 + 4 If fixed carbon is calculated upon a basis of true coal, dry and ash free, the following table may be used: FC B.T.U. .per Ib FC B. T. U. per Ib 100 14,500 68 15,480 97 14,760 63 15,120 94 15,120 60 14,580 90 15,480 57 14,040 87 15,660 54 13,320 80 15,840 51 12,600 72 15,660 50 12,240 Determination of Fuel Value by Means of the Calorimeter. The best laboratory method for the determination of fuel value is by' the use of one of the standard calorimeters. Practically all of these depend upon the measurement of rise in tempera- ture of water, caused by the combustion of fuel within a closed vessel or "bomb" immersed in the water. Combustion is best iCompt. rend., 135, 477 (1902). 318 QUANTITATIVE ANALYSIS effected by electrical ignition in an atmosphere of compressed oxygen since oxidation is complete and no reactions can occur other than those of ideal combustion. The original bomb calo- rimeter of Berthelot 1 has been improved and changed in such a manner as to make it a practical instrument for industrial as well as for purely scientific laboratories. Successful modifications are those of Mahler 2 and Emerson. 3 Parr 4 has also perfected a fuel calorimeter in which the oxidizing agent is sodium peroxide. The advantages of this instrument for industrial testing are chiefly due to the fact that it dispenses with the use of compressed oxygen and much of the accessory apparatus for filling the bomb. Charging and firing become a comparatively simple matter and the instrument may be operated by persons who have limited scientific training. The great fault of this and similar instru- ments comes from the reaction of the products of combustion with the excess of sodium peroxide or with sodium monoxide formed. Such reactions as the following occur: H 2 O+Na 0->2NaOH, Materials in the ash also react and form sodium salts. All of these reactions involve heat liberation or absorption and this cannot be exactly calculated because the composition of the coal is never exactly known. The best that can be done is to deter- mine experimentally approximate corrections which will apply to different classes of coal. The sum of these corrections may often be as large as 10 percent of the total rise in temperature during a determination and the uncertainty is so great as to render the instrument of questionable value for any but the most approxi- mate determinations. The instruments using compressed oxy- gen, while usually more expensive, are best for accurate fuel testing, even for the works laboratory. Fuel Value. If a calorimeter is available the heat .units should be determined experimentally. Calculation may also 1 Ann. chim. phys., [5] 23, 160 (1881); [6] 10, 433 (1887). 2 Chem. Zentr., 63, 889 (1892). a J. Ind. Eng. Chem., 1, 17 (1909). * J. Am. Chem. Soc., 22, 646 (1900); 29, 1606 (1907). FUELS 319 be made from the analytical results, using the formulas already given as well as others that have been proposed. Comparison with calorimetric data will indicate the degree of usefulness of the formulas. Following is a description of the Emerson calorimeter and also directions for making the determination of fuel value. Bomb. The bomb is made of steel, consisting of two cups joined by means of a heavy steel nut. The two cups are machined FIG. 74. Emerson's calorimeter. at their contact faces with a tongue and groove, the joint being made tight by means of a lead gasket inserted in the groove. The lining is of sheet nickel, platinum or gold, spun in to fit. The bomb is made tight with a milled wrench or spanner. The pan holding the combustible is of platinum or nickel, and the support- ing wire of nickel. Calorimeter. The jacket is a double walled copper tank, the space between the walls being filled with water. The calo- rimeter can is made as light as is possible, of sheet brass, nickel plated. Stirring Device. The stirrer is directly connected to a small motor and is enclosed in a tube to facilitate its action in circulating 320 QUANTITATIVE ANALYSIS the water. The stirrer is mounted on a post on the calorimeter jacket as is the thermometer holder. FIG. 75. Section of Emerson calorimeter. Ignition Wire. Unless ignition of the fuel requires a very high temperature a platinum resistance wire is suitable. For ignition of such substances as are used in determining the water equivalent of the calorimeter (naphthalene, cane sugar, etc.) or of anthracite FUELS 321 coal an iron wire is more certain in its action because it burns and produces a higher temperature. When iron wire is used a correction of 1600 calories per gram of wire is subtracted from the total calories obtained from the fuel combustion. This is the heat of oxidation of the iron. Formation of Nitric Acid. When coal is burned in air prac- tically all of the nitrogen is liberated in the elementary form. On account of the high concentration of oxygen in the calorimeter bomb a considerable portion of the nitrogen is oxidized and the products dissolve in the water which is formed by the combustion of hydrogen. A dilute solution of nitric acid is thereby formed. This gives rise to a positive error in the observation of fuel value, the magnitude of the error depending upon the extent to which nitric acid is formed. As a rule the error is small and it may be ignored for ordinary fuel testing but if a correction is to be made the nitric acid is titrated by standard base, at the end of the experiment. The heat of formation and solution of nitric acid from elemen- tary nitrogen is 230 calories per gram. It is convenient to use a standard solution of base, 1 cc of which is equivalent to 5 calories. The normality of such a solution is = 0.3450 N. 230X0.06302 The number of cubic centimeters of base required to titrate the nitric acid in the bomb after the combustion is multiplied by 5, the product being subtracted from the observed calories. Formation of Sulphuric Acid. A similar error results from the formation and solution of sulphuric acid. In ordinary combus- tion of coal, organic sulphur as well as the sulphur of pyrites is oxidized to sulphur dioxide, which leaves the furnace as a gas. On the other hand, in the calorimeter some sulphur trioxide is formed and this dissolves as sulphuric acid. The higher degree of oxidation of sulphur, as well as the solution of the acid that is formed, yield additional heat and this also should be subtracted from the observed calories. The application of this correction is not so simple as that for oxidation of nitrogen. A determination of total sulphur, such as is ordinarily made in the analysis of coal, does not give any 21 322 QUANTITATIVE ANALYSIS data as to the amount existing in the coal as inorganic sulphates which, obviously, do not develop heat through oxidation. Be- sides, not all of the remainder of the sulphur is oxidized to the highest form. As the correction is usually small it is scarcely advisable to attempt any calculations upon such an uncertain basis. Radiation. Radiation or absorption of heat by the calorim- eter may be avoided by making the calorimeter "adiabatic. " This may be done in a number of ways, three of which will be mentioned. (1) The water in the surrounding jacket may be heated by electrical means, so as to keep pace with the rise in temperature of the calorimeter water. This is the most satisfactory method, although somewhat complicated and expensive apparatus is required. (2) The water in the jacket may be warmed by chemical action. By Richards' method a basic solution is used to fill the jacket and an acid is run in from a burette at a rate which depends upon the rate of change in temperature of the calorim- eter water and upon the concentration of the acid. The acid solution may be standardized in terms of the number of calories liberated by the action of each cubic centimeter upon the base, in which case the proper rate of addition is more easily determined. (3) The jacket of the calorimeter may be evacuated, on the principle of the Dewar flask, the transfer of heat outwardly then being limited to that which occurs through conductivity of the glass of the jacket. This would appear to be the least trouble- some method but it has not worked well in practice. Radiation Corrections. If adiabatic conditions cannot be maintained several methods for making radiation corrections are available. (1) The combustion may be begun as far below atmospheric temperature as it is to end above it. By this means absorption of heat in the first half of the experiment would appear to balance radiation during the last half. This is the roughest sort of approximation and it would not serve for ordinarily accurate work. (2) The rate of change of temperature may be observed for a certain period before firing and for another period after the FUELS 323 calorimeter water has absorbed all of the heat from the bomb. The average of these rates is then considered to be the mean rate of absorption or radiation of heat for the entire experiment and if this is multiplied by the time elapsing between the firing and the maximum absorption the net gain or loss during the entire observation period is given. This method is very commonly employed and it gives a very close approximation to the true correction. (3) Observations are made in the same way as in method (2). In addition the time, a } required for six-tenths of the total rise in temperature is observed, also the time, 6, for the remaining rise. Instead of averaging the two radiation (or absorption) rates the preliminary rate, E\, is multiplied by a and the final rate, Rz, by b. The corrected rise is then T+Ria+R 2 b, where T = total rise, and Ri and R 2 are regarded as positive for falling temperatures and negative for rising temperatures. The observation of the time, a, is subject to some uncertainty when the temperature is rising rapidly and on this account the method is not so easily applied as is method (2). It will rarely be found that the difference between the corrections calculated by these two methods will differ by more than 0.2 percent and as this is well within the permissible variation method (2) is recommended for all but the most refined work. (4) The Regnault-Pfaundler method approaches theoretical accuracy more nearly than any of the methods already de- scribed. For a discussion of this method, see White : Gas and Fuel Analysis (International Chemical Series) page 224. Time -temperature Curves. Three types of time-tempera- ture curves are produced, according to whether the experiment is (a) begun and finished below room temperature, (6) begun below and finished above or (c) begun and finished above. These types are illustrated in Fig. 76. The relative slopes of the ends of the curves represent RI and R 2 . It will be observed that these slopes are easily determined in curve (6) but that it is especially difficult to decide as to what temperature should be taken as -the maximum produced by the fuel combustion, in the experiment represented by curve (a). 324 QUANTITATIVE ANALYSIS Conditions represented by curve (6) are to be obtained when possible. Determination: Heat of Combustion of Solid Fuels. Place the lower hall of the bomb in the holder, and the fuel pan in the wire support, after having wired the fuse wire according to Fig. 75. Extend the wire across the pan, allowing it to dip sufficiently to be in contact with the fuel, which is later to be placed in the pan. The wire must in no case touch the pan. The fuse wire should be placed in series with two 32-c.p. lamps in parallel when the 110- volt power circuit is used for firing. Time, Minutes FIG. 76. Time-temperature curves. The fuel used is sampled and powdered according to directions already given. Fill a weighing bottle with the prepared sample, and weigh accurately to one-tenth of a milligram. Pour from this into the pan in the bomb, until the pan is approximately half full. Weigh the bottle again, and the difference between the above weighings gives the net quantity of the fuel in the bomb. This weight should be greater than 0.5 gm and not more than 1.2 gm. For hard coal the maximum charge should be not greater than 1 gm. Hard coal should not be as finely divided as soft coal. The upper half of the bomb is now placed in position and tbje nut is screwed down as far as may be by hand, care being taken not to cross the threads. The shoulder on the upper half of the bomb, over which FUELS 325 the nut makes bearing contact, should be lubricated with oil. Extreme care should be taken that no oil or grease is deposited on the lead gasket. The bomb is now ready to be filled with oxygen. The nipple is coupled to the oxygen piping by means of the attached hand union and after the connection of the bomb to the oxygen piping is accomplished the hand set screw on the table is tightened. In handling the bomb, care should be taken not to tip or jar it, as fuel may be thrown from the pan. The spindle valve on the bomb is opened one turn and then the valve on the oxygen supply tank is very cautiously opened. The pressure gauge should be carefully watched and the tank valve so regulated that the pressure in the system shall rise very gradually. When the pressure reaches 300 Ib per square inch, the tank valve is closed and the spindle valve immediately afterward. The bomb should be im- mersed in water immediately to detect any possible leaks. The bomb is now ready for the calorimeter, which is prepared as follows: 1900 gm of distilled water, weighed or measured in a calibrated flask, is placed in the calorimeter can at a temperature about 1.5 below the jacket temperature (which should be in the proximity of the room temperature). The bomb is then placed in the calorimeter and the stirrer and thermometer are lowered into position as indicated by Fig. 75. The thermometer is immersed about 3 inches in the water. The bulb of the thermometer should not touch the bomb. The terminals of the electric circuit used for firing should now be attached. Care should be taken that neither the bomb nor the stirrer touches the sides of the can. The stirrer is now started and allowed to run 3 or 4 minutes to equalize the temperature throughout the calorimeter. Readings of the thermometer are now taken for 5 minutes (reading to 0.001 or 0.002 every minute) at the end of which time the switch is turned on for an instant only, which will be found sufficient to fire the charge. In course of a few seconds the temperature begins to rise rapidly and approximate readings are taken every minute until the rise becomes slow, more accurate readings then being taken. After a maximum temperature is reached and the rate of change of temperature is evidently due only to radiation to or from the calorimeter, the readings are continued for an additional 5 minutes, reading every minute. These readings before the firings and after the maximum temperatures are necessary in the computation of the cooling correction. The time elapsed from the time of firing to the maximum temperature should be, in no case, more than 6 minutes. When through with the run, replace the bomb in the holder and allow the products of combustion to escape through the valve at the top of 326 QUANTITATIVE ANALYSIS the bomb. Unscrew the large nut and clean the interior of the bomb. The inside of the nut should be kept greased, also the threaded part at the top of the lower cup. Immediately after each run, the lining of the bomb should be washed out with a cloth moistened with a dilute solution of ammonium hydrox- ide and then with water. When the apparatus, after using, is to be left for several hours or more before making another test, the linings should be removed and the inner surface of the bomb slightly coated with oil. This oil under the linings should be removed when next preparing the bomb for use, as an excess of it may be ignited with a possible resulting injury to the linings. Heavy Oils, Coke, Hard Coal, Etc. The determination of the heat of combustion of heavy oils, such as crude petroleum, and also of coke and extremely hard coals, is best made by mixing with a ready burning combustible, such as a high-grade bituminous coal or pure carbon. This auxiliary combustible facilitates the complete combustion of the whole mixture in the case of coke and hard coal, and with the heavy oil it acts as a holder and prevents rapid evaporation of the oil. The auxiliary combustible should be placed at the bottom of the pan and the coke, coal or oil sprinkled over it. The carbon or other auxiliary combus- tible should be dried with extreme care and carefully standardized as to the resulting rise in temperature per gram in the calorimeter when completely burned. Calculation. First plot a smooth curve, using temperatures as ordinates and time as abscissas. Use only the straight portions of the ends of the graph for calculating RI and E.2- The difference between the temperature at maximum and the temperature at firing gives directly the total rise in temperature in the calorimeter. To this rise a cooling correction must be applied, which is computed as follows: The change in temperature during the preliminary 5 minutes of reading, divided by the time (5 minutes) gives the rate of change of temperature per minute, due to radiation to or from the calorimeter, and also any heating due to stirring, etc. This factor is RI and in like manner the readings taken after tempera- ture change has become uniform give R2. The two rates of change of temperature give the existing conditions in the calorim- eter at the start and at the finish of the run. The algebraic signs of RI and R 2 will be (+) for falling temperatures and ( ) for rising temperatures. Therefore, the algebraic sum of the two FUELS 327 rates, divided by two, will give the mean value of the rate of change of temperature during the entire run, due to radiation or absorption by the calorimeter. This value multiplied by the time from firing to maximum will give the total cooling correction. The cooling correction is thus expressed : p I T> X time from firing to maximum temperature. 2 This quantity is either added to, or subtracted from, the appa- rent rise taken from the data of the run, according to its sign. The corrected rise of temperature divided by the weight of fuel used, will give directly the rise per gram of fuel. This rise per gram is multiplied by the weight of water plus the "water equivalent." This figure is furnished by the manufac- turers or it may be determined by use of a standard fuel, as naphthalene or cane sugar. The product is calories per gram of fuel, which is the result to be obtained. The result in calories per gram of fuel, multiplied by the factor 1.8 gives B.T.U. per pound of fuel. The final expression for fuel value is then - = cal per gm, where T = total rise from firing temperature to maximum, S = gm of coal, e = water equivalent of calorimeter, RI and R 2 having the significance already mentioned. In using this formula it must always be remembered that RI and R 2 are regarded as radiation rates and that if the temperature is rising they must be given negative signs. Since cal per gmX 1:8 = B.T.U. per Ib, if the latter quantity is desired in most cases the product (1900+e)X1.8 should be calculated at the beginning. Permissible differences in duplicate determinations: Same analyst Different analysts 0.3% 0.5% 328 QUANTITATIVE ANALYSIS GAS MIXTURES The separation and exact determination of gases may be accomplished by using various gravimetric and volumetric methods. Certain gases may be absorbed in suitable reagents, the absorption product being precipitated and determined gravi- metrically. Examples of this class of methods have been met in the determination of the halogens (page 122). Sulphur dioxide may be absorbed in a basic solution, oxidized by bromine and precipitated as barium sulphate. Carbon dioxide has already heen determined by absorption in a weighed solution of potassium bydroxide. Numerous other examples will suggest themselves. On the other hand many gases can be absorbed by reagents in which they can be determined volumetrically. For example chlorine may be absorbed by a solution of potassium iodide and the liberated iodine titrated by standard sodium thiosulphate (page 264) ; carbon dioxide may be absorbed in a standard solu- tion of a base and the solution titrated by a standard acid in presence of phenolphthalein ; sulphur dioxide may be absorbed and titrated by standard iodine solution, etc. For commercial mixtures of gases these methods are not often used because the time required for a complete analysis is too long. The analysis of such mixtures as illuminating gas, natural gas, producer or water gas, or chimney or mine gas must be made by more rapid methods even at a sacrifice of a degree of accuracy. The gases from a measured volume of the original mixture are absorbed in suitable reagents and the volume loss is measured. The results of the analysis are computed in percents by volume. A standard type of apparatus for gas volumetry and one that is to be found in most laboratories is that of Hempel. Gas Burette. The gas burette, in which the gas mixture is measured, is shown in Fig. 77. The measuring tube (a) is con- nected with a levelling tube (b), the gas being confined over water. In making a reading the water is brought to the same level in the two tubes so that the gas is measured at atmospheric pressure. A complete analysis may usually be completed in a time suffi- ciently short that no serious error is caused by barometric changes. Changes in temperature during the course of an analy- sis constitute the most serious sources of error. To make the FUELS 329 method even commercially accurate great care must be exercised in this regard. A quiet room in which no other work is being performed should be used. The operator must at all times avoid touching the burette or levelling tube directly with the hands or 1 FIG. 77. Gas burette with levelling tube. breathing upon them more than is necessary. Sometimes the burette is enclosed in a water jacket to guard against any but very slow changes. The burette may have a simple rubber tip and pinch cock at 330 QUANTITATIVE ANALYSIS the top or it may be closed by a glass cock. The latter is desir- able but is liable to become stuck by contact with basic reagents. It must be well lubricated and frequently used or loosened. The glass three-way cock at the bottom of the tube is not often used and is not required. Absorption Pipette. The apparatus in which the gases are absorbed is known as an " absorption pipette." The simplest form of the Hempel pipette is illustrated in Fig. 78. The reagent fills the lower bulb, the bent tube and the capillary tube. The latter is connected with the gas burette by means of a short bent capillary tube and when the gas mixture is forced into the pipette n FIG. 78. Hempel's simple absorption pipette. FIG. 79. Hempel's double or "com- pound" absorption pipette. the absorbent fills the upper bulb. Some reagents are rapidly altered and rendered inefficient by contact with air. Protection from such action is afforded by the compound pipette (Fig. 79), the second pair of bulbs being filled with water. It is sometimes necessary to insert solid reagents, such as sticks of yellow phos- phorus, copper wires for reducing cupric chloride, etc., or rolls of iron gauze or glass tubes for giving greater absorbing surface to the reagent. A pipette for solids and liquids has an opening at the bottom of the first bulb for the insertion of such materials. (Fig. 80.) In order to increase the rate of absorption modifications of the original pipette have been introduced. The gas is caused to FUELS 331 bubble through the reagent instead of being forced down over the latter. After absorption has been completed the remaining gas is drawn from the top by turning the three-way cock to com- municate with the upper part of the bulb. (See Fig. 81.) When transferring gases from the burette to the pipette it is necessary to avoid mixing the water of the burette with the reagent in the pipette because the latter is thereby diluted. It is still more im- portant that the entrance of reagents into the burette should be prevented because such contamination of the water would cause premature absorption of gases. In order that such mixing may FIG. 80. Hempel's double pipette, modified to admit solids. FIG. 81. Bubbling absorp- tion pipette. be avoided it is necessary that there be a neutral zone in the con- necting tubes, into which neither water nor reagent shall enter. If this part of the tube has any but a very small capacity there will be an appreciable error, due to the gas that is left in the tube each time. For this reason the connecting tubes are of capillary dimensions. One of the most serious disadvantages in the use of Hempel pipettes comes from the necessity for connecting and discon- necting each pipette in turn as the different gases are absorbed. To obviate this inconvenience many modifications have been 332 QUANTITATIVE ANALYSIS made in the direction of a composite apparatus that does riot require the interchange. The most important feature of such forms of apparatus is a permanent connection of the burette with the several absorption pipettes, communication being estab- lished with each in turn by special forms of stop cocks. This usually involves the use of longer capillary tubes and this in- creases the error already mentioned as inherent in connecting tubes. In apparatus designed for the analysis of chimney gases the feature of permanent connection must be combined with porta- bility because the analysis must usually be conducted at the plant. A modification of the Orsat apparatus is here illus- trated (Fig. 82). Solubility of Gases in Rea- gents. When water is used as the confining liquid in the gas burette and water solutions are used as absorbents in the absorp- tion pipettes it is impossible to avoid small errors, due to the solubility of the components of the gas mixture in water. If the gases are taken into a burette containing pure water, each gas dissolves and the volume is diminished after the total volume has been read. In order to avoid the disappear- ance of a part of the gases in this way, the water must have been previously saturated by allowing the gas to bubble through it. This does not entirely obviate the error because, as the mixture is drawn back into the burette for measurement after the removal of each constituent, the partial pressure of that constituent being reduced to zero, a part passes out of the solution in the burette and mixes with the remaining gases, the total observed volume being rendered too large. To FIG. 82. Orsat's apparatus (modi fied) for analysis of chimney gases. FUELS 333 illustrate this action, suppose that a mixture of oxygen, carbon dioxide and carbon monoxide is being analyzed. The water in the burette is first saturated with the mixture but the amount of each dissolved is a function of its partial pressure (concentration) in the mixture. The measured gases are passed into a pipette con- taining potassium hydroxide where the carbon dioxide is com- pletely absorbed, its partial pressure in the gases being reduced to (practically) zero. Upon passing the mixture of carbon monoxide and oxygen back into the burette a certain amount of dissolved carbon dioxide will be given up by the water and the volume will be somewhat larger than the sum of the volumes of the other two gases. Also where the mixture was confined over potassium hydroxide solution the latter dissolved small amounts of carbon monoxide and oxygen and some of these gases may be given up to mixtures later being analyzed. The calculation of the amount of error may thus become a compli- cated matter. The error is negligible, from the industrial stand- point, if the analysis is completed within a short period of time. Fuel and Lighting Gases. In illuminating gas the following constituents are determined: carbon dioxide, ethylene and its homologues, oxygen, carbon monoxide, hydrogen, methane and nitrogen. They are absorbed, in the order named, one after another, and the contraction in volume noted after each absorp- tion. Hydrogen and methane are determined by combustion and nitrogen is computed by subtracting the sum of the other gases from 100. The various absorbents for these gases will be discussed. Carbon Dioxide. A solution of any of the strong bases may be used for absorbing carbon dioxide. Potassium hydroxide possesses the advantage of large solubility and rapid absorption of gas and is almost always used for this purpose in gas analysis. Also potassium carbonate, formed by absorption of carbon diox- ide in potassium hydroxide, is more soluble in the basic solution than is sodium carbonate. A solution made by dissolving solid potassium hydroxide in twice' its weight of water (about 33 per- cent, by weight) is suitable, this being the same strength as that employed for gravimetric determinations. 100 cc of a 33 per- cent solution will absorb about four liters of carbon dioxide before it becomes inefficient. The potassium hydroxide used 334 QUANTITATIVE ANALYSIS should not be that which has been purified from alcohol solu- tion, because traces of alcohol are retained in the solid base and alcohol vapor or other organic vapors are given up to the gas. The solution may be used in either the single or double Hempel pipette or in any of the modified pipettes. If the Hempel pipette is used it should contain rolls of iron gauze in order to increase the surface of solution exposed. As the solution is forced down, leaving the gauze exposed, the film of solution retained upon the surface of the wires greatly increases the rate of absorption. Hydrogen sulphide will be included in the fraction absorbed by potassium hydroxide unless it has been otherwise removed. Its quantity is usually small. Carbon Monoxide. The most conveniently used absorbent for carbon monoxide is a solution of cuprous chloride. This salt is only slightly soluble in water and must be dissolved in either hydrochloric acid or ammonium hydroxide. Either solu- tion absorbs carbon monoxide with the formation of a rather unstable compound whose exact nature is unknown. The acid solution is made as follows: Mix 86 gm of cupric oxide and 17 gm of finely divided copper and slowly add to 1000 cc of a mix- ture of equal volumes of concentrated hydrochloric acid and water. Stir until the solid matter has dissolved, then place in bottles having bundles of copper wire reaching from top to bot- tom. Stopper the bottles and allow to stand until colorless. Cupric chloride, formed by dissolving cupric oxide in hydro- chloric acid, is reduced by copper to cuprous chloride : CuO+2HCl^CuCl 2 +H 2 0, CuCl 2 +Cu-2CuCl. 100 cc of this solution will efficiently absorb about 400 cc of carbon monoxide. Absorption takes place slowly and the gas must be shaken with the solution for some time or be allowed to bubble through it. The -double pipette must be used because cuprous chloride is readily oxidized in contact with air, cupric chloride being formed. Oxygen. Oxygen is absorbed by a solution of potassium pyrogallate or by yellow phosphorus. The former solution is prepared by dissolving 120 gm of potassium hydroxide in 80 cc FUELS 335 of water, cooling and placing in the double absorption pipette then adding 15 cc of a 25 percent solution of pyrogallic acid and mixing. Potassium hydroxide purified by alcohol should not be used. The solution will readily absorb about 200 cc of oxygen for each 100 cc of solution. It does not act rapidly at tempera- tures below 15. Yellow phosphorus may be used in the form of sticks which are placed in the double pipette for solids and liquids and kept covered with water. This absorbent possesses the great advan- tage of retaining its capacity for absorbing oxygen until the sticks have become completely used up. The product of the union of phosphorus and oxygen, phosphorus pentoxide, dissolves in water so that the surface of the sticks is always fresh. Absorp- tion becomes slow below 15, and traces of unsaturated hydrocar- bons of the ethylene series partially inhibit the absorption. For the latter reason phosphorus is not suitable for use in those forms of assembled apparatus for the analysis of chimney gases in which the ethylene hydro- carbons are not determined at all. Ethyleue and Its Homologues. These gases give higher illuminating power to the mixture of methane, hydrogen and carbon monoxide, gases which burn with a non-luminous flame. For this reason they are collectively known as "illuminants." Fuming sulphuric acid or bromine water may be used. Fuming sul- phuric acid reacts with members of the ethylene series of hydrocarbons, forming addition products as well as condensation products. These are either liquids or soluble solids and are therefore removed from the gas mixture. The absorption is not rapid and the acid should be shaken with the gas if the Hempel pipette, or one similar to it, is used. The single pipette is used, since a water seal in the second bulbs is inadmissible. . In order to increase the contact of gas with acid the pipette contains a third small bulb which is filled with glass beads. Contact of the acid with rubber connections must be avoided. 100 cc of fuming sulphuric acid will absorb about 800 cc of ethylene. FIG. 83. Fuming sulphuric acid pi- pette for unsatu- rated hydrocarbons. Gill's modification. 336 QUANTITATIVE ANALYSIS Bromine water absorbs ethylene and its homologues with formation of bromine addition compounds: C 2 H 4 +Br 2 -+C 2 H 4 Br 2 . Tt is somewhat more convenient to use than fuming sulphuric acid but does not absorb with great readiness. If excess of bromine is placed in the pipette the absorbing power is undi- minished until all of this bromine has been dissolved. Hydrocarbon Vapors. Gases formed by distilling coal often contain vapors of liquid hydrocarbons, chiefly benzene. These are partly absorbed by fuming sulphuric acid but may not be entirely removed. They may be absorbed in absolute alcohol and so determined. The absorbing power of absolute alcohol is not large and gases coming from the ordinary gas burette, being saturated with water vapor, soon diminish the efficieticy of the alcohol by imparting moisture to it. Dennis and O'Neill suggested 1 the use of a solu- tion of nickel sulphate in am- monium hydroxide. Neither this solution nor absolute alcohol is an entirely satis- factory absorbent. The de- termination of hydrocarbon vapors is frequently omitted, these vapors then being ab- sorbed along with unsaturated hydrocarbons. Hydrogen. The determi- nation of hydrogen is made by burning with oxygen, measuring the resulting contraction in volume, or by absorption in palladium. The combustion may be carried out by exploding the mixture of hydrogen and oxygen over mercury in a suitable pipette or the burning may be made to proceed more slowly. If a mixture of hydrogen with an excess of oxygen or air is burned the resulting water vapor condenses and only the excess of oxygen or air remains as gas. From the equation 2H 2 +O 2 -2H 2 1 J. Am. Chem. Soc., 26, 503 (1903). FIG. 84. Hempel's explosion pipette. FUELS 337 it is seen that two-thirds of the volume of the disappearing gas is that of hydrogen. Therefore, two-thirds of the contraction measured after cooling equals the volume of hydrogen. The explosion pipette is shown in Fig. 84. The confining liquid should not be water since larger quantities of gases will dissolve in it at the moment of explosion because of the momen- tary increase in pressure. Mercury is substituted for water and the pipette is so constructed as to permit altering at will the difference in level between the mercury in the two bulbs. If the ordinary single pipette were used it would be impossible to force gas into the pipette because of the great density of mercury and the consequent back pressure. Ignition is effected by con- necting with the secondary of an induction coil. FIG. 85. Pipette for the preparation of hydrogen. FIG. 86. Pipette for slow combustion. If pure hydrogen is mixed with pure oxygen and burned the explosion is too violent for safety. If the gas to be burned is rich in hydrogen it is mixed with air instead of with oxygen. One volume of hydrogen requires more than two and one-half volumes of air for complete combustion, allowing a small excess. Dilution with air is not necessary if the residual gas is poor in combustibles. On the other hand it may be necessary to enrich the gas, before burning, by adding a measured volume of pure hydrogen. This is conveniently generated from zinc and sul- phuric acid in a special pipette (Fig. 85). Combustion may also be effected by passing the mixture with oxygen through a heated capillary tube or by exposing the mix- 22 338 QUANTITATIVE ANALYSIS tore to a glowing platinum wire in the pipette arranged for slow combustion (Fig. 86). In using this pipette either of two methods of procedure may be followed : The hydrogen is placed in the pipette, the wire made to glow by the passage of a current and a measured volume of oxygen led in, or the hydrogen and oxygen are mixed in the burette and slowly brought into the pipette, in which the wire is glowing. In either case combustion occurs without explosion. Hydrogen may be separated from nitrogen and methane by absorption in palladium sponge which has been superficially coated with oxide. Absorption readily takes place at 100 and the hydrogen may be later removed by passing oxygen through FIG. 87. Palladium tube. the palladium, the hydrogen being thereby oxidized and palla- dium oxide again formed on the surface. A tube of the form shown in Fig. 87 is used. The enlarged part is filled with asbestos which has been coated with spongy palladium and the tube is connected directly with the burette at one side and with a pipette filled with water at the other. Upon passing the gases through two or three times the hydrogen is quantitatively ab- sorbed, a small amount being oxidized by the trace of palladious oxide, and a certain amount is also burned by oxygen of air which was already in the tube. Except for this small amount of oxygen, the shrinkage in volume gives directly the volume of hydrogen. The amount of air in the tube must be known. This may be determined by connecting with the gas burette and measuring the expansion between two temperatures. One- fifth of the total volume is taken as the contraction due to con- tained oxygen. The absorption of hydrogen by palladium is hindered by traces of hydrochloric acid. On this account the ammoniacal solution FUELS 339 of cuprous chloride should be used for the absorption of carbon monoxide if this is to be followed by palladium absorption of hydrogen. The explosion pipette gives fairly accurate results and is not" difficult to manipulate but requires a battery and an induction coil. It is subject to the disadvantage that only a small amount of gas may be used, on account of the relatively large volume of air that must be mixed with it in the pipette, so that the error in reading volumes is relatively large. Gill has devised a pipette 1 which overcomes this objection. The bulb in which the explosion is to take place is large enough to hold the entire residue from 100 cc of gas, together with the necessary oxygen for the combustion, and is made of quite heavy glass. Both the slow combustion pipette and the palladium tube permit the use of larger quantities of gas. Methane. Methane is determined by combustion, the pro- cedure being the same as for hydrogen. In the analysis of natural gas and illuminating gas, as well as many other com- mercial gas mixtures, hydrogen and methane will both occur in the residue after other gases have been absorbed. They must therefore be burned together unless hydrogen is to be absorbed by palladium. According to the equation CH 4 +20 2 -+C0 2 +2H 2 one volume of methane with two volumes of oxygen will produce one volume of carbon dioxide, the rest of the oxygen disappear- ing as condensed water vapor. The contraction is therefore twice the volume of the methane. Since a volume of carbon dioxide equal to that of the methane is produced a measurement of the former by absorption in potassium hydroxide will give a direct determination of the volume of methane. For the residue of hydrogen and methane, therefore, the procedure is as follows: An excess of air or oxygen is mixed with the gases and the mix- ture exploded. The gases are cooled and measured in the burette, the contraction being noted. Carbon dioxide is then determined by absorption in potassium hydroxide. The volume of carbon dioxide is equal to the volume of methane. Twice this 1 J. Am. Chem. Soc., 17, 771 (1895). 340 QUANTITATIVE ANALYSIS volume is the contraction due to the combustion of methane. This contraction subtracted from the total contraction leaves the contraction due to the combustion of hydrogen. Two- thirds of this contraction is equal to the volume of hydrogen. The volumes of hydrogen and methane so determined, multi- plied by the ratio of the total residue to the volume taken for explosion, gives the volumes of hydrogen and methane in the original gas. The following example will illustrate the calcula- tions involved: 100 cc of illuminating gas gave, after all absorbable gases had been removed, 65.2 cc of residue, this consisting of hydrogen, methane and nitrogen. 15 cc of the residue was mixed with air, the total volume then being 90.5 cc. After explosion the volume was 71.0 cc. Carbon dioxide was absorbed, the volume of the remaining gases being then 66.6 cc. Volume methane = volume carbon dioxide = 71. 066.6 = 4.4 cc. Contraction due to combustion of methane = 2X4.4 = 8.8 cc. Total contraction = 90.5 -71. 0=19.5 cc. Contraction due to combustion of hydrogen = 19.5 8.8 = 10.7 cc. 2 Volume of hydrogen = ^X 10.7 = 7.1 cc. o 65 2 Volume of methane in original gas = ' X4.4 = 19.1 cc. lo (\ Pv O Volume of hydrogen in original gas = ~^-X 7. 1 = 30.9 cc. lo ( volume of sample sum of volumei of all other gases. Since 100 cc of gas was taken for analysis, the volume of -the con- stituents will also be their percents by volume. Analysis of Illuminating Gas. For this exercise the Hempel appara- tus may be used or any of the modified pipettes or burettes may be substituted. The method of manipulation is not essentially different for the different forms of apparatus except in minor details and such variations will readily suggest themselves. Throughout the analysis avoid touching the body of the burette or the bulbs of the pipettes with Volume of nitrogen in original gas = I sum of volumes FUELS 341 the hands, or breathing upon them. Allow the sides of the burette to drain thoroughly each time before reading. Prepare water for the gas burette by allowing the gas to bubble through it for ten minutes. Fill the burette with this water, raise the levelling tube until the water flows out of the top of the burette, then close the upper cock. Place a rubber tube on a gas cock and allow gas to escape through it until all air is displaced. With the gas still running connect the tube with the top of the burette, open the burette cock and fill with gas until the 100 cc mark has been passed. Close the upper cock and detach from the gas supply. It is desirable that ex- actly 100 cc of gas be taken, measured at the prevailing pressure of the atmosphere. In order to do this allow the water to drain down the sides of the burette for 1 minute then raise the levelling tube, compressing the gas until the 100 cc mark is exactly reached. Now close the cock at the bottom of the burette or close a pinch cock which is placed on the rubber connecting tube. Open the upper cock momentarily and close again. This permits gas to escape until the pressure within the burette is the same as that of the atmosphere. Hydrocarbon Vapors. Place the pipette filled with absolute alcohol on the stand by the burette and connect with the burette by a bent capillary tube, having previously caused the alcohol to fill the lower bulb and the capillary up to a point near the top. Force the gas into the pipette, detach the latter and shake for 1 minute. Return the gas to the burette, allow the water to drain down the sides, adjust the levelling tube to provide atmospheric pressure and measure. Record the difference as "hydrocarbon vapors." Carbon Dioxide (and Hydrogen Sulphide). Attach the burette to the pipette containing potassium hydroxide solution, pass the gas into the pipette and directly back again. Measure and record the percent of carbon dioxide (including also hydrogen sulphide if present). Illuminants. Determine illuminants by absorption in fuming sul- phuric acid or bromine water, drawing back to the burette at once. Avoid the entrance of any water into the fuming sulphuric acid. Oxygen. Absorb oxygen by yellow phosphorus, allowing three minutes, or by potassium pyrogallate, shaking the pipette for three minutes.- If pyrogallate is used in a pipette containing rolls of iron gauze the shaking may be omitted. Carbon Monoxide. Absorb carbon monoxide in either acid or basic solution of cuprous chloride. The gas should be shaken with the cu- prous chloride solution for three minutes, then passed into the pipette containing potassium hydroxide to absorb vapors of hydrochloric acid. Hydrogen, Methane and Nitrogen. Pass all of the gas residue into the cuprous chloride pipette for storage, pour out the water from the 342 QUANTITATIVE ANALYSIS burette and replace with water that has been saturated with air. Deter- mine hydrogen and methane by one of the following described methods. Combustion by Explosion. Return 10 to 12 cc of the gas to the burette, measure accurately, then draw air into the burette until a total volume of nearly 100 cc is obtained. Do not attempt to obtain exactly 100 cc as there is danger of loss of gas during the adjustment of volume. Measure, then transfer the mixture of air and gas to the ex- plosion pipette, allowing water from the burette to enter and fill the capillary of the explosion pipette. Close the rubber connecting tube (which should have thick walls and be securely wired in place) with a screw clamp. Place the mercury reservoir bulb so that the mercury is at the same level as inside the explosion bulb, then connect the ter- minal wires with the secondary of an induction coil and cause a spark to pass. A flash will pass across the bulb and mercury will almost imme- diately begin to flow into the bulb, on account of the contraction of gas volume resulting from the combustion. At all times when the gas is in the explosion pipette the mercury must be so adjusted in level that a pressure much greater or less than that of the atmosphere is avoided. Return the gas to the burette, allow the water to drain down the sides, then measure. Absorb the carbon dioxide and remeasure. In order to be sure that an excess of oxygen was present the gas should be passed into the phosphorus or pyrogallate pipette. If no oxygen is found the explosion must be repeated with another sample of gas, using a larger proportion of air. Calculate the percents of hydrogen, methane and nitrogen by the method already discussed. Repeat with another portion of the residue in the cuprous chloride pipette. Slow Combustion. Use the pipette shown in Fig. 86. Measure about half of the residue which is stored in the cuprous chloride pipette and transfer this to the combustion pipette. If the residue is known to be chiefly methane not more than 25 cc should be used. If it is chiefly hydrogen more may be taken since hydrogen requires for com- bustion only half its own volume of oxygen. Fill the burette with pure oxygen, and measure accurately. Connect the terminals of the plati- num wire of the pipette with a current source and heat the coil to bright redness. Pass the oxygen into the combustion pipette but not so rapidly as to cause an explosion. When the combustion is completed transfer the entire gas mixture to the burette and record the volume and con- traction. Determine carbon dioxide, test for excess of oxygen and calculate exactly as in the case of explosion. Absorption of Hydrogen by Palladium, Followed by Combustion of Methane. If the palladium tube is to be used for absorption of hydro- gen the solution of cuprous chloride in ammonium hydroxide must have been used for absorption of carbon monoxide. The entire gas residue FUELS 343 is used. Connect the palladium tube with the burette on one side and a pipette filled with water on the other. The palladium tube should dip into a beaker of water which is kept nearly boiling. Pass the gases through the tube and back, repeating two or three times. Replace the hot water with water at the temperature of the room and again pass the gas through the tube to cool it. Determine the internal volume of the palladium tube as already directed and subtract one-fifth of this volume from the total contraction. The remainder is the volume of hydrogen. Determine methane by combustion by either of the methods already described. FIG. 88. Aspirator for sampling chimney gases. Chimney Gases. Ideal combustion of fuel gases or of coal should yield waste gases containing only carbon dioxide, water vapor, and nitrogen. In practice complete combustion is not secured without a considerable excess of air, and oxygen is there- fore found in the chimney gases. The presence of carbon monoxide is an indication of imperfect draught and incomplete 344 QUANTITATIVE ANALYSIS combustion while a large excess of oxygen shows that heat has been wasted in raising the temperature of unused air. For control work the determination of oxygen, carbon dioxide and carbon monoxide is sufficient and the portable form of apparatus (Fig. 82) is used. It contains a burette and three pipettes for these determinations. Many modifications of this apparatus will be found illustrated and described in the scientific journals and trade catalogues. To obtain the sample for analysis a porcelain or iron tube is inserted into the stack at the proper point. An aspirator is caused to draw a continuous stream of gas from the stack, the sample being removed by the burette as often as desired. The determination of the three gases is made as with the Hempel apparatus. Potassium pyrogallate should be used fgr the ab- sorption of oxygen because of the possible presence of traces of ethylene. CHAPTER XIII OILS, FATS AND WAXES BUBNING OILS The chemist's examination of fuel oils usually has more to do with the determination of certain physical constants than with the actual analysis. Petroleum products are cheaper than animal or vegetable oils and are, consequently, seldom adulterated with the latter. Animal and vegetable oils are rarely used for burning. The examination of the fuel oil, therefore, usually resolves itself into a determination of the fitness of the oil for the purpose for which it is to be used. The determinations may include specific gravity, flash point, burning point and fractional distillation. Specific Gravity. The relation between the specific gravity and the volatility of petroleum fractions is fairly definite, so that it is often possible to secure the correct oil by specifying only the specific gravity. This may be determined by means of a Westphal balance or a floating hydrometer. The latter is most conveniently used and is sufficiently accurate for most purposes. The specific gravity may be expressed in relation to water or in degrees Baume*. The system of Baume* is much used in commercial testing. In this system two scales are used, one being for liquids lighter than water, the other for liquids heavier than water. The first is applicable to all petroleum products and to most other oils and fats. In the original Baume* scale for liquids heavier than water the point to which the hydrometer sinks in a solution of sodium chloride, 15 percent by weight and at 15 C., was taken as 15. The corresponding point for- pure water was taken as and all other points were located by these two. For liquids lighter than water the scale has the point 10 for the density of pure water at 15 C. and the point corresponds to the density of a 10 percent solution of sodium chloride. 345 346 QUANTITATIVE ANALYSIS Several modifications of these scales have come into use and much confusion has resulted thereby. As the system is at present used in many industrial laboratories the following for- mulas may be used for converting specific gravity into Baume* degrees and vice versa. For liquids heavier than water: 145 15 5 where B = degrees Baume* and S = specific gravity at T-^^O' JLO. o For liquids lighter than water : 140 S= 130+B and B-f-,30. On account of the complexity of this system and the fact that it is entirely unnecessary it is unfortunate that it has become so generally used in chemical industries. Flash Point. The " flash point" is the temperature at which the oil gives off vapor rapidly enough that the mixture with air becomes explosive and will flash if a small flame is brought into the mixture. This is one of the most important tests to be applied to burning oils because it determines the degree of safety attending the use of the oil in enclosed vessels, such as lamps and burners of various kinds. In most of our States the lower limits of flash and burning points are specified for kerosene by legal restriction. The location of the flash point depends to a great extent upon the manner of confining and heating the oil. The mixture of vapor with air is explosive at any temperature if the concentra- tion of vapor is sufficiently great. Under ordinary circumstances the vapor is evolved so slowly that it escapes by diffusion before an inflammable mixture is obtained and it is only when the tem- perature is raised that rapid evolution of vapor results in the production of a mixture that will ignite. From this it will OILS, FATS AND WAXES 347 readily be seen that the flash point is lowered by rapid heating, by confinement of the vapor by covering the tester, as well as by too close contact of the test flame with the surface of the oil. It has therefore become necessary to regulate by law not only the tem- peratures of the flash point but also the exact form of the tester and the manner of heating. The following extract from the Indiana law of March 11, 1901, is an illustration of the manner in which these tests are governed by law. " The test shall be made in a test cup of metal or glass, cylindrical in shape, two and one-quarter inches in diameter and four inches deep (both measurements being made inside the cup) and this cup shall be filled to within one -quarter of an inch of the brim with the oil or other substance to be tested. The cup shall be placed in a water bath sufficiently large to leave a clear space of one inch under the cup and three-eighths of an inch around it, and in such a manner as to project about one-quarter of an inch above the water bath. The space between the cup and the water bath shall be filled with water of medium temperature and shall be heated by an alcohol lamp, with its flames so graduated that the rise in temperature, from 60 degrees Fahrenheit to the highest test temperature, shall not be less than two degrees per minute and shall, in no case, exceed four degrees per minute. A Fahrenheit thermometer shall be suspended in such a manner that the upper surface of its bulb shall be, as near as practicable, one-quarter of an inch below the surface of the oil undergoing the test. As soon as the temperature reaches the point of ninety-eight degrees Fahrenheit, the lamp shall be removed from under the water bath, and the oil shall then be allowed to rise to the temperature of one hundred degrees Fahrenheit by the residual heat of the water, and at that point the first test for flash shall be made as follows : A taper (hereinafter described) shall be lighted and the surface of the oil shall be touched with the flame of the taper (and it shall be lawful to apply this flame either to the center of the oil surface or to any or all parts of it) but the taper itself shall not be plunged into the oil. If no flash takes place at the temperature of one hundred degrees Fahrenheit, the lamp shall be placed under the water bath, and the temperature raised to one hundred and three degrees Fahrenheit, when the lamp shall be again withdrawn and the oil allowed to rise to one hundred and five degrees by the residual heat of the water, when the test shall be made by again applying the flame of the taper as hereinbefore specified; if no flash occurs the test shall be repeated as often as the oil gains five degrees in temperature, three degrees with the lamp under the water bath, and two with the lamp removed. These 348 QUANTITATIVE ANALYSIS tests shall be repeated until a flash is obtained. The one making the test shall further test the oil by applying the taper at every two degrees rise without removing the lamp or stirring; but if a flash is obtained by this means, by a less rise in temperature than five degrees herein re- quired, he shall at once remove the lamp, stir the oil, and immediately apply the flame. The taper used for testing may be of any wood giving a clear flame, and it shall be made as slender as possible, and with a tip no more than one-sixteenth of an inch in thickness. No taper or match with sulphur on it shall be used, unless the sulphur is first removed before lighting. When a taper is first lighted, it shall be applied to the oil immediately (that is to say, before an ash or coal has had time to form on the end of the taper beyond the end of the flame) and the flame shall be made to touch the oil, but the taper itself shall not be brought in contact with the oil; provided, that if the taper be so brought in contact with the oil, but not held there longer than for the space of one second, and the oil flashes, the test shall not thereby be vitiated, but the Supervisor of Oil Inspection shall immediately remove the lamp, and again test the oil by the flame without allowing the body of the taper to touch the oil. No oil or other substance, which, by the test herein described, flashes at any temperature below one hundred and twenty degrees Fahrenheit, shall be allowed to be sold, offered for sale, or consumed for illuminating purposes in this State. And it shall be lawful to sell for illuminating purposes any oil or oils herein described, to be consumed within this State, which shall bear a flash test of one hundred and twenty degrees Fahrenheit, as shown by said apparatus." The Indiana law is not specific in the matter of covering the tester and the inference is that the open tester is permitted. Burning Point The burning point ("fire test") is the tem- perature at which vapor is evolved with sufficient rapidity to sustain a continuous flame. It is determined by removing the cover, if one was used during the flash test, and continuing the heating after the flash point has been passed, applying the test flame until a temperature is reached where continuous flame results. The thermometer bulb is immersed in the oil and the temperature is always noted just before the application of the test flame, which should be as small as possible. Examination of Kerosene. Determine the specific gravity at 15 C. with a hydrometer float or a Westphal balance, reporting in the usual units and also in degrees Baume*, using the formula given on page 346 for calculation of degrees Baume". OILS, FATS AND WAXES 349 Determine the flash and burning points, using, preferably, the tester specified by the law of the state in which the oil was sold and following in detail the directions furnished with the instrument. If no such tester is available construct one as follows : Upon a small sand bath plaee a 3-inch porcelain dish, pressing the dish into the sand until the latter is within 1/4 inch of the top of the dish. Fill to the same height with the oil to be tested and suspend in the middle of it a thermometer. Cover the dish with a watch glass having a perforation for the ther- mometer and a notch at the side for the application of the test flame. Heat the oil so that the temperature shall rise at the rate of about 2 per minute. When the temperature has reached 85 F. begin testing and test for each two degrees rise in temperature by inserting a small flame (a gas flame 1/4 inch long) and immediately withdrawing it. The experiment should be performed where the light of the room is not strong and in a place free from air currents. The flash point is reached when a flash passes entirely across the dish. Remove the cover and continue the heating and testing until a permanent flame is sustained. This temperature is the "fire point" or "burning point." The method used with the form of apparatus just described will not give the same flash point as that obtained by another form of apparatus and cannot be used as a legal check where another form of tester is specified by law. It is here described because it will afford practice in the determination when no other tester is available. Elliott Tester. This instrument is also known as the New York Board of Health tester. It may be used either open or closed for the flash test but the closed test is preferable unless otherwise specified. The outer cup is filled with water to the mark placed on the inner surface. If no mark is found, entirely fill the cup with water then push the oil cup in as far as it will go, the excess water overflowing. Remove the oil cup and take out 10 cc more of water. This allows room for expansion during heating. Flash Test. Place the tester in a room which is free from air currents and in which the light is not bright. Insert the oil cup and carefully fill with the kerosene sample to within 5 mm of the shoulder, but allowing no oil to splash on the shoulder. Replace the glass cover, in which the thermometer is fixed at such height as to bring the bulb just beneath the surface of the oil. Heat the water in the bath without stirring, fast enough to cause the thermometer in the oil to indicate a rise of 2 to 3 F. per minute. When the temperature reaches 85 F. begin the tests. The test flame should be a gas jet, not over 5 mm long. In 350 QUANTITATIVE ANALYSIS making a test the flame is inserted at the notch in the cover so that it passes well into the cup but without, at any time, touching the surface of the oil. Withdraw the flame immediately. Repeat the test at every 2 rise in temperature, always reading the thermometer immediately before applying the flame. When a flash passes across the cup the flash point is reached. Fire Test. Remove the cover and suspend the thermometer in the same position that was used during the flash test. Continue the heating at a rate not exceeding 10 per minute, making the tests as before. When the vapor finally burns with a continuous flame the burning point is reached and the temperature indicated by the ther- mometer just before the final test is called the fire test. Fractional Distillation. Any fraction of petroleum now ap- pearing in commerce includes many different chemical com- pounds and can itself be separated by fractional distillation into other fractions having boiling points within still more narrow limits. The determination of the amount distilling between certain specific limiting temperatures yields information regard- ing the composition of the mixture. The results have little significance, however, unless the distillation is conducted in a standard apparatus and by a standard method. LUBRICATING OILS For purposes of lubrication either mineral, animal or vege- table oils or mixtures of these are used. Such an oil should have the proper viscosity for the purpose, should be free from acidity, should produce the minimum of gumming under continued use, and, if to be used as a lubricant for cylinders of internal com- bustion engines, it must be capable of undergoing distillation without the deposition of more than a very small percent of free carbon. This is analogous to the "fixed carbon" of coal. In many cases specifications provide against the presence of more than small amounts of animal or vegetable oils or even against any quantity, because of the gumming action that occurs by oxidation and because of the development of acids through par- tial hydrolysis of the oil. Viscosity. Viscosity is usually expressed either as a specific property with the viscosity of water considered as unity, or in terms of an arbitrary scale of one of the standard instruments. OILS, FATS AND WAXES 351 The exact determination of viscosity is a difficult process. For commercial purposes an approximate determination is all that is necessary. The various instruments that are used for the determination of viscosity of oils do not give the same results but when the arbitrary scale of a given instrument is used, com- FIG. 89. Engler's viscosimeter. parative results are obtained for different oils. The Engler viscosimeter 1 is illustrated in Fig. 89. The principle used in this and many other viscosimeters is that of measuring the time required for a given quantity of oil to flow through a standard orifice. 1 Z. angew. Chem., (1892) 725; J. Soc. Chem. Ind., 12, 291 (1893). 352 QUANTITATIVE ANALYSIS Determination. If the oil is not perfectly free from suspended solids, filter through muslin. If the oil is not dry, decant after long standing. Pour the oil into the inner cup until the points marking the required level are reached. Fill the outer cup with water to the mark on the inside and heat by the ring burner until both water and oil are at the desired temperature (25 unless otherwise specified). A wooden plug closes the gold-lined orifice in the bottom of the oil cup. When this is lifted note the time on a stop watch and allow the oil to flow out until 200 cc is measured in the graduated flask, noting the time when the graduation is reached. The viscosity is the number of seconds required for 200 cc of oil to flow out. The instrument is standardized by measuring the" time necessary for 200 cc of water to flow at 20. This should be 50 to 52 seconds. The relative viscosity is the ratio of the time required for the oil to that for water at the same temperature. Specific Gravity. Determine as with burning oils unless the viscosity is too high to permit the use of either of these methods. In the latter case a picnometer is to be used or the specific gravity is determined at higher temperatures. The special hydrometer designed by Sommer 1 may also be used for the determination of the specific gravity of highly viscous oils. This is illustrated in Fig. 90. The brass cup has a capacity of exactly 10 cc. It is filled with the oil at 20, the cap is screwed on and the cup is then suspended from the hydrometer float, which is placed in pure water at 20. The specific gravity is read on the stem of the float, at the position of the meniscus. Acidity. Shake a small amount of oil in a test-tube with warm water and test the water with litmus. If F J Q ' r , 9 '~ acidity is shown a weighed sample of oil is shaken with drometer for alcohol and the acids titrated with a standard alcoholic asphalt and solution of a base which is preferably tenth-normal viscous oils. potassium hydroxide. Separation of Saponifiable from Mineral Oils. The method of separation depends upon the difference in chemical nature between mineral oils and those of animal or vegetable origin. The former are mostly hydrocarbons while the latter are esters 1 J. Ind. Eng. Chem., 2, 181 (1910). OILS, FATS AND WAXES 353 derived from glycerine and small quantities of other higher alco- hols with fatty acids. The esters are saponifiable by bases and the resulting soaps are soluble in water while the unsaponified mineral oils easily dissolve in petroleum ether. Determination. Weigh a 100 cc Erlenmeyer flask, add about 10 gm of the oil and weigh again. Add 50 cc of an approximately half- normal solution of potassium hydroxide in alcohol, place in the neck of the flask a funnel having a stem not more than 5 cm long and warm on the steam bath for 30 minutes. Remove the funnel and evaporate, frequently blowing out the vapor, until the odor of alcohol disappears. The evaporation of alcohol may be hastened by inserting a glass tube in the flask so that the end is four or five centimeters above the liquid, attaching a pump and drawing air through the flask. The tube must be slanted downward outside the flask in order to prevent condensed alcohol from returning to the flask. Cool, add 50 cc of petroleum ether, stir thoroughly with the soap and rinse into a separatory funnel with petroleum ether, disregarding any soap that may adhere to the flask. Add to the ethereal solution in the separatory funnel an equal volume of water, shake and allow to separate completely. The water will dissolve the soap that was produced from animal or vegetable oils while the petroleum ether containing the mineral oil will form the upper layer. Separate and discard the water solution and then rinse the ethereal solution into the flask in which saponification was accomplished, having previously washed and dried the flask. Evapo- rate the petroleum ether by placing the flask in a steam bath from which the flame has been removed. The evaporation may be hastened by the same device as was used in evaporating alcohol from the soap. After all ethereal odor has disappeared the flask is cooled and weighed. This gives directly the percent of mineral oil in the sample, and this percent subtracted from 100 gives the percent of saponifiable oil. The method gives somewhat high results for saponifiable oils because some loss of mineral oil occurs during the extraction of the soap. Chill Test. The chill test is the determination of the tempera- ture at which turbidity appears because of the formation of crystals. This is the temperature at which the oil would tend to clog oil holes in bearings. It has little significance except where saponifiable oils are present because mineral oils do not crystallize upon cooling. Determination. A 4-ounce bottle having a wide mouth is half filled with the oil and a thermometer placed in it. The bottle is placed 23 354 QUANTITATIVE ANALYSIS in a freezing mixture and stirred continuously with the thermometer. When the liquid ceases to be perfectly clear the temperature is noted as the "chill test." Cold Test. This is the determination of the temperature at which the oil ceases to flow freely and at which it will therefore fail to be delivered to bearings from the oil cup. From a con- sideration of the composite nature of all oils it will be seen that both chill and cold tests can give results which are only approxi- mately constant. They are of service, however, in forming a basis for judging the fitness of oils for use within known tempera- ture ranges. Determination. The bottle containing the oil that was used in the chill test is placed in the freezing mixture and cooled until the oil becomes solid. It is then removed and allowed to warm by contact With the air, being stirred with the thermometer meanwhile. At intervals of two degrees rise in temperature the bottle is inverted. When the oil has become sufficiently fluid to flow from one end of the bottle to the other the temperature is noted as the "cold test." EDIBLE FATS AND OILS For an interesting discussion of the fat and oil industries reference may be made to an address by Lewkowitsch. l Composition. The chief constituents of animal and vegetable oils are esters derived from fatty acids and the triatomic alcohol, glycerine. Of the former the most important are palmitic, stearic and oleic acids, the first two being saturated acids, the last an unsaturated acid. The glycerides of these acids are respectively known as palmitin, stearin and olein and they have the following composition : Palmitin Stearin Olein In addition to these are esters of higher alcohols other than glycerine and of other saturated and unsaturated fatty acids, also in certain cases small amounts of free higher alcohols. The chief differences in properties of different oils are caused by varia- tions in the proportions of the constituent esters. Vegetable oils contain much palmitin while stearin predominates in animal 1 Bull. soc. chim., [41 6, 1 (1909); Am. Chem. J., 43, 428 (1910). OILS, FATS AND WAXES 355 oils. The more liquid oils contain more olein and esters of acids having smaller molecular weights. The true waxes differ chemically from the oils and fats in that they are not glycerides but are esters of mono- or diatomic alcohols with the higher fatty acids. These alcohols are either aliphatic or aromatic. Some examples of such esters are as follows: Cetyl palmitate, derived from palmitic acid and cetyl alcohol, CieHssOH; this is the chief constituent of spermaceti. Ceryl palmitate, the chief constituent of opium wax, is derived from palmitic acid and ceryl alcohol, C27H 5 50H. Myricyl palmitate occurs in beeswax. It is an ester of palmitic acid and myricyl alcohol, C 3 oH 6 iOH. Ceryl cerotate is the chief constitu- ent of Chinese wax. It is an ester of cerotic acid, C25H 5 iCOOH, and ceryl alcohol. The most important aromatic alcohols occur- ring in waxes are the isomeric alcohols cholesterol and phytos- terol, C26H 43 OH. These are found as esters of palmitic, stearic and oleic acids. Notwithstanding the differences in composition the task of separating and determining the percent of different oils in a mixture is a difficult and often impossible one, because of the fact that the same general compounds constitute the greater proportion of all fats and oils. The chemist must usually be satisfied if he can recognize single oils or, with the nature of a single oil known, determine the approximate extent and nature of adulteration. The differences in molecular weight and degree of saturation, the presence and percent of free alcohols or acids and the occasional occurrence of traces of unusual substances, characteristic of certain oils, constitute the bases of the tests used in the effort to identify an oil. The examination becomes there- fore not an analysis, in the usual sense, but a series of tests applied in order to gain information regarding the identity of a pure oil and, so far as is possible, the composition of a mixture. Certain physical and chemical " constants" are determined and compared with the constants obtained from oils of known purity. The chief obstacle to the use of such figures lies in the fact that, for a given kind of oil they are actually variable within certain limits. These limits may be very narrow, but since they do include a certain range it sometimes happens that the ranges for two or more oils overlap. Thus olive oil from Italy is not chemically 356 QUANTITATIVE ANALYSIS identical with olive oil from California. The soil, climate, variety of plant and method of expressing from the olive have their influence upon the properties of the various glycerides and other substances present in the oil. It is only when the ranges of variation do not overlap that it is easy to determine the identity of a single oil, although it often happens that while overlapping occurs with a single constant it does not occur with others. The significance of the various constants and their methods of determination will be described. Specific Gravity. In a general way the specific gravity of oils increases with the percent of (a) glycerides of unsaturated acids, (b) glycerides of soluble acids and (c) free fatty acids. Old oils also usually have higher specific gravities than the normal, on account of oxidation. The specific gravity of the waxes and of solid fats is usually higher than of liquid oils. These rules do not hold in all cases and the determination of specific gravity, like that of the other constants of oils, is made for com- paring with recorded data for the purpose of identification more often 'than for throwing light upon the chemical constitution of oils of known purity. Unfortunately there has been a great lack of uniformity in selecting conditions and modes of expression for specific gravities of oils as they are recorded in the literature. Temperatures of 15.5, 20, 25, 40, 60, 100 and others are commonly used. In favor of the higher temperatures it may be said that the fats and waxes are all liquid at these temperatures so that determina- tions may readily be made. The specific gravity has been vari- ously expressed as the weight of oil at t divided by the weight of the same volume of water at either t, 0, 4 or 15.5. These differences make the compilation of comparison tables difficult. However it has been found 1 that a fair degree of approximation may be made in correcting the specific gravity to another tem- perature by using the coefficient 0.0007 as the change for each centigrade degree. This is the average value for a considerable number of oils between temperatures of 15.5 and 98. Of course this does not remedy the lack of uniformity of expression, noted above. i Wright: J. Soc. Chem. Ind., 26, 513 (1907). OILS, FATS AND WAXES 357 For the determination use a picnometer, a Westphal balance or an accurately calibrated hydrometer. If a Westphal balance is used the plummet should be accurately calibrated at the temperature at which the balance is to be used. The thermometer in the plummet should bn compared with a standard thermometer. The picnometer method is recommended. 20 Determination at o^. Use a 25 cc specific gravity bottle (picno- meter) . Clean with chromic acid, followed by distilled water, then rinse with alcohol and dry in an oven at 100. Cool in the balance case (in which the air should be at a temperature not above 20) and weigh. Fill with distilled water which has been recently boiled to expel dissolved gases and cooled to a few degrees below 20. Insert the stopper and nearly immerse the stoppered bottle in a distilled water bath which is kept at exactly 20. After 30 minutes take off the drop of water from the tip of the stopper, remove the bottle and wipe perfectly dry with a clean towel but without warming the bottle to above 20. Place in the balance case and weigh after 15 minutes. Calculate the weight of contained water. Empty and dry the bottle inside and out, then fill with oil and ma- nipulate as before, calculating the weight of contained oil. This weight divided by the weight of contained water gives the specific gravity of 20 the oil at o' If the specific gravity has been determined at any other temperature or if it is desired to calculate the specific gravity at any temperature from the determination at 20, these changes may be made with a fair degree of accuracy by the use of the following formula: G = G' + 0.0007 (T'-T), where G = specific gravity at temperature T, G' = specific gravity at temperature T'. 20 20 Determination at -^. Multiply the specific gravity at ^ by the density of water at 20, as shown in the table on page 181. The 20 product is the specific gravity of the oil at -TO~" Determination at the Temperature of Boiling Water. Fill a 25 cc picnometer, dried and weighed as above described, with freshly boiled hot water. Nearly immerse in a bath of briskly boiling water and leave for 30 minutes, replacing evaporated water with boiling distilled water. Insert the stopper, previously heated to 100, remove the picnometer 358 QUANTITATIVE ANALYSIS from the bath, wipe dry, cool to room temperature and weigh. Cal- culate the weight of contained water. Fill the flask, dried at 100, with the dry, hot, freshly filtered fat or oil, which must be entirely free from air bubbles. Keep in the boiling water bath for 30 minutes then insert the stopper, which has been heated to 100 wipe dry, cool to room temperature and weigh. Calculate the weight of contained oil and from this and the weight of water contained at boiling temperature calculate the specific gravity of the oil at the temperature of boiling water. This determination is necessarily less accurate than the one at 20, on account of the difficulty involved in keeping the bath at any constant temperature. Superheating may easily occur with distilled water and less pure water will have a boiling point above 100. Variation in barometric pressure will also change the temperature of the bath so that it becomes necessary to carry out both parts of the experiment at the same atmospheric pressure. However the determination is sanctioned and has been made official by the Association of Official Agricultural Chemists. 1 The specific gravity at any temperature other than 20 may be determined by the method outlined for this temperature or it may be calculated from the determination at this temperature, using the formula given above. It should be understood that the figure desired for purposes of identification is the specific gravity at the temperature for which data may be found in the literature. Index of Refraction. The measurement of index of refraction is a valuable addition to the list of tests for oils. While not in all cases characteristic it will frequently serve to distinguish between certain possibilities when other tests are not conclusive. The refractive index increases with increasing molecular weight of the combined acids and with increasing unsaturation. If free fatty acids are present in an oil the refractive index will be lower than the normal value for the oil. In consequence of the latter fact one may expect to find abnormally low indices for old or rancid fats or oils. The selection of standard temperatures for the determination is highly desirable in order to make comparison data useful. Temperatures of 20 for oils and 40 or 60 for fats and waxes are suitable in most cases. For calculating the index of refrac- tion at any temperature from experimental results at another 1 J. Assoc. Off. Agr. Chem., Vol. II, No. 3, Pt. II, p. 299. OILS, FATS AND WAXES the formula of Tolman and Munson 1 may be used: R = R' + 0.000365(T'-T), where R = index of refraction at temperature T, R' = index of refraction at temperature T'. 359 FIG. 91. Abb6's refractometer. The coefficient 0.000365 is the average change of refractive index for 1 for a large number of common oils. The index of refraction is determined by the use of any of the standard instruments, such as the Abbe, Pulfrich, Zeiss butyro- refractometer or the immersion refractometer. 2 Of those named 1 J. Am. Chem. Soc., 24, 754 (1902). 2 For a discussion of the theories of refraction and of the various types of refractometers, see Shook: Met. Chem. Eng., 12, 572 and 630 (1914) and 13, 19 (1915). 360 QUANTITATIVE ANALYSIS the Abbe* refractometer is probably the most generally useful instrument for the laboratory because it may be used with either solids or liquids covering a wide range of refractive indices and because it does not require the use of monochromatic light. This instrument is shown in Fig. 91. A layer of the oil is enclosed between two prisms in such a manner that light rays enter it at an angle differ- ent from the normal, refraction resulting (Fig. 92). The instrument measures the angle of total reflection of the ray emerg- ing from the oil, the field being a divided light and dark one. Dispersion is cor- rected by a " compensator " consisting of two similar Amici prisms, of direct vision for the D-line and rotated simul- taneously, though in opposite directions, around the axis of the telescope by means of the screw head. In this process of rotation the dispersion of the compen- sator passes through every value from zero (when the refracting edges of the two prisms are parallel and on different sides of the optical axis) to double the amount of dispersion of a single Amici prism (the refracting edges being parallel and on the same side of the optical axis). The dispersion produced by the oil in the refractometer may thus be annulled by rotating the screw head of the com- pensator until the latter produces a dis- persion equal to that of the oil but in the opposite direction. The border line between light and dark fields then becomes sharp and distinct, even when white light is used for illumination of the refrac- tometer prisms. The scale is graduated to read directly the index of refraction. The prisms are enclosed in such a manner that water at any desired temperature may be circulated about them. The heating arrangement for the water is a special feature of the Zeiss FIG. 92. Path of rays in the Abbe refractometer. OILS, FATS AND WAXES 361 instruments. This is shown in Fig. 93. Water is caused to pass through the heating spiral and the refractometer under a constant pressure, the temperature being controlled by regulating the size of the burner flame and the rate of flow through the clamp C. Since the pressure of water in the labo- ratory mains is not constant the pressure is made independent by fixing the upper and lower levels by means of the overflow tubes in the vessels A and. The Zeiss " butyro-ref ractometer " is an instru- ment which uses the same arrangement of prisms as that of the Abbe instrument. It is made espe- cially for use in the examination of butter and has a purely arbitrary scale. Readings of the butyro-refrac- tometer can be converted into indices of refraction by use of a table furnished with the instrument. The chief disadvantage of this instru- ment is the absence of the compensator. The prisms are achromatized for pure butter and give no dispersion of white light when this fat is used. For all other liquids the line of division between the light and dark fields is indistinct and consists of a prismatic series of colors unless monochromatic light is used. Determination by Means of the Abbe Refractometer. Set up the refractometer in front of a window or a source of sodium light. Con- nect the heating apparatus as shown in the figure and adjust the flow of water and the height of the flame until the desired temperature (20 for oils, 40 or higher for fats and waxes) is attained. Open the prism so that the lower half is in a horizontal position and place two or three drops of oil or melted fat upon it, using a glass rod or pipette but avoiding scratching the prisms. Quickly close and lock the prisms, allow time for the temperature to become constant then adjust the compensator until the line of division of the field is sharply defined and FIG. 93. Zeiss' apparatus for heating refractometer prisms. 362 QUANTITATIVE ANALYSIS bring this line to the cross hairs. Read the index of refraction upon the scale. Clean the prisms by applying a mixture of equal volumes of alcohol and ether, using a tuft of absorbent cotton. (Ordinary cotton may con- tain grit.) Melting Points of Fats. From the fact that fats are not single, pure compounds it will be seen that they cannot have definite and sharp melting points and the observation will, therefore, be a somewhat arbitrary one. The official method 1 follows. Determination. Prepare an alcohol- water mixture of graduated density as follows: Boil, separately, water and 95 percent alcohol for 10 minutes to remove dissolved gases. While still hot pour the water into an 8-inch test-tube until it is almost half full. Nearly fill the tube with the hot alcohol, pouring down the side of the inclined tube to avoid too much mixing. If the alcohol be added after the water has cooled the mixture will contain so many air bubbles as to be unfit for use. Prepare discs of fat as follows: Allow the melted and filtered fat to fall a distance of 15 to 20 cm from a dropping tube upon a piece of ice or upon the*surface of cold mercury. The discs thus formed should be 1 to 1.5 cm in diameter and should weigh about 200 mg. Since a recently melted and solidified fat does not possess its normal melting point the discs should stand for 2 to 3 hours before testing. Place the test-tube containing the alcohol-water mixture in a tall beaker containing ice water, until cold. Drop the disc of fat into the tube and it will at once sink to a point where the density of the mixture is exactly equal to its own. Lower an accurate thermometer, which can be read to 0.1, into the test tube until the bulb is just above the disc, stirring very gently with the thermometer. Slowly heat the water in the beaker, stirring constantly by means of an air blast or some other device. When the temperature of the alcohol-water mixture has risen to about 6 below the melting point of the fat the disc will begin to shrivel and roll into an irregular mass. Now lower the thermometer until the fat particle is even with the center of the bulb. Rotate the thermometer gently and regulate the temperature so that about 10 minutes is required for the last increment of 2. As soon as the fat becomes spherical read the thermometer. This serves as a preliminary observation of melting point. Remove the tube from the bath and place in the latter a second tube of alcohol-water mixture. The test-tube is of sufficiently low 1 J. Assoc. Off. Agr. Chem., Vol. II, No. 3, Pt. II, p. 301. OILS, FATS AND WAXES 363 temperature to cool the bath to the desired point, ice water having been used for cooling. Add another disc of fat and regulate the temperature so as to reach a maximum of 1.5 above the melting point as already determined. Run still another determination, which should agree closely with the second. The disc should not be allowed to touch the sides of the tube in any determination. Melting Point of Fatty Acids ("Titer Test"). A preliminary saponi- fication of the fat and separation and washing of the resulting fatty acids is necessary for this determination. The author considers the time consumed in the entire experiment to be out of proportion to the value of the results, in most cases. If the determination is required it may be found described in the official methods. 1 Iodine Absorption Number. The iodine absorption number is the percent of halogen, expressed as iodine, absorbed by the fat or oil when subjected to the action of a halogen solution under specified conditions. The absorption takes place because of the presence in the oil of glycerides of unsaturated acids which contain double or triple bonded carbon atoms. This action is analogous to the addition of oxygen, forming saturated oxygen compounds which are often hard and resinous in their nature. Such absorption of oxygen from the air is known as "drying/' although the term is here misapplied since no real drying occurs. The determination of halogen absorption number is, in a general way, a measure of " drying" properties and serves for the distinction between the broad, general classes of "drying," "semi-drying" and "non-drying" oils. Of the unsaturated acids whose glycerides commonly occur in fats and oils the following important members may be mentioned: Oleic Acid, Ci8H 34 O 2 . The structure of this acid is sufficiently indicated by the formula CH 3 (CH 2 ) 7 CH=CH(CH 2 ) 7 COOH. Olein, the triglyceride of this acid, occurs to some extent in all oils and fats. Olein is liquid at ordinary temperatures and its presence in oils is responsible, in many cases, for their liquid character. Either oleic acid or olein will absorb two atoms of bromine or one molecule ,of iodine monochloride or mono- bromide, the double bonded carbon atoms thus becoming satu- rated. Similarly either oleic acid or olein might be expected to absorb one atom of oxygen and to give "drying" properties i J. Assoc. Off. Agr. Chem., Vol. II, No. 3, Pt. II, p. 302. 364 QUANTITATIVE ANALYSIS to a fat or oil but this action does not take place readily and most of the oils of pronounced drying properties are found to contain considerable quantities of simple or mixed glycerides of linolic or linolenic acids. Linolic Acid, Ci 8 H 32 O2, contains two pairs of doubly linked carbon atoms: This acid or its glyceride, linolin, will absorb four atoms of halogen or one molecule of oxygen. It gives marked drying properties to oils, linolin being abundant in linseed, soy bean and poppy seed oils. Linolenic Acid, CigHsoC^, probably to be represented^ CH 3 .CH 2 .CH = CH.CH 2 .CH = CH.CH 2 .CH = CH. (CH 2 ) 7 COOH. This acid possesses three sets of double bonds and will absorb six halogen atoms or three oxygen atoms. It occurs as simple or mixed glycerides in linseed oil and, together with linolic acid, plays the most important part in the hardening or " drying" of this oil when it is exposed to the air. An isomer, isolinolenic acid, also occurs as a constituent of the glycerides of drying oils. Clupanodonic Acid, CisH^gC^, has the following structure: CH 3 .CH 2 .CH = CH.CH 2 .CH = CH.CH 2 .CH = CH.CH 2 .CH II HOOC(CH 2 ) 4 .CH. Having four pairs of doubly linked carbon atoms it is able to absorb eight halogen or hydrogen atoms or four oxygen atoms. This acid will be mentioned in connection with the detection of fish oils. Ricinoleic Acid, Ci 8 H 34 O 3 , is hydroxyoleic acid and, like oleic acid itself, contains only one pair of doubly linked carbon atoms. It will not readily absorb oxygen from the air and it does not impart drying properties to an oil. It is, however, an important constituent of castor oil and will be mentioned later in the discussion of acetyl value. OILS, FATS AND WAXES 365 The five acids named above serve to illustrate the principle that only those unsaturated acids which contain more than one pair of doubly bonded carbon atoms are important from the stand- point of drying. Also an interesting, although perhaps un- expected fact is that trebly linked carbon atoms do not, under ordinary conditions, absorb halogens or oxygen to the point of complete saturation, only two atoms of halogen or one of oxygen adding to each such pair. Thus, the acids of the tariric series, C n H 2n -4O 2 , are isomeric with those of the linolic series. Tariric acid, Ci8H 32 02, is isomeric with linolic acid. Its unsaturated state is shown by the formula, CH 3 (CH 2 ) 10 C=C(CH 2 )4COOH. This and other acids of the series, or their glycerides, absorb two halogen atoms quite readily but are not oxidized upon ex- posure to air. They are therefore quite unimportant as con- stituents of the drying oils. After the treble linking has been half saturated by the absorption of halogens the remaining double bond becomes saturated but very slowly and this property is partly responsible for the fact that, to some extent, the iodine absorption number is a function of the time allowed for the re- action and of the nature of the halogen solution. Many of the methods for determining iodine absorption num- ber have been open to the objection that they permit more or less substitution in saturated compounds as well as addition to unsaturated compounds. Hubl's 1 method, formerly much used, is especially faulty in this respect. Hubl's solution is made by dissolving 26 gm of iodine in 500 cc of alcohol and 30 gm of mercuric chloride in 500 cc of alcohol, the two solutions being then mixed. The resulting solution probably contains 2 some mercuric chloriodide and iodine monochloride, the reaction being expressed as follows: HgCl 2 +l2->HgICl+ICl. The latter is the active constituent of the solution but its con- centration is relatively small, which accounts for the fact that much time is required for the absorption. The oil is dissolved in chloroform and allowed to stand with a measured volume of 1 Dingl. polyt. J., 263, 281 (1884); J. Soc. Chem. Ind., 3, 641 (1884). 2 Ephriam: Z. angew. Chem. (1895), 254. 366 QUANTITATIVE ANALYSIS the solution for three hours, after which the excess of iodine is titrated. Substitution takes place to a considerable extent and the amount of iodine absorbed varies with the time allowed. When hydrogen in a saturated ester is substituted by iodine, hydriodic acid is also formed. In the case of palmitin: By determining the amount of hydriodic acid so formed the amount of substitution may be determined but it is better to use one of the solutions suggested by Hanus and Wijs because these cause very little substitution. Hanus' 1 solution is made by dissolving iodine in glacial acetic acid and adding an equivalent weight of bromine. The active constituent is iodine monobromide, IBr. The oil is dissolved in chloroform. Wijs' 2 solution contains iodine monochloride, IC1, and is made by adding an equivalent amount of chlorine to a solution of iodine in glacial acetic acid. Either chloroform or carbon tetrachloride is used as the solvent for the oil. Both Hanus 7 and Wijs 7 solutions are more active than that of Hiibl, the absorption being completed in thirty minutes. The solutions are also more stable and need not be so frequently restandardized. The amount of substitution taking place is also much less and is practically zero with many oils. The solu- tion of Hanus is more conveniently made than that of Wijs and the method of Hanus will therefore be described. 3 Determination. Prepare an iodine monobromide solution as follows: Equivalent quantities of bromine and iodine are dissolved in glacial acetic acid in such a ratio as to make a solution somewhat more than fifth-normal, referred to total halogen. The glacial acetic acid is first tested to insure absence of reducing substances. A drop of sulphuric acid and two or three drops of tenth-normal potassium dichromate solu- tion are added to 10 cc of the acetic acid and the mixture is warmed. The yellow color should persist without the appearance of green chromium salts. 1 Z. nahr. Genussm., 4, 913 (1901). 2 Ber., 31, 750 (1898). 8 See a comparison of methods by Tolman and Munson: J. Am. Chem. Soc., 26, 244 (1903); 26, 826 (1904). OILS, FATS AND WAXES 367 Dissolve 13.6 gm of powdered iodine in 825 cc of glacial acetic acid, warming the flask if necessary. Cool, decant to insure that no particles of iodine remain undissolved, and mix. Measure from a burette 25 cc of the solution into a 250 cc Erlenmeyer flask, add 10 cc of 15 percent sodium iodide solution and 100 cc of water and mix. Titrate at once with tenth-normal sodium thiosulphate solution. From a small burette measure 3 cc of bromine into 200 cc of glacial acetic acid. Mix and titrate 5 cc of the solution against sodium thiosul- phate solution, adding sodium iodide and water as in the iodine titration. Calculate the volume of tenth-normal thiosulphate solution that would be equivalent to 800 cc of iodine solution, then calculate the volume of bromine solution that would be equivalent to this volume of thiosul- phate solution. Add this quantity of bromine solution to the iodine in a glass stoppered bottle and mix well. Prepare starch solution, also prepare and standardize a tenth-normal solution of sodium thiosulphate according to the directions given on page 263. Half fill a 20 cc weighing bottle with oil, place in it a piece of glass rod and weigh without the stopper. Carefully pour about 0.25 gm of the oil into a 500 cc bottle having a ground glass stopper, using the glass rod to assist in the transference. Reweigh and prepare two more samples in the same manner. Dissolve the weighed sample of oil in 10 cc of chloroform then add 25 cc of iodine monobromide solution, measuring from a pipette. Stop- per the bottle, mix and allow to stand for thirty minutes, shaking occasionally. The bottle should not be left in strong light. At the time that the iodine monobromide solution is measured into the oil solution, measure the same amount of solution into two bottles, containing the chloroform but no oil. Treat these in exactly the same manner as the solution containing oil. At the end of the absorption period add 10 cc of the sodium iodide solution which was used in standardizing the sodium thiosulphate solu- tion. Add 100 cc of water, washing down any iodine that may be on the stopper. Titrate the unabsorbed iodine with standard sodium thiosulphate, shaking constantly. When only a faint yellow remains add 1 cc of starch solution and finish the titration. At the last the bottle should be closed and shaken until all iodine remaining in the chloroform has been extracted by the potassium iodide. The temperature should be kept as nearly constant as possible throughout the experiment. From the volume of sodium thiosulphate required for the iodine solu- tion alone subtract that required for the oil and iodine solutions. The remainder is the volume corresponding to the absorbed iodine. Calcu- late the percent of iodine absorbed. 368 QUANTITATIVE ANALYSIS Acid Value. Fresh oils sometimes contain small amounts of free fatty acids produced during the process of extraction. Rancid fats and oils contain free acids as products of hydroly- sis of the glycerides composing them. The acid value is defined as the number of milligrams of potassium hydroxide required to neutralize the free fatty acids in 1 gm of oil or fat. Acidity is also sometimes expressed in terms of oleic acid as percent, or as "acid degree," which is cubic centimeters of normal base equivalent to the free acids in 100 gm of oil or fat. The determination of acid value is made for the purpose of determining the condition of the oil and its fitness for a given use, rather than for the purpose of identifying it, since the acid value is a variable within rather wide limits for any oil. Determination. Weigh 20 gm of oil or fat into a 200 cc flask and add 50 cc of 95 percent alcohol which has been made neutral to phenolph- thalein by a dilute solution of so.dium hydroxide. Heat to the boiling- point in a steam bath and agitate thoroughly. Titrate with a tenth-nor- mal solution of sodium or potassium hydroxide using phenolphthalein. Shake vigorously during the titration and add the standard solution until the pink color persists. Saponification (Kottstorfer) Number. The saponification number 1 is the number of milligrams of potassium hydroxide required to saponify 1 gm of oil or fat. Different oils show differ- ent saponification numbers because of variation in the molecular weight of the esters contained in them, those of relatively low average molecular weights requiring more base for the saponifica- tion of a given weight of oil than those of relatively higher mo- lecular weights. The variation is, however, not as great as is the case with iodine absorption numbers and the saponification num- ber is consequently not as valuable for use in identifying oils as is the iodine number. The comparatively small variation in saponi- fication number is due to the relatively small variation in the average molecular weight of the esters entering into the compo- sition of the oils. Of the whole list of the more common oils nearly all are chiefly composed of stearin, palmitin and olein, the molecular weights of these being 890, 806 and 884, respectively and their equivalent weights are one-third of these numbers. The saponification numbers of the pure esters would be as follows : 1 Z. anal. Chem., 18, 199 and 431 (1878). OILS, FATS AND WAXES 369 rn r for stearin ^95 X 1000=189, for palmitin ^9 X 1000 = 208, rs* and for olein ^rr X 1000 = 190. The greatest possible variation in the proportions of these three esters could make a difference of but 19 in the saponification numbers. The occurrence of appreciable quantities of esters of lower acids in certain oils causes a much greater deviation from the numbers given above. For example, butter fat is chiefly composed of the following glycerides in the approximate proportions indicated: butyrin, C3H 6 (C4H7O2)3, 7.0 percent; caproin, CsHXCeHnC^s, caprylin, CsEWCsHuC^s and caprin, CsHsCCioHigC^s, 2.3 percent; olein 37.7 percent; palmitin, stearin and glycerides of small quantities of other acids 53.0 per- cent. The calculated saponification number of butyrin is 554, of caproin 434, of caprylin 356 and of caprin 302. The presence of these esters of small molecular weight raises the saponification number to about 227, a number which serves to distinguish butter fat from a large number of other fats, particularly from oleomar- garine which has a saponification number of about 195. On the other hand the saponification numbers of true waxes are, on the whole, considerably lower than those of oils or fats. As noted on page 355, the characteristic difference between oils and fats, on the one hand, and waxes on the other is that the latter consist of esters of higher fatty acids with monohydric and di- hydric alcohols instead of with the trihydric alcohol, glycerol. These esters have higher equivalent weights than those of the glycerides and the saponification numbers are correspondingly lower. Thus, spermaceti consists chiefly of cetin (cetyl palmitate), CieHasO.CO.CisHsi, mixed with smaller proportions of other esters and possibly of free alcohols. Cetyl palmitate is an ester of palmitic acid and the monohydric cetyl alcohol. Its theoret- fft ical saponification number is 117 ( = |oQXlOOO) and this gives spermaceti the actual saponification number of 122 to 129, which at once serves to characterize it as a true wax. Beeswax may be noticed as another example. This wax is largely composed of a mixture of cerotic acid, C 2 5H 5 iCOOH, and myricyl palmitate, C 3 oH 6 iO.CO.Ci 5 H3i. The saponification 24 370 QUANTITATIVE ANALYSIS r number of free cerotic acid is 141 ( = ^r=r X 1000) and that of myricyl palmitate is 83 ( = ^^ XI 000). It might therefore be expected that beeswax would have an abnormally low saponi- fication number and this also on account of the presence of about 10 to 15 percent of unsaponifiable hydrocarbons. The actual saponification number is found to be about. 94. Water solutions of potassium hydroxide act upon oils very slowly because of the small solubility of most oils in water. Hot alcohol dissolves oils more readily and alcoholic solutions of potassium hydroxide are therefore used for the saponification. Commercial alcohol contains aldehydes which are changed by potassium hydroxide into resinous bodies, a dark red solution being produced and the basic concentration being diminished. The alcohol should therefore be purified by first heating with a stick of potassium hydroxide in a flask fitted with a reflux con- denser and placed on a water or steam bath, then distilling. Insoluble Acids (Hehner Value) and Soluble Acids. The determination of the saponification number may be conveniently combined with the determination of soluble acids and insoluble acids. Among the most important of the acids of smaller molecular weight than oleic acid combined as glycerides are butyric, caproic, caprylic and capric acids, discussed above. These acids are soluble in water, the solubility decreasing as the molecular weight increases, so that, while butyric acid is infi- nitely soluble, capric acid dissolves to the extent of 1 part in 1000 parts of boiling water. The next acid in the series, lauric acid, is almost insoluble while the next member, myristic acid, is practically insoluble. An approximate separation of the lower acids from the higher ones may be accomplished by saponifying the oil, decomposing the resulting soap with sulphuric acid and washing the fatty acids with water. The percent of insoluble acids is called the Hehner value. 1 An inspection of the formula for a typical triglyceride, as that of palmitin, C 3 H 5 (OCi 6 H3iO)3, shows that the acid residue comprises the greater part of the compound. Also since the variation in the molecular weights of the three acids, palmitic, 1 Z. anal. Chem., 16, 145 (1877). OILS, FATS AND WAXES 371 stearic and oleic, which make the greater part of the acids of most oils and fats, is small as compared with the molecular weights themselves, it is not to be expected that there would be a large variation in either the Hehner value or the percent of soluble acids. The former has an average value of about 95 and the latter of considerably less than 1. Therefore these numbers are without any great significance in most cases and their determina- tion will give little assistance in the task of identifying most oils. A few exceptions- to this statement should be noticed. Butter has already been mentioned as containing unusually large quantities of butyric, caproic, caprylic and capric acids. Consequently its Hehner value falls to 88-90 and its percent of soluble acids rises to about 5. Other notable exceptions are cocoanut, palm nut, croton and porpoise oils. Practically, it is in these cases only that the determination of soluble and insoluble acids will be of any great use. Determination. Prepare a tenth-normal solution of sodium or potas- sium hydroxide in boiled and cooled water, using phenolphthalein in standardizing. Purify two liters of alcohol by heating on a steam bath for thirty minutes with about 10 gm of sodium hydroxide, using a reflux condenser. Distill and make 1000 cc of a solution of 40 gm of potassium hydroxide in the alcohol. The potassium hydroxide should be as nearly free from carbonate as is possible. Allow the solution to stand until the small amount of potassium carbonate that is always present has settled out, then decant into another bottle. The concentration does not remain constant for long and the solution need not be standardized until it is used for saponifying the oil. Prepare also a half-normal solution of hydrochloric acid in water. Saponification Number. Select two ordinary flasks of 250 cc capacity having, if possible, necks of slightly larger diameter at the top than at the bottom, though this feature is not essential. Clean with alcohol. Weigh into each flask about 5 gm of oil or fat, using a small bottle and glass rod as in the determination of iodine number. Add to each flask 50 cc of the alcoholic solution of potassium hydroxide from a calibrated pipette or burette, place in the neck of the flask a funnel having a short stem and warm on the water bath until the alcohol boils, though it should not be evaporated more than is necessary. The oil is usually saponified in about thirty minutes. A homogeneous solution must be produced, so that no separation will occur when boiling is interrupted. Measure 50 cc of the alcohol solution of potassium hydroxide into each 372 QUANTITATIVE ANALYSIS of two other flasks, for standardization. While saponification of the oils is proceeding titrate these solutions with the half-normal acid, using phenolphthalein. Cool the flasks in which the oil was saponified, add a drop of phenolphthalein and titrate the excess of base with half-normal acid, deduct from the volume used for 50 cc of the base in the standard- ization and calculate the saponification number. Preserve the neutral solution for the determination of soluble and insoluble acids. Soluble Acids. Evaporate the alcohol by placing the flasks upon a water bath and drawing air through them as explained on page 355. When the odor of alcohol has entirely disappeared add enough standard acid to make the total amount, including that used in titrating the excess of base, 1 cc more than the volume that is equivalent to the 50 cc of potassium hydroxide used in saponifying the oil. It is necessary to be very careful about the removal of all alcohol at this point. The acids that are classed as insoluble are considerably more soluble in alcohol. Consequently any alcohol that might be left with the soap would cause an error in the separation of soluble from insoluble acids. Connect a reflux condenser and warm on the water bath until the insoluble fatty acids have melted and separated from the water solution. Add hot water to bring the liquid within about 2 cm of the top of the neck of each flask, again allow the insoluble acids to separate, then cool in ice water. Carefully detach the cake of insoluble acids and pour the cold solution through a filter into a flask of 1000 cc capacity. Replace the cake of acids in the flask, fill with hot water, separate as before and filter. Repeat the treatment once more and do not discard the insoluble acids. In some cases the insoluble acids will not solidify, even at 0, on account of the preponderance of oleic acid, a liquid acid. In such cases the entire liquid is poured into an already wet filter and the flask and the insoluble acids on the filter are washed with cold water. The combined filtrates now contain the excess of half-normal acid (1 cc) and the soluble acids of the oil, besides potassium chloride, glycerine and other alcohols, etc. Titrate the acids with tenth-normal base in presence of phenolphthalein. From the volume of standard base used deduct the volume equivalent to 1 cc of the standard acid and cal- culate the percent of soluble acids as butyric acid. The arbitrary assumption that butyric acid is the only soluble acid present is merely a convenience. Insoluble Acids (Hehner Value}. Allow the cake of insoluble acids to dry on the filter paper for twelve hours at the temperature of the room, then transfer to a small weighed dish. Wash thoroughly with warm alcohol the paper and the flask in which saponification was accomplished, allowing the solution to run into the dish. Evaporate the alcohol ou the OILS, FATS AND WAXES 373 steam bath and dry to constant weight at 100. From 2 to 5 hours dry- ing will usually be required. If the insoluble acids did not solidify the filter is pierced and the acids are allowed to run into a weighed dish. The flask and paper are washed thoroughly with hot alcohol, this running into the dish. Evapo- rate and dry as with solid acids. Calculate the percent of insoluble acids. Reichert Number and Reichert-Meissl Number. There is no sharp line of division between the fatty acids volatile with steam and those not volatile and it is not possible to effect more than a very approximate separation by a method of distillation unless this is continued for a very long time. On the other hand fairly constant proportions of acids may be distilled if the method is rigidly standardized. In this way figures may be obtained that have a value in identifying certain oils and fats. The determination is made chiefly in the examination of butter and its substitutes. Pure butter contains volatile acids to the extent of nearly 10 percent of the total fatty acids. The saturated acids that have been classed as " soluble" (to and including capric acid) are the only ones of the series that may be distilled without decomposition. They are therefore known as " volatile" acids while the higher acids (above lauric) decom- pose when distilled and are therefore called " non-volatile." Lauric acid distills with steam but is slightly decomposed. Al- though the volatile acids boil at temperatures higher than 100 they can be distilled with steam. The boiling points of the more commonly occurring " volatile" acids are as follows: Acid Boiling point, degrees Butyric 162 3 Caproic 200 Caprvlic 236 Capric 270 The method proposed by Reichert 1 and the modifications of this method by Meissl 2 have been extensively adopted. It should be understood that neither method gives the correct percent of volatile acids but simply the proportion that will be 1 Z. anal. Chem., 18, 68 (1879). 2 Dingl. polyt. J. f 233, 229 (1879); Chem. Zentr., 10, 586 (1879). 374 QUANTITATIVE ANALYSIS distilled under certain stated conditions. The Reichert Number is the number of cubic centimeters of tenth-normal base required to titrate the adds obtained from 2.5 gm of oil or fat by Reichert's distillation process. The Reichert-Meissl number is the same as the Reichert number except that 5 gm of oil or fat is used. The Reichert-Meissl number is not exactly double the Reichert number. The Reichert-Meissl number of most oils, fats and waxes is less than 1 and the determination will be of little service in identifying these oils. The following oils are exceptional in this respect. Oil or fat Reichert-Meissl number Butter fat 28 Cocoanut 7 Croton 13 Mocaya 7 Palmnut 5 Porpoise 47 Determination. Prepare the following reagents : (a) Sodium hydroxide solution in water, 50 percent by weight. (b) Alcohol, 95 percent, redistilled from sodium or potassium hydroxide. (c) Sulphuric acid, 1 part concentrated acid in 5 parts water. (d) Potassium hydroxide, approximately tenth-normal, standardized against standard acid, using phenolphthalein as indicator. If the sample is either real or imitation butter it will contain water and curd. Melt and keep at 60 until the fat has separated and, if necessary, filter the fat through a dry paper placed in a hot-water funnel. If the sample is an oil it may usually be weighed without treatment. Ordinary flasks of 200 cc capacity, are cleaned and dried. The oil or melted fat is dropped in from a weighed bottle until 5 gm, measured to within one drop, is obtained. The oil must not be left on the neck of the flask. Record the exact weight. Add 10 cc of alcohol and 2 cc of 50 percent sodium hydroxide solution, connect with a reflux condenser and heat upon the steam bath until the oil is saponified. Remove the con- denser and evaporate the alcohol as in the determination of soluble acids. Add 135 cc of recently boiled water and warm on the water bath until solution is complete, then cool. Add two or three pieces of pumice stone or about 1 gm of crushed porcelain to prevent bumping, then add 5 cc of the diluted sulphuric acid. Again attach the reflux condenser OILS, FATS AND WAXES 375 FIG. 94. Kjeldahl's distilling tube. and heat on the steam bath until the acids form a clear layer. Connect the flask with a Kjeldahl or Hopkins distilling tube and a condenser and distill over a flame at such a rate that 110 cc shall be obtained in approximately thirty minutes. The distillate is received in a flask which is graduated to contain 110 cc. Mix the distillate, and filter through a dry filter to remove traces of insoluble acids carried over by the steam, receiving the filtrate in a flask graduated to con- tain 100 cc. Titrate 100 cc of the filtrate with standard potassium hydroxide. Make the proper cor- rection for the fact that only 100 cc of the distillate was used, also cor- rect the number of cubic centimeters of standard potassium hydroxide used, in case this solution was not exactly tenth-normal or in case the sample weight was not exactly 5 gm. The result is the Reichert-Meissl number. Polenske Value. 1 One of the very important constituents of some butter substitutes is cocoanut oil, a pure white vegetable fat having a pleasant taste and a consistency which is about the same as that of butter. Its Reichert-Meissl number is somewhat lower than that of butter, as is shown in the table on page 374. Salkowski 2 noticed that the volatile acids obtained from cocoanut oil in the Reichert-Meissl distillation contained much larger quantities of acids insoluble at 15 than do the volatile acids from butter. Butyric acid comprises from 60 percent to 70 percent of the volatile acids from butter and this acid is soluble in water in all proportions. The volatile acids from cocoanut oil contain larger quantities of caproic, caprylic, capric and lauric acids, 3 these being almost insoluble at 15. The Polenske value (called by its author the "new butter value") is the num- *Z. Nahr. Genussm., 7, 273 (1904); J. Soc. Chem. Ind., 23, 387 (1904). 2 Z. anal. Chem., 26, 581 (1887). 3 Elsdon: Analyst, 38, 8 (1913). Percents of acids here reported are caproic 2, caprylic 9, capric 10, lauric 45, myristic 2, palmitic 7, stearic 5, and oleic 2. 376 QUANTITATIVE ANALYSIS ber of cubic centimeters o] decinormal base required to titrate the insoluble acids obtained in the Reichert-Meissl distillation. The Polenske value for pure butter varies from 1.5 to 3.0, while that for cocoanut oil varies from 16 to 18. It is necessary to avoid the use of alcohol in the saponification of the fat and therefore the determination of Reichert-Meissl number must be modified if the two determinations are to be combined. Polenske's modification is essentially as follows: Determination. Saponify 5 gm of the fat by heating in a 300-cc round flask, using a reflux condenser. For the saponification use 20 gm of glyc- erol and 2 cc of a 50 percent solution of sodium hydroxide in water. When saponification is complete dissolve the soap in 135 cc of recently boiled water and add 25 cc of dilute sulphuric acid (50 cc of concentrated acid in 1000 cc of solution) and a small amount of crushed porcelain or pumice. Connect with a condenser by means of a Kjeldahl or Hop- kins distilling tube and distill into a flask which is graduated at 100 cc and 110 cc; the distillation should proceed at such a rate that 110 cc passes over in about 20 minutes. When the distillate reaches the 1 10 cc mark on the flask replace the latter by a 25 cc cylinder and stop the distillation. Immerse the flask in water at 15 and allow to remain for 15 minutes. The level of the water must be above the 110 cc mark on the flask. Mix the contents of the flask and pass through a dry, 8 cm filter and, if desired, determine the Reichert-Meissl number, using 100 cc of the filtrate. Rinse the 110-cc flask but without removing any of the insoluble acids adhering to it. Wash the filter three times with 15 cc of water, this water having previously been used for washing the con- denser, cylinder and flask. Dissolve the insoluble acids from the con- denser, cylinder and filter, using three successive portions of neutral 90 percent alcohol and allowing the solution to run into the 110 cc flask. Titrate the alcoholic solution with decinormal potassium hydroxide solution, using phenolphthalein, and calculate the Polenske value. Acetyl Value. Compounds containing a hydroxyl group will readily combine with acetic anhydride, acetic acid and an acetyl compound being produced. This takes place with an oil con- taining free higher alcohols or hydroxy-acids, the latter either in the form of esters or of free acids. The general reaction may be thus shown: RCHOHCOOH + (CH 3 CO) 2 0-^RCHOCH 3 COCOOH + CH 3 COOH, ROH + (CH 8 CO) 2 0->ROCH 3 CO + CH 8 COOH. OILS, FATS AND WAXES 377 For example lanopalmic acid forms acetolanopalmic acid: C 15 H 30 OHCOOH + (CH 3 CO) 2 O^Ci 5 H 3 oOCH 3 COCOOH + CHsCOOH. After washing out the excess of acetic anhydride the amount absorbed may be determined by saponifying the oil with an alcohol solution of potassium hydroxide, evaporating the alcohol, adding standard sulphuric or hydrochloric acid to liberate the acetic and fatty acids and either distilling the acetic acid or washing out with water, then titrating. The reactions illustrated by the case of aceto-lanopalmitin are 3Ci 5 H3oOHCOOK+3CH 3 COOK, 2CH 3 COOK+H 2 SO4^CH 3 COOH+K 2 S04. Effect of Soluble or Volatile Acids. It should be noticed that whether the distillation or the filtration process is employed, the standard base required to finally titrate the acid will include that equivalent to acids other than acetic. That is, the distilla- tion process will yield a distillate of acetic acid and volatile or- ganic acids while the filtration process will yield a filtrate con- taining acetic acid and soluble organic acids. The close relation between soluble acids and volatile acids has already been dis- cussed (page 373). To correct for the presence of these acids in the solution containing the acetic acid one may either subtract the volume of base used in the determination of soluble (or vola- tile) acids, or a different method may be used. As a rule this correction will be small but with oils showing a high soluble- acid number or Reichert-Meissl number, failure to apply the cor- rection may result in serious error. Benedikt and Ulzer 1 proposed first saponifying the oil and then liberating the fatty acids by the addition of sulphuric acid. After washing the fatty acids they are acetylated and the excess of acetic anhydride removed. The acids are then titrated in cold alcoholic solution, under which circumstances the carboxyl alone reacts with the base. The acetylated soap is then heated 1 Monatsh., 8, 41 (1887). 378 QUANTITATIVE ANALYSIS with alcoholic potassium hydroxide when the acetyl radical is saponified. A titration of the excess of base gives the acetyl value. This method avoids the interference of soluble fatty acids but, as was shown by Lewkowitsch 1 it is subject to another error in the fact that acetic anhydride also reacts, to a small extent, with non-hydroxylated fatty acids forming acetic acid and fatty acid anhydride: 2C 1 5H3iCOOH+(CH3CO) 2 0-^(Ci5H3iCO) 2 0+2CH 3 COOH. Palmitic acid Acetic anhydride Palmitic anhydride Acetic acid In cold alcoholic solution these anhydrides are not at once saponi- fied, part remaining until the treatment with hot potassium hydroxide solution, being then saponified and giving rise to a positive error in the calculation of acetyl value. The most desirable method is to acetylate the oil/ wash free from acetic acid, saponify, liberate the fatty acids and acetic acid from the soap and then either distill or filter, titrating the acids of the distillate or filtrate and making the proper correction for volatile or soluble acids. The " acetyl value" is defined to be the number of milligrams of potassium hydroxide required to combine with the acetic acid lib- erated from 1 gm of acetylated fat or oil. Certain oils are charac- terized by unusually high acetyl values. Castor oil is the most noteworthy of these, having a value of about 150. Another class of oils having high acetyl values is composed of " blown" or ''oxidized" oils. By blowing air through oils at somewhat ele- vated temperatures (70 to 115) the viscosity and specific gravity are considerably increased and they become suitable for use as lubricating oils. The chemical changes that take place are not thoroughly understood but oxidation is known to occur. This is partly due to combination with unsaturated acids (evi- denced by a diminished iodine absorption number) and partly to the formation of hydroxyl radicals from hydrogen. The latter change results in an increased acetyl value and this may even reach a number as great as that for castor oil. The large variation in acebyl values recorded in the table on page 379, adapted from a similar table by Lewkowitsch, 2 will 1 Proc. Chem. Soc., 6, 72 (1890). 2 J. Soc. Chem. Ind., 16, 503 (1897). OILS, FATS AND WAXES 379 indicate the value of this determination for the identification of certain oils and fats. In other cases the determination will have little value. Oil or fat Acetyl value (average) Butter fat 0.0 Castor 149 5 Colza 16.6 Cotton seed 21.5 Croton 19.9 Fish 41.0 Linseed 6 9 Maize 8.2 Olive. 13.5 Shark liver 17.8 Abnormal Variation in Acetyl Values. Certain abnormalities in acetyl values should be noticed and due allowance made in specific cases. Since acetic anhydride is absorbed by the hydroxyl radical it might be expected that free acids, free alcohols or partially hydrolyzed glycerides or other esters would show such absorption and that their occurrence in oils or fats would cause these to ex- hibit unusually high acetyl values. This is found to be the case and, since the three classes of substances named above are the direct products of hydrolysis, it follows that rancid oils or fats will not give normal acetyl values. For example, hydrolysis of stearin will yield free stearic acid, together with distearin, monostearin or glycerol, according to the degree of hydrolysis: Each of these reactions produces a hydroxylated compound, which is capable of combining with acetic anhydride. The acetyl values of these substances are as follows: Compound ' Acetyl value Distearin Monostearin .... Glycerol 84.2 253.9 772.0 380 QUANTITATIVE ANALYSIS The free acids will also combine with acetic anhydride to a varying degree and this property will still further increase the acetyl value of rancid materials. The measurement of add value is a convenient method for determining this property. In case high acid values have been obtained the proper correction should be made in the acetyl value as the latter has been determined experimentally. In other words it is only in fresh oils that acetyl values can be used with certainty for identification purposes. The most important application of this determination is in the identification of castor oil. This oil is nearly pure ricinolein, a glyceride of ricinoleic acid. The latter is hydroxylated oleic acid, CH 3 (CH 2 ) 5 CH.OH.CH 2 .CH = CH(CH 2 ) 7 COQH, and the glyceride, ricinolein, has a theoretical acetjd value of 159.1. Its abundance in castor oil gives the latter an actual acetyl value of about 148, a value which is far above that of any other natural oil, only blown oils approaching it in this respect. Lastly may be mentioned the occurrence of certain quantities of free alcohols, especially in the waxes which have, on this ac- count, appreciable acetyl values. Cholesterol, C 2 7H 4 60H, in fats, oils and waxes of animal origin, and its isomers, the phytosterols, in vegetable oils, etc., are the most important of such alcohols. The method here described for the determination of acetyl value is essentially that of Lewkowitsch and adopted as a pro- visional method by the Association of Official Agricultural Chemists. 1 It involves the acetylation of the oil before saponi- fication and includes the soluble or volatile fatty acids, if calcu- lated according to the " official " method. Much confusion would be avoided if the true acetyl value were recorded instead of this acetyl-soluble acid value. In the following exercises the true acetyl value will be calculated. Determination. Place about 20 gm, approximately weighed, of oil or fat in a 100 cc flask, add an equal volume of acetic anhydride, insert a short-stemmed funnel and boil gently for two hours. Cool and pour into 500 cc of water contained in a beaker. Pass a current of car- 1 U. S. Dept. of Agr., Chem. Bull. 107, 142. OILS, FATS AND WAXES 381 bon dioxide into the beaker through a fine orifice of a glass tube and boil for 30 minutes. At the end of this time siphon out the water layer and repeat the treatment with water and boiling until the water is no longer acid, as shown by a litmus test. Separate the acetylated oil in a separa- tory funnel, filter in a drying oven and dry. Weigh accurately 2 to 4 gm of the acetylated oil into a flask and saponify according to the method used in determining the saponification number, measuring the alcohol solution of potassium hydroxide accu- rately and running blank determinations for standardization. Evap- orate the alcohol and dissolve the soap in water. Add standard sul- phuric acid in a quantity exactly equivalent to the potassium hydroxide added, warm to melt the fatty acids and filter through a wet paper. Wash with boiling water until the washings are no longer acid, testing with litmus paper by barely touching a corner to the bottom of the funnel. The combined filtrate and washings are titrated with tenth- normal base. Subtract the volume of base already found to be equiva- lent to soluble acids and calculate the true acetyl value according to the definition of this number. Maumene Number and Specific Temperature Reaction. All oils and fats react with concentrated sulphuric acid, heat being evolved. The reactions are complex and cannot be expressed by a simple equation but oxidation occurs to a considerable degree. The heat evolution varies with different oils and is, to some extent, characteristic. The Maumene number 1 is the number of centigrade degrees rise in temperature caused by mixing 10 cc of concentrated sulphuric acid with 50 gm of oil. A small variation in the proportion of water in the acid causes a con- siderable variation in the heat evolved and to this extent the figures recorded by different investigators are not comparable because " concentrated sulphuric acid," as obtained commer- cially, is not a substance with any definite percent of water. In order to eliminate the errors due to variation in water a determination may be made, using the same amount of acid but substituting 50 gm of water for the oil. The ratio Rise in temperature with oil Rise in temperature with water is known as the "specific temperature reaction." 2 That this 1 Compt. rend., 36, 572 (1852). 2 Thomson and Ballantyne: J, Soc. Chera. Ind., 10, 233 (1891). 382 QUANTITATIVE ANALYSIS number is not subject to variation as is the Maumene" number is shown by the following table in which the specific temperature reaction is multiplied by 100. Kind of oil Sulphuric acid of 95.4 percent Sulphuric acid of 96.8 percent Sulphuric acid of 99 percent Maumene 1 No. Sp. temp, reaction Maumene' No. Sp. temp, reaction Maumen6 No. Sp. temp, reaction Olive Rape 36.5 49 34 104.5 95 127 88 270 39.4 95 44.8 58 96 124 Castor Linseed. . . . 37 89 125.2 269 Water. . 38.6 100 41.4 100 46.5 100 Determination. Place a beaker, about 5X1.5 inches, inside one that is about 6X3 inches and pack the open space between * with wool, asbestos or cotton. Cover the beakers with a piece of cardboard through which passes a thermometer. Weigh into the inner beaker 50 gm of oil. Bring concentrated sulphuric acid to the same temperature as that of the oil and then add under a hood, 10 cc of this acid, stirring thoroughly with the thermometer. When the acid is all in, place the thermometer in the center of the oil-acid mixture and note the highest point attained by the mercury. The total rise in temperature is the Maumen6 number. Determine also the specific temperature reaction as follows: Clean the inner beaker and introduce 50 cc of water. Add 10 cc of acid as before and note the rise in temperature. The Maumen6 number divided by this rise is the specific temperature reaction. The drying oils often develop so much heat that active foaming results. Such oils should be first diluted with petroleum oils or olive oil and the proper correction made in the temperature rise. Qualitative Reactions. If simple and reliable qualitative tests were known for all of the oils, it is not likely that the work outlined in the preceding pages would often be carried out. It has already been explained that comparatively few such tests are known because of the similarity in the composition of the various animal and vegetable oils. Aside from the mere varia- tion in the proportion of the various glycerides, free alcohols and free acids, there are certain constituents of certain oils that will give color reactions which are characteristic. A few of those that are reliable will be described. In most cases these tests OILS, FATS AND WAXES 383 should accompany the determination of the analytical constants, rather than be substituted for them. Resin Oil. Polarize the oil in a 200-mm tube. If the oil is too dark in color for this purpose it may be diluted with petroleum ether and the proper correction made in the reading. Resin oil has a polarization in a 200-mm tube of from +30 to +40 on the sugar scale (Schmidt and Haensch) while other oils read between +1 and 1. Cotton Seed Oil: Halphen Test. 1 Mix carbon disulphide containing about 1 percent of sulphur in solution, with an equal volume of amyl alcohol. Mix equal volumes of this reagent and the oil and heat in a bath of boiling, saturated solution of sodium chloride for 1 to 2 hours. In the presence of as little as 1 percent of cotton seed oil a character- istic red color is produced. Lard and lard oil from animals fed on cotton seed meal will give a faint reaction for cotton seed oil. The unknown constituent which gives the color apparently is assimilated by the animal without change. A negative result does not prove the absence of cotton seed oil because heating the oil for 10 minutes at 250 renders it incapable of giving the color. Arachis (Peanut) Oil. Modified Renard 2 Test. This test is based upon the fact that about 5 percent of " crude arachidic acid" may be isolated from arachis oil, whereas stearic acid is the highest acid that occurs in any considerable quantity in most oils and fats. " Crude arachidic acid'*' is a mixture of true arachidic acid, C 2 oH 4 oO2, and lignoceric acid, C 2 4H 4 8O 2 . The oil is first saponified and the excess of base is neutralized with acetic acid. Lead acetate is then added and the lead soaps of the higher acids are separated from those of the lower acids by washing with ether, in which lead soaps of the soluble acids dissolve. The insoluble soap is decomposed by hydro- chloric acid and the resulting palmitic, stearic, arachidic, ligno- ceric and traces of other higher fatty acids are extracted with ether, which is then evaporated. Finally the acids are dissolved in 90 percent alcohol and the solution is cooled to 15. Stearic and palmitic acids remain in solution and the higher acids crystallize, leaving a saturated solution containing 0.00025 gm of " crude arachidic acid" in each cubic centimeter of alcohol. 1 J. pharm. Chim., [61, 6, 390 (1897). 2 Z. anal, chem., 12, 231 (1871); Compt. rend., 73, 1330 (1871). 384 QUANTITATIVE ANALYSIS By applying this solubility correction the approximate weight of arachis oil in a mixture can be calculated. The test is made as follows: Weigh 20 gm of the oil into a 250 cc Erlenmeyer flask. Saponify with a solution of potassium hydroxide in alcohol as directed in the discussion of the determination of saponification number. Add a drop of phenol- phthalein and exactly neutralize with 5 percent acetic acid and wash into a 500-cc flask containing a boiling mixture of 100 cc of water and 120 cc of a 20 percent lead acetate solution. Boil for a minute and then cool the precipitated lead soap by immersing the flask in water, occa- sionally giving it a whirl to cause the soap to stick to the sides of the flask. After the flask has cooled, the water and excess of lead acetate can be poured off and the soap washed with cold water and with 90 percent (by volume) alcohol. Add 200 cc of ether, cork and allow to stand for some time until the soap is disintegrated; heat on the^ water bath, using a reflux condenser, and boil for about five minutes. In the oils most of the soap will be dissolved, while in lards which contain much stearin, part will be left undissolved. Cool the ether solution of soap to from 15 to 17 and let stand until all the insoluble soaps have crys- tallized out (about twelve hours). Filter upon a Biichner funnel or a folded filter and thoroughly wash the insoluble lead soaps with ether, then wash them into a separatory funnel by means of a jet of ether from a wash bottle, alternating at the end of the operation, if a little of the soap sticks to the paper, with hy- drochloric acid, 10 percent solution. Add sufficient of this dilute hydro- chloric acid so that the total volume of the acid layer amounts to about 200 cc and enough ether to make the volume of the ether layer 150 to 200 cc and shake vigorously for several minutes. Allow the layers to separate, run off the acid layer and wash the ether once with 100 cc of dilute hydrochloric acid and then with several portions of water until the washings are no longer acid to methyl orange. If a few undecom- posed lumps of lead soap remain (indicated by solid particles remaining after the third washing with water) break these up by running off almost all of the water layer and then 'add a little concentrated hydrochloric acid, shake and then continue the washing with water as before. Distill the ether from the solution of higher fatty acids and dry the latter in the flask by adding a little absolute alcohol and evaporating on the steam bath. Dissolve the dry fatty acids by warming with 100 cc of 90 percent alcohol (volume) and cool slowly to 15, shaking gently to aid crystallization. Allow to stand at 15 for 30 minutes. If arachis oil has been present arachidic acid will separate from the solution as crystals. Filter, wash the precipitate twice with 90 percent alcohol OILS, FATS AND WAXES 385 and then with 10 cc portions of 70 percent alcohol, care being taken to keep the acid crystals and the washing alcohol at definite temperature in order to be able to apply the solubility corrections given below. Dissolve the arachidic acid upon the filter with boiling absolute alcohol, evaporate to dry ness- in a weighed dish and weigh. Add to the weight found 0.0025 gm for each 10 cc of 90 percent alcohol used in the crystallization and washing. 20 times the corrected weight of acid will be the approximate weight of arachis oil in the sample used. The crystals should be tested qualitatively by determining the melting point, which should be 71 to 72. The crystals should also be examined under the microscope. As little as 5 percent of peanut oil may be detected by this method. Sesame Oil: Baudouin Test. 1 Dissolve 0.1 gm of finely powdered sugar in 10 cc of hydrochloric acid (sp. gr. 1.20), add 20 cc of the oil to be tested, shake thoroughly for a minute, and allow to stand. The aqueous solution separates almost at once. In the presence of even a very small admixture of sesame oil this is colored crimson. Some olive oils give a slight pink coloration with this reagent, but they are not hard to distinguish if comparative tests with sesame oil are made. The color was thought by Villa vecchia to be due to a reaction of a constituent of sesame oil with furfurol, the latter being produced by the interaction of sugar with hydrochloric acid. Furfurol was accordingly substituted for sugar and hydrochloric acid and the method somewhat modified as follows: Sesame Oil: Villavecchia Test. 2 Add 2 gm of furfurol to 100 cc of alcohol (95 percent) and mix thoroughly 0.1 cc of this solution, 10 cc of hydrochloric acid (sp. gr. 1.20) and 10 cc of oil by shaking them together in a test-tube. The same color is developed as when sugar is used, as in the Baudouin test. Villavecchia explained this reaction on the basis that furfurol is formed by the action of levulose and hydro- chloric acid and he therefore substituted furfurol for sucrose. As fur- furol gives a violet tint with hydrochloric acid it is necessary to use the very dilute solution specified in this method. Fish and Marine Animal Oils in Mixtures with Vegetable Oils. Practically all of these oils have very considerable " drying" properties, as shown by their iodine absorption numbers. They are characterized by the presence of glycerides 1 J. Assoc. Off. Agr. Chem., Vol. II, No. 3, Pt. II, 314. 2 J. Soc. Chem. Ind., 12, 67 (1893); 13, 69 (1894). 25 386 QUANTITATIVE ANALYSIS containing the unsaturated clupanodonic acid, whose formula and properties are mentioned on page 364. The peculiar " fishy" odor of these oils is probably due to the presence of this acid. Absorption of bromine by unsaturated acids or their glycerides produces bromides of limited solubility and high melting point. Octobromstearin, obtained from clupanodonin, melts at a higher temperature (above 200) and has a lower solubility than hexa- bromstearin, obtained by brominating linolenin, and this also differs in a similar manner from tetrabromstearin, obtained from linolin. Therefore the separation of Octobromstearin from brominated fish and blubber oils provides a means for detecting marine animal oils in the presence of vegetable oils. The test is performed as follows: Dissolve in a test-tube about 6 gm of the oil in 12 cc of a mixture of equal parts of chloroform and glacial acetic acid. Add bromine, drop by drop, until a slight excess is indicated by the color, keeping the solution at about 20. Allow to stand for 15 minutes or more and then place the test-tube in boiling water. If only vegetable oils are present the solution will become perfectly clear, while fish oils will remain cloudy or contain a precipitate of insoluble bromides. Color Reactions. A large number of qualitative tests, based upon certain color reactions, have been proposed and considerably used in the past for the detection of various oils. Color reactions produced by adding concentrated nitric or sulphuric acids may be mentioned. Almost without exception these have been found to be unreliable and they will not be described here in detail. The "elaidin test" is worthy of brief mention. This is based upon the conversion of liquid olein into its solid isomer, elaidin, by the action of nitrous acid, the change being from oleic acid of olein to elaidic acid of elaidin: CH 3 (CH 2 ) 7 CH CH 3 (CH 2 ) 7 CH II - II HC(CH 2 ) 7 COOH HOOC(CH 2 ) 7 CH Linolin, linolenin or clupanodonin are not thus affected and the test serves to distinguish between liquid non-drying oils and OILS, FATS AND WAXES 387 drying oils. It has been used to a considerable extent in testing the purity of olive oil but must be performed under strictly standardized conditions if it is to have even a qualitative value. Examination of an Oil or Fat Whose Identity is Unknown. For the purpose of identifying an oil or fat of unknown character the complete chemical and physical examination is made unless its identity can be determined unmistakably by a qualitative test. With all of the constants determined a comparison is then made with all available data contained in published analyses of oils and fats and a reasonable agreement with such data will generally fix the identity of the unknown oil. In making com- parisons the most important figure is the iodine number because this will serve to classify the oil at once as a drying, semi-drying, or non-drying oil, the approximate ranges for these somewhat arbitrary divisions being as follows: Oils Iodine Number Drvinsr 200 and higher to 120 Semi-drying 120 to 95 Non-drying. 95 to 70 and lower The choice will, by this means, be narrowed down to a limited list of oils or fats. The remaining constants are then considered, one by one, and each comparison will narrow the choice still further. At the last all available qualitative tests are made in order to confirm the results of comparative tests or to aid in making a final decision. It sometimes happens that the figures obtained in the examination of the unknown oil will not all correspond, even to a reasonable degree, with the recorded data for any of the common oils. This may be the result of (1) errors in the determinations, (2) adulteration, or (3) a real abnor- mality of the oil which is being examined. The first case should be at once excluded by repeating the determination of constants in which lack of agreement is observed. A careful inspection of all data may serve to indicate certain oils which, by addition to one that most nearly resembles the oil under examination, would change the " constants" in the manner observed. The matter of commercial values should also be considered in this. 388 QUANTITATIVE ANALYSIS OIL CONSTANTS The following figures represent average values as determined on many samples of the oils indicated, and by many different chemists. Exact agreement should not be expected. Specific gravity and index of refraction are calculated for the temperatures chosen, from results reported for various temperatures. Oil Sp. gr. at 20 Ind. refr. at 20 Iodine Sapon. R.M. Acety Maumene Sp. temp. Hemp seed Henbane Linseed . . . 0.919 0.936 929 1.477 1 482 148 138 186 192 171 193 o 4 97 127 331 Poppy seed Soy bean 0.922 922 1.475 1.480 1 476 138 140 127 195 193 193 88 90 72 220 167 Tune 936 1 503 161 193 Walnut 922 1 478 145 195 102 985 1 552 105 178 Cotton seed Croton Grape seed 0.920 0.947 0.932 1.472 1.478 1.473 109 103 96 194 212 185 0.8 13 5 13 26 82 63 170 Maize 0.920 1.475 119 190 5 7 83 180 Rape 912 1 472 98 175 5 15 60 135 920 1 474 105 191 1 2 6 65 155 915 1 470 95 191 1 13 53 110 916 1 471 90 193 5 9 47 116 Castor 961 1 478 86 184 1 5 148 46 84 Hazel nut Olive Pistachio 0.913 0.914 0.915 1.469 1.469 1.471 86 83 90 192 190 191 1.0 0.6 3 11 36 43 91 Quince Cod liver Herring 0.919 0.921 0.927 1.471 1.481 113 168 133 182 180 177 0.5 0.6 5 115 258 Menhaden Neat's foot Salmon 0.927 0.912 1.480 1.467 1 479 156 73 161 191 195 183 1 1.5 22 126 52 306 95 Seal 925 1 475 147 192 16 92 278 Sheep's foot 0.914 74 195 50 Whale 92*> 1 476 134 190 92 157 Caoao butter Cocoa nut fat Japan wax Myrtle wax at 60 0.929 0.897 0.744 964 at 60 1.450 1.441 1.450 1 444 36 9 10 3 194 253 227 209 0.5 7 3 11 28 Palm 890 1 451 54 199 1 2 18 25 59 Palm nut Beef tallow Butter 0.921 0.916 902 1.443 1.451 1 440 15 42 38 246 196 227 5 0.7 27 5 5 5 Chicken 893 65 193 Goose 896 1 452 65 193 2 Hare 1 452 102 201 Horse 888 1 455 78 196 1 8 50 105 Lard 0.905 1 454 55 196 0.4 3 27 Mutton tallow 914 1 451 40 193 Beeswax 938 1 450 g 94 4 15 Carnaiiba wax Spermaceti Sperm oil Wool wax 0.964 0.901 0.851 0.913 1.463 1.440 1.465 (20) 1.473 13 4 86 23 87 129 135 100 1.3 55 3 5 23 51 97 OILS, FATS AND WAXES 389 connection, since any commercial material is adulterated, if at all, by a cheaper material. After the decision as to the identity of the oil has been made or has been limited to two or three possible oils, consult a good reference book for a complete discussion of these oils and make any additional tests that may be there suggested. For this purpose are to be recommended Lewkowitsch's Chemical Analysis of Oils, Fats and Waxes, 5th edition, volume 2, and Allen's Commercial Organic Analysis, volume 2, part 1. The figures in the table on page 388 are given for the purpose of comparison. They are gathered from various published analyses of the more common oils and fats. Exact agreement should not be expected. For more extensive tables consult Lewkowitsch: Chemical Analysis of Oils, Fats and Waxes, and the technical journals. Hardened Oils. Under any circumstances the analytical investigation of oils and fats offers difficulties that are often serious. The problems of the analyst are now increased many fold by the large development of the industry of hydrogenation of liquid oils. It has been seen that the most important difference between oils and fats lies in the larger proportion of olein in the former and of stearin and palmitin in the latter. Olein differs from stearin only in that it contains one unsaturated double bond in each oleic acid residue; the problem of saturating this group by the insertion of hydrogen, thus forming stearin, is one that has oc- cupied the attention of chemists for many years. At the present time the hydrogenation of the cheaper liquid oils (e.g., cottonseed, corn and peanut) to form edible fats is an industry that has at- tained large proportions. While this process changes liquid oils to solid fats, it will also make a corresponding change in any analytical constants or tests that depend upon the degree of unsaturation, as well as in the physical properties of the oil. Linolin, linolenin and clupanodonin will be changed to stearin. Consequently the halogen absorption number, drying properties, specific gravity, refractive index and temperature reactions will be materially altered, as will also the odor and the general appearance and consistency. It has been stated that fish oils probably owe their characteristic odor to glycerides containing 390 QUANTITATIVE ANALYSIS clupanodonic acid, while the somewhat similar odor of linseed oil is due to gycerides of linolenic acid. It is interesting to note that these odors are entirely lost through hydrogenation and that the oils are no longer recognizable by any ordinary tests. Many individual tests for other oils, such as the Halphen reaction for cottonseed oil and the Renard test for sesame oil, fail in the hydrogenated product. From one standpoint it might appear that the determination of what oils originally formed the raw materials for the "hard- ended" product is not a necessary one for the analyst to solve, since the properties of the finished product are, after all, the ones that have the chief practical interest for us. Yet it may some- times happen that the identity of the original oil, or the proof that a hydrogenating process has been employed may have a legal or other significance and the development of a series of suitable tests is very desirable. Analytical chemistry has made little progress in this direction. The application of delicate tests for metals (nickel, palladium, etc.) that are used as catalyzers in the hardening process, may sometimes serve to show that the material is a hardened oil, rather than a natural fat. Other than this one can say very little. But this knowledge of the nature of the changes caused by hydrogenation should serve to make the analyst more cautious than he might otherwise be when interpreting the results of his tests of oils or fats of unknown origin. CHAPTER XIV WATER The chemical examination of water may be made to determine its fitness for drinking or for industrial uses, such as steam pro- duction, laundering, textile industries, etc. It is not necessary that a complete analysis should be made for all of these purposes because not all substances occurring in water are equally impor- tant in the different applications of the water. Natural waters often contain substances that are objectionable if they are to be used industrially and these substances are, for the most part, inorganic salts and, occasionally, acids. Most of such inorganic materials are without appreciable effect upon the human system and the examination for potability is rather directed toward the detection of pollution by sewage. On this account it becomes necessary to treat the subject of water analysis in two distinct divisions. Industrial Analysis. By far the largest industrial consumption of water is for the production of steam and for this reason the chemist is more often called upon for the analysis of water to determine its fitness for steaming than for any other industrial purpose. Pure water, however desirable it may be for use in the steam boiler, is not a natural product. Water from streams and other surface origins contains mineral and organic substances derived from the surface soil as well as inorganic compounds derived from springs which feed the stream. Water from wells contains whatever mineral matter is common to the region through which it has flowed. Even rain water contains organic matter and ammonia and may develop organic acids when stand- ing. Some of the compounds contained in water are com- paratively unobjectionable because their action is slight. It is to be remembered, however, that in steam boilers the tempera- ture is higher than 100 because of the increased pressure. At a pressure of 100 pounds per square inch the boiling-point of 391 392 QUANTITATIVE ANALYSIS water is 164 and at 200 pounds per square inch the boiling-point is 194. At these temperatures the chemical activity of many dissolved substances is very much augmented. According to their effects upon boiler steel the constituents of natural waters may be classified as corrosives, incrustants and foam producers. Corrosives. Any soluble compound that can dissolve iron at high temperatures will give rise to pitting of the boiler, especially when the steel is not of uniform composition. Corrosives com- monly occurring in water are chlorides, nitrates, and sulphates, particularly of the alkaline earth metals, and free carbonic acid. Free inorganic acids are of rare occurrence and absolutely unfit a water for steaming without preliminary treatment. A small amount of acid will cause corrosion for an indefinite period because of the ready hydrolysis of iron salts. A cycle of re- actions takes place as, follows: Fe+2HCl->FeCl 2 +H 2 , 6FeCl 2 +3O-+4FeCl 3 +Fe 2 O 3 , FeCls+3H 2 0-+Fe(OH)8+3HCl. A metal chloride which is easily hydrolyzed will also produce continuous corrosion: MgCl 2 +2H 2 0-^Mg(OH) 2 +2HCl, Fe+2HCl-+FeCl 2 +E 2 , etc. Nitrates are equally injurious, although they seldom occur in more than small concentration. Sulphates are somewhat less corrosive and free carbonic acid still less so. Incrustants. Any substance that can be precipitated by heat- ing or evaporation of water is, in a sense, an incrustant. The steam boiler as a power producer is also a machine for continuous concentration of water solutions, since 'fresh, impure water is continually added and only vapor is removed. Strictly speaking only those substances which adhere to the boiler plate when they are precipitated are classed as incrustants because only these are particularly objectionable. These are carbonates of calcium and magnesium and calcium sulphate. In presence of considerable quantities of these materials certain other compounds, such as WATER 393 silicic acid, iron oxide and aluminium oxide, may be included with the scale and then become incrustants. Calcium and magnesium carbonates are not dissolved as such in water but are present as bicarbonates, having been dissolved from the mineral carbonates by carbonic acid. CaCO 3 +H 2 C0 3 -Ca(HC0 3 ) 2 , MgC0 3 +H 2 CO 3 ^Mg(HCO 3 ) 2 . When the water is heated reactions which are the reverse of these take place and the normal carbonates are precipitated: Ca(HCO 3 ) 2 -+CaC0 3 +H 2 O+CO 2 , Mg(HC0 3 ) 2 ^MgC0 3 +H 2 0+C0 2 . These carbonates adhere to the boiler plate, the greatest amount of precipitation occurring over the heating surface. The scale thus formed, although comparatively loose, hinders the trans- mission of heat from the steel to the water and causes local super- heating. The result is a loss of efficiency and injury to the boiler. Although these substances occur in the water as bicarbonates, they are arbitrarily calculated as normal carbonates because the latter are precipitated when the water is heated. Calcium sulphate precipitates only when continued evapora- tion of the water concentrates it to the point of saturation. Pre- cipitation then causes the formation of a scale that is much more serious in its effects than the scale of carbonates, because it is compact and adheres firmly to the boiler. While carbonate scale can be largely removed by occasionally blowing off the water, calcium sulphate can be loosened only by the use of hammer and chisel. On this account calcium sulphate is one of the most objectionable incrustants of all compounds found in natural waters. Foam Producers. Carbonates of sodium and potassium in- crease the surface tension of water to such an extent that the result is foaming or " priming" as steam is taken from the boiler. Some of the alkali waters of the West contain large quantities of these salts. Expression of Results. The systems used in the calculation of results of water analysis were discussed in connection with the determination of hardness of water (page 231). It is convenient 394 QUANTITATIVE ANALYSIS to work with 1000 cc of water or simple fractions of this quantity and to express results as milligrams of dissolved substance per liter oj water. These figures may be changed to grains per gallon by multiplying by the factor 0.0583 +. The analysis of the water solution will be made by means of methods which give metals and acid radicals as the result of separate determinations. It was at one time customary to calcu- late these as basic and acid anhydrides, as is still done in the analysis of minerals. There arises the same difficulty that is experienced in mineral analysis, viz.: that salts of hydracids cannot be expressed as oxides. A much better rule is to calculate all constituents as positive or negative radicals. Hypothetical Compounds. There is still current among industrial chemists and engineers a custom of making a second calculation of compounds supposed to exist in the- water. Most natural waters are highly dilute solutions of mineral matter. In such a solution most of the compounds are highly ionized and all possible combinations of radicals as compounds are present to some extent, no matter what compounds were originally dissolved by the water. It is evident, therefore, that any list of compounds calculated from the results of the analysis will be entirely fanciful, so far as the actual condition of the solution is concerned. The basis of such a calculation was formerly the supposed affinity possessed by the different radicals for each other. If the radicals commonly occurring in water are arranged in order of decreasing base and acid character the following series will be obtained: Positive Radicals Negative Radicals Potassium Sodium Calcium Magnesium Chloride Nitrate Sulphate Carbonate Based upon the assumption that the combination of these radicals will follow from their relative affinities the mathematical procedure would be to calculate the maximum amount of potas- sium chloride that could be formed, taking the excess of either potassium or chlorine as combined with the next radical of oppo- WATER 395 site sign (either sodium or the nitrate radical) and so on, down the list. If the analysis has been accurately carried out and if all substances present have been determined the positive and negative radicals should be found in equivalent quantities, with a very slight excess of either magnesium or of the carbonate radical after the calculation is finished. This excess is the result of cumulative errors in the determination of the various radicals existing in the water and also of the occasional omission of small quantities of radicals other than those above named. Silicic acid, iron and aluminium are not included in the calcula- tion of hypothetical compounds because the colloidal nature of their hydroxides causes nearly complete, though indefinite, hy- drolysis of any salts that might originally have been present. They are therefore, according to custom, reported as oxides and this conventional method is sometimes responsible for the appear- ance of a slight excess of negative radicals in the final report. If ammonium salts are present in any considerable quantity, as in sewage effluents or factory wastes, a failure to deter- mine this radical also will result in an apparent excess of negative radicals. The customary method of calculating hypothetical com- pounds is not based upon scientific principles, as has already been shown. There is a certain justification for such a calculation, on account of the fact that when a water is heated and evapo- rated there will be produced the least soluble compounds of all that might be formed from the various radicals present. Through a certain coincidence, this would leave the radicals com- bined in about the same manner as is indicated by the conven- tional calculation. Heating in the boiler will produce the maxi- mum possible quantities of normal carbonates of calcium and magnesium. Which of these carbonates is least soluble at high temperatures is not definitely known because of difficulties en- countered in the determination of solubility. It is assumed, however, that if the radical of carbonic acid is not present in quantity sufficient to form carbonates with all of the calcium and magnesium, calcium, rather than magnesium, will ultimately remain to form the sulphate as the water is evaporated. Calcium sulphate is certainly next to the carbonates of calcium and mag- nesium with respect to its insolubility and it will precipitate 396 QUANTITATIVE ANALYSIS when evaporation within the boiler concentrates it to the point of saturation. After these three compounds have been formed, the method of combining the remaining radicals is quite imma- terial because they will not precipitate in any form, on account of the large solubility of salts of the alkali metals. In order to emphasize the real basis for any calculation of compounds we shall reverse the order of radicals given on page 394 and calculate combinations in the following order: Positive Radicals Negative Radicals Magnesium Calcium Sodium Potassium Carbonate Sulphate Nitrate Chloride This will give precisely the same result as the calculations from the original order, unless there is found to be an excess of either positive or negative radicals. In this case the excess will be found to be either potassium or the chloride radical instead of magnesium or the carbonate radical, but the excess should be small enough to be insignificant in either case. On account of the fact that sodium and potassium have very little significance in most boiler waters (the negative radicals and the metals that take part in scale formation being of chief importance) and because they occur in relatively small amounts in most waters, the determination of potassium is frequently omitted and calculations are based upon the assumption that sodium is the only alkali metal present. The conventional method of calculating hypothetical com- pounds is illustrated in the example given below. The analysis of a ground water gave the following results: Milligrams per liter Silica. 5 01 Oxides of iron and aluminium Sodium 3.53 6.02 Potassium 5 26 Calcium 75 41 Magnesium 24 19 Chloride radical 4 52 Nitrate radical 34 Sulphate radical 32.91 Carbonate radical 160.21 WATER 397 Following is the calculation of compounds: 30 24.19+59.65 = 83.84 = MgCO 3 . 160.21 - 59.65 = 100.56 = (CO 3 )" remaining; ~^ X 100.56 = 67.20 = Ca<> (CO 3 )" remaining; 100.56+67.20= 167.76 = CaCO 3 . 75.41 -67. 20 = 8.21 = Ca remaining; ||^|x8.21 = 19.69 = (S0 4 )"oCa remaining; ZO.Oo 8.21 + 19.69 = 27.90 = CaSO 4 . 32.91 - 19.69 = 13.22 = (SO 4 )" remaining; X6> 02 = 12.56 = (S0 4 )"oNa; 6.02+ 12.56 = 18.58 = Na 2 SO 4 . 13.22 - 12.56 = 0.66 = (SO 4 )" remaining; on -JA ||^ X 0.66 = 0.53 = K=c=(S0 4 )" remaining; 0.66 + 0.53 = 1.19 = K 2 SO 4 . 5.26 -0.53= 4.73 = K remaining; 0.34+0.21 = 0.55 = KNO 3 . 4.73 -0.21= 4.52 = K remaining; 2g^X4.52 = 4.10 = Cl'-K remaining; 4.52+4.10 = 8.62 = KCl. 4.52 4. 10 = 0.42 = 01' remaining. This excess of chlorine represents experimental errors as is explained above. Since "milligrams per liter" multiplied by 0.0583 gives "grains per gallon" the complete statement of compounds is as follows. 398 QUANTITATIVE ANALYSIS Formula for compounds Milligrams per liter Grains per gallon SiO 2 5 01 292 Fe 2 O 3 +Al 2 O 3 KC1 3.53 8.62 0.216 0.502 KNO 3 . . . 55 032 K 2 SO 4 1 19 069 Na 2 SO 4 CaSO 4 18.58 27.90 1.082 1.626 CaCO 3 . . 167 76 9 775 MgCO 3 83.84 4.885 The use of the conversion factors as in the above illustration fits only the analysis that was used as the example. In the case of other waters certain factors might be the reciprocals of those used above or they might involve different pairs of equivalent weights. All possible combinations should be calculated as a preliminary exercise and the factors with their logarithms recorded with the list of factors in the note book. It should be noted that for water analysis the conversion factors are not multiplied by 100, since percents are not to be calculated. Problem 73. Calculate the conversion factors used above into the corresponding mixed numbers by carrying out the divisions indicated. Also calculate the following additional factors and their logarithms and record all of these in the note book. The first factor is given as an example. Found Required Factor Log. Found Required Factor Log. Mg C0 3 SO 4 2.467 0.3922 C0 3 Ca Mg NO 3 Na Cl K Ca CO 3 SO 4 Ca SO 4 Mg NO 3 Na Cl K ' Na CO 3 NO 3 Ca SO 4 Mg NO 3 Na Cl K K CO 3 Cl Ca SO 4 NO 3 Mg Na Cl K WATER 399 Industrial Analysis of Water. Measure accurately in a calibrated flask enough water to give, upon evaporation, 0.5 gm to 1 gm of residue. Add 2 cc of concentrated hydrochloric acid and evaporate in a platinum or porcelain dish on the steam bath. Heat the residue at a temperature below redness until organic matter is removed. Silicious Matter. Add 1 cc of concentrated hydrochloric acid to the residue and then add about 20 cc of hot water. Warm and stir until all soluble matter has dissolved then filter on an extracted filter paper. Wash well with hot water until the combined filtrate and wash- ings amount to about 75 cc. Fold the filter paper and burn in a weighed platinum crucible. The residue is reported as "silicious" if it is white and does not weigh more than 5 mg, otherwise it may contain appreciable amounts of metal oxides or silicates. If the weight is greater than 5 mg, add to the residue a drop of sulphuric acid and then volatilize the silica by warming with 1 cc (more if necessary) of hydro- fluoric acid. Ignite the residue and weigh. Report the loss as silica. Dissolve any remaining residue in concentrated hydrochloric acid and add to the main solution. Oxides of Iron and Aluminium. Drop into the solution a very small bit of litmus paper and carefully add dilute ammonium hydrox- ide until the solution is slightly basic. Boil gently to flocculate the hydroxides of iron and aluminum and to remove any unnecessary excess of ammonium hydroxide. Filter and wash with hot water until free from chlorides. Ignite and weigh and report as oxides of iron and aluminium. Usually iron is not present in quantity sufficiently large to make its separate determination important. If this is desired the method given on page 288 may be used and the equivalent amount of ferric oxide sub- tracted from the combined oxides, the remainder being aluminium oxide. Calcium. Add 1 cc of dilute ammonium hydroxide to the filtrate and washings from the iron and aluminium hydroxides and then precipitate and determine the calcium with the precautions mentioned in the discus- sion on page 81. Report as calcium. // potassium is to be determined, add to the solution 2 cc of concen- trated sulphuric acid and evaporate in a weighed platinum dish to dryness. Heat the residue carefully to evaporate excess of sulphuric acid and then more strongly to expel ammonium salts, finally heating for a short time until the dish is dull red. If sulphuric acid fumes do not appear upon heating, excess is not present and more should be added before the stronger heating. Weigh and record the weight of sulphates of sodium, potassium and magnesium. Dissolve the combined 400 QUANTITATIVE ANALYSIS sulphates and dilute the solution to 250 cc in a calibrated volumetric flask. Magnesium. Fill a dry 100-cc volumetric flask with the solution of sulphates and rinse into a Pyrex beaker. Determine magnesium as directed on page 112. Notice that only 0.4 of the original sample was used for this determination. Potassium. Use 100 cc of the sulphate solution that was obtained just before the determination of magnesium. Evaporate in a platinum dish over the steam bath adding the necessary quantity of chlorplatinic acid before crystallization of salts begins. Determine potassium by the Lindo-Gladding method, page 102. Here also 0.4 of the original quantity of sample was used. Sodium. Calculate the weight of potassium sulphate equivalent to the potassium chlorplatinate found and also the weight of magnesium sulphate equivalent to magnesium pyrophosphate found. Multiply the sum of these weights by 2.5 and subtract the product so obtained from the total weight of combined sulphates already, found. The remainder is sodium sulphate. From this calculate sodium. // potassium is not to be determined the weighed residue of sulphates is dissolved and the solution is diluted to about 100 cc in a 250 cc beaker of Pyrex or similar resistance glass. Magnesium is then determined in the entire solution. From the weight of magnesium pyrophosphate found calculate the weight of magnesium sulphate equivalent to it and subtract this from the weight of combined sulphates, already found. The remainder is assumed as the weight of sodium sulphate, from which sodium is calculated. Sulphates. Use 100 cc of water unless a qualitative test shows the presence of only a small concentration of sulphates, in which case 500 cc or more should be evaporated to about 100 cc. Add 0.5 cc of concen- trated hydrochloric acid and precipitate by barium chloride, carrying out the precipitation and treatment of the precipitate as directed on page 95. Calculate milligrams per liter of the sulphate radical. Chlorides. Make 500 cc of a standard solution of pure sodium chloride, 1 cc of which contains 0.001 gm of chlorine. Make 1500 cc of a solution of silver nitrate such that 1 cc is calculated to be equivalent to about 0.000505 gm of chlorine and standardize against the sodium chloride solution as follows : Measure 25 cc of the standard sodium chlo- ride solution into a 4-inch porcelain casserole or into a beaker placed on a white background. Add 1 cc of a 5 percent solution of potassium chromate (from which chlorides have been precipitated by the addition of a slight excess of silver nitrate) and then titrate with silver nitrate solution until the first permanent red tint of silver chromate appears. This is the end point of the reaction between silver nitrate and sodium WATER 401 chloride. Calculate the dilution necessary to make 1 cc of the silver nitrate solution equivalent to 0.0005 gm of chlorine and dilute 1000 cc to the required volume by adding distilled water from a burette. Determine chlorine in the water by titrating 100 cc or more of water by standard silver nitrate, using 1 cc of potassium chromate as the indicator. In case chlorides are present in very small concentration it may be necessary to use more than 100 cc and to evaporate to this volume. The end point of the titration is much more easily observed if a rather heavy precipitate of silver chloride is present at the end. If, therefore, the concentration of chlorine in the water is small it is ad- vantageous to add first a measured volume (10 to 25 cc) of standard sodium chloride solution, deducting this volume from the total volume of silver nitrate required. Calculate milligrams per liter of the chloride radica . Nitrates. Use one of the methods described on pages 427 and 428 for the sanitary examination of water. Calculate milligrams per liter of the nitrate radical. The concentration of nitrates in most waters is too small to be of consequence in steaming but this is not always the case and the determination should not be omitted. Carbonates. Titrate 100 cc of the water by fiftieth-normal hydro- chloric acid, using methyl orange as indicator. More or less than 100 cc of water may be necessary if the concentration of carbonates is very small or very large. Enough should be used to require 25 to 45 cc of standard acid. Calculate milligrams per liter of the carbonate radical (COs)". This is based upon the customary conventional assumption of normal carbonates instead of bicarbonates. The determination of carbonates by titration with standard acid is the determination of alkalinity as described on page 231 except that hardness is arbitrarily calculated as calcium carbonate while in this connection the carbonate radical is calculated. Calculate milligrams per liter of the hypothetical compounds using the method illustrated on page 394. Calculate the corresponding grains per U. S. gallon. Treatment. If the chief incrustants of a water are bicarbonates of calcium and magnesium these may be largely precipitated by heating the feed water by means of the exhaust steam. This kind of treatment is limited in its application on account of the short time allowed for settling. Other incrustants than those mentioned are not removed by this process. Calcium hydroxide will precipitate bicarbonates and this is 20 402 QUANTITATIVE ANALYSIS the cheapest agent available for this purpose. It also possesses the advantage of leaving no by-products in the water: Ca(HC0 3 ) 2 +Ca(OH) 2 ->2CaC0 3 +2H 2 0. Mg(HCO 3 ) 2 +Ca(OH) 2 ^CaCO 3 +MgC0 3 +2H 2 0. Sodium carbonate will also precipitate bicarbonates as well as other salts of calcium and magnesium. It, however, leaves in the water the corresponding salt of sodium and this is objection- able. Examples of the reactions are expressed by the following equations : Ca(HC0 3 ) 2 +Na 2 C0 3 -^CaC0 3 +2NaHCO 3 , CaSO 4 +Na 2 CO 3 -CaCO 3 +Na 2 SO 4 , . CaCl 2 +Na 2 C0 3 ->CaC0 3 +2NaCl. The best procedure is to treat the water first with the amount of calcium hydroxide necessary to react with bicarbonates, allowing a small excess, and then with sufficient sodium carbonate to precipitate all remaining calcium and magnesium. The reactions should be carried out in tanks large enough to provide water for the plant, allowing the necessary time for settling. The initial cost of the purifying plant is almost the only cost because of the relatively low cost of the small amount of lime and soda ash necessary. In calculating the necessary treatment the purity of the lime or lime water must be known and also that of the soda ash. The results may be expressed according to custom as pounds of reagent per 100,000 gallons of water. The quantity of (C0 3 )", expressed as grains per gallon, is multiplied by the fraction equivalent weight of CaO x equivalent weight of (CO.)"' the TeSult bemg *""* f ^ ^ quired for one gallon of water. One pound avoirdupois contains , grains CaO per gallon X 100,000 7000 grains. Therefore- ^T^TT - = pounds of calcium oxide required for 100,000 gallons of water. Instead of this the calculation may be made to gallons of lime water per 100,000 gallons of water, where a saturated solution of calcium hydroxide is first made and its concentration determined. WATER 403 The amount of sodium carbonate necessary for the precipita- tion of other salts of calcium and magnesium may be calculated in a similar manner. An inspection of the statement of the analysis as given on page 395 will show that calcium and magnesium as carbonates may be precipitated by lime water, while calcium as sulphate may be removed by sodium carbonate. Since no other bicarbonates are present the amount of calcium oxide required may be calculated directly from the amount of carbonate radical. 56X160. 21X0. 0583 X 100,000 _ 10/< ft 60X7000 Therefore 124.6 pounds of calcium oxide will be required to precipitate calcium and magnesium carbonates. 53X27.90X0.0583X100,000 68X7000 18.1. Therefore 18.1 pounds of sodium carbonate will be required to remove calcium sulphate from 100,000 gallons of water. The remaining salts are classed as corrosives and they cannot be removed by chemical treatment. Problem 74. Calculate the treatment for the water already analyzed in the laboratory. Boiler Compounds. Many commercial mixtures are now on the market, designed to be mixed with feed water as it enters the boiler to prevent the formation of scale or to loosen scale already formed. Most of these mixtures are solutions of sodium carbon- ate or sodium hydroxide with or without the addition of tannin or some other colloidal organic compound. The base precipitates bicarbonates but this precipitation occurs within the boiler where it would occur if no base had been added. The base is a strong corrosive and may be highly injurious to the boiler if added in . excess. Colloidal bodies, such as tannin, starch or dextrine, have a certain loosening effect upon the scale already present and to some extent prevent the formation of compact 404 QUANTITATIVE ANALYSIS scale. This action can produce only temporary relief, and a preliminary treatment such as has already been described is much cheaper and better. Examination of Water for Sanitary Purposes. The examina- tion of water to be used for drinking may be made to determine the quantity of mineral salts that have a supposed medicinal value or to determine its potability. The examination for the first purpose will follow the lines of methods already described for boiler water or it may be extended to include other substances, such as lithium, free carbonic acid, hydrogen sulphide, etc. It is practically certain that none of the mineral waters that are exploited in a commercial way contains enough of any salt or gas to have any appreciable effect upon the human system and that the beneficial effects that are noticed by those who take treatment at the mineral springs are due largely or entirely to other causes, such as enforced dieting, bathing, relaxation from care, good exercise and the drinking of plenty of water. This being the case it is apparent that the chemist's report on the analysis of mineral waters can have little value except as a commercial document. If this analysis is demanded the methods already given may be used, proper modifications being made in quantity of water and reagents according to large variations in the percent of mineral matter present. Additional determinations will be described. Free Carbonic Acid. Water containing free carbonic acid will rapidly lose carbon dioxide and it therefore becomes necessary to make this determination as soon as the water sample is taken or to preserve the sample in tightly closed bottles entirely filled with water. The determination depends upon the fact that as sodium carbonate is added to carbonic acid in presence of phenol- phthalein a color change occurs at the moment that all of the car- bonic acid has been used in forming sodium bicarbonate, sodium carbonate having a basic reaction toward phenolphthalein. The end point is shown when the following reaction is completed : Na 2 CO3+H 2 C03-+2NaHCO 3 . Determination. Calculate the normality of a solution of sodium car- bonate so made that 1 cc is equivalent to 0.001 gm of carbon dioxide. Make 1000 cc of such a solution, using sodium carbonate prepared by WATER 405 heating recrystallized sodium bicarbonate to about 300 until it ceases to lose weight. Do not allow the solution to come into contact with air more than is necessary. Titrate 100 cc of water as rapidly as possible with the standard sodium carbonate solution and report milligrams per liter of carbon dioxide. The report is often made as volume of gas per unit volume of water at some specified temperature as, for example, cubic inches of gas per gallon of water, at 60 F. Reference to tables of density of carbon dioxide will give the necessary data for this calculation. Hydrogen Sulphide. The determination of hydrogen sulphide must be made as soon as the water is taken and for the same reasons that apply to carbon dioxide. Titration is made with standard iodine solution according to the reaction: Determination. Make a solution of iodine, 1 cc of which is equivalent to 0.001 gm of hydrogen sulphide. Standardize by titrating with stand- ard sodium thiosulphate. Titrate 100 cc of the water with the standard- ized iodine solution, using starch solution as indicator. Calculate milligrams per liter of hydrogen sulphide, also cubic inches per gallon. Iron. The quantity of iron in water is usually too small to make ordinary volumetric methods desirable for its determina- tion and a more sensitive colorimetric method is substituted. The iron is obtained in the ferric state and is treated with potas- sium thiocyanate. The red color so produced is compared in tubes with that formed by a standard iron solution. A distinc- tion may be made between ferrous and ferric iron by using potas- sium ferricyanide instead of potassium thiocyanate. In this case a blue color is produced by ferrous iron and no visible color re- sults from the reaction of the small amount of ferric iron usually present. Determination of Total Iron. Prepare the following reagents: 1. Standard Iron Solution. Calculate the weight of ferrous am- monium sulphate required for 1000 cc of solution, 1 cc of which shall contain 0.0001 gm of iron. Dissolve this quantity of the salt in about 50 cc of distilled water and add 20 cc of dilute sulphuric acid. Warm 406 QUANTITATIVE ANALYSIS slightly and add potassium permanganate solution until the iron is completely oxidized, using the smallest possible excess. Dilute to 1000 cc. 2. Potassium Thiocyanate. Dissolve 20 gm of the salt in 1000 cc of distilled water. 3. Dilute Hydrochloric Acid. Dilute the concentrated acid with an equal volume of distilled water. The acid must be free from nitric acid. 4. Potassium Permanganate. Make 500 cc of a solution approxi- mately fifth-normal. Evaporate 100 cc of the sample to dryness, or use the residue left after the determination of solids. With silt-bearing waters the quantity of iron is sometimes so great that it is necessary to use as little as 10 cc of the sample. With such waters evaporation should be made in the presence of 5 to 10 cc of concentrated hydrochloric acid to effect complete solution of the iron. If the sample of water contains much organic matter, destroy this by ignition, taking care not to prolong the ignition so as to render the iron too difficultly soluble. Cool the dish and add 5 cc of dilute hydrochloric acid to moisten the whole of the inner surface of the dish. Place the dish on the steam bath for two or three minutes and again moisten the whole inner surface by allowing the hot acid to flow over it. Add 5 to 10 cc of distilled water to rinse down the sides of the dish, and again place on the steam bath for about three minutes. The hot acid solution is washed from the dish with distilled water into a 100-cc Nessler tube (see page 417). Filter the sample if necessary, carefully washing the filter paper with hot water. Add a drop or two of potassium permanganate solution to oxidize the iron to the ferric condi- tion. The color of the permanganate should persist for at least 5 minutes; if not, add more permanganate solution, a drop at a time. To the cooled solution 10 cc of potassium thiocyanate solution is added, and the volume made up to 100 cc and well mixed. Immediately compare the resulting color with that in a series of standards prepared side by side with the sample in 100-cc Nessler tubes in which amounts of standard iron solution ranging from 0.05 cc to 4 cc are first diluted with water to about 50 cc. 5 cc of dilute hydro- chloric acid and a drop or two of potassium permanganate are added to each tube of standard solution and all are diluted to 100 cc. The number of standards needed is governed by the quantity of iron likely to be present in the sample examined. Potassium thiocyanate is added to each of the standard solutions at the same time that this reagent is added to the samples of water under WATER 407 examination. Comparison of the sample -with the standards, which are made up to 100 cc after adding the thiocyanate and mixing, should be made immediately. Potability. The examination of water to determine its suit- ability for drinking (potability) is practically always directed toward the question as to whether pollution by sewage has occurred. This examination is quite different in principle from any of the processes already studied, in that the substances actually determined are nearly or quite harmless and are sig- nificant only as they point to the probable presence of patho- genic bacteria. The chemical examination goes no farther than the determination of certain chemical compounds which always accompany sewage and which therefore indicate a danger in the use of the water for drinking because disease-producing micro- organisms also generally accompany sewage. Since this is the case it might be supposed that the examina- tion might be more properly made by the bacteriologist, who determines directly the presence or absence of bacteria in ab- normal numbers and who also makes a direct test for B. coli communis, an organism that practically always accompanies faecal discharges and is therefore found in all water polluted by waste matter from human organisms. If the results of the bacteriological examination were unfailing this examination would probably suffice for all cases. It should be noted, however, that the bacteriologist is also striving for indirect rather than direct results. It is not practicable to make a direct examina- tion of the species of every organism found in order to test for the presence of actual pathogenic forms and reliance is generally placed upon the two factors noetd above, i.e., the concentration of bacteria (number per cubic centimeter) and the presence or absence of B. coli communis. If, for some reason, conditions were temporarily unfavorable to the growth of bacteria at the time the sample was taken, the number of organisms might be so reduced as to cause no suspicion of the real condition of the water. It is conceivable that 'the chance entrance of antiseptic substances into sewers or streams or the action of sunlight and air should bring about this result and B. coli communis might also be entirely absent. This might, for instance, happen as a result of mixing with factorywastes such as those from the manu- 408 QUANTITATIVE ANALYSIS facture of paper and textiles and from plating, bleaching and other chemical industries. In such a case the water would be passed by the bacteriologist when it should be condemned. On the other hand such influences as those mentioned would not eliminate the chemical products of putrescent sewage and the chemical examination would be likely to show pollution. From this examination the water would be condemned. The conclu- sion is that neither method of examination is infallible and both should be used wherever much importance attaches to the results. If one method must be omitted it is preferable that this should be the bacteriological method, provided that sufficient data are at hand for the proper interpretation of the results of the chemical examination. Interpretation. In order properly to interpret thq results of the analysis it is necessary to know the normal condition of the particular class of water under examination because all of the substances occur normally in practically all waters and their proportions vary with the source of the water. For example, chlorides occur in all ground and surface waters and the amount is governed largely by soil conditions. Near the sea shore chlorides occur in ground waters in large quantities. Similar conditions obtain for nitrogen in all of its forms and for organic matter and total solids. The necessary data for the interpreta- tion of a single analysis cannot well be collected by individuals without great expenditure of time and labor. They are gener- ally obtained as a result of organized efforts of state and city boards of health and of scientific societies. The details of manipulation in the analytical work later described, as well as the directions for taking samples and making the physical examination, are essentially those recommended by the Joint Committee on Standard Methods of Water Analysis of the American Public Health Association, American Chemical Society and Association of Official Agricultural Chemists. The report of this Committee forms a most valuable contribution to the scientific phases of the subject, not only in the matter of unification of practice but also in the guidance that it affords in a selection of the best methods now available. The third edi- tion of the report is printed as a special volume to be obtained from The American Public Health Association. WATER 409 For the directions for determinations other than those here described reference should be made to the complete report. Collection of Samples: Quantity. The minimum quantity neces- sary for making the ordinary physical, chemical and microscopical analyses of water or sewage is two liters; for the bacteriological ex- amination, one hundred cubic centimeters. In spe'cial cases larger quantities may be required. Bottles. The bottles for the collection of samples must be of halt, clear, white glass, and they must have glass stoppers. Earthen jugis or metal containers must not be used. Sample bottles must be carefully cleansed each time before using. This may be done by treating with sulphuric acid and potassium dichromate, or with a basic solution of potassium permanganate, and afterward with a mixture of oxalic and sulphuric acids, then thoroughly rinsing with water and draining. When clean, the stoppers and necks of the bottles are protected from dirt by tying cloth or thick paper over them. For shipment they are packed in cases with a separate compartment for each bottle. Wooden boxes may be lined with indented fiber paper, felt or some similar substance, or provided with spring corner strips, to prevent breakage. Lined wicker baskets also may be used. Bottles for bacterial samples, besides being washed, must be sterilized with dry heat for one hour at 160 or in an autoclave at 115 for fifteen minutes. For transportation they may be wrapped in sterilized cloth or paper, or the necks may be covered with tin foil and the bottles put in tin boxes. When bacterial samples must of necessity stand for twelve hours before plating, bottles holding more than four ounces must be used. The bottles used for chemical samples may be sterilized and the samples so collected used for the bacteriological analysis. When bacterial samples are not plated at the time of collection they are kept on ice at a temperature not higher than 15 and preferably as low as 10. Time Interval between Collection and Analysis. Generally speaking, the shorter the time elapsing between the collection and the analysis of a sample, the more reliable will be the analytical results. Under many conditions, analyses made in the field are to be commended, as data so obtained are frequently preferable to those made in a distant laboratory after the composition of the water has changed en route. The allowable time that may elapse between the collection of a sample and the beginning of its analysis cannot be stated definitely, as it depends upon the character of the sample and upon other conditions, but the following may be considered as fairly reasonable maximum limits under ordinary conditions: 410 QUANTITATIVE ANALYSIS Physical and Chemical Analysis. Ground waters 72 hours Fairly pure surface waters 48 hours Polluted surface waters 12 hours Sewage effluents 6 hours Raw sewages 6 hours Microscopical Examination. Ground waters 72 hours Fairly pure surface waters 24 hours Waters containing fragile organisms . . Immediate examination Bacteriological Examination. Samples kept at less than 10 24 hours If sterilized by the addition of chloroform, formaldehyde, mercuric chloride, or some other disinfectant, samples for chemical and micro- scopical examination may be allowed to stand for longer periods than those indicated, but as this is a matter which must vary according to local circumstances, no definite procedure is recommended. If unsterilized samples of sewage, sewage effluents, and highly polluted surface waters are not analyzed on the day of their collection, caution must be used in regard to the organic contents, which frequently change materially upon standing. The gaseous contents of samples, especially dissolved oxygen and carbonic acid, should be obtained immediately, in accordance with the directions given in connection with each determination. Representative Samples. Care should be taken to secure a sample which is truly representative of the liquid to be analyzed. In the case of sewages this is especially important, in view of the marked variations in composition which occur from hour to hour. Frequently satisfactory samples can be obtained only by mixing together several portions col- lected at different times or in different places the details as to collecting and mixing depending upon local conditions. Physical Examination: Temperature. The temperature of the sample should be taken at the time of collection, and should be expressed in Centigrade degrees, to the nearest degree or closer if for any reason more exact data are required. For obtaining the temperature of water at various depths below the surface the thermophone is recommended. Turbidity. The turbidity of water is due to suspended matter, such as clay, silt, finely divided organic matter, microscopic organisms, etc. The increasing use of filters for the purification of water and sewage has made this determination one of great importance. WATER 411 The standard of turbidity is that adopted by the United States Geological Survey, namely, a water which contains 100 parts of silica per million in such a state of fineness that a bright platinum wire one millimeter in diameter can just be seen when the center of the wire is 100 millimeters below the surface of the water and the eye of the ob- server is 1.2 meters above the wire, the observation being made in the middle of the day, in the open air, but not in sunlight, and in a vessel so large that the sides do not shut out the light so as to influence the results. The turbidity of such water is taken as 100. Preparation of Silica Standard. Dry Pear's "precipitated fuller's earth" and sift it through a 200-mesh sieve. 1 gm of this preparation in one liter of distilled water makes a stock suspension which contains 1000 parts per million of silica and which should have a turbidity of 1000. Test this suspension, after diluting a portion of it with nine times its volume of distilled water, with a wire to ascertain if the silica has the necessary degree of fineness and if the sus- pension has the necessary degree of turbidity. If not, correct by adding more silica or more water as the case demands. Standards for comparison are prepared from this stock suspension by dilution with distilled water. For turbidity readings below 20, standards of 0, 5, 10, 15 and 20 are kept in gallon bottles made of clear white glass; for readings above 20, standards of 20, 30, 40, 50, 60, 70, 80, 90 and 100 are kept in 100-cc Nessler tubes approximately 20 mm in diameter. Comparison of the water under examination with the standards is made by viewing them sidewise toward the light, looking at some object and noting the distinctness with which the margins of the object can be seen. The standards must be kept stoppered and both sample and standards thoroughly shaken before making the comparison. In order to prevent any bacterial or algal growths from appearing in the standards, a small amount of mercuric chloride may be added to them. Platinum Wire Method. This method requires a rod with a platinum wire having a diameter of 1 mm inserted in it about 25 mm from the end of the rod, and projecting from it at least 25 mm at a right angle. Near the end of the rod, at a distance of 1.2 meters from the platinum wire, a wire ring is placed directly above the wire through which, with his eye directly above the ring, the observer shall look when making the examination. The rod is graduated as follows : The graduation mark of 100 is placed on the rod at a distance of 100 mm from the center of the wire. Other graduations are made ac- cording to the table on page 412. These graduations are the ones used 412 QUANTITATIVE ANALYSIS to construct what is known as the U. S. Geological Survey Turbidity Rod of 1902. Procedure. Push the rod down into the water vertically as far as the wire can be seen and then read the level of the surface of the water on the graduated scale. This will indicate the turbidity. The following precautions should be taken to insure correct results: Observations should be made in the open air, preferably in the middle of the day and not in direct sunlight. The wire must be kept bright and clean. Waters which have a turbidity above 500 are diluted with clear water before the observations are made, but in case this is done the degree of dilution used should be stated and should form a part of the report. Turbidity, parts per million Vanishing depth of wire, mm Turbidity, parts per million Vanishing depth of wire, mm 7 1095 * 70 - 138 8 971 75 130 9 873 80 122 10 794 85 116 11 729 90 110 12 674 95 105 13 627 100 100 14 587 110 93 15 551 120 86 16 520 130 81 17 493 140 76 18 468 150 72 19 446 160 68.7 20 426 180 62.4 22 391 200 57.4 24 361 250 49.1 26 336 300 43.2 28 314 350 38.8 30 296 400 35.4 35 257 500 30.9 40 228 600 27.7 45 205 800 23.4 50 187 1000 20.9 55 171 1500 17.1 60 158 2000 14.8 65 147 3000 12.1 WATER 413 The wire method is used for testing the degree of fineness of the standard silica, and this degree of fineness shall be such that when added to distilled water in an amount equal to 100 parts per million, the wire observed under standard conditions can be just seen at a depth of 100 mm below the surface of the water. Expression of Results. The results of turbidity observations are expressed in whole numbers which correspond to parts per million of silica, and recorded as follows: Turbidity between 1 and 50, recorded to nearest unit. Turbidity between 51 and 100, recorded to nearest 5 Turbidity between 101 and 500, recorded to nearest 10 Turbidity between 501 and 1000, recorded to nearest 50 Turbidity between 1001 and above, recorded to nearest 100 Coefficient of Fineness.- The number obtained by dividing the weight of suspended matter in the sample (in parts per million) by the turbidity is called the coefficient of fineness. If greater than unity it indicates that the matter in suspension in the water is coarser than the standard; if less than unity, that it is finer than the standard. Color. The color of water may form an important indication of pollution. There is little of value to be obtained from a quantitative measurement of color although the determination is discussed at length in the report of the Committee on Standard Methods. Odor. The observation of the odor of cold and hot samples of surface waters is very important, as the odors are usually con- nected with some organic growths or with sewage contamination or both. The odor of ground waters is often caused by the earthy con- stituents of the water bearing strata. The odor of a contami- nated well water is often decisive evidence of its pollution. A study of the organisms of water is an invaluable adjunct to the physical and chemical examination of water. Certain organ- isms can be distinguished by their odor, as, for example, the " fishy" odor of Uroglena the " aromatic" or "rose geranium" odor of Asterionella and the " pig-pen" odor of Anabcena. Determination. Observe and record the odor, both at room tempera- ture and at just below the boiling-point, as follows: Cold Odor. Shake the sample violently in one of the collecting bottles, when it is about half or two-thirds full and when the sample is at room 414 QUANTITATIVE ANALYSIS temperature (about 20). Remove the stopper and test the odor at the mouth of the bottle. Hot Odor. Into a 500 cc Erlenmeyer flask pour about 150 cc of the sample. Cover the flask with a well-fitting watch glass, place on a hot plate and bring the water to just below boiling. Remove the flask from the plate and allow it to cool for not more than five minutes. Then shake with a rotary movement, slip the watch glass to one side and test the odor. . Expression of Results. Express the quality of the odor by some such descriptive epithet as the following, which for purposes of record may be abbreviated : v vegetable a aromatic g grassy f fishy e earthy c free chlorine m moldy M musty d disagreeable P peaty s sweetish S hydrogen sulphide Express the intensity of the odor by a numeral prefixed to the term expressing quality, which may be defined as follows: Numerical value Term Approximate definition 1 2 None Very faint Faint No odor perceptible. An odor that would not be ordinarily detected by the average consumer, but that could be detected in the laboratory by an experienced observer. An odor that the consumer might detect if his attention 3 4 5 Distinct Decided Very strong. . . were called to it, but that would not otherwise attract attention. An odor that would be readily detected and that might cause the water to be regarded with disfavor. An odor that would force itself upon the attention and that might make the water unpalatable. An odor of such intensity that the water would be absolutely unfit to drink. (A term to be used only in extreme cases.) Chemical Examination. The following determinations may be made: Total solids, chlorine of chlorides, albumenoid nitrogen, total organic nitrogen, nitrogen of ammonia or ammonium salts, nitrogen of nitrites, nitrogen of nitrates, total organic matter, dissolved oxygen and poisonous metals. Besides the chemical analysis certain purely physical tests may be made such as temperature, color, odor and turbidity. These determinations have just been described. WATER 415 Total Solids. This is taken as the residue obtained when a measured volume of water is evaporated. The general character of the solids may be sometimes noted, also the amount of loss suffered by igniting in air and the odor and amount of charring afford an indication as to the quantity and character of solid organic matter. Determination. Ignite and weigh a platinum dish, then evaporate in it 100 cc of water, using the steam bath. The dish need not be large enough to hold the entire 100 cc at one time. A small dish is better. Heat the residue at about 103 for one-half hour. Cool in the desiccator and weigh. Report the increase in weight as milligrams per liter of total solids. Heat the dish at low redness until all organic matter is burned. The change in weight is loss on ignition. Suspended Matter. Filter a portion of the sample through a very fine, close filter paper or a well-formed asbestos mat in a Gooch crucible, rejecting the first 15 to 25 cc. Determine the total dissolved solids in the filtrate as already directed. This, subtracted from the solids of the original sample, gives the total suspended matter. Chlorine of Chlorides. Chlorine occurs to some extent in all natural waters. It is found to a much larger extent in sewage where it enters chiefly as sodium chloride of urine and faeces, and of household wastes. Sewage polluted streams or wells, therefore, always carry abnormally large quantities of chlorine. Also the chlorine content may be increased by ocean vapors carried inland by natural deposits or by factory wastes. If the normal chlorine content exceeds 20 mg per liter, the determina- tion will have little significance from the sanitary standpoint. Determination. Use the method described on page 400. Report milligrams per liter of chlorine. Nitrogen in Various Forms. Human faeces contains large quantities of nitrogen while urine has a normal content of about 0.85 percent of nitrogen. The entrance of sewage therefore im- parts abnormal concentrations of nitrogen to water. This nitrogen is at first practically all in the form of organic com- pounds and of urea. Part of the organic nitrogen is readily converted into ammonia by oxidizing with potassium per- 416 QUANTITATIVE ANALYSIS manganate in basic solution. This part is known as "albume- noid nitrogen" because it is contained in albumenous bodies. The action of certain forms of bacteria (chiefly anaerobic) causes the putrefaction of organic matter and this cleavage of complex compounds results in the formation of ammonia from the nitrogen. Part or all of this ammonia may combine with acids to form ammonium salts. All such nitrogen is known as " nitrogen of free ammonia," whether this be of really free ammonia or of ammonium salts. Where sewage or water polluted by it is exposed to air and sunlight the simpler organic compounds produced by putrefaction are subjected to oxidation, this being promoted by other forms of bacteria (aerobic). Ultimately the organic compounds are completely oxidized. The two processes, putrefaction and oxidation, are made the basis of the septic process of water puri- fication. In the operations of water analysis the changes in the forms of nitrogen are most important. "Free" ammonia is oxidized to nitrous acid which usually remains combined as nitrites of metals or of ammonium. Further oxidation produces nitric acid or nitrates, the final stage in the series. The analytical estimation of the nitrogen in different forms in water offers a valuable indication, not only as to the probability of pollution but also concerning the present condition. The presence of abnormal quantities of "albumenoid nitrogen" indicates the presence of unchanged sewage and the probable presence of dangerous micro-organisms in their most virulent condition. "Free ammonia" in considerable quantities shows that the raw sewage has become fermented and that it must have been largely diluted in the time that has elapsed since the entrance of sewage. Nitrites are very readily oxidized and will not be found in more than traces unless free ammonia is also present. Abnormal quantities of nitrates, unless these are of inorganic origin, are the result of complete oxidation of organic matter and this must have required time and continued action of air and sunlight. If all forms of nitrogen are found in abnormal quantities continuous pollution is occurring. All of these figures are highly significant in view of the fact that the same influences that promote the decomposition and oxidation of nitrogenous organic matter also combine for the partial or complete steriliza- WATER 417 tion of the water. It is not, by any means, to be concluded that water which has been polluted by sewage but in which the latter has become completely oxidized is necessarily safe for drinking. Indeed if the analysis shows pollution, even at a remote source or time, the water should be condemned as dangerous. The degree of danger is still indicated and the indication will prove of value. As compared with most other substances ordinarily considered in quantitative analysis the different forms of nitrogen occur in extremely slight concentrations. Unusually delicate reactions must be used in order to cause the figures to have any value. Ordinary gravimetric or volumetric processes are rarely used in this connection but very sensitive color reactions are made the basis for the comparison of the water with color standards. Free Ammonia is made evident by the brown color produced when a solution of potassium mercuriodide, K 2 HgI 4 , is added. This solution is known as "Nessler's reagent," from the name of the discoverer of the reaction. 1 The compound that is produced when ammonia is added is a complex substance, thought to have the composition Hg 2 NI. It is an intensely colored brown substance of small solubility and gives a visible color in water containing one part of nitrogen as ammonia in ten million parts of water. The process of determining free ammonia is one of comparing the color produced by adding Nessler's reagent to water with that produced by the reagent with a standard solution of ammo- nium chloride. The comparison is made in tubes of colorless glass, the two that are being compared having the same cross section so that the same length of column is placed in the line of vision. The color is observed by looking vertically downward through the tubes at a white surface placed at an angle in front of a window so as to reflect the light upward. If Nessler's reagent is added directly to water containing organic matter, iron or aluminium, a precipitate is produced and an accurate color comparison is impossible. One of two pre- liminary treatments may be used. The free ammonia may be separated by distillation and the distillate then " Nesslenzed " 1 Z. anal. Chem., 7, 415 (1868). 27 418 QUANTITATIVE ANALYSIS or reagents may be added to the water sample to precipitate inter- fering substances and " direct Nesslerization" may be employed. Distillation is preferable, but where apparatus or time is limited direct Nesslerization may be useful. Direct Nesslerization. For precipitating organic matter use is made of the power of flocculating colloids for adsorbing this material, which is itself chiefly colloidal. Cupric sulphate is added to the water which is then made basic by the addition of potassium hydroxide. The precipitating cupric hydroxide so clarifies the water that direct Nesslerization is practicable. Instead of adding cupric sulphate a solution of magnesium chlo- ride may be substituted. Colloidal magnesium hydroxide accom- plishes the same result as does cupric hydroxide. If the water already contains much magnesium it is unnecessary to add even magnesium chloride. Boiling the water with potassium hydrox- ide will cause the precipitation of magnesium hydroxide. If hydrogen sulphide is present it will cause the precipitation of mercuric sulphide when Nessler's reagent is added. This inter- ference is prevented by the addition of lead acetate before the removal of colloids by cupric hydroxide. The chief objection to direct Nesslerization is the tendency of the precipitating colloids to adsorb small amounts of ammonia. However, the process is recommended for raw sewages, sewage effluents and highly polluted surface waters. Whether direct Nesslerization or distillation processes are used for free ammonia either an accurately prepared standard solution of an ammonium salt or a standard color solution of a permanent nature is required. This must have a very slight concentration and it is best made by successive dilutions of a more concentrated solution. The solvent used is water that has been shown to be free from ammonia by a test with Nessler's reagent. The labora- tory supply of distilled water is often free from ammonia. If it is not it may be purified by the addition of basic potassium permanganate solution and distilling. After the distillate no longer gives a test for ammonia it is collected and kept in well- stoppered bottles. For a permanent color standard mixtures of potassium chlor- platinate and cobalt chloride are recommended. By properly varying the relative concentrations of the two salts, solutions are WATER 419 obtained in which the color accurately corresponds with that of Nesslerized ammonia solutions of known concentrations. These solutions are to be preferred to the standard ammonium chloride solutions in laboratories where many determinations are to be made, because of their permanency. Albumenoid Nitrogen cannot be determined by direct Nessleri- zation. It is determined after the distillation of free ammonia by adding to the residue a basic solution of potassium permanganate and distilling. The organic matter is oxidized and remaining nitrogen is distilled and Nesslerized. As the ratio of nitrogenous organic matter to the ammonia obtained by distillation is decidedly variable in sewages and other substances containing much nitrogenous organic matter, albumenoid nitrogen results on such materials are less accurate than total organic nitrogen, obtained by the Kjeldahl process. Therefore in sewage work, including analysis of influents and effluents of purification plants and the water of highly polluted streams, it is recommended by the joint committee that deter- minations of total organic nitrogen be substituted for determina- tions of albumenoid nitrogen. For ground waters and surface waters containing but little pollution, the albumenoid nitrogen is approximately one-half the organic nitrogen; accordingly the continuance of albumenoid nitrogen determinations for this class of work is approved. All determinations of nitrogen must be made in a laboratory in which the air is free from ammonia. Determination. Prepare the following reagents: 1. Nessler's Reagent. Dissolve 25 gm of potassium iodide in the minimum quantity of cold water. Add a saturated solution of mercuric chloride until a slight but permanent precipitate persists. Add 200 cc of 50 percent solution of potassium hydroxide made by dissolving the potassium hydroxide and allowing it to clarify by sedimentation before using. Dilute to 500 cc, allow to settle and decant. This solution should give the required color with ammonia within five minutes after addition, and should not precipitate with small amounts of ammonia within two hours. 2. Basic Potassium Permanganate. Pour 600 cc of distilled water into a porcelain dish holding 1500 cc, boil 10 minutes and turn off the gas. Add 8 gm of potassium permanganate and stir until dissolved. 420 QUANTITATIVE ANALYSIS Add 400 cc of 50 percent clarified solution of potassium or sodium hydroxide and enough distilled water to fill the dish. Boil down to 1000 cc. Test this solution for albumenoid ammonia by making a blank determination. Correction should be made accordingly. 3. Ammonia-free Water. Test the laboratory supply of distilled water by rinsing a clean Nessler tube several times, filling to the mark and add- ing 2 cc of Nessler 'B reagent. Cover and allow to stand for 5 minutes. If the color produced at the end of this time is more intense than that of the diluted Nessler's reagent at first, the water must be purified. In this case add 10 cc of basic potassium permanganate solution to each 1000 cc of distilled water and distill, using a tin or aluminium condenser if one is available. After the distillate ceases to give a test for ammonia it is collected in a clean, glass-stoppered bottle, which is first rinsed with the distillate and the rinsings tested for ammonia. 4. Standard Solution for Color Comparisons. Use either {a) or (b). a. Ammonium Chloride Solution. Dissolve 3.82 gm of ammonium chloride in ammonia-free water and dilute to 1000 cc. Dilute 10 cc of this to 1000 cc with ammonia-free water. 1 cc contains 0.00001 gm of nitrogen. b. Platinum Solution and Cobalt Solution. Weigh 2 gm of potassium chlorplatinate, dissolve in a small amount of distilled water, add 100 cc of concentrated hydrochloric acid and make up to 1000 cc. Weigh 12 gm of cobalt chloride and dissolve in distilled water; add 100 cc of concentrated hydrochloric acid and make up to 1000 cc. Nitrogen of Free Ammonia. A 750-cc Kjeldahl digestion flask, connected with a tin or aluminium condenser in such a way that the distillate may be conveniently delivered from the condenser tube directly into the Nessler tubes, is freed from ammonia by boiling dis- tilled water in it, until the distillate shows no further traces of free ammonia. When this has been done, empty the distilling flask and measure into it 500 cc of the sample, or a smaller portion diluted to 500 cc with ammonia-free water. Apply heat so that the distillation will be at the rate of not more than 10 cc nor less than 6 cc per minute. Collect four Nessler tubes of the distillate, 50 cc to each portion; these contain the free ammonia to be measured as described below. Use only Nessler tubes which do not show a variation of more than 6 mm (0.25 inch) in the distance which the graduation mark (50 cc) is above the bottom. The tubes should be of clear white glass, with pol- ished bottoms. The residue from the distillation is immediately used for the determination of albuminoid nitrogen as described below. The measurement may be made either by (1) comparison with Ness- lerized solutions containing known quantities of nitrogen as ammonium WATER 421 chloride, or (2) comparison of the Nesslerized distillates with permanent standards. Comparison with Ammonia Standards. Prepare a series of 16 Nessler tubes which contain the following numbers of cubic centimeters of the standard ammonium chloride solution, diluted to 50 cc with ammonia- free water, namely: 0.0, 0.1, 0.3, 0.5, 0.7, 1.0, 1.4, 1.7, 2.0, 2.5, 3.0, 3.5, 4.0, 4.5, 5.0, and 6.0. These will contain 0.00001 gm of nitrogen for each cc of the standard solution used. Nesslerize the standards and also the distillates by adding approxi- mately 2 cc of Nessler's reagent to each tube. Do not stir the contents of the tubes. Have the temperature of the tubes practically the same as that of the standards, otherwise the colors will not be directly comparable. Compare the color produced in these tubes with that in the standards by looking vertically downward through them at a white surface placed at an angle in front of a window so as to reflect the light upward. Allow the tubes to stand for at least 10 minutes after Nesslerizing before mak- ing the comparison. In case the color obtained by Nesslerizing the distillates is greater than that of the darkest tube of the standards, mix the contents of the tube thoroughly and pour out half of the liquid, making up the remainder to the original volume with ammonia-free water, then make the color comparison and multiply the result by two. If, after pouring out half of the liquid, the color is still too dark, repeat this process of division until a reading can be made. In case the color of the distillates is too high, this process may be shortened by mixing together all of the distillates from one sample before making the comparison, subsequently taking an aliquot portion for comparing with the standards. After the readings have been made and recorded, add together the results obtained by Nesslerizing each portion of the entire distillate from each sample. Calculate milligrams per liter of nitrogen as free ammonia in the sample. Comparison with Permanent .Standards. Prepare standards by put- ting varying amounts of potassium chlorplatinate and cobalt chloride solutions (page 420) in Nessler tubes, filling up to the mark with distilled water as follows: 422 QUANTITATIVE ANALYSIS Equivalent volume of standard ammonium chloride, cc Platinum solution, cc Cobalt solution, cc 0.0 1.2 0.0 0.1 1.8 0.0 0.2 2.8 0.0 0.5 4.7 0.1 0.7 5.9 0.2 1.0 7.7 0.5 1.4 9.9 1.1 1.7 11.4 1.7 2.0 12.7 2.2 2.5 15.0 3.3 3.0 17.3 4.5 3.5 19.0 4 5.7 4.0- 19.7 7.1 4.5 19.9 8.7 5.0 20.0 10.4 6 20.0 15.0 7.0 20.0 22.0 It is necessary to use tubes which have the 50 cc mark not less than 20 nor more than 22 cm above the bottom. These standards may be kept for several months if protected from dust. The method of cal- culating results is practically the same as with the ammonia standards. Albumenoid Nitrogen. Interrupt the distillation (made as already described) after the collection of the distillate for free ammonia. Add 50 cc or more of basic potassium permanganate solution and conduct this distillation until at least four portions of 50 cc each and preferably five portions of the distillate have been collected in separate tubes. Have enough permanganate solution present to insure the maximum oxidation of the organic matter. These distillates contain the albu- menoid nitrogen as ammonia, measurement of which will be made as described in connection with nitrogen as free ammonia. Nitrogen of Free Ammonia by Direct Nesslerization. Prepare the following solutions. 1. A 10 percent solution of cupric sulphate. 2. A 50 percent solution of sodium hydroxide. If hydrogen sulphide is present in the water prepare also : 3. A 10 percent solution of lead acetate. 50 cc of the sample to be tested, mixed, if necessary, with an equal volume of ammonia-free water, is placed in a Nessler tube and a few WATER 423 drops of cupric sulphate solution added. After thorough mixing 1 cc of the sodium hydroxide solution is added and the solution again thor- oughly mixed. The tube is then allowed to stand for a few minutes, when a heavy precipitate should fall to the bottom leaving a colorless supernatant liquid. Nesslerize an aliquot portion of this liquid. If hydrogen sulphide is present add a few drops of lead acetate solu- tion before the addition of potassium hydroxide. Organic Nitrogen. This determination may be conveniently com- bined with that of ammonia nitrogen, by the procedure for water and the first procedure for sewage, described below. Procedure for Water. Distill the ammonia from 500 cc of the sample exactly as already directed for the determination of nitrogen of free ammonia, page 420. This will usually involve the loss of 200 cc of the sample. Cool somewhat and rinse into a Kjeldahl digestion flask, unless such a flask was used for the first distillation. Add 5 cc of concentrated sulphuric acid which is free from ammonium sulphate, also a small piece of ignited pumice, dropped in while hot. Mix and digest over a free flame, using a suitable apparatus for removing the sulphuric acid fumes (see Figure 116, page 514). The digestion should be continued until copious fumes are evolved and the liquid is finally colorless or very pale yellow. Remove from the flame and add to the hot solution potassium permanganate crystals until a heavy green precipitate per- sists. Cool and dilute to about 300 cc with ammonia-free water. Make basic by adding 10 percent sodium hydroxide which has been made free from ammonia by boiling for a short time. Distill the ammonia and Nesslerize as directed for the determination of nitrogen of free ammonia. First Procedure for Sewage. Use 100 cc or less of the sample, according to the amount of ammonia expected. Dilute to 500 cc with ammonia- free water in the distilling flask and distill the free ammonia, Nesslerizing the distillate. Cool somewhat and add 5 cc of nitrogen-free concen- trated sulphuric acid and 1 cc of 10 percent copper sulphate solution. Digest the solution for a half hour after it has become colorless or pale yellow. Add 0.5 gm of potassium permanganate crystals to the hot acid solution, cool, transfer to a 500 cc volumetric flask, dilute to the mark and mix. By means of a pipette transfer 10 cc of this solution to a Kjeldahl distilling flask and dilute to about 300 cc with ammonia- free water. Make basic with 10 percent sodium hydroxide solution, distill and Nesslerize. (With some samples direct Nesslerization may be employed.) Second Procedure for Sewage. Omit the separation of ammonia nitrogen and determine this and organic nitrogen together. Determine the ammonia nitrogen on a separate sample by direct Nesslerization as 424 QUANTITATIVE ANALYSIS directed on page 422 . The difference between these results is the organic nitrogen. Nitrogen as Nitrites. It has already been explained that nitrites will not normally occur in more than traces in water because of the readiness with which they oxidize. In order to give this determination any significance it is necessary to use a very delicate test. Use is made of the ready action of nitrous acid with aromatic amines, forming diazo compounds, and of the latter with naphthylamine, forming azo dyes of intense color- ing power. When water containing nitrites is acidified and sulphanilic acid (p-amidobenzenesulphonic acid) is added there is formed the anhydride of p-diazobenzenesulphonic acid, thus: HN0 2 +C 6 H 4 < ->C 6 H 4 < +2H 5 0. \S0 3 H X S0 3 / If to this solution a-amidonaphthalene is added, an azo dye, a-amidonaphthaleneazobenzene-p-sulphonic acid, is produced. /N = N x ,N = NC 10 H 6 NH 2 C 6 H 4 < > +C 10 H 7 NH 2 ->C 6 H 4 < X S0 3 / \S0 3 H This dye possesses a very intense red color and one part of nitro- gen as nitrite can be detected in 1,000,000,000 parts of water. The amino compounds entering into these reactions are not easily soluble and their soluble salts are used. The hydrochlo- rides may be employed but the reactions appear to proceed more rapidly if the acetates are used. Determination. Prepare the following reagents: 1. Sulphanilic Add. Dissolve 4 gm of the purest sulphanilic acid in 500 cc of 5-normal acetic acid (sp. gr. 1.041) or in a mixture of 950 cc of nitrite-free water and 50 cc of concentrated hydrochloric acid. This is a practically saturated solution. 2. a-amidonaphthalene Acetate or Hydrochloride. Dissolve 2.5 gm of solid a -naphthylamine in 500 cc of 5-normal acetic acid or in 1000 cc of nitrite-free water containing 8 cc of concentrated hydrochloric acid. Filter the solution through washed absorbent cotton or an alundum crucible. 3. Sodium Nitrite, Stock Solution. Dissolve 1.1 gm of silver nitrite in nitrite-free water. Precipitate the silver with sodium chloride solution and dilute the whole to 1000 cc. WATER 425 4. Standard Sodium Nitrite Solution. Dilute 100 cc of solution (3) to 1000 cc and dilute 10 cc of the resulting solution to 1000 cc with sterilized nitrite-free water. Add 1 cc of chloroform and preserve in a sterilized bottle. Calculate and record the weight of nitrogen in 1 cc of this last solution. 5. Fuchsine Solution. 0.1 gm per liter of water. Measure 50 cc or 100 cc of the water to be tested into a Nessler tube. These Nessler tubes should be of clear, white glass, with the graduation mark not varying more than 6 mm in its distance above the bottom. At the same time make a set of standards by diluting various volumes of the standard nitrite solution in Nessler tubes to 50 or 100 cc with nitrite-free water, for example, 0, 1, 3, 5, 7, 10, 14, 17, 20 and 25 cc. Add 2 cc of reagents (1) and (2) to each 100 cc of the sample and to each standard. Mix and allow to stand 10 minutes. Compare the samples with the standards. Do not allow the samples to stand over one-half hour before being compared. Make a blank determination in all cases to correct for the presence of nitrites in the air, the water and the reagents. Dilute all samples which develop more color than the 30 cc standard before comparing. Mixing is important. Calculate milligrams per liter of nitrite nitrogen. Permanent standards may be prepared by matching the nitrite standards as above made against dilutions of the fuchsine solution. Fuchsine standards have been found to be sufficiently accurate for waters high in nitrite and for sewage. Such standards should be checked once a month and should be kept out of bright sunlight to avoid bleaching. Nitrogen as Nitrates. When a soluble aromatic sulphonic acid is mixed with nitrates and sulphuric acid the nitric acid so liberated acts upon the aromatic compound and produces nitro- derivatives which are faintly yellow in most cases. If a base is now added the sulphonate is formed and this is much more in- tensely colored. These reactions are applied to the determina- tion of nitrates in water by what was originally known as Spren- gePs method. 1 The method was further modified by Grandval and Lajoux. 2 A measured volume of water ife evaporated to dryness, sodium carbonate having been first added if the water contains free acid. The dry residue is treated with a small amount of a phenolsul- 1 Pogg. Ann., 121, 188 (1863). 2 Compt. rend., 62, 101 (1885). 426 QUANTITATIVE ANALYSIS phonic acid, the mono-nitro derivative being formed. The reagent is made by heating phenol with sulphuric acid in the pro- portions indicated below. These interact with the formation phenol-o-p-disulphonic acid. The reaction of this acid with ni- tric acid results in the formation of o-nitrophenol-o-p-disulphonic acid: C 6 H 3 .OH.(S0 3 H) 2 +HN03->C6H 2 .OH.N0 2 .(S03H) 2 +H 2 0. Treatment with a base produces the highly colored sulpho- nate, e.g., C 6 H 2 .OK.N0 2 .(S0 3 K) 2 . Three important sources of error may render impossible a determination of nitrate by the phenolsulphonic acid method or may cause incorrect results to be obtained. These may be enumerated as follows: Interference of Organic Matter.- If the water contains any considerable amount of organic matter, as is always the case with surface streams, sewage or waters contaminated by sewage, the addition of sulphuric acid will cause a charring of the organic matter and the color comparison cannot be made accurately because of the resulting brown coloration. Of course the same interference will result from any color that may have been in the sample before treatment. Organic matter may also cause the reduction of the nitrates during evaporation. This interference may be prevented by first removing the organic matter by means of colloidal aluminium hydroxide. Interference of Chlorides.- If the sample contains more than about 30 mg of chlorine as chlorides per liter there is a possibility of reactions occurring between chlorides and nitrates during evapo- ration, resulting in the reduction of nitrates. This may be avoided by the addition of silver sulphate solution to the slightly acidified sample, silver chloride being precipitated. Interference of Nitrites. Nitrites likewise cause variable results to be obtained unless certain precautions are taken. During the evaporation of water containing nitrites some of the latter will be decomposed and nitrogen lost, while some may be i Chamot and Pratt: J. Am. Chem. Soc., 31, 922 (1909); 32, 630 (1910); 33, 366 and 381 (1911). WATER 427 oxidized to nitrates. On account of the uncertain extent to which these reactions occur it is necessary either to remove the nitrites entirely or to oxidize them quantitatively to nitrates. .The latter is accomplished by treatment of the sample with hydrogen peroxide. All of these interferences may be avoided by using a method based upon the reduction of nitrates to ammonia by nascent hydrogen. The sample is made basic and concentrated by boil- ing. By this means all of the ammonia, free or combined, as well as all of the nitrogen as nitrites is removed. Nascent hydro- gen is then generated by adding aluminium to the basic solution. The resulting ammonia is later distilled and determined by Nesslerization. Determination. For the phenolsulphonic acid method the following reagents will be required: 1. Phenoldisulphonic Acid. Dissolve 25 gm of pure white phenol in 150 cc of concentrated sulphuric acid. Add 75 cc of fuming sulphuric acid containing 15 percent of "free" sulphur trioxide, stir well and heat for 2 hours at about 100. 2. Sodium or Potassium Hydroxide Solution. This solution should be made approximately 12 normal. About 5 cc will then be required to neutralize 2 cc of the phenolsulphonic acid. 3. Standard Nitrate Solution. Dissolve 0.72 gm of pure recrystal- lized potassium nitrate and dilute to 1000 cc with distilled water. Evaporate cautiously 10 cc of this solution in a dish placed on a steam bath. Moisten the residue quickly and thoroughly with 2 cc of phenol- sulphonic acid, dissolve and dilute to 1000 cc. Calculate the weight of nitrogen in 1 cc of the last solution. 4. Standard Silver Sulphate Solution. Dissolve 4.397 gm of silver sulphate, free from nitrate, in nitrate-free water and dilute to 1000 cc. If a good grade of silver sulphate has been used this solution will require no other standardization for the purpose for which it is here to be used. Otherwise it should be standardized against pure sodium chloride. 1 cc should be equivalent to 1 mg of chlorine. 5. Standard Sulphuric Acid Solution. A solution, N/50, standardized by titration against pure sodium carbonate (pages 224 and 231). The solution must be free from nitrates. 6. Aluminium Hydroxide.- This must be freshly prepared. From a solution of nitrate-free aluminium sulphate or alum precipitate hy- droxide by adding dilute ammonium hydroxide. Filter and wash several times. The water and ammonium hydroxide must be free from nitrates. 428 QUANTITATIVE ANALYSIS The aluminium hydroxide so prepared is used without drying and before it has had time to change to the crystallqidal form. Measure into an evaporating dish enough sample to furnish not more than 0.01 mg of nitrate nitrogen. 100 cc will be suitable for ordinary unpolluted waters. Add sufficient N/50 sulphuric acid to make the water nearly neutral, as determined by a separate titration with methyl orange as indicator. (See the determination of carbonates, page 398.) Now add silver sulphate solution in quantity sufficient to precipitate all but about 0.1 mg of chlorine. (This determination is described on page 400.) This treatment may be omitted if the quantity of sample used contains less than about 30 mg of chlorine. If silver sulphate has been added or if the water sample is perceptibly colored heat the mixture to boiling, add a little aluminium hydroxide, stir, filter and wash with small amounts of nitrate-free hot water. Evaporate over the steam bath and add 2 cc of phenoldisulphonic acid, rubbing with a glass rod to insure intimate contact. If the residue becomes packed or appears vitreous because of the presence of much iron, heat the dish on the steam bath for a few minutes. Dilute the mixture with distilled water and add slowly solution (2) until the maximum color is developed. Transfer to a Nessler tube, filtering if necessary. Compare the color with that of standards made by adding 2 cc of the basic solution to various amounts of the standard nitrate solution and diluting to the mark in Nessler tubes. The following amounts of standard are suggested: 0.5, 1.0, 1.5, 2.0, 4.0, 6.0, 8.0, 10.0, 15.0, 20.0 and 40.0 cc. These standards may be kept for several weeks without deterioration if the tubes are kept corked to prevent evaporation or contamination. The amount of nitrite nitrogen that will remain after evaporation is not sufficient to alter materially the results unless present in excess of 1 mg per liter. In case such quantity is found the nitrite is oxidized to nitrate by repeatedly heating with a few drops of hydrogen peroxide which is free from nitrate. Proper correction in the nitrate nitrogen figure must then be made. Calculate milligrams per liter of nitrate nitrogen. For the reduction method for nitrates prepare the following reagents : 1. Sodium Hydroxide Solution. Dissolve 50 gm of the purest ob- tainable sodium hydroxide in 250 cc of distilled water, add several strips of aluminium foil and leave over night. Evaporate to 200 cc by boiling. 2. Aluminium Foil. Use strips about 10 cm long, 0.33 mm thick and 6 mm- wide. Place 100 cc of the water in a 300 cc casserole or dish. Add 2 cc of sodium hydroxide solution and boil until the volume is 20 cc. Using WATER 429 nitrogen-free water, rinse into a test-tube about 15 cm long and 3 cm in diameter. The volume of the solution should now be about 75 cc. Add a strip of aluminium foil and close the tube with a rubber stopper through which passes a bent glass tube, the shorter end of which is flush with the lower end of the stopper, the longer end dipping beneath the surface of distilled water in a beaker. This serves as a trap to pre- vent the entrance of oxygen. Allow to stand for four hours or more. If the supernatant liquid is then clear and colorless, Nesslerize at once, otherwise rinse into the apparatus used for ammonia determinations, dilute to 250 cc, distill and Nesslerize the distillate. Calculate milligrams per liter of nitrate nitrogen. Required Oxygen. Besides the indirect estimation of organic matter through the determination of nitrogen in its various forms a more direct estimation may be made by oxidizing with standard potassium permanganate. It is readily seen that no calculation of organic matter can be made as a result of such a titration because the great variety of organic substances present in polluted water gives rise to a great variety of reactions. On the other hand the calculation of the oxygen required to oxidize all reducing agents in the water gives a fair, though inexact idea of the amount of organic pollution. Differences in procedure will cause the reduction of varying quantities of potassium per- manganate and it is therefore necessary rigidly to standardize the method. It is also necessary to correct the. results according to the amount of nitrites, ferrous salts and hydrogen sulphide, if these are found in considerable quantities. The determination is carried out by treating a measured volume of water with an excess of standard potassium perman- ganate solution, at a specified temperature, and titrating the excess after a stated' period. If the solution is heated during the treatment the excess of permanganate is determined by adding a measured excess of a standard solution of oxalic acid or ammo- nium oxalate, titrating the excess by standard potassium per- manganate solution. If cold treatment is used the excess of potassium permanganate cannot be determined in this manner because oxalic acid reduces permanganates very slowly unless heated to at least 60. In this case the excess of potassium permanganate is reduced by adding potassium iodide and titrat- 430 QUANTITATIVE ANALYSIS ing the liberated iodine with standard sodium thiosulphate solution : 2KMn0 4 +10KI+8H 2 SO4->6K 2 S04+2MnS0 4 +8H 2 0+10I; 2Na 2 S 2 O 3 + 2I-+Na 2 S 4 O 6 + 2NaI . The method involving digestion at 100 6 will be described. Determination. Prepare the following reagents: 1. Sulphuric Acid. Dilute the concentrated acid with three volumes of distilled water. Add potassium permanganate until a faint pink color persists after standing for several hours. 2. Standard Potassium Permanganate Solution. Calculate the weight of crystallized potassium permanganate required for 1000 cc of a solu- tion, 1 cc of which shall be equivalent to 0.1 mg of oxygen. Dissolve this weight of salt and dilute to 1000 cc. Standardize as follows : Add 10 cc of this solution and 10 cc of solution (1) to 100 cc of distilled water in a flask, immersing the flask in boiling water for 30 minutes. This destroys the oxygen -consuming capacity of the distilled water. Add 10 co of solution (3) and then potassium permanganate solution until a faint pink color persists. Now add 10 cc more of the oxalate solution and titrate with the permanganate solu- tion for the standardization of the latter. Calculate its value in milli- grams of available oxygen per cubic centimeter. 3. Ammonium or Sodium Oxalate Solution. Use the purest obtain- able salt. Make a solution of which 1 cc is equivalent, as a reducing agent, to 0.1 mg of oxygen. Measure 100 cc or less of the water into a 200-cc flask, add 10 cc of solution (1) and 10 cc (exactly measured) of solution (2) and immediately place the flask in a bath of boiling water, the water level of which is kept above the level of the contents of the flask. Digest for exactly 30 minutes. Remove the flask and add exactly 10 cc of solution (3). Titrate with standard potassium permanganate solution and calculate milligrams per liter of required oxygen. If 10 cc of permanganate solution is insufficient for complete oxidation repeat the digestion with a larger quantity. At least 5 cc excess of permanganate should remain after the digestion. Interfering Substances. If oxidizable mineral substances, such as ferrous iron, sulphide or nitrite, are present in appreciable quantities corrections should be applied as accurately as possible by suitable procedures. Direct titration of the acidified sample while cold, using a three-minute period of digestion, serves this WATER 431 purpose quite well for polluted surface waters and fairly well for purified sewage effluents. Few raw sewages containing no trade wastes need such a correction but raw sewages containing "pickling" liquors do need it. If the sample contains both oxi- dizable mineral compounds and gaseous organic substances the latter should be driven off by heating, the sample being allowed to cool before applying this test for the correction factor. If such corrections are made the fact should be stated, with the amount of correction. CHAPTER XV STEEL AND ALLOYS STEEL AND CAST IRON The impurities contained in iron and steel usually form a very small portion of the total mass. Wrought iron and steel often contain a total of less than one percent of elements other than iron while even pig iron does not often contain as much as ten percent of other elements. It is therefore not custorriary to make determinations of the percent of iron but rather of the small amounts of other elements, which give certain very important properties to the metal in which they are contained. Elements occurring in iron and steel and commonly determined are carbon, silicon, phosphorus, sulphur, manganese and titanium. In alloy steels for special purposes determinations are also made of tung- sten, nickel, chromium, molybdenum, vanadium and copper. Exact and Rapid Methods. For the determination of each element there are available certain well known methods and these are continually being revised and supplemented by other newer methods. Considerable experience is therefore necessary if the analyst is to be able to intelligently select the method best adapted to his purpose. There is a certain distinction to be made between what may be classed as " exact" methods and others that are more properly called " rapid" methods. Thus a determination of carbon in steel, made by an approved exact method, may require at least two hours and sometimes longer while a less exact determination might be made by another method in ten or fifteen minutes. Conversations with works chemists will often give the student the erroneous impression that the longer methods are impracticable and are taught in colleges but not used in practice, while the rapid methods are improperly neglected in the students' college courses. It is true that more emphasis is laid upon the exact method, as a rule. If 432 STEEL AND ALLOYS 433 the science and careful manipulation involved in the longer method are properly appreciated and learned the student will have no difficulty in learning the shorter and less exact method after he enters his professional career. It is highly important that one should understand the proper place of each class of methods. In the steel works samples may be taken from the melted iron as it runs from the blast furnace or from the steel ladles which receive the product of the steel furnaces. These samples are taken directly to the works labora- tory where the analysis must be made very quickly in order to furnish information which will serve as a guide in mixing charges for the steel furnace or for properly disposing of or modifying the product of a given furnace. The results of such an analysis do not often serve as a guarantee to the steel consumer but rather as a check upon the various stages in the process of steel manu- facture. For this reason rapid methods are quite suitable for the purpose. When the steel is placed upon the market as a finished product the case is quite different. Modern industrial development has created new and rigorous requirements regarding the quality of steel entering into machinery and structural work. It becomes necessary for the steel manufacturer to guarantee the percents of the elements in his steel within very narrow limits and a method of analysis that will not give results having a high degree of accuracy is quite useless for this purpose. Standard Methods. An inspection of the methods for analysis of steel as practised in the various works laboratories will show that while the several standard methods are quite universally used, many variations have been introduced by individual chem- ists. Each laboratory usually has its methods described and specified and these must be rigidly followed by all chemists working in that laboratory. There is, of course, much difference of opinion concerning the relative merits of different methods and it is inevitable indeed it is even desirable that modifications should be made whenever any improvement is seen to be possible. It is also 'true, however, that many modifications of good methods have made poor methods because the modifications have been made without an adequate knowledge of the scientific principles underlying the analytical process. Many chemists have con- 28 434 QUANTITATIVE ANALYSIS fidence in their methods when this confidence is based upon little more than the ability to obtain close agreement of duplicate determinations. The error involved in such conclusions has been discussed in an earlier section. Many of the analytical methods for iron and steel have been in use for a long time. Some of these have been retained in practically their original form and still bear the names of the chemists who first proposed them. Others have been so modified that they bear little resemblance to the original method. Sampling. Analysis may be required of either works samples, taken from the metal as it runs from the furnace, or of the finished product. Samples of the first class are dipped from the melted metal by means of a small ladle and are poured onto a clean iron plate or into a small iron mold. The sample is crushed, if a brittle product like pig iron, or drilled if steel. Pieces of already solidi- fied metal are drilled to obtain a sample for analysis. The outer case should be first removed because it may contain iron oxide or sand, or the percent of carbon may have been lowered by oxida- tion. The drill should be set to make as fine drillings as possible and if powder is at the same time produced it should be well mixed with the larger pieces before weighing for analysis. Solution and Evaporation. Steel and iron analysis involves many operations of dissolving and of evaporating solutions. The work of the iron and steel laboratory must usually be done as quickly as possible and the analyst must therefore give considerable attention to the best manipulation, from the stand- point of speed and accuracy. Dissolving metals in strong acids is always attended with the evolution of disagreeable and often poisonous fumes and good draught hoods are absolutely essential for carrying on this work. The evolution of gases and the boiling of solutions for evaporation will occasion loss of the dissolved matter unless proper attention is given to the prevention of such loss. When evaporation is not to follow solution of the sample an Erlenmeyer flask is usually the best vessel for the purpose. Loss by spat- tering is thus reduced to a minimum. Casseroles should be used if the solution of the sample is later to be evaporated. These are covered with glasses until the sample has all dissolved and evolution of gases is completed. The cover glasses are then STEEL AND ALLOYS 435 rinsed down and removed and the solution is placed on a steam bath or hot plate or it is held over a free flame and rotated con- tinuously by hand. The choice of method to be used for evaporating will depend upon the requirements of the case. If time does not press and work may be so fitted together as to carry on a large number of analyses at a time, evaporation on the steam bath will prove to be a convenient and safe process, except for solutions of high boiling points such as those containing sulphuric acid. If the opposite is true and speed is a matter of prime importance, evaporations must be hurried but the resulting danger of loss through bumping and spattering requires that the casserole should be held in the hand and kept in continuous motion. This constant agitation considerably increases the rate of evapora- tion, particularly because of the spreading of the solution over the sides of the casserole and the consequent increase in the effective surface. Of course this process is carried out with an uncovered casserole as otherwise little advantage would be derived from forced boiling. Standard Samples. In the description of the volumetric determinations that follow, methods are given for the standardi- zation of the solutions against suitable primary standards. In many of these cases it is also convenient to standardize the solu- tion against a steel or iron in which the percent of the element in question is accurately known , the standard sample being weighed and treated exactly as is the sample that is being analyzed. This procedure possesses a further advantage in that it auto- matically corrects for any deviation from the theoretical course of the reactions occurring during the preparation of the solution or during the titration itself. The steel laboratory may prepare and carefully analyze its own standard samples. It is also possible to obtain most of the necessary standard steels and irons from the Bureau of Standards. These standard samples may also be used for checking the accu- racy of gravimetric methods and in this way they will serve for control work. Carbon. The most important element occurring in steel is carbon. This is because it is the element which makes possible the formation of steel by imparting to iron the capability of being 436 QUANTITATIVE ANALYSIS hardened by suitable heat treatment. The development of alloy steels has lately brought nickel, chromium, vanadium, tungsten and other metals into prominence as constituents of special steels, but without carbon, alloys of these elements with iron would be of little value. The effect of carbon upon iron with which it is combined is to increase the tensile "strength and hardness and to decrease the ductility. Carbon is present in steel chiefly as a carbide, FeaC, although small quantities may occur as free carbon. In cast iron large quantities of carbon are free, particularly in gray cast iron. A more extended discussion of the properties of steel, as dependent upon the condition of the carbon, will be taken up later (page 478) . The determination of total carbon is the only carbon determina- tion that is usually required in steel analysis. In 4 the analysis of cast iron determinations also of free and combined carbon may be required. The determination of total carbon is generally made by a combustion process, the carbon being oxidized and the resulting carbon dioxide determined, although oxidation in solu- tion has been employed, the oxidizing agent being chromic acid. The details of the combustion processes vary widely. Fine drillings of steel may be burned directly or a preliminary separa- tion of the carbon may be made. The apparatus for combustion may be a furnace and combustion tube or a special form of closed crucible, through which air and oxygen may be passed. The resulting carbon dioxide may be measured at an accurately observed temperature and pressure and its weight calculated or it may be absorbed in a basic solution and either weighed or the excess of base determined by titration. Direct Combustion. Iron or steel may be completely burned in oxygen at 900 to 1100. The method is desirable because it avoids a rather tedious process of preliminary solution, nitra- tion and washing. For the combustion there is required a tube of quartz or porcelain, 24 inches long and having an inside diameter of 3/4 inch. It must be protected from contamination by iron oxide by placing an alundum cylinder in the section which is to hold the boat. The combustion furnace may be heated by gas or electricity and should be 12 to 14 inches in length. The combustion is carried out in a manner quite similar to the combustion of coal (page 311). The long furnace and STEEL AND ALLOYS 437 tube there used are not required in this case because no volatile hydrocarbons are produced and long contact of the gases with cupric oxide is not necessary. The small amount of carbon monoxide that may be formed at first is completely oxidized by passing the mixture with oxygen over a small amount of cupric oxide or through platinized asbestos. The platinum black in the latter case catalyzes the combination of carbon monoxide with oxygen. The train of apparatus necessary for the gravimetric determina- tion is as follows: (1) Oxygen tank, (2) bubble tube of 30 percent potassium hydroxide solution, (3) tube of soda lime to retain spray from the potassium hydroxide solution and to complete the removal of carbon dioxide, (4) combustion tube containing (4a) space of about 3J inches inside the furnace, lined with an alundum cylinder to hold the combustion boat, (4b) platinized asbestos to the end of the furnace, leaving the projecting ends of the tube empty, (5) U-tube filled with granular zinc or with glass beads moistened with chromic acid solution for the re- tention of oxides of sulphur, (6) U-tube filled with calcium chlo- ride and (7) absorption apparatus for carbon dioxide, discussed below. The method of preparing and assembling this apparatus should be clear from the discussion of the determination of carbon dioxide in carbonates (page 129) and of the determination of carbon and hydrogen in coal (page 311). Absorption Apparatus. The nature of the absorption ap- paratus (7) will depend upon the method chosen for the final determination. This may be either gravimetric or volumetric in principle. One of the following variations is recommended: (a) Absorption in Weighed Bulbs of the Geissler or a Similar Type, Filled with 33 Percent Potassium Hydroxide Solution and Carrying a Prolong Filled with Calcium Chloride. The manipulation of these bulbs is carried with precautions similar to those observed in the determination of carbon dioxide in carbonates. (See pages 133 to 136.) (6) Absorption in Weighed Tubes Carrying Soda Lime and Calcium Chloride or Phosphorus Pentoxide. The Fleming tube (figure 95) is satisfactory for this purpose and a determination can be made very rapidly by this method. In filling, the part (s) is to contain, first, a loose plug of asbestos then alternating 438 QUANTITATIVE ANALYSIS layers, about one-fourth inch thick, of 20-, 40- and 60-mesh soda lime. Part (d) is filled with either phosphorus pentoxide on glass wool or calcium chloride. (If phosphorus pentoxide is used at this point it must also be substituted for calcium chloride in tube (6) of the absorption train.) Phosphorus pentoxide should not be used after it has become visibly moist. On this account the charging of tubes with this sub- stance should not be undertaken on days when the humidity is high as it is then im- possible to work rapidly enough to prevent absorption of considerable quantities of moisture from the air. To prepare phos- phorus pentoxide for this purpose a loose ribbon of glass wool is first spread on a clean piece of paper and the dry material is sifted over the ribbon in an even layer. The edges of the glass wool are then turned over and the material is rolled into the proper shape to fit loosely into the tube. By this means the drying agent is given the maximum exposure to gases passing through the tube. (c) Absorption in Standard Barium Hydroxide Solution, the Unused Base being Titrated in Presence of Phenolphthalein after the Absorption is Finished. The Meyer tube (figure 96) is suitable for this purpose and it possesses the important merit of being easily emptied and rinsed just before the titration. The drying tube (6) should be omitted in this and all other volumetric methods as the drying of gases entering the carbon dioxide absorption tube is quite unnecessary. A saturated solution of barium hydroxide in carbon dioxide-free water is prepared and kept as a stock solution. This will contain approximately 3.9 gm of barium hydroxide, calculated as the anhydrous base, in each 100 cc of solution and it will be approxi- mately 0.45 normal. 550 cc of this solution is diluted to 1000 cc. The resulting solution is then approximately 0.25 normal, FIG. 95. Fleming absorption tube. STEEL AND ALLOYS 439 so that 1 cc is equivalent to about 0.0015 gm of carbon. This solution should be kept in a bottle which is provided with a siphon or with an outlet at the bottom and protected in such a manner as that entering air shall be drawn through a tube of soda lime or of saturated barium hydroxide solution, to remove carbon dioxide. As carbon dioxide is absorbed in barium hydroxide solution barium carbonate is formed and almost entirely precipitated. One liter of a saturated solution of barium carbonate in water contains 0.022 gm (corresponding to 0.00134 gm of carbon) at 20. Therefore it is possible to titrate the excess of unused FIG. 96. Meyer absorption bulbs. base after the absorption is finished, using a standard acid of concentration equivalent to that of the barium hydroxide and without nitration, provided that the solution was previously saturated with barium carbonate and that the titration is carried out rapidly and with stirring, so as to avoid resolution of the precipitate by local excess of acid. Phenolphthalein is used as the indicator. The first condition, above, is usually met by the fact that commercial barium hydroxide always contains small amounts of carbonate and that more is formed by carbon dioxide in the water that is used for preparing the standard solution. (d) Absorption in Standard Barium Hydroxide Solution, the Unused Excess being Titrated after Filtration and Washing with Carbon Dioxide-free Water. This procedure has no ad- vantage over that outlined as (c). Absorption of carbon dioxide from the air takes place during filtration and washing of the precipitate. Also the addition of wash water that is not saturated with barium carbonate causes resolution of traces of the precipitate. The second error tends to compensate the 440 QUANTITATIVE ANALYSIS first but a method that depends for accuracy upon mutual com- pensation of variable errors is not ideal. (e) Absorption in Saturated Barium Hydroxide Solution, the Precipitated Barium Carbonate being Removed by Filtration, Washed, Dissolved in an Excess of Standard Acid and the Unused Excess of Acid Titrated by Standard Base, with Methyl Orange as Indicator. 1 The barium hydroxide solution that is used in this method may be much more concentrated than that used in methods (c) and (d) because it is not to be titrated. On this account absorption is more certain when the gas is bubbled rapidly through the solution and in this lies the only important advantage over the other volumetric methods. The sources of error noted under (d) obtain here also. Drying tubes are omitted from the absorption tr,ain in volu- metric methods (c), (d) and (e). Combustion, Preceded by Solution of the Iron and Separation of the Carbon. Iron or steel dissolves easily in a solution of the double chloride of potassium and copper. The cupric chloride is the active agent and the double salt is used only because it is more easily purified and preserved. The reactions are : Fe+2CuCl 2 ->FeCl 2 +2CuCl and Fe 3 C+6CuCl 2 -3FeCl 2 +6CuCl+C. Free carbon is also left undissolved. The residue contains organic compounds, formed during the process of solution, and the total residue cannot, therefore, be weighed directly for the determina- tion of total carbon. If the solution is not well stirred or if not enough cupric potassium, chloride is used, copper will separate, returning to the solution upon stirring or addition of more of the solution of cupric salt: Fe+CuCl 2 -+FeCl 2 +Cu, These reactions illustrate the principle of replacement of one metal of a salt solution by another which has a higher solution tension. The solution of cupric potassium chloride must contain hy- 1 Cain: J. Ind. Eng. Chem., 6, 465 (1914). STEEL AND ALLOYS 441 drochloric acid in order to prevent the precipitation of cuprous chloride, a substance having very small solubility in water. If too much acid is present carbon may be lost through the forma- tion of hydrocarbons during the process of solution. This is typified by the hypothetical reaction shown by the following equation : 2Fe 3 C + 12HCl-*6FeCl 2 +C 2 H 2 +5H2. Concerning the choice of methods it may be said that direct combustion is much more rapid and is accurate if the metal is in a proper state of division. The method of preliminary solution is fully as accurate, except for high-speed tool steels, and is safer if the nature of the steel is not accurately known. There is little to choose between the gravimetric method, weighing the absorp- tion bulbs before and after the absorption of carbon dioxide, and the volumetric methods. The so-called " moist combustion " processes and the methods involving measuring the volume of carbon dioxide evolved, while attractive in principle, are trouble- some in execution and are subject to large errors unless great care is exercised. These are therefore little used. Determination by Direct Combustion. The apparatus train described on page 437 is assembled, drying tube (6) being used only in the gravi- metric modifications. If a high-pressure oxygen cylinder is used to supply this gas a special control valve must be provided. The entire apparatus is first tested for leaks. With the carbon dioxide absorption tube temporarily removed the furnace is brought to a temperature of about 1100 while oxygen is passed through slowly for one-half hour. This preliminary heating to expel traces of moisture and organic matter may be dispensed with if the apparatus has been in use. Unless the apparatus has been in continuous use one or more blank determinations must be made in order to correct for any constant small gain of carbon dioxide from imperfect apparatus or reagents. The determination of carbon in steel follows the blank. The different modifications, with blanks, will be referred to by the letters that have already been used in the discussion of the principles of these methods. Method (a). The potassium hydroxide bulbs are carefully wiped and the outlets are closed by short pieces of rubber tubing bearing glass plugs. They are allowed to stand in the balance case for 10 minutes, after which the rubber tubes are removed and the bulbs are weighed. 442 QUANTITATIVE ANALYSIS Crucible forceps tipped with rubber should be used for handling the bulbs in this and all subsequent operations. After weighing, the bulbs are inserted in the train and oxygen is passed through at the rate of about three bubbles per second for 15 minutes. At the end of this period the gas flow is stopped and the bulbs are removed, plugged and placed in the balance case. After 10 minutes the rubber caps are re- moved and the bulbs are immediately weighed. If this blank determi- nation gives a change of more than 0. 1 mg in the weight of the bulbs it must be repeated until the change drops to zero or becomes constant within this limit. When this condition is reached the determination of carbon may be made. While the blank determinations are running, prepare an alundum boat by placing a layer of granular alundum (grade RR) on the bottom and sides, then igniting for 5 minutes. After this has cooled about 2 gm of steel sample is weighed on a counterpoised glass, brushed into the boat and distributed in a uniform layer. Insert the weighed bulbs into the train and adjust the flow of oxygen to the same rate as that used in the blank. Without stopping this flow, open the end of the combustion tube next to the oxygen tank and insert the boat containing the sample, pushing it by means of a wire into the alundum thimble which is next to the platinized asbestos. Close the tube as quickly as possible and carefully twist the stopper into the end so that no leak can occur. The combustion of the steel begins almost immediately and is usually completed within a very short time. Now continue the passage of oxygen for 15 minutes, at the end of which period remove the absorption bulbs, stopper, place in the balance case and weigh, without the rubber caps, after 10 minutes. In the meantime another sample of steel should have been made ready in a second boat and another absorption bulb should have been weighed. As soon as the first bulb is removed the second is inserted in the train. The combustion tube is opened as before, the first boat is drawn out and the second is inserted. A continuous series of determinations may be made in this way without stopping the flow of oxygen or cooling the furnace. Weighings may be made while the combustion is proceeding. Bulbs may be used without refilling until exhausted, following the rule given on page 133. From the weights of samples and of carbon dioxide calculate the per- cent of carbon in the steel samples. Method (b). The manipulation is exactly like that of method (a) The rate of flow of oxygen is judged by bubbling through the potassium hydroxide tube following the oxygen tank. Method (c). A saturated solution of barium hydroxide is first pre- pared by warming and stirring the solid base with recently boiled water, STEEL AND ALLOYS 443 using a ratio of 70 to 100 gm of base to 1000 cc of water according to the purity of the barium hydroxide obtainable. Cool to room temperature and siphon into a bottle which is then closed with a rubber stopper. Dilute 550 cc of this solution to 1000 cc with distilled water, mix and place in a bottle which is provided with a guard tube of soda lime and a siphon or similar outlet. Prepare a solution of hydrochloric acid, 1 cc of which is equivalent to 0.002 gm of carbon, standardizing against pure sodium carbonate in presence of methyl orange. Rinse the Meyer bulbs with boiled water, then measure into them from a burette or an automatic pipette attached to the bottle 50 cc of the dilute solution of barium hydroxide, first discarding the few drops that are in the outlet of the measuring instrument. Add to the bulbs from a graduated cylinder enough cold, carbon dioxide-free water to bring the liquid just to the lower edge of the upper bulb when the gas is flowing. The quantity necessary should be determined, once for all, so that it may be added without delay in subsequent determinations. With the furnace already heated, connect the bulbs in place while the oxygen is flowing and conduct the blank experiment as in method (a), to the end of the absorption period, then disconnect the bulbs without stopping the flow and rinse the solution into a 250 cc Erlenmeyer flask by means of boiled water, paying no attention to the precipitate. Add a drop of phenolphthalein and titrate at once (not too rapidly but stir- ring vigorously) with standard hydrochloric acid, to the disappearance of the pink color. When this point is reached the volume of required acid is read. The pink color will return after standing, due to the gradual resolution of barium carbonate, but this is not considered in the titration. At the time when the absorption bulbs are removed from the train of apparatus another tube, charged like the first, is inserted for use in the next blank determination, the barium hydroxide solution having been measured just before inserting the bulbs. Run this blank deter- mination like the first. While the gas is flowing for the second blank the steel sample should be weighed and placed in the prepared boat, as in method (a). When the blank is finished the first set of absorption bulbs, charged as before, is substituted and the boat with the steel sample is inserted into the combustion tube. Continue the combustion as directed for method (a). While this is proceeding the base from the second blank is titrated. This titration should agree with the first to within 0.1" cc of standard acid. A continuous series of determinations may be made without stopping the flow of oxygen or cooling the furnace. Enough blanks must be 444 QUANTITATIVE ANALYSIS run to obtain agreement of titrations and at least two determinations of carbon should be made on each sample of steel. The volume of acid required in the determination is subtracted from that used in the blanks and the remainder is multiplied by the carbon equivalent of the standard acid. From this and the weight of sample calculate the percent of carbon in the steel. Method (e). The saturated solution of barium hydroxide whose preparation was described for method (c) is used. About 25 cc is placed in the Meyer bulbs and carbon dioxide-free water is added so as to fill all but the upper bulb when the gas is flowing. The blank experiment and the carbon determination are conducted as in method (c), with the following exceptions : Barium hydroxide need not be accurately measured, although approxi- mately equal volumes should be used in all experiments. A graduated cylinder is suitable for measuring the solution but it should first be rinsed with boiled water. At the end of the absorption the solution is filtered rapidly on a paper filter and the Meyer tube, the precipitate and the paper are washed several times with cold, carbon dioxide-free water. The precipitate is then dissolved by adding exactly 25 cc of tenth-normal hydrochloric acid (or the solution described for method (c)) the solution being caught in the Meyer tube so that all adhering barium carbonate is dissolved. The paper is well washed with hot water and the combined filtrate and washings rinsed into an Erlenmeyer flask. The excess of acid is then titrated by tenth-normal sodium hydroxide, using methyl orange as indicator. Determination by Combustion, Preceded by Solution. Prepare a solution of cupric potassium chloride containing 500 gm of the crystal- lized salt and 75 cc of concentrated hydrochloric acid in 1000 cc of solu- tion. Filter through ignited asbestos into the bottle. Weigh 1 gm of the steel or iron drillings, place in a 250-cc beaker and add 100 cc of the cupric potassium chloride solution. Stir until the metal is all dissolved, warming to about 65. If many determinations are to be made a stirring machine is desirable. Filter the solution through a Gooch crucible or a carbon tube (shown in Fig. 67, page 250). The asbestos used in the filter must have been previously ignited to remove all organic matter. Wash with warm (50) dilute hydrochloric acid until the washings are free from color, then with cold water until free from chlorides. It is desirable that most of the water be removed from the filter and carbon, although the latter need not be absolutely dry. After partial drying by means of the pump and dry- ing oven the asbestos and carbon are carefully removed and placed in a combustion boat, using a small pair of forceps. This operation should STEEL AND ALLOYS 445 be performed over a sheet of white, glazed paper. The inside of the crucible or carbon tube is carefully wiped clean, using a tuft of ignited asbestos. The combustion and subsequent determination are carried out as in the direct combustion process except that directly following the combus- tion tube there is inserted a U-tube containing a saturated solution of ferrous sulphate, acidified by sulphuric acid. If traces of hydrochloric acid are retained by the filter or carbon this is partly oxidized, chlorine being produced, and partly carried over without change. Chlorine is absorbed and reduced by the ferrous sulphate while small quantities of hydrochloric acid are absorbed by the water of the solution. Free (Graphitic) Carbon. When steel or iron containing both free and combined carbon is dissolved in nitric acid of specific gravity 1.2 the combined carbon passes into solution as hydro- carbons, the graphitic carbon being left as an insoluble residue. The latter may be separated by filtration and used for the deter- mination of free carbon. If it is to be weighed directly the silica which is also left must be removed by washing with potassium hydroxide solution, then with water. A better method is to wash the carbon and silica free from iron salts and acids and then determine by combustion, exactly as in the case of total carbon. Graphitic carbon may also be determined by difference, sub- tracting the combined carbon from the total carbon. Determination. Weigh 1 gm of pig iron or 10 gm of steel and dis- solve in nitric acid, specific gravity 1.2, using 15 cc of acid for each gram of sample. Filter through ignited asbestos in a Gooch crucible or a carbon tube and wash with dilute hydrochloric acid, then with hot water until free from chlorides. Burn in the combustion tube used for total carbon and determine in the same way. Combined Carbon. -The percent of combined carbon may be determined indirectly by subtracting the percent of graphite from that of total carbon. The only reliable method for the direct determination of combined carbon is that of Eggertz. 1 This method depends upon the fact, noted in the discussion of free carbon, that when steel or iron containing combined carbon is dissolved in dilute nitric acid the combined carbon forms soluble organic compounds which impart a color to the solution, the inten- X Z. anal. Chem., 2, 433 (1862); Chem. News, 7, 254 (1863). 446 QUANTITATIVE ANALYSIS sity of which varies with the percent of combined carbon. The solution is then compared in tubes with the solution of a stand- ard steel whose carbon content is known, the unknown per- cent being then calculated. It will be seen later, when a more extended study of carbon conditions is taken up, that combined carbon may exist in more than one physical state, although probably always present as the carbide FeaC. This difference in physical state is influenced by the presence of other elements and also by the mechanical and thermal treatment which the steel has received. The color of the acid solution is affected by all of these factors and it therefore becomes necessary to use for a standard steel one in which not only the percent of combined carbon is known to be approximately the same as that of the steel being examined but also one that has nearly the same rjercents of other impurities and that has been subjected to the same thermal and mechanical treatment. All of these factors cannot well be known in general testing and the method is therefore of little value for this class of work. Its chief value is to the steel works chemist who knows in every case the nature of the steel with which he is dealing and who is thereby enabled to select his standard steel with due regard to all of the variable factors. Determination. Treat the standard steel and the steel being ex- amined as follows: Weigh 1 gm of the drillings and dissolve in a beaker in 30 cc of nitric acid whose specific gravity is 1.2 and which is free from chlorine. Warm the acid toward the end of the process, to complete the solution. Filter to separate free carbon and silica, receiving the filtrate in a 100-ec volumetric flask. Wash the residue, dilute to the mark and mix. Transfer 30 cc of the solution of lighter color to an Eggertz tube, which is a tube graduated from 1 cc to 30 cc and having an internal diameter of about 1 cm. Add the darker solution to another similar tube until the color of the two tubes appears to be equal, viewed from above. In case the color is very dark, less solution must be used or the color observed from one side, or else the color of the darker solution is lightened by dilution. Calculate the percent of combined carbon. Silicon. Silicon occurs in all steels, generally in quantities less than 0.3 percent. Certain silicon steels contain as much as 20 percent. Cast iron contains as much as 3 percent of silicon. Silicon occurs as a silicide which is probably to be represented by STEEL AND ALLOYS 447 the formula FeSi, this forming a solid solution with the remainder of the iron. Inclusions of slag also may contain silicon as silicates of iron and manganese. Silicon has little effect upon the me- chanical properties of steel but is desired in cast iron because of its tendency toward throwing carbon out of its combination with iron, thus forming gray iron which has a greater fluidity when melted than does white iron, and which is therefore better suited for foundry purposes. When iron silicide is dissolved in nitric acid the silicon is entirely converted into silicon dioxide, largely in the state of colloidal silicic acid. If the silicic acid is dehydrated and the resulting silicon dioxide made insoluble by heating with acids it may be separated by filtration. As obtained from pig iron the silicon dioxide so obtained contains all of the free carbon of the iron. This is removed by ignition. In the adaptation of this process to the quantitative determination of silicon in iron and steel the chief difficulties encountered are due to the tendency of silica to change from the gel to the sol and also to incomplete washing of the silica. In order to assist in the separation of silica in an insoluble condition Drown suggested 1 the addition of sulphuric acid to the solution during the evaporation to render silica insoluble. This materially shortens the time required fora determination as otherwise the solution must be evaporated and heated for some time in order to completely separate the silica. During the washing of the silica, if pure water is used, iron salts hydrolyze and insoluble basic salts are retained by the filter. Alternate washing with water and hydrochloric acid will remove all but traces of iron salts and a correction may be made for these by the common process of volatilization of silica by hydrofluoric acid. Determination. Prepare a mixture of 375 cc of concentrated nitric acid, 125 cc of concentrated sulphuric acid and 500 cc of water or as much of this mixture as is needed. Weigh 1 gm of pig iron or 5 gm of steel and dissolve by warming in a casserole or platinum dish with 75 cc of the acid mixture. The solution is evaporated by agitation of the un- covered casserole over a flame until pronounced fumes of the sulphuric acid appear. Allow the solution t:> cool, then add 10 cc of dilute hydro- chloric acid and 50 cc of water. Warm until iron salts are dissolved 1 Trans. Am. Inst. Min. Eng., 7, 346 (1879). 448 QUANTITATIVE ANALYSIS then filter and wash alternately with hot dilute hydrochloric acid and water until practically free from iron. Ignite in a platinum crucible, cool and weigh. Volatilize the silica by treatment with sulphuric acid and hydrofluoric acid (see page 293), and from the loss in weight calcu- late the percent of the element silicon. Sulphur. Sulphur occurs in iron and steel as ferrous sulphide, FeS, unless manganese is present, in which case it forms mangan- ous sulphide MnS. Ferrous sulphide is itself brittle. It also shows a tendency toward the formation of envelopes surrounding the crystalline grains of steel, reducing their cohesion and result- ing in " shortness," particularly when hot. Sulphur is therefore said to cause "red shortness" of steel. Manganese sulphide usually occurs as small rounded masses instead of envelopes and it is therefore much less objectionable than ferrous sulphide. The steel maker therefore relies upon manganese Ho correct largely the bad effects of sulphur, although the latter should not be present in steel in quantities greater than 0.05 percent. In the best steel its quantity is much less than this. The determination of sulphur may be accomplished by oxida- tion to sulphuric acid, followed by precipitation as barium sulphate, or by evolution methods; in the latter the metal is dissolved in hydrochloric acid, ferrous sulphide or manganous sul- phide forming hydrogen sulphide. The latter is distilled into some absorbing solution and subsequently determined by gravi- metric or volumetric methods. Oxidation Method. Steel or iron dissolves more readily in dilute nitric acid than in the concentrated acid and the former is therefore used for dissolving the sample for nearly all other deter- minations. But the dilute acid will not serve for dissolving the metal as a preliminary to the gravimetric determination of sulphur because a part of the sulphur will be evolved as hydrogen sulphide and will then escape. Concentrated nitric acid com- pletely oxidizes the sulphide to sulphate. 6FeS+24HNO 3 ->2Fe 2 (S04)3+2Fe(NO 3 )3+18NO+12H 2 O. The sulphur is then precipitated as barium sulphate. The separation from the large amount of iron involves some difficulty. Unless a considerable excess of acid is present basic ferric salts, products of hydrolysis, are retained by the precipitate STEEL AND ALLOYS 449 of barium sulphate. If too much acid is present the precipitation of barium sulphate is incomplete. For the solubility of barium sulphate in hydrochloric acid, see page 92. Nitric acid must not be present at all because of its effect upon the occlusion of iron salts by the precipitate. Silica must be separated by evaporation and nitration before the precipitation of barium sulphate, and during the evaporation and heating that are necessary for this purpose there is danger of loss of sulphur through decomposition of ferric sulphate: In order to prevent loss of sulphur trioxide by this means, a small amount of sodium carbonate is added before the evapora- tion. This immediately forms sodium nitrate or chloride (hy- drochloric acid also having been added) and this reacts during the evaporation, thus: Fe 2 (SO 4 ) a + 6NaCl->2FeCl 3 + 3Na 2 SO 4 . Sodium sulphate is not decomposed by moderate heating. Determination. Weigh 5 gm of drillings or powder into a casserole. Place under a hood and add 50 cc of concentrated nitric acid which is free from sulphuric acid. Action may not begin at once unless the cas- serole is warmed but after the metal begins to dissolve the action may become violent. In this case the casserole should be placed in cold water. In the later stages it may again be necessary to heat the cas- serole. Add 1 gm of sodium carbonate, free from sulphate, and evapo- rate to dryness, holding the casserole over the flame and giving it a rotary motion to prevent bumping and to hasten evaporation. When the residue is dry, heat for 15 minutes at a temperature just below red- ness, then add 30 cc of concentrated hydrochloric acid, and again evaporate to dryness and heat as before. Cool, add 30 cc of concen- trated hydrochloric acid, warm until all iron salts are in solution, then evaporate in the same manner as before until ferric chloride begins to crystallize. Add a very small amount of hydrochloric acid to redissolve these crystals, then add 25 cc of water, filter and wash with water, then with a very small amount of hot, dilute hydrochloric acid, repeating the water and acid washing until the paper, silica and carbon are free from the red or brown stains of ferric chloride. Finally wash with hot water until the volume of the filtrate is about 200 cc. If this residue is large in quantity it will contain an appreciable amount of sulphur. 29 450 QUANTITATIVE ANALYSIS In this case transfer the paper with the residue to a platinum crucible, burn until paper and carbon have been removed and fuse with 2 gm of sodium carbonate. Cool, dissolve the fusion in dilute hydrochloric acid, using no more than is necessary, and wash into the main solution. Heat to boiling and add, a drop at a time and stirring continuously, 10 cc of 10 percent barium chloride solution. Digest at a temperature near the boiling point for 30 minutes, then allow to stand for 2 hours. Filter and wash, alternately with dilute hydrochloric acid and water, until the filter and precipitate are white and finally with water until free from chlorides. In this washing use as little hydrochloric acid as possible. Ignite the paper and precipitate in a platinum crucible at a low temperature and weigh the barium sulphate. If the ignited precipi- tate is not white some iron oxide is contained in it. In this case add 1 gm of sodium carbonate, fuse, dissolve the fusion in water and dilute hydro- chloric acid and precipitate as before. Calculate the percent of sulphur in the sample. Evolution Method. The determination of sulphur by evolu- tion depends upon the decomposition of metallic sulphides by hydrochloric acid, the resulting hydrogen sulphide being distilled and absorbed in another solution. The absorbing solution may form an insoluble sulphide with the hydrogen sulphide or it may oxidize the latter to sulphuric acid which is then determined gravimetrically. Absorbents of the first class are basic solutions of salts of lead, cadmium or silver. Absorbents of the second class are bromine in hydrochloric acid, potassium permanganate and hydrogen peroxide. A solution of cadmium chloride in excess of ammonium hydroxide is to be preferred. The precipitate of cadmium sulphide may be washed, dried and weighed, but it is better to decompose it with hydrochloric acid and titrate the liberated hydrogen sulphide with standard iodine solution. The solubility of cadmium sulphide in hydrochloric acid is not large unless the resultant hydrogen sulphide is removed, as in this case by oxidation. The evolution method may be performed in less time than the oxidation method. It has been shown by Phillips to be inaccu- rate, 1 however, for white pig iron because of the formation of organic sulphur compounds, of which methyl sulphide, (CHs^S, was isolated. Such sulphides are difficult to expel from the 1 J. Am. Chem. Soc., 17, 891 (1895). STEEL AND ALLOYS 451 evolution flask and require as much as two hours of boiling, during which time air, carbon dioxide or hydrogen is drawn through the apparatus. Phillips found that the organic sulphides could be decomposed by passing the vapors through a tube, heated to redness. The additional time necessary for the ex- pulsion of organic sulphides from the evolution flask makes the method impracticable for white pig iron. For steel and gray pig iron, containing relatively low percents of combined carbon, the method is satisfactory. Determination. Prepare an approximately fiftieth-normal solution of iodine by dissolving the calculated weight of iodine and twice its weight of potassium iodide or sodium iodide in 1000 cc of water. Stand- ardize just before making the sulphur determination by titrating against a standard solution of sodium thiosulphate or of sodium arsenite (As 2 3 and half its weight of NaOH). The iodine dissolves rather slowly unless well powdered. It is well to decant the solution into another bottle in order to avoid the possibility of particles of undissolved iodine changing the concentration of the solution after standardization. Determination. Use a 300-cc flask, having a round bottom, for the evolution flask. Connect, through a 2-hole rubber stopper, a 100- cc separatory funnel and a short tube, bent at a right angle with the flask. The separatory funnel should reach to the bottom of the flask and should have the bottom turned up, as in the apparatus for the determination of carbon dioxide in carbonates. The short exit tube is connected with another tube which reaches to the bottom of a cylinder having a capacity of about 100 cc, in which is placed 50 cc of a solution made as follows: Cadmium chloride 10 gm, water 375 cc, concentrated ammonium hydroxide 625 cc. This cylinder is similarly connected with a second cylinder containing the same kind of solution. Both cylinders should stand in a large beaker of cold water. Instead of the two cylinders a Meyer tube (figure 96) may be used. Weigh into the evolution flask 5 gm of steel or iron drillings, close the flask and place 75 cc of hydrochloric acid (1:1) in the separatory funnel. Admit the acid fast enough to cause a rapid evolution of hydrogen. Finally add all but about 5 cc of the acid and warm to assist the solution. When the metal is all dissolved boil for 5 minutes at a rate that will permit absorption of the hydrogen sulphide. This boiling should completely expel hydrogen sulphide and hydrogen from the flask. Disconnect the delivery tube, then remove the source of heat and rinse the tubes, allowing the washings to run into the absorption cylinders. 452 QUANTITATIVE ANALYSIS If the tubes contain any cadmium sulphide wash with dilute hydrochloric acid and then with water but do not agitate the solution. Rinse the contents of the absorption cylinder into a 500-cc beaker, dissolving adhering precipitate by means of dilute hydrochloric acid, allowing this solution to run immediately into the main body of solution. Usually all of the hydroge"n sulphide is absorbed in the first cylinder and the contents of the second need not be used if no trace of yellow cadmium sulphide appears in it. Add water until the volume is about 300 cc, then add dilute hydrochloric acid until the liquid is distinctly acid in character, stirring gently meanwhile. The disappearance of turbidity is sufficient indication of an acid reaction. Rapid stirring and undue agitation will cause a loss of hydrogen sulphide. Add 1 cc of starch solution and ti- trate at once with decinormal iodine solution. Calculate the percent of sulphur in the sample. Phosphorus. The proportion of phosphorus in steel of satisfactory quality is not usually higher than 0.1 percent and is frequently required to be less. Acid open hearth and acid Bessemer steel contain larger quantities of phosphorus than steel made by basic processes. Phosphorus occurs in steel as the phosphide Fe 3 P. Its effect is to cause brittleness of the steel, this being at least partly due to the promotion of coarse granulation. The determination of phosphorus in iron or steel may follow either gravimetric or volumetric methods. In any case the final determination must be preceded by separation from the relatively large excess of iron. The separation is usually made by either a modification of the method of Fresenius 1 known as the "acetate method," or the molybdate method of Sonnenschein. 2 Acetate Method. This method of separating iron and phos- phorus depends upon the relatively large solubility of ferrous acetate as compared with that of basic ferric acetate and ferric phosphate. The iron is first reduced entirely to the ferrous condition by sulphurous acid, then either a small amount reoxi- dized by bromine or a small amount of ferric chloride is added. The solution is now made slightly basic, then an excess of acetic acid and water is added. A precipitate forms, consisting of ferric phosphate and basic ferric acetate, the latter being present 1 J. prakt. Chem., 45, 258 (1848), 2 Ibid., 53, 339 (1851). STEEL AND ALLOYS 453 in very small quantity. The larger part of the iron has remained in solution as ferrous acetate and is separated by filtration. This method necessarily leaves a small amount of iron in the phosphorus precipitate. In order to separate this, advantage is taken of the fact that small quantities of iron are not pre- cipitated by ammonium hydroxide if organic acids are present. Either citric acid or ammonium citrate is added and the phos- phorus is precipitated by " magnesia mixture" in presence of ammonium hydroxide. The ionization of ferric citrate is so small that the solubility product of neither ferric hydroxide nor basic ferric citrate is attained. The acetate method is accurate if carefully performed, but is complicated in detail and is more liable to fail than the next method to be described. Molybdate Method. The molybdate method of separating iron and phosphorus depends upon the insolubility of ammonium phosphomolybdate and the solubility of iron in nitric acid. The iron or steel is dissolved in nitric acid, carbon is oxidized by potas- sium permanganate, the solution is nearly neutralized and a solu- tion of ammonium molybdate in nitric acid is added. The pre- cipitate of ammonium phosphomolybdate is separated by filtration and is then treated according to the method which has been selected for the final determination. The removal of carbon by oxidation is necessary in order that precipitation shall be complete. The determination of phosphorus may now be made (1) by drying and weighing the yellow precipitate of ammonium phos- phomolybdate, (2) by measuring its volume, (3) by titrating its molybdic oxide by means of a standard base, (4) by reducing its molybdic oxide to molybdenum sesquioxide and titrating by standard potassium permanganate solution, or (5) by dissolving the yellow precipitate in ammonium hydroxide and precipitating as magnesium ammonium phosphate. If method (1) is to be followed it is necessary that care be exercised in precipitating the ammonium phosphomolybdate in order that its composition may be constant. Precipitated under the conditions later described its composition is represented by the formula (NH 4 )3P04.12MoO 3 . The composition is somewhat altered by variation in temperature, excess of ammonium 454 QUANTITATIVE ANALYSIS molybdate, excess of nitric acid and time of precipitation. The precipitate may contain also small amounts of free molybdic acid, especially if too much nitric acid is present, or of ammonium silicomolybdate if silicon has not been removed. This method of direct weighing is not often followed. Method (2) is a rapid but inaccurate method. The precipita- tion is carried out in a pear-shaped bulb having a graduated stem. The precipitate is packed into the stem by centrifugal action and its volume is read and converted into weight percent by a previously determined factor. Method (3) was suggested by Pemberton. 1 In this method the yellow precipitate is dissolved in an excess of a standard solution of potassium hydroxide or sodium hydroxide, the excess being then titrated by a standard acid solution, phenolphthalein being used as the indicator. The reaction between the phosphomolyb- date and the base is as follows: 12H 2 0. Upon the addition of standard acid, phenolphthalein changes color when the excess of base has been neutralized and the following reaction has occurred: (NH 4 ) 3 P0 4 + HC1^(NH 4 ) 2 HPO 4 + NH 4 C1. Twenty-three equivalents of base have therefore apparently been used at the end point and in order to express this fact in one equa- tion the reaction is often represented as follows: 2(NH 4 ) 3 P0 4 .12Mo0 3 + 46KOH->2(NH 4 ) 2 HP0 4 + (NH 4 ) 2 Mo0 4 +23K 2 MoO 4 +22H 2 O. This is seen to be really a direct titration of molybdic acid instead of a titration of phosphorus and it is therefore an indirect estima- tion of phosphorus and can be correct only in case the composition of the precipitate is constant. It is also essential that no free molybdic acid should be present with the phosphomolybdate. There is some difference in opinion concerning the accuracy of this method. If the precipitation is carefully performed it is probably as accurate as the gravimetric method (5). 1 J. Chem. Soc., 15, 382 (1893); 16, 278 (1894). STEEL AND ALLOYS 455 Determination by Pemberton's Method. Prepare the following reagents : (a) Acid Solution of Ammonium Molybdate. Dissolve 100 gm of molybdic acid in a mixture of 144 cc of ammonium hydroxide (specific gravity 0.90) and 271 cc of water. Pour this solution, slowly and with vigorous stirring, into a mixture of 590 cc of concentrated nitric acid (specific gravity 1.42) and 1148 cc of water. Allow to stand at a tem- perature of about 40 for several days and then decant from sediment and preserve in glass-stoppered bottles. (6) Standard Potassium Hydroxide Solution, 1 cc of which is equiva- lent to 0.1 mg of phosphorus. This should be as nearly free from carbonates as possible and is made as follows : Dissolve 2 percent more than the calculated quantity for 1000 cc, dilute to 100 cc and add 1 cc of a saturated solution of barium hydroxide. Stopper the flask and allow to stand until the precipitate of barium carbonate has settled. Decant and dilute to 1000 cc. Standardize by titration against solution (c), using phenolphthalein. Adjust so that 1 cc is equivalent to 0.1 mg of phosphorus. (c) Standard Hydrochloric Acid Solution, equivalent in concentra- tion to the standard base; use boiled water. (d) Potassium Permanganate Solution, 1.5 gm in 100 cc. (e) Potassium Nitrate Solution, 1.0 percent. Weigh 2 gm of iron or steel into a 250-cc Erlenmeyer flask and add 100 cc of nitric acid (specific gravity 1.13) and warm until the sample is dissolved (see note on page 457: ''Interference of Titanium"). Boil to expel oxides of nitrogen, then add 10 cc of solution (d) and boil until the combined carbon is completely oxidized and the excess of potassium permanganate is decomposed, as is made evident by the disappearance of the pink color. Dissolve the precipitated manganese dioxide by boiling with about 1 gm of ferrous ammonium sulphate. Add ammonium hydroxide with vigorous stirring. The last part of this operation must be conducted with care because if much ferric hydroxide is allowed to form it will not readily redissolve, even though the solution still contains an excess of acid. Redissolve the precipi- tate by the addition of the least necessary quantity of nitric acid. Place a thermometer in the flask and warm to a temperature of 60 to 65 by placing the flask in a water bath and then add 40 cc of freshly filtered ammonium molybdate solution, stir well and allow to stand 15 minutes. Filter immediately and wash flask and precipitate with solution (e) until the washings are neutral to phenolphthalein but with- out attempting to remove all precipitate from the flask. Transfer the paper and precipitate to the flask in which the precipita- tion was made and add enough standard solution of potassium hydrox- 456 QUANTITATIVE ANALYSIS ide to dissolve the precipitate. Dilute to about 75 cc with recently boiled water, add a drop of phenolphthalein and titrate the excess of base with standard acid solution. Calculate the percent of phosphorus in the steel. Method (4). This also is an indirect method for the determina- tion of phosphorus, since it also depends upon reactions of molyb- denum oxides, rather than of phosphorus. The precipitate of ammonium phosphomolybdate, obtained as in method (3), is dissolved in ammonium hydroxide, the solution is acidified with sulphuric acid and zinc is then added. Molybdenum trioxide, MoOs, is reduced to molybdenum sesquioxide Mo 2 03, which is again oxidized by titration with standard potassium perman- ganate solution. Method (5). This is one of the most reliable of air methods if carefully performed, since its accuracy does not depend, in any way, upon the composition of the yellow precipitate. A some- what less acid solution of ammonium molybdate may be used and this keeps better than the solution required for volumetric proc- esses. The yellow precipitate of ammonium phosphomolybdate is dissolved in ammonium hydroxide and the phosphorus is then precipitated as ammonium magnesium phosphate by the addition of a solution of magnesium chloride. The ammonium mag- nesium phosphate is ignited and weighed as magnesium pyro- phosphate. Potassium permanganate cannot be used for the oxidation of carbon since it would later form a precipitate of ammonium manganese phosphate. Determination. Prepare the following solutions: (a) Acid Solution of Ammonium Molybdate. Dissolve 100 gm of molybdic acid in a mixture of 144 cc of ammonium hydroxide (specific gravity 0.90) and 271 cc of water. Pour this solution slowly and with vigorous stirring into a mixture of 490 cc of concentrated nitric acid (specific gravity 1.42) and 1148 cc of water. Allow to stand at a tem- perature of 30 to 40 for several days and then decant and preserve in glass-stoppered bottles. (b) Ammonium Citrate Solution. Dissolve 50 gm of citric acid in water, add 350 cc of ammonium hydroxide (specific gravity 0.90) and dilute to 1000 cc. (c) Ammonium Hydroxide Solution containing 2.5 percent of ammonia. (d) " Magnesia Mixture " Dissolve 55 gm of crystallized magne- STEEL AND ALLOYS 457 slum chloride and 140 gm of ammonium chloride in water, add 130 cc of ammonium hydroxide (specific gravity 0.90) and dilute to 1000 cc. (e) Ammonium Nitrate Solution, 10 percent. Dissolve 1 to 2 gm of steel in 20 cc of nitric acid (specific gravity 1.2) in a casserole, cover and boil until nitrogen oxides are expelled. Evaporate to dryness on the steam bath or by agitating over a flame. Heat for 15 minutes over the direct flame in order to oxidize organic matter, formed from combined carbon. Cool, add 30 cc of concentrated hydrochloric acid and heat to dissolve iron oxide. Evaporate with stirring until ferric chloride begins to crystallize but do not allow salts to dry on the sides of the casserole. Add 10 cc of concentrated nitric acid, boil to expel chlorine, dilute to 75 cc and filter into a 250-cc flask. (If titanium is present, see note below: "Interference of Tita- nium.") Wash the silica and carbon on the paper with 2 per cent nitric acid and water until the iron is all removed, as made evident by the disappearance of brown stains. Dilute the filtrate to about 100 cc and add dilute ammonium hydrox- ide solution very slowly and with vigorous stirring until a small amount of precipitate remains undissolved. Redissolve this in concentrated nitric acid, immerse the flask in water and warm to about 60. Add 50 cc of ammonium molybdate solution (a), shake and allow to remain at a temperature of 65 for an hour. Filter and wash with solution (e) until no brown stains remain on the paper. It is not necessary to re- move all of the precipitate from the sides of the flask at this point, but it must be well washed. Place the flask in which precipitation was made under the funnel and dissolve the precipitate by adding about 25 cc of ammonium citrate solution (6). Wash the paper thoroughly with hot water. Rotate the flask until all of the precipitate is dissolved from the sides then nearly neutralize with hydrochloric acid. Transfer to a beaker, dilute to 100 cc, add 10 cc of magnesia mixture, slowly and with vigorous stirring. After the solution has stood for 30 minutes add slowly ammonium hydroxide of specific gravity 0.90 in quantity equal to 1/9 of the total volume of solution. Allow to stand for two hours, filter and wash with dilute ammonium hydroxide solution. Ignite in a platinum or porcelain crucible until white and weigh the magnesium pyrophosphate. Calculate the percent of phosphorus in the steel. Interference of Titanium. If titanium is present, as it fre- quently is in pig iron and sometimes in steel, the phosphorus will not all be recovered by any of the methods already described because the action of acids upon iron leaves an insoluble double 458 QUANTITATIVE ANALYSIS salt of phosphoric acid, titanic acid and iron. In this case the residue of silica, carbon, ferric phosphotitanate, etc., obtained by filtration of the acid solution of iron, is ignited in a platinum crucible to burn organic matter, the silica is volatilized by mois- tening with a drop of sulphuric acid and adding 1 cc of hydro- fluoric acid, and the residue is then fused with about 2 gm of sodium carbonate. Sodium phosphate, ferric oxide and sodium titanate, Na 2 TiOs, are formed. Sodium phosphate is dissolved in water and added to the principal solution of the iron in nitric acid. Sodium titanate and ferric oxide are insoluble in water. Titanium. Titanium is often present in pig iron as an impurity, being derived from the iron ores. There is now an increasing use of titanium, in the form of iron-titanium " alloys," as an agent to promote sound castings and sound steel. Its effect is to reduce oxides, combine with nitrogen and sulphur and thus to prevent blow holes and flaws by the formation of a solid oxide or nitride which enters the slag. If this action were ideal there should be no titanium remaining in the metal at the end of the process but this is not always the case and determinations of titanium may be required. Titanium is now also used to some extent as an essential constituent of finished alloy steels. It has already been stated that much of the titanium will remain as an insoluble compound with iron and phosphorus when iron is dissolved in acids. This is freed from carbon by ignition and from silica by treatment with sulphuric acid and hydrofluoric acid. The titanium in the acid solution of the sam- ple is recovered by neutralizing the excess of acid, reducing the iron to the ferrous state by sodium thiosulphate or sulphurous acid and precipitating titanic acid by boiling. Titanic acid is an irreversible colloid (see page 19) and becomes insoluble when its solution is boiled for some time. This precipitate is removed by filtration and added to the residue already in the crucible. The whole is then fused with sodium carbonate and the sodium acid titanate, NaHTiOs, and ferric oxide are separated from sodium phosphate by dissolving the latter and filtering. One of two methods of procedure may now be adopted: (a) The insoluble sodium acid tifcanate is fused with potassium pyrosulphate, forming titanic acid. Sulphuric acid and water are added, the titanic acid forming a colloidal sol. The iron also STEEL AND ALLOYS 459 passes into solution as ferric sulphate. This is reduced to the ferrous condition by sulphurous acid or ammonium acid sulphite, the solution is largely diluted and boiled, when the sol is floccu- lated, titanic acid again passing into the irreversible gel. This is separated by filtration, washed and ignited and weighed as tita- nium dioxide. (6) The sodium titanate in the crucible is dissolved in hot, dilute sulphuric acid, transferred to a color comparison tube (a Nessler cylinder or similar tube) and treated with hydrogen peroxide. Titanium is oxidized by hydrogen peroxide to the hexavalent condition and forms an intensely yellow solution. The color is compared with that produced by a standard titan- ium solution in a similar tube. Determination. Weigh from 2 to 5 gm of iron or steel and place in a casserole. Add 50 cc of concentrated hydrochloric acid, cover and warm until the metal is dissolved. Filter, wash twice with hot water, transfer to a platinum crucible and burn the paper and all carbon. Add a drop of sulphuric acid and about 3 cc of hydrofluoric acid and finally heat to expel the acids and silicon tetrafluoride. To the filtrate containing most of the iron add dilute ammonium hydroxide, slowly and with continuous stirring, until a small amount of ferric hydroxide remains undissolved. Redissolve this in hydro- chloric acid, leaving the solution with a small excess of acid. Add a. 20-percent solution of sodium thiosulphate until the red color of ferric chloride disappears and sulphur begins to precipitate. Dilute to about 400 cc, add 20 gm of sodium acetate and 50 cc of 30-percent acetic acid and boil for 15 minutes or until precipitation of titanic acid seems to be complete. Filter and wash two or three times with hot, 1-percent acetic acid and place the paper and precipitate in the crucible contain- ing the main portion of titanium. Burn the paper and carbon then add about 5 gm of sodium carbonate and thoroughly fuse. Cool, place the crucible in a beaker and cover with hot water. When the fusion is entirely disintegrated, filter and wash the sodium titanate and iron oxide with 1-percent sodium carbonate solution. Proceed by method (a) or (6), below. (a) Gravimetric Method. 1 Return the paper containing the washed residue of sodium titanate to the crucible in which the fusion was made. Add 10 gm of potassium pyrosulphate and heat gently, avoiding loss by effervescence. Gradually raise the temperature until the crucible is 1 Blair: The Chemical Analysis of Iron, 8th ed., 172. 460 QUANTITATIVE ANALYSIS finally red and keep at this temperature until all the iron oxide is dis- solved. Cool, add 15 cc of concentrated sulphuric acid and heat until the entire contents of the crucible have becomo liquid. Cool and pour, slowly and with stirring, into 400 cc of water contained in a 500-cc beaker. If basic ferric salts precipitate, redissolve in hydrochloric acid. Add 50 cc of a 20-percent solution of sodium thiosulphate. Filter if not clear, nearly neutralize with ammonium hydroxide, redissolve any precipitate that may have formed and add a clear solution contain- ing 20 gm of sodium acetate and 150 cc of 30-percent acetic acid. Boil and filter the titanic acid. Wash three times with 1-percent acetic acid, transfer the paper and precipitate to a porcelain or platinum crucible and burn the carbon, finally igniting for five minutes over the blast lamp. Weigh the titanic oxide, Ti0 2 , and calculate the percent of titanium. (b) Colorimetric Method. Dissolve the residue in the 4 crucible by heating with dilute sulphuric acid, place the filter paper in a beaker and pour the sulphuric acid upon it. Heat until the sodium titanate is dissolved then remove the paper, rinse thoroughly and rinse the contents of the crucible into the beaker. Transfer to a Nessler tube, filtering if not clear, and dilute to the mark. Prepare a standard solution of titanic acid as follows: Ignite 1 gm of the purest obtainable titanic acid in a weighed platinum crucible, cool and weigh. Dissolve the titanic acid in dilute sulphuric acid, rinse into a 1000-cc volumetric flask and dilute to the mark. Mix well, transfer to a dry glass-stop- pered bottle and record the concentration of the solution. Into four similar tubes, having the same capacity and the mark at the same height as the first tube, containing the titanium from the sample, place 1 cc, 3 cc, 5 cc and 10 cc, respectively, of the standard titanium selu- tion and dilute these to the mark. Add to each of the five tubes so prepared 5 cc of hydrogen peroxide. Compare the color of the tube containing the sample with that of the four tubes of standard, looking vertically downward through the tubes toward a white surface, placed near a window. As a result of these comparisons, limits will be found for the concentration of the sample tube. Prepare other tubes of solutions whose concentrations lie between these limits, until an equality of intensity is obtained. Calculate the percent of titanium in the sample. Manganese. Manganese is found in certain quantities in practically all iron and steel. At least traces of this metal are derived from iron ores while larger quantities are intentionally added, either to correct the undesirable effects of other elements rvr -fry 01 STEEL AND ALLOYS 461 or to add desirable properties of its own. It has already been stated that manganese overcomes the tendency of sulphur to render steel " red-short." Manganese also has an effect upon carbon, exactly the opposite of that of silicon, which is to increase the formation of graphitic, or free, carbon. Manganese, on the other hand, increases the tendency of carbon to remain combined with iron as the carbide, Fe 3 C. With cast iron it thus favors the formation, of "white" iron. Part of the carbon also combines with manganese to form a carbide, Mn 3 C, which is very hard and brittle. On this account the addition of manganese to steel in quantities above 0.50 percent renders the steel increasingly hard. Such steel is frequently used for apparatus that must resist abrasion, such as ore crushers and grinders. Several excellent methods are in use for the determination of manganese in iron and steel. The underlying principles of the most important of these will be discussed and details will be given for some. All methods involve (a) the separation of the manganese from the large excess of iron and (b) the determi- nation of the separated manganese. Bismuthate Method. Schneider 1 discovered the reaction upon which this method is based and the details of the method have since been modified. The method is now recognized as one of the most accurate and easily applied of all methods now in use. It was adopted as a standard method in 1907 by the Committee on Standard Methods for the Analysis of Iron, of the American Foundrymen's Association. 2 This method is based upon the oxidation of bivalent manganese to heptavalent manganese by sodium bismuthate. The solution of pig iron or steel in an acid contains manganese as a manganous salt. This is oxidized to sodium permanganate by sodium bis- muthate, the bismuth being reduced to the trivalent condition. Sodium bismuthate is derived from bismuth pentoxide and is a salt of the hypothetical acid HBiO 3 . The reaction between sodium bismuthate and manganous nitrate may be represented thus: 5NaBiO3+.2Mn(NO 3 )2+14HN03->2NaMn04+5Bi(NO3)3+ 3NaN0 3 +7H 2 0. 1 Dingl. polyt. J., 269, 224 (1888); Monatsh. 9, 242 (1888) 2 J. Am. Chem. Soc., 29, 1372 (1907). 462 QUANTITATIVE ANALYSIS The method usually involves a solution in nitric acid. Metzger and McCrackan 1 proposed a sulphuric acid solution, the reaction taking place as had already been shown by Schneider. Determination. Prepare a solution of potassium permanganate, 1 cc of which contains 1 mg of manganese. Standardize against ferrous ammonium sulphate by the method given on page 250. Prepare also a solution of ferrous ammonium sulphate or sodium arsenite (page 451) in recently boiled water equivalent in concentration to the permanganate solution, the former containing 50 cc of concen- trated sulphuric acid in each 1000 cc. This solution will slowly oxidize, even if stoppered when not in use, and it will be necessary to obtain its relation to the permanganate solution at the time the determination of manganese is made, by a blank determination. For pig iron, which contains much silicon and free carbon, weigh 1 gm of drilled or crushed sample, dissolve in 50 cc of nitric acid of specific gravity 1.13 and filter, receiving the filtrate in a 250-cc flask. Wash the paper and residue with dilute nitric acid until free from iron. For steel, weigh 1 gm of drillings, placing in a 250-cc flask, and dissolve in 50 cc of nitric acid of specific gravity 1.13. For steels with excep- tionally high manganese content modify the sample weight accord- ingly. For either pig iron or steel proceed with the clear solution as follows : Add about 0.5 gm of sodium bismuthate which is free from manganese. Heat until the permanganate, which is formed at first, is decomposed by nitric acid and the pink color disappears. This insures the oxida- tion of organic matter from combined carbon. Add enough powdered ferrous ammonium sulphate to redissolve any precipitated manganese dioxide and boil until all nitrogen oxides are expelled. Cool and add 0.5 gm of sodium bismuthate and agitate. Add 50 cc of 3-percent nitric acid and filter through an ignited asbestos filter in a Gooch crucible or a carbon tube, washing the excess of sodium bismuthate with 50 cc of 3-percent nitric acid. Immediately add from a burette the ferrous ammonium sulphate or sodium arsenite solution until the permanganate is reduced and an excess of the ferrous salt is present. Titrate this excess by means of standard potassium permanganate. The ratio of concentrations of reducing solution and potassium permanganate solution should be determined each day, or more often if it changes rapidly. A direct titration should not be made because the presence of sodium bismuthate exerts a slight disturbing effect due to partial oxidation of iron. In order to correct this error the blank i J. Am. Chem. Soc., 32, 1250 (1910). STEEL AND ALLOYS 463 determination is made by treating the solutions as they are treated in the determination of manganese. Measure 25 cc of nitric acid having the same concentration as the acid used in dissolving the sample and 50 cc of 3-percent nitric acid. Add 0.25 gm of sodium bismuthate. Heat to boiling, cool and filter through asbestos into a 250-cc flask, washing with 50 cc of 3-percent nitric acid. Add 35 cc, accurately measured, of ferrous ammonium sulphate or sodium arsenite solution and titrate immediately with standard potassium permanganate solution. Calculate the number of cubic centimeters of potassium permanganate solution equivalent to 1 cc of reducing solution and record this as the value of the secondary standard. Calculate the percent of manganese in the sample. Ford's Method. * The separation of manganese from most of the iron is accomplished, in this method, by precipitating the former metal from an acid solution, using nitric acid and potas- sium chlorate. The nitric acid must be quite free from nitrous acid as the latter will redissolve manganese dioxide: MnO 2 +HN0 2 +HNO3-+Mn(NO 3 )2+H 2 0. Manganese is oxidized to the dioxide and precipitates, carrying a small amount of iron with it. The precipitate is filtered on asbes- tos that has been washed with acids and ignited, is washed and dissolved in sulphurous acid or ammonium acid sulphite and the excess of sulphur dioxide is removed by boiling. The iron is reoxidized by bromine and is then precipitated as basic acetate by boiling with acetic acid or ammonium acetate and water. In the filtrate from this precipitate the manganese is precipitated as manganese ammonium phosphate and ignited to the form of manganese pyro phosphate by a method described on page 114. The Ford-Williams Method. Williams proposed 2 the precipi- tation of manganese dioxide by Ford's method, following this by the volumetric determination of the manganese by adding a measured excess of ferrous ammonium sulphate or sodium arsenite and titrating the excess by means of standard potassium permanganate solution. The time necessary for a complete determination is thereby much shortened and the method is 1 Trans. Am. Inst. Min. Eng., 9, 397 (1879). 2 Ibid., 10, 100 (1880). 464 QUANTITATIVE ANALYSIS fully as accurate as Ford's method. In neither method is the iron precipitated, so that large samples may be used when the percent of manganese is low. The reaction between manganese dioxide and ferrous sulphate is expressed by the following equation: MnO 2 +2FeSO4+2H 2 S0 4 ->MnS04+Fe 2 (S0 4 ) 3 +2H 2 0. An objection to both the Ford and the Ford- Williams methods is in the difficulty that is experienced in filtering the manganese dioxide when the steel or iron contains much silicon. A. P. Ford and Bregowsky showed 1 that the addition of a few drops of hydrofluoric acid to the solution before filtering eliminated the silica without materially injuring the beaker. If this is not done it is necessary to evaporate such solutions to dryness with hydrochloric acid in order to render silica insoluble. Determination. Prepare the following solutions: (a) Potassium Permanganate Solution, 1 cc of which is equivalent to 0.001 gm of manganese by the reaction just given and assuming that the excess of ferrous sulphate is to be titrated by potassium permanganate. (6) Ferrous Ammonium Sulphate Solution, approximately equivalent in concentration to the potassium permanganate solution and contain- ing 50 cc of concentrated sulphuric acid in each 1000 cc. Dissolve about 5 gm of sample in a 250-cc beaker in 75 cc of nitric acid of specific gravity 1.2. Evaporate until the solution becomes viscous then add 75 cc of concentrated nitric acid and 5 gm of potassium chlorate. The nitric acid must be free from nitrous acid as indicated by the absence of a brown coloration. If not perfectly colorless, draw a current of air through the acid until all oxides are removed. After the addition of concentrated nitric acid and potassium chlo- rate, boil the solution for 15 minutes, remove the flame and add about 5 or 6 drops of hydrofluoric acid, dropping it near the center of the beaker and mixing it well with the solution at once. Boil for 10 minutes to expel silicon tetrafluoride and the excess of hydrofluoric acid, then add 1 gm of potassium chlorate and boil again until chlorine oxides are no longer evolved. Filter on a pad of acid-washed and ignited asbestos and wash two or three times with concentrated nitric acid which is free from nitrous acid. The precipitate need not all be removed from the beaker but all must be washed. Remove as much acid as possible 1 J. Am. Chem. Soc., 20, 504 (1898). STEEL AND ALLOYS 465 from the filter by suction, then wash the beaker and residue on the filter with cold water until the washings are free from acid. Remove the asbestos pad and manganese dioxide to the beaker in which the latter was precipitated, wiping the interior of the Gooch cru- cible or filtering tube with a tuft of asbestos. Measure into the beaker enough ferrous ammonium sulphate or sodium arsenite solution to dis- solve completely the manganese dioxide and stir until solution is complete. Titrate at once with standard potassium permanganate solution. Measure 35 cc of standard reducing solution into another beaker and titrate with standard potassium permanganate solution. Calcu- late the number of cubic centimeters of potassium permanganate solution equivalent to 1 cc of the standard reducing solution and record this as the value of the secondary standard. Calculate the percent of manganese in the sample. Volhard's Method. This method is discussed on page 253. Applied to iron and steel it is somewhat difficult of execution because of the large amount of ferric hydroxide that is produced when zinc oxide is added. Either the bismuthate or the Ford- Williams method is to be preferred to it. Acetate Method. This title really applies only to the sepa- ration of manganese and iron and it is, in practice, followed by any one of several methods of determination. In the discussion of the acetate method for separating iron and phosphorus (page 452) it was noticed that the object is to precipi- tate the phosphorus as ferric phosphate, along with the least- possible excess of basic ferric acetate, leaving the greater portion of the iron in solution in the ferrous condition. In the acetate method for manganese separation the object is to precipitate all of the iron as basic ferric acetate, leaving the manganese in solu- tion to be subsequently determined as manganese pyrophosphate or manganese tetroxide, or by one of the volumetric processes. The method is not applicable to the use of more than 1 gm of sample because of the difficulty that is experienced in the filtra- tion of the large quantity of colloidal basic ferric acetate and in washing this precipitate free from manganese. Walters' Method.-^-Marshall showed 1 that ammonium per- sulphate, in presence of silver nitrate, oxidizes manganese from 1 Chem. News, 83, 76 (1901). 30 466 QUANTITATIVE ANALYSIS the bivalent to the heptavalent condition, thus producing per- manganates from manganese salts. His interpretation of the reaction is as follows: (NH 4 )2S208+2AgN0 3 -Ag 2 S 2 O 8 +2NH 4 NO3, Ag 2 S 2 8 +2H 2 O-*2H 2 S0 4 +Ag 2 O2. According to this the silver peroxide, formed momentarily in small amount, is responsible for the oxidizing action: 5A g2 2 +2Mn(N0 3 )2+6HN0 3 _,2HMn04+10AgN03+2H 2 0. Walters applied these reactions to the quantitative determina- tion of manganese in iron and steel. 1 The sample is dissolved in nitric acid, silver nitrate and ammonium persulphate are added and the intensity of color is compared with that produced by a standard steel in which the manganese has been determined by another method. The relative volumes required to produce the same intensity of color in the two provide a basis for the calcula- tion of the percent of manganese in the sample. Determination. Weigh 0.2 gm of the sample and the same amount of a standard steel of known manganese content, placing in different test tubes having a capacity of 50 cc. Add to each 10 cc of nitric acid (specific gravity 1.2) and immerse the tubes in hot water until solution is complete and brown oxides of nitrogen are expelled. Add 15 cc of a solution containing 0.02 gm of silver nitrate. Immediately add 1 gm of ammonium persulphate and continue warming until the pink tint of permanganic acid is fully developed, which will require about 1 minute. Remove the tubes from the bath while oxygen is still being evolved and place in cold water. When the solutions are cool rinse into 50-cc volumetric flasks, dilute to the mark and mix. If free carbon is present allow it to settle, then pour the solution whose color is less intense into a color comparison tube (a Nessler or similar tube). Fill a burette with the other solution and measure this into a second tube until the color has the same intensity, viewed from above when placed over a white surface near a window. From the relative volumes of the two solutions calculate the percent of manganese in the sample. Peter's Method. 2 This is an older method than that of Wal- ters and is also a colorimetric method, the manganese being oxi- 1 Chem. News, 84, 239 (1901). 2 Ibid., 33, 35 (1876). STEEL AND ALLOYS 467 dized to permanganic acid by lead peroxide in presence of nitric acid. It is necessary to remove the excess of lead peroxide before comparing the colors and this constitutes the greatest objection to the method. The reaction is expressed as follows : 5Pb0 2 +2Mn(N0 3 )2+6HN0 3 ->5Pb(N0 3 ) 2 +2HMn04+2H 2 0. Deshay's Method. Manganese is here oxidized by lead per- oxide, as in Peter's Method, followed by titration of the perman- ganic acid by a standard solution of sodium arsenite. 5Na 3 As0 3 +2KMnO4+6HNO 3 ->5Na3As04+2Mn(NO3) 2 + 3H 2 O+2KNO 3 . Moore and Miller suggested 1 the separation of iron and man- ganese by precipitating iron as ferric hydroxide by the addition of pyridine. The separation is quantitative, although it has not yet been applied to iron and steel analysis. Tungsten. When tungsten is alloyed with iron in steel it has the effect of retarding the change from hard to annealed steel. If sufficient tungsten is present (3 or 4 percent or more) the change to soft steel is almost entirely prevented. "Self -hard- ening" steels, invented by Mushet and often given his name, con- tain 4 to 12 percent of tungsten, 2 to 4 percent of manganese and 1.50 to 2.50 percent of carbon. They are called self -harden- ing because, when subjected to the ordinary process of slow cool- ing for annealing, they retain their hardness, even though they have been cooled from a temperature near the melting-point. Taylor and White, in 1906, developed a process for imparting a remarkable degree of toughness to self-hardening steels, so that they can be used for steel-cutting tools that are used for such rapid cutting that they become red hot and yet do not lose their hardness or toughness. Such steels are known as " high-speed tool steels." The carbon percent is usually less than 0.75 and they contain from 5 to 25 percent of tungsten, 2 to 10 percent of chromium and less than 0.50 percent of manganese. The ther- mal treatment of " high-speed " steels consists of heating to near the melting-point and then cooling in a blast of air. The determination of tungsten in steel must include separation from iron, silicon, carbon, phosphorus and usually chromium, * J. Am. Chem. Soc., 30, 593 (1908). 468 QUANTIT ATIVE ANALYSIS since the latter metal is now generally associated with tungsten in tool steels. A gravimetric method is usually employed, tung- sten being weighed as tungstic oxide, WOs. The details of the following method are mainly those given by Johnson. 1 Determination. Dissolve 2 gm of sample in 30 cc of sulphuric acid (1 : 3) in a 250-cc beaker or casserole, heating to hasten solution. Add 60 cc of nitric acid (specific gravity 1.2) and digest at a temperature near the boiling-point until the residue is yellow and is free from black particles. Filter and wash free from iron by means of dilute sulphuric acid. The filter paper now contains the main portion of the impure tungstic acid. Transfer to a weighed platinum crucible and burn the paper at a low temperature. Recover the tungsten from the filtrate by precipitating by means of cinchonine solution (25 gm of cinchonine in 200 cc o 1 : 1 hydro- chloric acid). Filter and wash free from iron by means of the cincho- nine solution. Burn the paper in the crucible containing the main portion of tungstic oxide and weigh, then add potassium acid sulphate to the extent of about twenty times the weight of residue in the crucible. Heat, gradually at first, finally raising the temperature until the cru- cible is dull red and keep at this temperature until yellow particles of tungstic oxide are dissolved. Silica will remain undissolved. Cool and place the crucible in a 250-cc beaker containing 100 cc of 10-percent ammonium carbonate solution and warm until disintegration of the mass is complete. Filter and wash the residue with 1-percent ammonium carbonate solution until the washings are free from sulphates. Ignite the paper in the same crucible as that used first and weigh. This gives the weight of ferric oxide, chromium oxide and silica and this subtracted from the weight of the total residue gives the weight of tungstic oxide, WOs. .Calculate the percent of tungsten in the sample. Chromium and Nickel. Alloy steels containing nickel and chromium, also chromium and tungsten, are now of considerable commercial importance. Both nickel and chromium increase the hardness, tensile strength and elastic limit of steel and de- crease the ductility but slightly, if at all. Nickel also lowers the temperature at which quenched steel is softened by slow cooling. In commercial nickel steels the nickel is not often present to the 1 Chemical Analysis of Special Steels, Steel-Making Alloys and Graphite, 64. STEEL AND ALLOYS 469 extent of more than 3.50 percent, the carbon content being not more than 0.50 percent. Chrome steels usually contain not more than 3 percent of chromium and less than 1 percent of carbon. Various combinations of chromium, nickel, carbon, and iron produce chrome-nickel steels of great strength and harden- ing power. Chromium is found in many steels, with or without nickel, vanadium or tungsten, and particularly in high-speed tool steels, in which it is added in quantities as high as ten percent. Cain's method 1 for the separation of chromium and for the determina- tion of the former involves precipitation of chromium as hydrox- ide, by the addition of barium carbonate to a slightly acid solu- tion in which all of the iron is in the ferrous condition. Under these circumstances the small concentration of hydroxyl ions, produced by hydrolysis of the slightly soluble barium carbonate, is sufficient to exceed the solubility product with the chromium ion but not with the ferrous ion, although ferric hydroxide would precipitate if iron were allowed to oxidize. The precipitated chromium hydroxide, accompanied by the slight excess of barium carbonate, is oxidized to sodium chromate by fusion with sodium carbonate and potassium nitrate, the mass later being dissolved in hot water and treated with hydrogen peroxide to complete the oxidation. From this solution the chromium is then precipitated as lead chromate. This is re- moved and dissolved in hydrochloric acid and chromium is titrated by standard ferrous ammonium sulphate solution. Determination. In a covered 250-cc Erlenmeyer flask dissolve an amount of steel that will give 0.06 to 0.07 gm of chromium, using about 10 cc of concentrated hydrochloric acid for each gram of steel. As soon as solution is complete dilute to 100 cc with hot water, nearly neutralize with saturated sodium carbonate solution and add a sus- pension of barium carbonate in slight excess. Boil vigorously for 15 minutes, adding small amounts of barium carbonate suspension every two or three minutes, keeping the flask covered meanwhile. About 1 gm excess of barium carbonate is -all that should be present at the last. The precipitate will be green, with a slight admixture of white barium carbonate, but little or no brown ferric hydroxide should be in evidence. Remove from the source of heat and as soon as the precipitate has 1 J. Ind Eng. Chem., 4, 17 (1912). 470 QUANTITATIVE ANALYSIS settled, filter rapidly and wash twice with hot water. Remove as much water as possible by suction then place the paper and precipitate in a platinum crucible and carefully burn the paper. Add 2 gm of sodium carbonate and 0.25 gm of potassium nitrate and fuse for 20 minutes. Support the inverted and inclined crucible on a glass triangle which has glass legs about 2 cm long fused to the corners, place in a 250-cc beaker, cover with boiling water and digest until the mass has com- pletely disintegrated. Filter into a flask, add 2 cc of hydrogen peroxide to the filtrate and boil for 10 minutes. The excess of hydrogen peroxide is thus expelled. Cool and transfer to a 250-cc separatory funnel. Add nitric acid, specific gravity 1.2, slightly more than enough to decompose the carbonate, allowing the carbon dioxide to escape by inverting the separatory funnel and opening the stop cock, shaking vigorously at the last. Trans- fer to a 250-cc beaker, barely neutralize with sodium hydroxide solu- tion, then add nitric acid, specific gravity 1.2, 2 cc for^each 100 cc of solution. Add 20 cc of 20 percent lead acetate solution to the cold chromate solution, stirring vigorously. Filter off the lead chromate on a Gooch crucible and wash three or four times with cold water. Transfer the asbestos and all of the pre- cipitate to a 400-cc Erlenmeyer flask and add enough hot dilute hydro- chloric acid to decompose all of the precipitate. Cool, dilute to 150 cc with recently boiled and cooled water and add from a pipette exactly 50 cc of approximately tenth-normal ferrous ammonium sulphate solu- tion. Immediately titrate the excess of ferrous salt, using a standard solution of potassium dichromate and potassium ferricyanide as an outside indicator (see page 258) or standard potassium permanganate without an additional indicator. Either a tenth-normal solution or a standard solution, 1 cc of which is equivalent to 0.002 gm of chromium is suitable for this purpose. At the same time make a blank determination by titrating 25 cc of the ferrous solution against the standard oxidizing solution. Sub- tract the volume of the latter as used in the chromium determination from twice the volume used in the blank. From the remainder calculate the percent of chromium in the steel. Glyoxime Method for Nickel. 1 This method, now almost universally considered the best and most accurate for the determi- nation of nickel, is based upon the extremely slight solubility of iTschugaeff: Ber., 38, 2520 (1905); Brunck: Z. angew. Chem., 20, 1844 (1907); Ibbotson: Chem News, 104, 224 (1911); Bogoluboff: Stahl u. Eisen (1910), 458. STEEL AND ALLOYS 471 the nickel salt of dimethylglyoxime in ammoniacal alcohol. The reaction is represented thus: H 3 C v /CH 3 H 3 C v /CH 3 H 3 C V /CH 3 + NiCl 2 -+ NOH HON Iron, copper, cobalt, chromium, manganese, tungsten, vana- dium, etc., the metals commonly found in steels, do not interfere since tartaric acid is added to prevent the precipitation of their hydroxides. Determination. Prepare the following solutions: (a) Dimethylglyoxime. A clear solution of 1 gm in 100 cc of 98 percent alcohol. (6) Tartaric Acid. A clear solution of 50 gm in 100 cc of water. Use 1 gm of ordinary nickel steel. If less than 3 percent or more than 5 percent of nickel is present modify the sample weight accordingly. Dissolve in a covered casserole in 10 cc of nitric acid, specific gravity 1.2. Remove the cover and evaporate to dryness, then heat carefully to decompose nitrates. When fumes are no longer evolved cool and add 20 cc of concentrated hydrochloric acid, warming until solution is complete. Evaporate to dryness but do not heat strongly. Redis- solve by warming with 5 cc of dilute hydrochloric acid, dilute and filter to remove silica. (This precipitate may be used, if desired, for the determination of silica.) Wash the paper with hot water containing a small amount of hydrochloric acid. To the filtrate add 14 cc of tartaric acid solution (b) for each gram of steel used. Make the solution slightly basic by the addition of dilute ammonium hydroxide. About 20 to 30 cc will be required. Any precipitate of basic ferric tartrate should be dissolved by adding more ammonium hydroxide. If the solution is still turbid when distinctly basic more tartaric acid should be added. Now make the solution slightly acid with hydrochloric acid, heat nearly to boiling and add 20 cc of dimethylglyoxime solution (a) and then make very faintly basic with dilute ammonium hydroxide. The odor of ammonia should be faint after blowing away the vapors above the solution. The nickel salt precipitates at once in the form of fine red needles. Allow the covered solution to stand for an hour over a steam radiator or in a warm place, then test for completion of precipitation by adding 10 cc more of glyoxime solution. Filter on a weighed Gooch or alundum crucible which has been dried at 120, wash free from iron with hot water and dry to constant weight at 120. Calculate the percent of nickel in the steel. 472 QUANTITATIVE ANALYSIS Ether Method. The separation of nickel and chromium from iron may be conveniently made by the ether method of Rothe. 1 This is based upon the fact that from a solution in hydrochloric acid having a specific gravity between 1.100 and 1.105 (containing 21 to 22 percent of acid) ether will extract all but traces of ferric chloride, leaving chlorides of chromium, nickel, copper, manganese, aluminium and cobalt in the water solution. The apparatus shown in Fig. 97 may be used for the separation. The manipulation of the apparatus is described below. After the separation of iron the solution in water is boiled to remove dissolved ether, after which several methods are available for the determination of the metals in this solution. The method described below is probably as easy of execution and as accurate as any of these. In this method chromium is precipitated as chromium hydroxide which is ignited and weighed as chromium sesquioxide, Cr 2 O3. Nickel is deposited electrolytically. If copper is present it may also be separated electrolytically before nickel is deposited. Determination. The quantity of steel to be taken for analysis will depend upon the percents of nickel and chromium present. If 1 to 3 per- cent of either metal is contained in the steel about 2.5 gm will be sufficient. If one metal is present in about this proportion and the other in but traces it may be necessary to use two samples, making the determination of the metal whose percent is small upon a larger sample. Weigh the proper quantity of sample and place in a casserole. Dis- solve in 30 cc of concentrated hydrochloric acid and 10 cc of concen- trated nitric acid, adding the latter acid cautiously. Evaporate to dry- ness and heat for a short time then redissolve in dilute hydrochloric acid and filter to remove silica. Evaporate until the liquid thickens, due to the separation of ferric chloride crystals. This gives an acid of approxi- mately correct composition for the ether separation. Transfer the solu- 1 Mitt.kgl. tech. Versuchs. (1892), 132; J. Soc. Chem. Ind., 11, 940 (1892). FIG. 97. Appa- ratus for separation of iron from other metals by solution in ether. STEEL AND ALLOYS 473 tion to bulb A of the apparatus shown in Fig. 97, the lower cock being turned to close both bulbs; rinse the casserole with hydrochloric acid having a specific gravity 1.100, until the bulb is nearly half full of solu- tion. Nearly fill the bulb with ether, close the upper cock and mix gradually by shaking, cooling under the tap to avoid rise in temperature, which would qause reduction of ferric chloride by ether. Allow the apparatus to stand until perfect separation of ether and water has occurred, then carefully turn the lower cock to establish communica- tion between the bulbs, allowing the lower water solution to run into bulb B. Introduce about 10 cc more of hydrochloric acid into A, mix and separate as before. Turn the lower cock to allow the aqueous solution in B to run into a 250-cc beaker. Boil the solution until the odor of ether has disappeared. Add a few drops of bromine to oxidize iron and manganese, make slightly basic with ammonium hydroxide and boil. The precipitate consists of a small amount of ferric hydroxide together with chromium hydroxide, aluminium hydroxide and manganese peroxide, but it is diffi- cult to wash it free from the soluble nickel and copper salts. To make the separation complete, redissolve the precipitate in hydrochloric acid and reprecipitate by means of bromine and ammonium hydroxide. Set aside the solution for the determination of nickel. Chromium. Transfer the paper and precipitate to a platinum cru- cible and burn the paper. Add 5 gm of sodium carbonate and 0.5 gm of potassium nitrate and fuse until effervescence occurs. Place the cru- cible on its side in a beaker and cover with water. Heat until the mass is disintegrated, filter and wash with 1-percent sodium carbonate solu- tion. The residue consists of ferric oxide, some manganese oxide and aluminium oxide and it is discarded. The solution contains sodium chromate and potassium chromate and some sodium manganate. Evaporate the solution nearly to dryness after adding 5 gm of potas- sium nitrate and enough ammonium hydroxide to give a distinct odor. Dilute to 100 cc and filter, thus removing aluminium hydroxide and man- ganese dioxide. Boil the filtrate to remove ammonia then add a slight excess of hydrochloric acid and 5 gm of sodium acid sulphite. This reduces sodium chromate to chromium chloride. Boil to remove all sulphur dioxide, add a slight excess of ammonia, boil for one minute and filter the precipitate of chromium hydroxide. Wash with water until free from chlorides, burn the paper, ignite the precipitate and weigh as chromium sesquioxide, Cr 2 3 . Calculate the percent of chromium in the sample. Nickel. The solution obtained above contains nickel and copper if these were present in the steel Add 10 cc of concentrated sulphuric acid and evaporate until the heavy fumes of sulphuric acid appear, 474 QUANTITATIVE ANALYSIS thus removing chlorides. Cool, dilute and deposit the copper and nickel by means of the current, using the method described in connection with the analysis of a nickel coin, page 163. If copper is known to be absent the solution may be made basic with ammonium hydroxide im- mediately after evaporating with sulphuric acid and diluting, the nickel then being deposited at once. Calculate the percent of copper and of nickel in the steel. Vanadium. Small amounts of this element are added to many plain carbon, as well as alloy steels. Vanadium removes dis- solved gases from the steel and thus promotes the formation of sound ingots and castings. Most of the vanadium which acts in this manner is removed in the slag but small quantities (usu- ally not more than 1 percent) remain in the steel and serve to raise the elastic limit and ductility, particularly in steels contain- ing chromium. Vanadium may be titrated by standard potassium permanga- nate solution, with resulting oxidation from vanadyl salts, such as vanadyl sulphate, VOS04, derived from the tetroxide, to van- adic acid, H 3 V04, derived from the pentoxide. It is necessary first to separate the vanadium from iron, chromium and manga- nese. The separation from most of the iron can be accomplished by a method analogous to that used for separating chromium from iron, already described for the chromium determination, where barium carbonate was used to precipitate chromium hydroxide from a solution containing all of the iron in the fer- rous condition. In the separation of vanadium from iron cad- mium carbonate is used instead of barium carbonate, on account of the desirability of using a sulphuric acid solution of the steel. The precipitate of hydroxides of chromium and vanadium, together with the excess of suspended cadmium carbonate, is dissolved in sulphuric acid and the cadmium is precipitated by hydrogen sulphide. From the solution containing vanadium, chromium and perhaps a little iron, the latter two metals are pre- cipitated by electrolysis, using a mercury cathode. The vanadium is then reduced by sulphur dioxide and titrated by standard potassium permanganate solution. The equivalent weight of vanadium is calculated from the equation 5VOS04+KMn04+HH 2 0-+5H 3 VO4+KHSO 4 +MnS04 + 3H 2 S0 4 . STEEL. AND ALLOYS 475 Cain's method 1 is based upon the principles just discussed. The apparatus used by Cain for the electrolysis of the solutions is shown in Fig. 98. The bulb is filled with mercury to within 2 mm of the top of tube A, this tube being completely filled. Instead of this apparatus an ordinary beaker or cylinder, with a platinum wire fused into the bottom, may be used. A rotating anode is employed. Determination. Prepare a fiftieth- normal solution of potassium permanga- nate. Standardize against sodium oxalate or ferrous ammonium sulphate and calcu- late the vanadium equivalent. Or stan- dardize against a standard vanadium steel in which the vanadium content is accurately known, using the method for the vanadium determination as given below. Weigh 5 gm of steel into a 300-cc Erlen- meyer flask and dissolve in 50 cc of 10 percent sulphuric acid, keeping the flask covered. If insoluble matter remains filter rapidly into a flask and wash with hot water two or three times. Stopper the flask to prevent oxidation of the ferrous sulphate. Burn the paper in a platinum crucible, add 1 gm of potassium pyrosulphate or sodium pyrosulphate and fuse over an ordinary burner for 5 minutes. Cool, dissolve the fusion in hot water and add to the main solution. Nearly neutralize with a saturated solution of sodium carbonate, then add finely powdered cadmium carbonate in small portions at intervals of 4 or 5 minutes, boiling between times and keeping the flask covered. 15 to 20 minutes boiling is usually sufficient and a gram or two of cadmium carbonate should finally remain. The precipitate should contain very little ferric hydroxide. Filter rapidly and wash the flask and precipitate twice with hot water, without attempting to remove the precipitate adhering to the flask. Discard the filtrate, if clear, and place the precipitating flask under the funnel. 1 J. Ind. Eng Chem., 3, 476 (1911). FIG. 98. Cain's apparatus for electrode position. 476 QUANTITATIVE ANALYSIS Dissolve the precipitate on the filter in the minimum quantity of nearly boiling 10 percent sulphuric acid and wash the paper with ho't water. Boil the solution in the flask until all precipitate is dissolved from the sides of the flask, then cool and nearly neutralize with dilute ammonium hydroxide. The faintest excess of acid should remain to prevent precipitation of iron hydroxide by boiling. Pass a rapid current of hydrogen sulphide through the boiling solution. When the cadmium sulphide is all precipitated, filter and wash two or three times with hot water. Concentrate, if necessary, to 60 or 70 cc and rinse into the electrolyzing vessel, which should contain about 200 gm of mercury. Electrolyze, using a rotating anode, at a pressure of 6 to 7 volts. After 15 or 20 minutes test a few drops of the solution on a white plate with potassium ferricyanide solution. When no test for iron is given the chromium also is tolerably certain to be removed, unless an unusually large amount is present. , When all chromium and iron have .been deposited, stop the rotation of the anode and lower it to within a very short distance of the mercury. Remove the solution through the stopcock of the apparatus shown in Fig. 98 or by any convenient siphon or suction arrangement if a beaker or cylinder has been used. When the surface of the remaining solution is barely above the anode and while the current is still passing, rinse down the sides of the vessel and continue to add water fast enough to keep the anode covered while the solution is being removed. As the electrolytes become more diluted the resistance of the solution rises. When the ammeter indicates no current the washing may be considered as complete. To the solution containing vanadyl sulphate add 5 cc of dilute sul- phuric acid, heat to 75 and add permanganate solution from a pipette until a slight excess is indicated by the color. Pass a current of sulphur dioxide into the boiling solution for 5 minutes, then pass a current of carbon dioxide through until sulphur dioxide is expelled. Filter, if necessary, through a Gooch filter and wash with hot water. Cool to 75 and titrate against a fiftieth-normal solution of potassium permanga- nate. Calculate the percent of vanadium in the steel. Oxygen. Steel in the liquid state dissolves considerable quantities of oxygen and nitrogen from the air, as well as hydro- gen from dissociated water vapor. Upon cooling these gases are released from the solution and give rise to flaws known as "blow holes." Oxygen, however, chemically combines with the iron as it cools, forming ferrous and ferric oxides and it is in this way retained. This is even more objectionable than nitrogen or STEEL AND ALLOYS 477 hydrogen because the oxides of iron render the steel brittle and they also form gases when the steel is heated, by combination with carbon of the steel. Oxygen may be determined by Ledebur's method. 1 This consists in heating the finely divided sample in a current of pure, dry hydrogen, absorbing and weighing the water produced. Several important sources of error must be avoided. The hydro- gen must be absolutely free from water, carbon dioxide, hydrogen sulphide, and oxygen. Traces of the latter gas are removed by passing through a preliminary heating tube containing platinized asbestos, then through a water absorbent. Carbon dioxide and hydrogen sulphide are removed by passing through potassium hydroxide solution. Moisture is also absorbed before the gas enters the tube containing the sample. The sample must be quite free from oil and should be taken by very slowly drilling or milling. If the metal is heated by cutting it will be superficially oxidized. The sample is placed in a combustion tube for heating with hydrogen. A second furnace may be used for the preliminary heating of the hydrogen or the tube containing platinized asbes- tos may be placed in the furnace containing the main tube. For the sample a silica tube 3/4 inch by 30 inches is suitable. For the preliminary heating a silica tube 1/4 inch by 12 inches will serve unless it is to be placed in the furnace with the larger tube, in which case its length must be the same as that of the larger tube. The apparatus is set up in the following order: (1) Hydrogen generator of the Kipp or similar type, charged with zinc and hydrochloric acid (1:1). (2) Absorption bottle half filled with 33-percent potassium hydroxide solution. (3) Absorption bottle half filled with concentrated sulphuric acid. (4) Silica tube, 1/4 inch in diameter, containing platinized asbestos for 6 inches of its length. (5) U-tube filled with dry calcium chloride. (6) Silica tube, 3/4 inch in diameter, to hold the boat containing the sample. (7) Two U-tubes filled with dry calcium ciiloride, the first (a) for absorption of the water produced by the combina- tion of the oxygen of the sample with hydrogen, the second (b) to ^eitfaden fur Eisenhuttenlaboratorien, 6th ed., 122; Stahl u. Eisen, 2, 193 (1882). 478 QUANTITATIVE ANALYSIS act as a guard tube. (8) An aspirator similar to that used for carbon dioxide determinations. Prepare another U-tube (9) containing calcium chloride, close the side. tubes and do not con- nect in the apparatus. For the directions for filling absorption tubes and setting up the apparatus, see the determination of carbon dioxide in carbonates, page 129. Determination. Set up the apparatus as already indicated but do not connect the aspirator. Pass a rapid stream of hydrogen through the apparatus for 30 minutes to displace all of the oxygen, then weigh and insert the absorption tube (7 a) and heat the tubes to bright redness. Continue the passage of gas for 30 minutes then remove the tube (7a), wipe clean, close the side tubes with rubber tubes and plugs, leave in the balance case for 10 minutes, remove the plugs and weigh. The increase in the weight of the tube in the blank determination should not be more than 2 or 3 mg. Run a second blank determination or more if the gain is not constant within 0.5 mg. Allow the combustion tube to cool and, in the meanwhile, weigh 20 to 30 gm of finely divided steel sample and place in a boat which is 6 inches long and 1/2 inch wide. With the stream of hydrogen still passing quickly open the end which is nearest the absorption tube 7a, insert the boat, push it to the middle of the tube, and quickly insert the stopper. By opening this end of the tube the current of hydrogen prevents the entrance of oxygen, which would entirely vitiate the results of the determination. Insert the weighed absorption tube, heat the silica tubes to redness and main- tain this temperature for 30 minutes with the hydrogen passing through at the rate of about six bubbles per second in the tube containing sul- phuric acid. At the end of this period disconnect the absorption tubes (Id) and (76) from the silica tube, connect tube (9) with the unguarded end of (7a) and connect (76) with the aspirator. Draw air through the tubes rapidly until about 500 cc has passed, as measured by the water that has run out of the aspirator. Close the side tubes of (7a) with rubber tubes and plugs and stand in the balance case. At the end of 10 minutes remove the rubber tubes and plugs and weigh the absorp- tion tube. From the weight of water so found calculate the percent of oxygen in the sample. Treatment of Steel. The property which makes iron the most generally useful of all metals is the property of combining with various quantities of other elements in such a way that its physical and mechanical properties are varied over a wide range. Scarcely STEEL AND ALLOYS 479 second to this property is that of. undergoing important changes in character as the compounds with carbon are subjected to dif- ferences in thermal treatment. The latter property is possessed by all irons containing carbon. These include the crude products of the blast furnace, " pig iron," and the various grades of the more refined product, known as "steel." The fact that a very small percent of carbon gives iron the capability of being hardened by suddenly cooling from high tem- peratures has been known for a long time. The beginning of an understanding of why this is true came when it was found that certain structural changes in steel take place with thermal treatment. The entrance of the microscope into the field of metal testing marked the beginning of a new age for steel, an age of the development of the scientific principles underlying thermal and mechanical treatment. The chemist's analysis is no longer expected to tell the entire history of the steel. As a result of the analysis we know the composition but this tells us only what the steel maybe made to do. The microscope, following this, tells what the steel has been made to do. This is a development of vast importance to the user of steel. When the manufacturer of steel articles buys his bars, sheet, rods, billets or forgings from the manufacturer of the steel itself, he is chiefly concerned with the composition, with respect to the various elements that are combined with iron to make the commercial steel, because he knows the composition that is necessary for his particular purpose. When he has made the steel into the forms necessary for his manufactured article of commerce he is still more seriously con- cerned with the properties that have been given, or are to be given, by the careful thermal treatment that is necessary. The analysis can be of little or no assistance at this point. The per- centage composition is not materially changed by thermal treat- ment, except in those forms of combined chemical and thermal treatment, known as "case-hardening" or "cementing." The microscopic anatomy of the steel is, on the contrary, very pro- foundly changed and these recognizable changes are so intimately associated with changes in physical and mechanical properties that the microscope is, at this point, the most important in- strument available for testing. This branch of testing is known as "metallography." 480 QUANTITATIVE ANALYSIS A thorough discussion of the principles of thermal treatment and of the metallography of steel would require a volume in itself. In the next following pages a brief outline of these principles will be given, with directions for a limited number of experiments, which will serve to illustrate the main points. Thermal Changes. If steel containing about 0.05 percent of carbon is allowed to cool from a high temperature and the rate of cooling is followed by means of a sensitive pyrometer, it will be noticed that the previously uniform rate of cooling is interrupted at about 875, the temperature remaining stationary for a short 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 Percent of Carbon FIG. 99. Recalescence curve for steel. time or the cooling being at least retarded. At about 750 another interruption is noticed and at 690 still a third. The exact location of these points depends upon the composition of the steel and also upon the rate of cooling, more rapid cooling lowering the point of change These interruptions in the cooling process are due to certain internal changes which involve the evolution of heat and the temperature at which they occur are therefore called points of "recalescence" or "critical points." If a series of steels having gradually increasing percents of carbon is treated as just described it will be noticed that the upper STEEL AND ALLOYS 481 critical point is lowered as the carbon percent increases, while the lower two critical points remain practically unaltered until a steel containing about 0.42 percent of carbon is reached, when the upper two points merge. As the carbon percent is further increased this new point is still further lowered until the steel having 0.85 percent carbon is reached when all three changes are merged. In order to distinguish the various points of recalescence the lowest is denoted by Ar i} the second by Ar 2 and the highest by Ar 3 . The relations between carbon percent and points of recales- cence are shown in Fig. 99, which is somewhat idealized. In steels containing more carbon than 0.85 percent, Ari remains at the same temperature while a new higher point is again noticed, increasing carbon raising the location of this point. Sauveur denotes this by the symbol Ar cm . Allotropism of Iron. The occurrence of these points of recal- escence shows that internal changes are taking place while the steel is cooling. These changes might, conceivably, be either physical or chemical, or both. The fact that Ar 3 and Ar 2 are noticed in the purest iron that can be manufactured indicates that at least a part of the heat evolution is due to allotropic changes. That such changes are not the only ones occurring at the critical points of steel is shown by the fact that the reca- lescence in pure iron is very faint while in medium and high carbon steel it is very pronounced, actual glowing being noticed at Ari. Iron is now believed to exist in at least three allotropic modifica- tions as follows: 7-iron exists from the melting-point down to Ar s . It crystal- lizes in the cubic system, the prevailing forms of crystals being octahedra. Its electrical resistance is about ten times that of a-iron and it is non-magnetic. Its hardness is somewhat less than that of 0-iron but greater than that of a-iron. It dissolves iron carbide to the extent of 25.5 percent, corresponding to 1.7 percent of carbon. /3-iron is normally present between Ar s and Ar z until these points merge, when it disappears. It crystallizes in the cubic system with cubes as the prevailing forms. It is feebly magnetic and is very hard. It has little or no solvent power for carbon. a-iron exists below Ar 2 . It also crystallizes in the cubic system. 31 482 QUANTITATIVE ANALYSIS Its electrical resistance is less than that of either /3-iron or 7-iron and it is strongly magnetic. It is the softest form of iron. Influence of Sudden Cooling. The influence of carbon upon the magnitude and location of the thermal changes of iron would indicate that it also plays a part in the changes occurring at the points of recalescence. This is confirmed by many careful chemical and microscopical examinations of polished and etched steel samples. The question very naturally arises as to how a microscopical or chemical examination can be made of any but a-iron, since the other forms do not normally exist at ordinary temperatures. Such examination is made possible by the fact that sufficiently sudden cooling partially or entirely arrests the change from 7-iron to /3-iron and from /3-iron to a-iron, with the accompanying changes in the form of carbon. This, as will later be shown, is the property upon which all thermal' treatment of steel is based. If, then, a sample of iron is heated to a tempera- ture above Ar s and suddenly cooled by quenching in cold water, mercury, liquid air, etc., y-iron is retained as an abnormal substance because at ordinary temperatures the molecular mobility of iron is too small to permit any changes in structure which would have taken place at higher temperature if sufficient time had been allowed. Similarly /5-iron may be retained by quenching from a temperature between Ar s and Ar 2 . It has also been shown that carbon is retained in the form which is nor- mal to the quenching temperature. This method of preventing changes that normally occur, retaining abnormal structures and physical states, opens the way for an investigation of the condition of steel at high temperatures. Proximate Constituents of Slowly Cooled Steel. Pure iron is never obtained in industrial practice and is seldom desired, since steel is the form which so readily lends itself to modification in its properties to suit the most varied requirements. Ferrite. Certain commercial articles that approach pure iron in character are known as "ingot iron," a product of the open- hearth process as used for steel making, but without recarburizing the iron. If a small section of such iron is given a high polish and is then subjected to the action of an etching agent, such as nitric acid or tincture of iodine, and the sample is then placed under a microscope it will be found that the substance that to the STEEL AND ALLOYS 483 naked eye appeared quite bright and homogeneous is really made up of numbers of granules somewhat resembling crystals in form. The structure is really crystalline but the crystals are so hampered in the process of formation that the word "granule" better expresses their appearance. The action of the etching solution has been to attack the grains more vigorously at the lines of juncture and therefore to bring their outlines into relief. A pho- tomicrograph of such a section is shown in Fig. 100. The carbon- less iron thus appearing as granules is found to be a distinct con- stituent of all steel containing less than 0.85 percent of carbon. This constituent is called "ferrite" and it is what has already been described as a-iron. Some of the properties of ferrite have been described in the discussion of a-iron. In addition it may be said that it is quite ductile, giving an elongation of about 40 percent in 2 inches, that it has the power of combining with carbon to form other constituents of steel and that its tensile strength is about 50,000 pounds per square inch. It cannot be hardened by sudden cool- ing except to the extent that is indicated by the partial retention of 7-iron. Cementite. If a piece of steel containing more than 0.85 per- cent of carbon is similarly polished, etched and examined under somewhat high magnification it will exhibit certain shaded areas and also bright lines, their extent increasing with increasing carbon percent. Such a section is shown in Fig. 101. The bright constituent is iron carbide, FesC, already mentioned in con- nection with the analysis of steel. As a distinct crystalline constituent of steel, iron carbide is known as "cementite." The properties of cementite are quite different from those of ferrite. Cementite cannot be obtained in the pure condition because, as its formula indicates, it contains 6.67 percent of carbon, and steel of this composition cannot be prepared without the presence of manganese or other elements or without the separation of graphitic carbon. According to Sauveur 1 the ten- sile strength of cementite is about 5000 pounds per square inch and its ductility (represented by elongation) is practically 1 The Metallography and Heat Treatment of Iron and Steel, 2d edition 138. 484 QUANTITATIVE ANALYSIS zero. It is the hardest constituent of steel and cannot be made appreciably harder by sudden cooling. Pearlite. In the description of cementite it was stated that steel containing more than 0.85 percent of carbon shows certain shaded areas in addition to the bright cementite, when it is polished, etched and highly magnified. (Fig. 101.) These shaded areas increase in extent as the carbon approaches 0.85 percent from either above or below this figure. If the carbon content is practically 0.85 percent the relatively bright areas of cementite and of ferrite are both absent. If the carbon content is less than 0.85 percent ferrite appears in addition to the shaded portion. Both ferrite and cementite are distinct, homogeneous entities, the one an element, the other a definite chemical com- pound. The darker substance is neither element nor compound but is a composite of both, so constituted that it reflects light in a way suggesting mother-of-pearl. On this account it is known as "pearlite." The peculiar iridescence of pearlite is evident without magnification of the polished specimen. If an unhardened steel is prepared as already described and examined under high magnification the structure of pearlite is shown quite clearly to be that of alternating plates or laminae of a dark and a light material. This is shown in Figs. 102 and 103. These plates have been shown to be chemically free iron and iron carbide. That is, they are the materials that have been described as ferrite and cementite, although the latter names are generally reserved for distinct granules of these materials, not constituents of pearlite. The laminations of pearlite show dis- tinctly because the etching agent attacks free iron more readily than iron carbide, thus bringing the former into relative shadow. The laminae are usually bent or contorted. One of the most characteristic of the properties of pearlite is its constancy in composition, the percent of free iron being 87.26 and that of iron carbide 12.74. This is calculated from the percent of carbon in cementite (6.67) and in pearlite (0.85). The percent of iron carbide in pearlite is therefore 85X100 - cTgy = 12.74. Pearlite has a hardness between that of ferrite and of cementite; its tensile strength is about 125,000 pounds per square inch and its ductility is represented by an STEEL AND ALLOYS 485 elongation of about 10 percent in 2 inches. It is the only one of the three constituents of unhardened steel that possesses the power of hardening when suddenly cooled and it is due to pearlite that steel exhibits this remarkable and useful property. Relation between Structure and Carbon Percent. It would follow from what has been said concerning the chemical and physical composition of ferrite, cementite and pearlite that the relative areas of these substances appearing in the polished and etched section of slowly cooled steel would give at least a fair indication of the percent of carbon. This is seen to be the case by an inspection of Figs. 104, 105 and 106. An estimate of the carbon percent may be made from the microscopic appearance, with an accuracy of 0.10 percent for steels containing less than about 0.6 percent of carbon. With more carbon than this it becomes more difficult to judge the percent. For this purpose a magnification of 50 to 150 diameters is suitable. If the steel has not been slowly cooled the microscope can give no indication of carbon percent. Austenite. If a small piece of steel is heated to a temperature considerably above Ar 3 and cooled very suddenly by quenching, the microscope does not reveal the presence of either ferrite, pearl- ite or cementite. Instead the mass has assumed a fairly definite crystalline appearance, in which no constituent can be distin- guished as different from another. The steel has also been made much harder than formerly, the degree of hardness depending upon the percent of carbon. The single crystalline substance composing the mass of hardened steel is called "austenite," after Roberts-Austen the English metallurgist. The bright portions of Fig. 107 are austenite. Austenite contains carbon in any proportion up to 1.7 percent. It is non-magnetic and thus contains iron in the 7 modification. It is not the hardest constituent of hardened steel but is harder than either ferrite or pearlite. Relation of Ferrite, Cementite, Pearlite and Austenite to the Critical Points of Steel. It has been shown that slowly cooled steel contains either ferrite and pearlite, pearlite alone, or cement tite and pearlite, according to whether the steel contains less than 0.85 percent, exactly 0.85 percent or more than this amount of carbon, also that steel cooled with sufficient rapidity contains 486 QUANTITATIVE ANALYSIS only austenite, no matter what the percent of carbon may be. The change of austenite into the other constituents when the steel is slowly cooled begins normally at the temperature denoted by the line ABCD of Fig. 99, that is at the point Ar 3 , for the steel of any particular composition. If less than 0.85 percent of carbon is contained in the steel the formation of ferrite begins at Ar s and continues until Ari is reached, when the austenite then remaining changes completely into pearlite. Since pearlite has a definite and constant composition it will be seen that austenite must have lost enough iron between Ar 3 and Ari to leave austenite contain- ing 0.85 percent of carbon. If more than this percent of carbon is contained in the steel austenite loses iron carbide at the point Ar cm and continues to lose it until Ari is reached, when again pearlite is formed of what remains. Steel as a Solid Solution. This behavior of slowly cooling steel suggests the behavior of ordinary solutions where separation of the constituents occurs at tolerably definite temperatures, depending upon the percentage composition of the solution. For a solution of any two substances in each other there exists a percentage composition for which the lowest possible freezing point is noticed. If more of either constituent is present there is a higher freezing point and at this point the excess of this constituent freezes and separates, leaving a solution which also freezes at or near the lower point already noticed. At the temperatures of separation there also occur thermal changes which are usually in the nature of evolution of heat. The solu- tion having the lowest freezing point of all solutions of a given pair of components is known as the "eutectic" solution. In these and a great many other respects, steel resembles liquid solutions. In fact the resemblance is so close that it can no longer be doubted that austenite is a solution of y-iron and iron car- bide in each other. To be sure, austenite is solid while most common solutions are liquid, but from the physical standpoint the mere state of aggregation is a point of secondary importance in the consideration of solutions. Austenite possesses the prop- erty of physical homogeneity, common to liquid and all other solu- tions. It changes completely into the resolution product, pearlite, at Ari if of eutectic composition. If not of this composition it begins to lose excess of either iron or iron carbide, as ferrite or STEEL AND ALLOYS 487 cementite, at Ar 3 , Ar^ or Ar cm , just as the liquid solution would lose the excess of solvent or solute if not of eutectic composition. Coincident with the separation of ferrite or of cementite allo- tropic changes also occur in the separated iron. Martensite. It is difficult entirely to prevent the change of austenite into the constituents normal to lower temperatures by the process of sudden cooling.' If the quenching medium is one that will cause cooling of extreme rapidity the change may be almost entirely arrested. Such media are ice-water, ice and salt solution or liquid air. These media are not generally used in commercial processes of hardening, partly because of intense mechanical stresses that are set up in the steel by such sudden cooling. These stresses will frequently result in cracking the piece. Therefore austenite does not make up the mass of com- mercially treated steels. Instead a new formation, called "mar- tensite" (after Martens, German metallurgist), makes its appear- ance. The true nature of martensite is not, even now, thoroughly understood. It was formerly thought to be the stable solid solution itself, but this view is no longer held. While opinions still differ, martensite is probably a transition product between austenite, on the one hand, and pearlite on the other and it is a solid solution of iron carbide in iron; the iron is not here in the 7 condition, as in austenite, but in either the /?, the a or both ft and a conditions, so that martensite is magnetic and is even harder than austenite. Like austenite, its hardness varies with the percent of carbon in the steel, i.e., with the percent of dissolved iron carbide in the martensite. The crystalline appearance of martensite is rather characteristic. The dark needles of Figs. 107 and 108 are martensite. It consists of intersecting needles which usually ar- range themselves at 60 angles, etching darkly with nitric acid. Being a transition substance and not in real equilibrium at any temperature, but merely a metastable condition retained by rapid cooling, martensite is frequently associated with austenite, or with troostite or sorbite, two other transition constituents presently to be described. Relation between Location of the Thermal Critical Points and the Rate of Cooling or Heating. Reference has already been made to the influence of rapid cooling upon the resolution of austenite into its components, the change being entirely pre- 488 QUANTITATIVE ANALYSIS vented by sufficiently rapid cooling. Even when the rate of cooling is comparatively slow a certain lag is noticed in the trans- formation, so that the critical point is lowered as the rate of cool- ing is increased. The exact location of Ari, Ar 2 and Ar 3 , there- fore, depends upon this rate of cooling, as well as upon the percent of carbon. Similarly the changes in structure observed upon heating steel are affected by the rate of heating, the critical points being raised. This " thermal hysteresis" is sometimes large, as in the case of steels containing nickel or chromium which retard the transformation. In ordinary work the difference between the points observed upon heating and those upon cooling is not greater than 30. In order to distinguish between the two points, the transformation temperatures observed when the steel is heated are denoted by the symbols Ac\, Ac Z) Ac 3 and Accm. For the theoretical point where Ac and Ar would coincide if the heating and cooling were infinitely slow, the symbols AI, A 2 , etc., are used. Relation between Hardening and Annealing of Steel and the Constituents Already Described. The following summary of the properties of the five constituents of steel already described will form a basis for an understanding of the true nature of harden- ing and annealing of steel. Above A 3 steel is essentially austenite, a solid solution of iron carbide in 7-iron. Austenite is a stable substance at all tempera- tures above A* and is very hard. Between A s and AI austenite spontaneously loses ferrite or cementite, according to the percent of carbon in the steel, these substances forming definite masses or granules, separate from the remaining austenite. Upon cool- ing to.Ari the austenite remaining changes spontaneously into pearlite which remains mixed with the granules of ferrite or cementite already formed. Ferrite is relatively very soft and ductile, cementite is very hard and pearlite combines the prop- erties of ferrite and cementite, being moderately hard and tough. Normally, then, steel is hard above A 3 and soft below AI, the de- gree of hardness in both regions being conditioned by the per- cent of carbon. The latter condition should be easily understood. Below AI low carbon steel is largely ferrite, the softest of all constituents of steel. Medium carbon steel contains more pearlite while STEEL AND ALLOYS 489 high carbon steel contains cementite and no ferrite. The hard- ness of these thre._ substances increases in the order named. Above As austenite exists and, being a solution of a soft and a very hard substance in each other, its hardness will naturally increase with the percent of the hard constituent. Steel possesses the properties normal to temperatures below A i only in case it has cooled very slowly. Even ordinary liquid solutions may be considerably supercooled without the normal dissolution taking place. Steel, already solid, possesses much less molecular mobility and requires even more time for such changes to take place. If this time is denied a metastable condition is obtained and this persists at ordinary temperatures because the molecular rigidity of the mass makes structural changes impossible. It is not, however, possible to cool austenite with sufficient rapidity to prevent absolutely its transformation and an intermediate product, martensite, always appears to some extent. This is, however, even harder than austenite and also contains /?- and a-iron. The effect of the commercial hardening process is now clear. It consists in heating steel to a temperature above Ac$ and sud- denly cooling by quenching, austenite or martensite or both being retained, according to the nature of the quenching medium. The annealing or softening process is simply reheating a hardened steel to a temperature just above Ac$, then cooling slowly so that the softest constituents may make their appearance. The details of these processes will naturally vary according to the degree of hardness desired. Some of these details are later dis- cussed along with tempering and with quenching media. Troostite.- Since the transformation of austenite into its segregated components requires an appreciable period of time it is but natural to expect that variation in the rapidity of cooling would result in varying degrees of imperfection in the formation of these separate components. It does not necessarily follow that these more or less imperfectly formed components will have dis- tinct physical properties, other than in the matter of crystal formation, but two fairly distinct transformation products in addition to martensite have been recognized. "Troostite" (after the French chemist, Troost) is the stage following marten- site in the transformation of austenite. Its constitution is not 490 QUANTITATIVE ANALYSIS known and much difference of opinion still exists with regard to it. The Committee on Nomenclature of the Microscopic Sub- stances and Structures of Steel and Cast Iron of the International Association for Testing Materials 1 says of troostite that it is "an uncoagulated conglomerate of the transition stages. The degree of completeness of the transformation represented by it is not definitely known and probably varies widely. 77 Its hard- ness varies with the percent of carbon but, in general, it lies between that of martensite and that of sorbite (the next stage in the transformation). Its tensile strength is greater than that of sorbite or pearlite, its ductility less. It occurs in granular masses, usually associated with martensite and sorbite, some- times also with pearlite. It is distinguished by its property of coloring more darkly than either martensite or sorbite; the lack of crystalline character also distinguishes it from martensite. Troostite is shown as the very dark portions of Fig. 108. Sorbite.' The next stage after troostite, in the resolution of the solid solution into its components, is sorbite, so called in honor of Sorby. Sorbite is to be regarded as imperfectly formed pearlite. The distinctly laminated appearance of the latter is lacking, sorbite being granular and apparently amorphous. It is somewhat less ductile than pearlite but its higher tensile strength and elastic limit make it a desirable constituent of struc- tural steels. It is the principal constituent of most oil-hardened steels. Fig. 109 shows a sorbitic steel. Influence of Method of Cooling. Quenching Media. If the resolution of austenite into its ultimate transformation products were an instantaneous change, requiring, for instance, a very small fraction of a second, it is doubtful whether any quenching medium could cool it so suddenly as to prevent the practically complete resolution. That the transformation does require an appreciable period of time is shown by the retention of austenite in very suddenly cooled steel. That the rapidity of cooling depends largely upon the nature of the quenching medium is shown by the appearance of the transition stages, martensite, troostite and sorbite, in steel cooled from above, or within, the critical range by quenching in such media as warm water, oil or air. ., 6th Congress, Intern. Assoc. Test- Mat., paper II 8 , p. 18 (1912). STEEL AND ALLOYS 491 Considering the gradation of properties of the series beginning with austenite and ending with pearlite it will be understood that the properties of hardened steel may be varied at will over a wide range by a proper selection of the quenching medium. Thus heating to a temperature above Ac$ and quenching in cold water will produce little more than martensite, with its high degree of hardness, tensile strength, and elastic limit, but great brittleness. Quenching in cold water from a temperature near Ar z , but below it, will produce martensite, associated with more or less ferrite or cementite and also some troostite or sorbite or both, the latter two products imparting a higher degree of toughness, as opposed to the brittleness of martensite, but being also responsible for a decreased tenacity and increased ductility. Quenching in the same medium from a temperature somewhat above Ar L , but nearer to this point than to Ar 3 , will increase the proportion of sorbite and ferrite and decrease that of, martensite and troostite, re- sulting in a further change ; n properties in the direction indicated. If the quenching medium is oil, slower cooling results and the constitution of the quenched steel is such as to indicate greater transformation than is the case with water. In other words, the tendency is now toward the sorbite end of the series and away from austenite. Cooling in quiet air or in an air blast will result in still slower cooling and a still greater percent of sorbite will now be formed, even pearlite appearing if the piece is large enough to cool slowly. Similar considerations will apply to various other quenching media, such as ice water, liquid air, mercury, salt solutions, alco- hols, mixtures of alcohols and water, etc. The exact reasons for the different rates of cooling produced by different media are not definitely known. Different scientists have held that the rate of cooling depends upon (a) the temperature of the cooling bath, (b) its specific heat, (c) its conductivity for heat or (d) its heat of vaporization. No doubt all of these factors are important. The number of substances commercially used for cooling baths is not as large as might be expected, water, oil and air supplying most of the needs. Tempering by Regulated Cooling. Extremely hard steel is demanded for certain purposes but for most structural work, 492 QUANTITATIVE ANALYSIS tools, etc., the brittleness accompanying extreme hardness is undesirable and it becomes necessary to sacrifice a certain degree of hardness and tenacity for a greater degree of toughness. From what has been said concerning the effect of varying the rate of cooling a method would here seem to be available for securing the desired combination of properties. To " temper" the steel (the property of hardness is ''tempered") it should be necessary to heat the piece above Acs and then to cool by quenching in the medium that is found by experience to permit the formation of the transition constituents that will give the desired properties. This can, in fact, be done and this method is actually used to a considerable extent as oil-hardening. Tempering by Reheating. A method of tempering that is found to be easily controlled is that of reheating the thoroughly hardened piece to a temperature where the metastable austenite (or in commercial practice, usually martensite) is partially con- verted into the transition products, troostite or sorbite. While the metastable austenite or martensite may be retained for an indefinite period at ordinary temperatures, it becomes partially converted into its more stable products long before the tempera- ture reaches Ac\. Transformation begins as low as 100 and takes place to a greater extent as the temperature is raised toward Aci which, it will be remembered, is about 700. If it be sup- posed that the hardened steel is never brought quite to Aci, the degree of tempering produced will vary directly with the temperature of the piece during the tempering process. After the proper temperature is reached the steel may be permanently fixed in its tempered condition by again cooling. In practice it is usually quenched but this is not essential to the tempering process and is practised only for the purpose of saving time. The tempering process serves also to relieve internal stresses and thus to diminish the danger of cracking. The advantages of this method of tempering over the method of regulated cooling will readily be seen. The rate of cooling, in the latter method, must be so accurately related to the degree of hardening required that it is often difficult to regulate the process with the required refinement. On the other hand there is no difficulty in first completely hardening the piece, subse- quently raising its temperature to a point, experimentally deter- STEEL AND ALLOYS 493 mined to be correct, then cooling. The method of tempering by reheating possesses another advantage, in that no ferrite, pearlite or cementite can be formed and the structure and properties of the steel must therefore be more uniform than is the case when regulated cooling from above the critical range has occurred. The best method of observing the temperature is by a good py- rometer, preferably of the thermoelectric type. In fact, scientific tempering must be regulated in this way. The method of " tem- pering by color" has been much practiced in the past and con- tinues to be practiced in cases where extreme accuracy is not required. This method depends upon the peculiar progressive change in color, noticed upon the surface of a previously polished piece of steel as it is heated to various temperatures below 500. The colors are due to films of oxides of different composition and stability at different temperatures. It should be noted that one of the purposes of either method of tempering is to remove stresses always existing in fully hard- ened steel. Granulation.- Up to this point in the discussion of thermal treatment emphasis has been placed upon the identity and phys- ical properties of the constituents of steel. Scarcely second to these in importance is the question of size of granules. No matter what may be the strength of the individual particles composing a piece of material, the strength of the piece as a whole will also depend largely upon the degree of coherence of the particles. If the various surfaces separating adjacent crystals (cleavage planes) are relatively large in area the piece will suffer permanent rupture more readily than if they are small. This being the case, the strength of the piece will vary inversely as the size of the granules, orientation of all crystals within a given granule being the same. The temperature of steel and the length of time during which a given temperature is maintained determine the size of the gran- ules. At temperatures below Ai any existing granulation will remain unchanged for an indefinite period of time. If the tem- perature is now raised to Ac 1} the destruction of existing granula- tion is begun by the partial formation of austenite. Complete reformation of grains does not take place at this point unless the steel is of eutectic composition ("eutectoid steel") because until Ac 3 is reached a certain amount of free ferrite or cementite is 494 QUANTITATIVE ANALYSIS normal to the steel. For eutectoid steel, complete destruction of existing granules occurs and new granules begin to form, Ac 3 and Aci coinciding for such steel. For steel containing less than 0.85 percent of carbon ("hypoeutectoid steel") or more than this amount of carbon ("hypereutectoid steel") new granulation of austenite begins the moment Aci is reached and these gran- ules continue to grow in size until Ac z is passed, complete absorp- tion of free ferrite or cementite being here accomplished. This explains the fact that it is difficult to produce thoroughly satis- factory refinement of grain in steel which is very low or very high in carbon. Ac z and Aci are so widely separated (Fig. 99) that the new system of granules has grown to an undesirable extent by the time Ac z , the point of complete destruction of old granules, is attained. The growth of granules increases in speed at higher tempera- tures and granules continue to grow so long as the steel is held at a temperature above A\. From these considerations the fol- lowing rules of procedure will be understood: 1. To refine the grain of a coarsely granulated piece of steel, reheat the piece until Ac$ is passed, then cool immediately, slowly if the piece is to be annealed, suddenly if it is to be hard- ened, whether or not hardening is to be followed by tempering. 2. All thermal treatment for the purpose of hardening or an- nealing should be carried out at the lowest temperature that will permit the desired constitutional changes and the time consumed in the treatment should be as short as possible, in order to avoid coarse granulation and consequent weakness.' The effect of long heating at high temperatures and of reheat- ing to refine the grain is shown in Figs. 110 and 111. These photomicrographs clearly show that a good piece of steel may be rendered absolutely unfit by careless or ignorant treatment, and also that many of such pieces may be restored by reheating. Overheating. Serious overheating of medium or high carbon steel produces another effect in addition to simple coarsening of grains. Crystals of austenite, which is the solid solution existing after Ac$ has been passed, possess a characteristic triangular arrangement. Figure 107 will illustrate this. If austenite grains have been permitted to grow to abnormal sizes this arrangement is not entirely broken up in the pearlite and ferrite (or cementite) SBwB?'<&R *.* > ^^!V* l ^'* > ^*s < '^S ****' -X* ^K- * -* r^afcT .'^*. j*' tw -y I BF>^VB^^^X^^ s< 1 | 1 ' 'I J o ' a * M i ^fl | F g o> TJ 1 Q " * Id ! is |x W "o I -2 t t I 4d CO 1 !'J c3 T ci ' I C5 STEEL AND ALLOYS 495 that are formed on slow cooling. These abnormally large grains, separated by planes instead of irregular, interlocking curved surfaces, give rise to additional weakness of structure. Because this structure is produced by long " soaking" at high temperatures of ingots which have just been poured from the steel ladles, Howe has called this phenomenon "ingotism." Figure 112 shows a steel which has been overheated in this manner. The cure for this structure is to be found in reheating until Ac 3 is passed and cooling immediately, as explained on page 494. If overheating has been serious, more than one reheating may be necessary. Proper Temperature for Thermal Treatment Indicated by Composition. Throughout the discussion of thermal treatment reference has been made to the critical points of steel as indicated by Fig. 99. On account of the pronounced slope of the graph of Az it will be seen that a knowledge of the percent of carbon is absolutely essential to the proper application of thermal treat- ment, unless a delicate pyrometer can be used for locating the critical points. It is quite easy to determine the location of AI by the latter method but at the upper critical points the evolution of heat is small and only a delicate instrument will detect it. An inspection of the figure will show that there is a difference of nearly 200 between the proper temperatures for hardening or annealing two steels of 0.05 percent and 0.90 percent, respectively, of carbon. It must be remembered the curves of figure 99 apply only to plain carbon steel. Other alloyed metals produce important changes in the location of the transformation points. Uneven Carbon Distribution ; Streaks. Aside from questions concerning the physical and chemical condition of carbon as affecting the physical properties of steel, it is important that the carbon should be distributed uniformly throughout the piece, in order that stresses to which the steel may be subjected in service shall not result in rupture of the weaker parts. For instance, it is of little use to give a piece thermal treatment to impart a high tensile strength to the main portion if here and there are masses of ferrite grains that have not responded to the treatment. These weaker localities may start a failure which will immediately result in rupture of the piece, merely because the elastic limit of 496 QUANTITATIVE ANALYSIS these spots has been exceeded by a stress which has been well below the elastic limit of the main portion of the mass. If carbon segregation has resulted through poor mixing of the melted steel, overheating or other causes, an irregular distribution of high and low carbon areas will result. If the steel is later forged or rolled the ferrite masses become flattened and the longitudinal section will show streaks instead of irregular masses (Fig. 1 1.3) . In neither case can the trouble be remedied by simply reheating past Ac s . It is true that at this point the ferrite and carbide have passed completely into mutual solution and new grain systems are generated. But the resulting austenite does not become uniform in composition in a short time. As might be expected the diffusion of molecules if iron and iron carbide in the solid solution is slow, as compared with diffusion in liquid solutions. The remedy for this condition lies in a prolonged "soaking," perhaps of several hours' duration, at a temperature above Ac*. Naturally, this also causes very coarse granulation but after the irregular distribution has been overcome a reheating, followed by immediate cooling, will break up the coarse system of granules. It should be noted that if the ferrite streaks contain slag threads it is difficult or impossible to restore a normal structure by thermal treatment. Sulphur Prints.- In the discussion of the occurrence of sul- phur in steel (page 448) it was stated that this element is found as ferrous sulphide or as manganous sulphide. Either of these compounds will yield hydrogen sulphide upon treatment with an inorganic acid. The distribution of sulphide particles in a steel may then be shown by making use of this reaction. A sec- tion is first polished as though it were to be prepared for etching. A piece of ordinary photographic paper is then soaked in dilute sulphuric acid and laid, face down, on the polished surface and carefully smoothed out. The liberated hydrogen sulphide forms brown silver sulphide in the paper. After a suitable exposure the paper is lifted from the steel surface and placed in an ordinary "hypo" fixing bath. The entire operation is carried out in a dim light. Sulphur prints are used to determine roughly the relative amounts of sulphur in various steel samples. A still more im- STEEL AND ALLOYS 497 portant use is in detecting any segregation of sulphide, since segregation of one element is, in a general way, indicative of segregation of other elements and this is invariably a source of weakness in the steel. Case Hardening. The treatment of steel to give it its maxi- mum hardness also produces brittleness. This may be of small importance if the piece is sufficiently massive or if it is to be sub- jected to no great shock or stress in service. It often happens, however, that small pieces are required to resist abrasion on the surface and still must have sufficient toughness to withstand shock or other stresses. Such is the case with gears and other small pieces of machinery. No combination of hardening and tempering is suitable for producing a piece of steel that will comply with both requirements. It will be remembered that either, hardened or annealed high carbon steel is harder and more brittle than low carbon steel treated in the same manner. If a piece of low carbon steel could be given a surface or case containing a higher percent of carbon the problem would be solved. This can be done by making use of the well known power of iron for absorbing carbon when the two materials are at temperatures above the upper critical point, As. This power is probably due to the penetration of gases, such as carbon monoxide or cyanogen, derived from carbonaceous materials with which the steel is in contact. The piece of " mild " steel is packed in the carburizing material and heated to a temper- ature just above Ac s , the length of exposure depending upon the depth of case required and varying from 30 minutes to several hours. The percent of carbon acquired by the case may be as high as 2.5 but it is not usually higher than 0.90. The materials used for case hardening are wood charcoal, bone charcoal, ground raw bone, leather scraps, scraps of horns and hoofs, etc. Potassium cyanide or ferrocyanide is sometimes used for quickly producing a very thin and hard case. Thermal Treatment of Case-hardened Articles. The temper- ature necessary for case hardening depends upon the percent of carbon already contained in the steel, since the piece must be heated above Ac s . The subsequent thermal treatment is even more important. Protracted heating above the critical range has given both case and core a coarsely granulated structure. In 32 498 QUANTITATIVE ANALYSIS order to refine the grain the piece must be reheated to just above Acz and then quenched but this process, which would have been simple before the production of the case, is now complicated by the fact that Ac s for the core is considerably higher than for the case. If, for example, the core contains 0.10 percent and the case 0.90 percent of carbon, reference to Fig. 99 will show that Ac* for the core is more than 100 higher than for the case. If the piece is heated to refine the core the case, momentarily refined as its critical range is passed, becomes again coarse through exposure to the higher temperature. To remedy this defect a double treatment is employed. The piece is first reheated to refine the core and then quenched in oil or water. It is then reheated to the lower temperature for refining the case and again quenched. Any desired tempering may follow. Effect of Working. The. mere act of forging or rolling steel into the required form does not, in itself, alter the constitution of the steel. The temperature at which such work is performed has, on the other hand, an important influence upon the size and form of the component granules. "Hot working" is carried out at temperatures above the critical range, "cold working" below it. Cold working has no other effect upon the granules than to flatten and elongate them. Hot working, on the other hand, having to do with steel in the form of the solid solution, has the effect of breaking up existing granulation and of preventing the growth of other granules. The granulation that will be observed in the finished piece must have been produced after work was finished and this makes the finishing temperature the really important consideration. Reference to the paragraph on "Granulation," page 493, will point the rule that the finishing temperature should be slightly above Ar z but as near to this point as is pos- sible, if subsequent reheating is not to be applied. If the article is to be reheated for hardening, annealing or refinement, the finishing temperature is not particularly important. Fatigue. Under the influence of repeated alternating stresses, applied for long periods, metals will finally rupture, even though the single stresses may not have reached the elastic limit of the material at any time. Single stresses, if below the elastic limit of 'the piece, cause a minute amount of slipping between adjacent crystals, the amorphous enveloping films probably providing STEEL AND ALLOYS 499 the slipping surfaces in most cases. As a result of a single stress or of a relatively small number of stresses this action would pass entirely undetected. But as the application of the stress is repeated many times the slipping occurs over and over again and it gradually becomes concentrated in the parts where the stress is greatest or the structure weakest. Rupture finally becomes apparent and the piece ultimately breaks. The phenomena resulting from the application of repeated alternating stresses are generally designated by the term "fatigue" and they must be considered in designing parts for machinery or other uses where the parts will be under severe alternating stresses, such as come from intense and rapid vibration. Failures due to fatigue are fairly common and are well'under- stood. But in this connection a fallacy has found wide accept- ance among engineers to the effect that vibration under stress induces coarse crystallization of the material. Many supposed instances of this may be found recorded in the literature. It is commonly stated that bridge structures, for example, become coarsely crystallized on account of the severe vibratory effect of vehicles and horses passing over them and that this coarsening finally results in failure of the material. Parts of machinery, as of rapidly moving engines, are sometimes thought to change similarly. Belief in this fallacy is due to an imperfect understanding of the nature of the molecular changes occurring in metals, and particularly in steel, and of the causes underlying these changes. It may be stated positively that changes in crystal- lization or granulation cannot take place in steel at temperatures below Acij except those of tempering or cold working. Neither tempering nor cold work has occurred in the cases just cited. The effect of cold work would be to distort existing grains, rather than to cause a growth of grains. Undoubtedly many cases of failure under alternating stresses have occurred, which have afterward been found to be due largely to coarsely granu- lated material but no cases are on record where the microscope has been used before as well as after the breaking has occurred and where any change in size of crystals or of granules has been observed. The opposite has been proved many times experimentally. 500 QUANTITATIVE ANALYSIS Slag. The occurrence of slag particles in wrought iron is almost universal, as a complete separation of iron from slag cannot be accomplished by mechanical working. The modern commercial form of nearly pure iron is found in ingot iron, a prod- uct of the open-hearth process carried out without subsequent recarburization of the metal. Iron prepared by this process should be nearly free from slag but small, scattered particles will be seen, even in the best of such metal as well as in the various grades of steel. Slag may be seen most readily by examination of unetched sections under the microscope, at a magnification of 50 to 100 diameters (Fig. 114). Larger masses are also readily visible in etched sections. Ordinary slag is composed, for the most part of silicate and sulphide of iron and manganese. Large particles will show a characteristic mottled effect, like that of Fig. 115. Apparatus for Work in Metallography. A detailed description of the apparatus, or detailed directions for manipulation of the apparatus that is necessary for metallographic investigation of the results of thermal treatment would be entirely beyond the scope of this book. In the following discussion and the directions for a limited number of experiments it is assumed that the labora- tory is fitted with the necessary apparatus and that the instructor will provide detailed instructions. An outline of the necessary steps in the preparation and examination of samples is given and some experiments illustrating the important principles of thermal treatment are described. Experimental Furnace. A small furnace which may be quickly heated and cooled and whose temperature may be easily con- trolled is suitable for experimental treatment. Electrically heated furnaces are most convenient and the temperature should be observed by means of a sensitive pyrometer which may be of the recording or the indicating type. The furnace chamber need not be larger than 2X2X3 inches. Cutting and Polishing Machines. It is not often possible to polish and examine large specimens and a hack saw is almost a necessity in the metallographic laboratory. This may be a small hand saw or one of the more expensive mechanical saws which may be purchased at a nominal price. Before the microscopic examination can be made it is necessary STEEL AND ALLOYS 501 that a high polish be given to the surface examined. Here again, polishing may be done by hand, but if much work is to be done a polishing machine is almost a necessity. Such a machine should provide at least four polishing surfaces. The final polish is given by a very fine powder, such as rouge, but it is quite im- practicable to polish the original, rough surface with such a powder. Instead, the piece is successively polished with powders of increasing fineness, each polish being applied at a direction perpendicular to the next preceding one and the application being continued until all scratches made by the preceding opera- tion have been removed. The following series of polishes may be used, proceeding toward the finest: (1) fine grinding surface of alundum, emery or carborundum, (2) fine emery powder on broadcloth or canvas, (3) tripoli on broadcloth, (4) rouge on broadcloth. The polishing powders are kept in water in separate dishes and are applied to the polishing cloth, wet, by means of brushes while the machine is running. It is highly important that the powders should never be mixed or applied to the wrong surfaces. The dishes should be kept covered to exclude dust. Etching. The microscopic examination will reveal little if ap- plied to the brightly polished surface of steel, although graphitic carbon may easily be seen in cast iron by this means. It is then necessary to bring the components into direct relief by the use of some agent which will color them differently or attack the grain boundaries. Very many such etching agents have been used, examples being alcoholic solution of nitric acid and of picric acid, concentrated nitric acid immediately followed by running water, tincture of iodine and others. The period of application will depend upon the agent used and upon the nature of the sample. High carbon steels etch more rapidly than low carbon steels and darken rapidly on account of the preponderance of pearlite. The etching solution must be thoroughly removed by rinsing, at the end of the etching process, and the sample dried by rinsing with alcohol and holding in an air blast. If the sample is to be preserved it must be protected from oxidation by a thin coat of lacquer, this being removed by alcohol before the microscopic examination. Examination. The microscope that is to be used for metallo- graphic purposes must be of a special type because reflected, 502 QUANTITATIVE ANALYSIS instead of transmitted, light must be used. A reflector is in- serted in the microscope, just above the objective, and illumina- tion is produced by artificial light from a Welsbach or arc light. For the examination to determine the approximate carbon per- cent, as well as to note the condition of granulation, it is best to use the low-power lenses. For this purpose a magnification of 50 to 150 diameters is convenient. This is sufficient to show readily the separate granules of pearlite and ferrite or cementite, or to determine whether the steel is hardened, and to indicate the relative area of pearlite in annealed pieces. For the closer examination to determine the identity, form of crystallization, lamination, etc., of the separate constituents a magnification of 400 to 600 diameters or, in special cases, even higher magnifi- cation is desirable. If a record of the results of the examination is to be kept a photographic camera should be attached to the microscope in such a manner that it may be swung into place after the visual examination has 'been made. A special contrast or " process" plate works best for making the negative. The light must usually be passed through ray filters but the nature of these will depend entirely upon the nature of the light and of the specimen under examination. Other Apparatus. For a complete study of the relation of physical properties to metallographic structure, access should be had to physical testing machines of the usual sort. A Brinell hardness tester or a Shore scleroscope, as well as a set for deter- mining thermal transformation points also are very desirable. For description of these and interpretation of the results obtained from physical tests reference must be made to special books on these subjects. 1 Exercise: Determination of Structure with Variation in Carbon Percent. Select a series of annealed simple carbon steels having ap- proximately the following percents of carbon: 0, 0.10, 0.30, 0.50, 0.70, 0.90, 1.20, or as many of these as may be obtained. Cut a small piece from each sample, of such form as to provide one plane surface 1/2 to 1 Sauveur: The Metallography and Heat Treatment of Iron and Steel. Howe: The Metallography of Steel and Cast Iron. Rosenhain: Physical Metallurgy. Bullens: Steel and Its Heat Treatment. STEEL AND ALLOYS 503 3/4 inch square, although the form of outline of this surface is not im- portant. Grind this surface until it is plane, also slightly round its edges to prevent cutting the polishing cloth. Polish by hand or on a machine, beginning with the coarsest of the surfaces and finishing on the finest, polishing with each powder in a direction perpendicular to the last polishing and continuing each operation until all scratches left by the preceding operation have disappeared. The polishing heads must be kept wet and gentle pressure used, to avoid heating. The appearance of a film of oxide on the polished surface is an indication of too little water or too great pressure. The powder and water should always be applied in such a way that the sample does not "drag" on the polishing surface. Wash the sample and the hands with each change to a finer polishing surface. - When the polishing is finished wash the sample, but without rubbing the polished surface, then rinse with alcohol. If drops of water are permitted to remain on the surface, spots of oxide will appear in a few minutes. Use an etching solution of 10 percent concentrated nitric acid in absolute alcohol. Pour this solution into a shallow dish and immerse the specimen in the solution . with the polished surface up. The length of exposure necessary will vary with the percent of carbon, 3 to 10 seconds being required. The steels containing more than about 0.50 percent of carbon will visibly darken. The lower carbon steels will simply become frosted on the surface. The exposure is easily learned by experience and if some are etched too little they may be re- treated. If etched too deeply the etched surface is removed by polish- ing and a new application of the solution is made. At the end of the etching process remove the specimen and at once rinse thoroughly with running water. Rinse off the water by alcohol and dry in a blast of clean air. The piece is now ready for examination. Examine the pieces in order of carbon content and, if possible, make photomicrographs under the direction of the instructor. Use first a magnification of about 100 diameters and observe the increase of the darker pearlite as the percent of carbon increases, f errite decreasing and finally disappearing when eutectoid steel is reached, cementite appear- ing in steels of hypereutectoid composition. Under a magnification of 500 to 1000 diameters, carefully study the structure of the individual granules of ferrite, pearlite and cementite. If the sample of nearly carbonless iron is wrought iron, particles of slag will be noticed. Also either wrought iron or "ingot iron" may show numerous "etching pits" within the granules of ferrite. The latter are due to the crystalline character of ferrite and indicate the boundaries of smaller crystals within the granules. 504 QUANTITATIVE ANALYSIS Hardened Steel. Heat all of the samples to temperatures just above Acs and immediately quench in cold water. The addition of 10 percent of alcohol will improve the quenching bath. To determine the location of Ac 3 refer the carbon content to Fig. 99. The samples containing about 0.50 percent and 0.90 percent, respectively, of carbon are now to be examined. Polish, etch, and examine these samples as in the case of the un- hardened pieces. Note the disappearance of granules, as observed by low-power magnification, and the characteristic crystalline appearance of martensite. Austenite may be produced by heating the steel to a higher temperature and cooling in ice water. Properly Annealed Steel. Reheat the one of the hardened pieces which contains 0.50 percent of carbon to just above Acs and immediately cool in the furnace. When 600 has been reached the piece may be removed and quenched. Polish, etch, and examine, noting size of gran- ules. Photograph if possible. Improperly Annealed Steel. Reheat the annealed sample to a tem- perature between 900 and 1100 and keep it at this temperature from 1 to 2 hours, then cool in the furnace. If desired it also may be quenched after Ar\ is passed. Examine and note the excessively coarse granula- tion, cracks frequently appearing. If the temperature has been allowed to reach 1400 or 1500 the characteristics of burnt steel may be ob- served. Photograph, if possible, and compare with the photograph of the same piece, properly annealed. Refinement of Grain of Improperly Annealed Steel. Reheat the coarsely granulated (but not burnt) steel to a temperature just above Ac^ and immediately cool in the furnace, quenching, if desired, after Ari is passed. Examine and note the refinement of grain. Tempered Steel. Reheat the various other pieces of hardened steel to about 200, 250, 300 and 350, respectively, quenching in water as soon as the required temperature is reached. Polish, etch and examine and make an effort to identify troostite and sorbite in these tempered pieces. Case-hardened Steel. Cut a small piece of low carbon steel (about 0.02 percent to 0.05 percent carbon) and pack in raw bone, bone char- coal or any of the other case-hardening materials, using a small cast-iron box which may be covered. Heat to 900 for about 3 hours then cool in the furnace in order to leave the piece soft enough to permit cutting by the hack saw. Cut entirely across the piece, so that both case and core will appear in the section, polish, etch and examine. Note the differ- ent appearances of core and case, indicating more carbon in the latter. The case will contain more pearlite or it may even contain cementite. STEEL AND ALLOYS 505 Physical and Mechanical Tests. If the laboratory is equipped with the various appliances for physical and mechanical tests of metals the work may be made even more interesting and instruc- tive by comparing and correlating these properties with thermal treatment, chemical composition and microscopic structure. BKASS AND BRONZE The analysis of brass and bronze, as of most other alloys, must be made by using methods of different classes for the differ- ent constituent metals. It seldom happens that it is desirable to employ all gravimetric or all volumetric or electrolytic methods for the various metals of a given alloy. Neither is it always practicable to determine all of the metals from a single solution of the alloy as is done in a systematic qualitative analysis. In- stead, the method that can be best adapted to each individual metal is applied and if the other metals do not interfere a separate sample is used in each case. Thus while the brasses may be considered as typical of a fairly large class of similar alloys it should not be understood that the methods here outlined will necessarily be the best for other alloys in which the same metals are found. The presence of additional metals, or even variation in the proportions of the same metals, will serve to make modi- fications desirable. Analysis of Brass. The samples may be' weighed on counterpoised glasses. If the exact specified weights are taken the calculations will be simplified. Tin. Weigh samples of 2 gm each into porcelain casseroles or plati- num dishes. Add 25 cc of nitric acid of specific gravity 1.2, cover and digest until the alloy has dissolved, leaving only a white residue of metastannic acid, H 2 Sn03. Digest for 30 minutes at nearly boiling temperature, replacing evaporated acid if necessary. Cool, dilute to 75 cc and filter through paper, receiving the filtrate in a casserole. Transfer all of the precipitate to the paper and wash with hot water, reserving the filtrate and washings for other metal determinations. Carefully burn the paper in a weighed porcelain crucible and ignite the precipitate for 15 minutes over a Me*ker burner. Cool and weigh. If the amount of stannic oxide is relatively small this residue may be white or nearly so. It may then be calculated directly to tin. If this is not the case and the stannic oxide is contaminated with copper oxide 506 QUANTITATIVE ANALYSIS and stannic phosphate it must be purified. To do this add 0.5 gm each of sodium carbonate and sulphur, cover and fuse over a small flame until the excess of sulphur is removed, as evidenced by the absence of a sul- phur flame above the crucible. Cool and digest in 50 cc of boiling water. Polysulphides of tin and of the traces of lead and copper are formed by the fusion and these partly dissolve in the solution of sodium poly- sulphide. Add a little powdered sodium sulphite to reduce the sodium polysulphide to monosulphide, the solution then becoming faintly yellow. The monosulphides of copper and lead now precipitate and are removed by filtration, washed, dissolved in dilute nitric acid and added to the filtrate from the first tin precipitate. Acidify the tin solution with acetic acid and pass hydrogen sulphide through until the stannous sulphide is completely precipitated, then filter and wash once or twice. Transfer to a porcelain crucible and care- fully burn the paper. Roast for some time in the inclined crucible then heat more strongly until the tin sulphide is converted completely into oxide. Weigh as SnC>2 and calculate the tin. Lead. Prepare a porcelain Gooch crucible filter by washing the asbes- tos felt with dilute sulphuric acid, then with hot water and finally with alcohol, drying in the oven for a short time, then heating for 15 minutes at 700, preferably in an electric furnace whose temperature is measured by a pyrometer. Cool and weigh. Place the filtrate and washings from the metastannic acid in a casse- role and add 5 cc of concentrated sulphuric acid. Hold in the hand over a flame and evaporate, agitating constantly, until dense fumes of sulphur trioxide appear. Cool, add 35 cc of water and boil gently for 1 minute to dissolve soluble sulphates of copper and zinc. Filter in the weighed Gooch crucible and wash three times with 5-cc portions of 15 percent sulphuric acid, receiving the filtrate and washings in a 250-cc volumetric flask. Remove the receiving flask and set aside with the solution for the copper determination. Wash the lead sulphate on the filter with 50 percent alcohol to remove sulphuric acid, discarding the washings. Dry the crucible in the oven for 10 minutes then heat at 700 for 20 minutes. Cool and weigh the lead sulphate and calculate the percent of lead in the brass. Copper. Since copper is one of the principal elements of alloys of this class the sample of 2 gm is too large for this determination. Dilute the solution to the mark on the volumetric flask and mix well. Care- fully pipette 50 cc into the beaker in which electrolysis is to take place and add 1 gm of ammonium nitrate. This, by reaction with sulphuric acid, produces a small amount of nitric acid but without increasing the total acidity of the solution. Connect the anode and the weighed cathode with the source of current in the usual way and add water STEEL AND ALLOYS 507 until the cathode is covered. Mix by stirring, then conduct the elec- trolysis and subsequent treatment of copper as directed on page 156, keeping the voltage below 2.5 to avoid the deposition of zinc. Calcu- late the percent of copper in the alloy. If the electrolysis produces a small deposit of lead peroxide on the anode, carefully wash this and dissolve in a small amount of concentrated hydrochloric acid. Determine this lead as sulphate and add to the percent already found. Zinc. The solution from which the copper has been deposited will represent enough sample to serve for the zinc determination in the analy- sis of brass but for bronzes, in which zinc is a minor constituent if it occurs at all, it may be necessary to use a larger aliquot portion of the filtrate from tin, electrolyzing to remove copper, as before. Prepare a 10 percent solution of diammonium acid phosphate by dissolving the salt in cold water and making barely basic to phenolph- thalein by adding dilute ammonium hydroxide, drop by drop. The addition of ammonia converts any monoammonium salt that may be present into the diammonium salt: NH 4 H 2 P04+NH40H->(NH 4 ) 2 HP04+H 2 0. Make the zinc solution slightly basic with ammonium hydroxide, using a drop of litmus solution as indicator. Add 1 cc of 10 percent acetic acid to the solution containing zinc; this should be enough to change the litmus to red. Drop in '25 cc of the phosphate solution and keep nearly boiling for 30 minutes, stirring occasionally. Zinc ammonium phosphate (ZnNH^PCX) precipitates first in an amorphous condition, changed by heating and stirring to the crystalline modification. Cool and filter through a Gooch crucible or an alundum crucible that has been dried at 105 and weighed. Wash the precipitate with 1 percent diammonium phosphate solution until the washings test free from sulphates, then several times with 50 percent alcohol. Dry at 105 and weigh the zinc ammonium phosphate, from which zinc is calculated. Instead of weighing zinc in this form the material may be filtered in an alundum or Gooch crucible that has been ignited for 15 minutes (conveniently heated in an electric furnace at about 800) and weighed. The precipitate is washed as already directed and it is then heated as was the empty crucible for 30 minutes. The zinc is then weighed as zinc pyrophosphate : 2ZnNH 4 P0 4 --Zn 2 P 2 7 +2NH3+H 2 0. The percent of zinc is calculated from this weight. 508 QUANTITATIVE ANALYSIS ANTI-FRICTION METALS Soft bearing metals, used for lining journal bearings, are usually alloys containing tin, copper, lead and antimony in dif- ferent proportions, with occasional additions of other elements. They are often collectively designated " Babbitt metal," al- though this name properly applies only to the alloy having the approximate composition: tin 90, antimony 7, copper 3. Many of the commercially used anti-friction metals have the tin partly or entirely replaced by lead. In addition small amounts of zinc may be present, so that a complete quantitative analysis must include the determination of these metals, as well as of any others that may be shown by a qualitative analysis. Tin and Antimony. The general methods outlined for brass and bronze will apply to these alloys also. An exception must be made in the case of tin, which cannot be weighed directly as oxide on account of the presence of antimony. Upon treatment of the alloy with nitric acid the metastannic acid which is formed from the tin is accompanied by antimonic acid, H 3 Sb04, or an indefinite mixture of this with hydrated antimony trioxide. The precipitation is not complete and total antimony is therefore not weighed with the tin. Volumetric methods are suitable for these elements. The alloy is dissolved in concentrated sulphuric acid, tin forming stannic sulphate and antimony dissolving as antimonious sul- phate. In this solution antimony is titrated by potassium per- manganate, antimonic acid being produced by the oxidation: 5Sb 2 (S0 4 )3+4KMn04+2H 2 0->10H 3 Sb04+2K 2 S04 +9H 2 SO 4 +4MnS0 4 . Tin is then reduced by heating with elementary antimony (in presence of hydrochloric acid) to stannous salts and is then titrated with standard iodine solution: SnCl 2 +l2-SnCl 2 I 2 or SnCl 2 +2HCH-I 2 ->SnCl4+2HI. Analysis of Tin-base Bearing Metal. Antimony and Tin. Prepare a solution of potassium permanganate, either tenth-normal or of such concentration that each cubic centimeter is equivalent to 0.005 gm of STEEL AND ALLOYS 509 antimony. Prepare also a standard iodine, approximately tenth-normal. Standardize these solutions against tin and antimony of known purity, following the methods that are now to be described for the determination of these elements in the alloy, using about 0.2 gm, accurately weighed, of each metal for the purpose. Weigh about 0.4 gm of the alloy and brush into a 500-cc Pyrex Kjel- dahl digestion flask. Add 10 cc of concentrated sulphuric acid and heat until all of the alloy is dissolved and all sulphur dioxide expelled. Cool and add 50 cc of water and 10 cc of concentrated hydrochloric acid, then warm until the solution is clear or nearly so. Cool and add 100 cc of water and 25 cc of concentrated hydrochloric acid. Cool in running water and titrate at once with standard potassium permanganate solu- tion, adding dropwise, but rapidly and with constant agitation. The end point will be sharp but the pink color will fade after a short time, on account of reduction of the permanganate by hydrochloric acid. Calculate the percent of antimony in the alloy. Rinse the titrated solution into a 500-cc flask, using for the purpose 50 cc of concentrated hydrochloric acid. Add about 1 gm of powdered tin-free antimony and insert a rubber stopper carrying two bent tubes. One of these extends beneath the surface of the solution. The other ends just below the stopper and has an outside connection with a 50 or 100-cc pipette. Warm on the steam bath for 15 minutes, then remove and connect the first named tube with a carbon dioxide generator which will furnish a rapid stream of gas. The pipette should dip into a beaker of distilled water. This excludes outside air from the flask. Start the current of carbon dioxide then heat the solution in the flask and boil over a flame for 5 minutes, which should serve to reduce all of the tin. Remove the flame and cool the flask in cold water but keeping carbon dioxide running fast enough to prevent back suction of water from the beaker. When the solution is cold, carefully remove the stopper and rinse down the tube with recently boiled water, then add 1 cc of starch solu- tion and titrate at once with standard iodine solution, whirling to mix the solutions but avoiding violent churning with air. Calculate the percent of tin in the alloy. Lead and Copper. Proceed as with brass and bronze, using sample weights according to the approximate percents of these metals in the alloy. CHAPTER XVI AGRICULTURAL MATERIALS FERTILIZERS Many elements naturally occurring in soils are extracted and used in small quantities by plants. Certain others are necessary to the growth of plant life and are demanded in greater abundance. With the growth of the knowledge of soil chemistry the addition of deficient elements to the soil has become a commercial matter and the analysis of fertilizers has become a necessary part of the chemist's work, not only for the purpose of placing a correct estimate upon the commercial value of the fertilizer but also to provide a basis for the intelligent application of the fertilizer to the soil that lacks it. Substances added to the soil -to promote plant growth belong either to the class of plant foods or to that of correctives. The most valuable elements belonging to the first class are nitrogen, phosphorus and potassium. An example of the second class is calcium carbonate, added to correct excessive acidity of certain soils. It is generally true that substances of the second class also provide plant food but this is usually incidental to the main purpose for which they are added. The Association of Official Agricultural Chemists has, for a number of years, carried on a systematic study of the action of fertilizers, their relative values as plant food and the methods for determining the essential constituents. Until 1915 the analyt- ical methods that were given official sanction were published by the Government as Bulletin 107, Bureau of Chemistry, Depart- ment of Agriculture. Since the date mentioned they have been published as a part of the proceedings of the society. 1 Practi- cally all of the methods given in the following pages are essen- 1 J. Assoc. Off. Agr. Chem., Pts. II of Vol. I, No. 4 and of Vol. II, Nos. 1, 2 and 3. 510 AGRICULTURAL MATERIALS 511 tially the official methods, although this rule is not universally followed. Preparation of Samples. Reduce the gross sample by rolling and quartering to an amount sufficient for analytical purposes. About 100 grams will usually be convenient. Transfer to a sieve having circular openings 1 mm in diameter and sift, breaking the lumps with a soft rubber pestle. Grind in a mortar the part remaining on the sieve until the particles will pass through. Mix thoroughly and preserve the sample in tightly stoppered bottles. Grind and sift as rapidly as pos- sible to avoid loss or gain of moisture during the operation. Moisture. The moisture of a fertilizer is, in most cases, a substance without any value whatever and its determination is made with this in view. Determination. Heat 2 gm of the prepared sample for 5 hours at 100. If the sample is of potassium salts, sodium nitrate or ammonium sulphate, heat from 1 to 5 gm at about 130 until the sample ceases to lose weight. Calculate the percent of moisture. Nitrogen. Although nitrogen is so abundant as an elementary constituent of the atmosphere it is one of the most costly of all the elements that are required for plant growth. This is because its chemically inert character makes its fixation and assimilation by plants a difficult matter. Plants do not abstract gaseous nitrogen from the atmosphere although the roots of leguminous plants support certain bacteria whose action is to oxidize at- mospheric nitrogen to nitric acid, this being then fixed in the soil by forming nitrates with such basic materials as calcium carbonate. Nitrogen may be added as an artificial fertilizer in the form of ammonium salts, nitrites, nitrates or organic ni- trogenous materials. Crude chloride and sulphate are the more common forms of ammonium salts used, ammonium sulphate being obtained during the process of gas manufacture. Nitrites are little used as commercial fertilizers. Nitrates are the most common and probably the most valuable of the various nitrog- enous materials. In the past, the chief source of nitrates has been the " Chile saltpetre" beds of South America. The artifi- cial fixation of atmospheric nitrogen has been successfully ac- complished by two processes: (1) When calcium carbide is heated with pure nitrogen to 512 QUANTITATIVE ANALYSIS 700-800, in the presence of a small amount of calcium chloride or calcium fluoride, calcium cyanamide, CaNCN, is formed. This substance decomposes in the soil, forming first cyanamine, then urea and finally ammonium carbonate. CaNCN+C0 2 +H 2 O->CaC0 3 +H 2 NCN, H 2 NCN + H 2 O-CO (NH 2 ) 2 , CO(NH 2 ) 2 +2H 2 0-+(NH 4 ) 2 CO 3 . (2) Under the influence of an electrical discharge nitrogen and oxygen combine directly, forming nitric oxide. This occurs to some extent during electrical storms, which explains the occur- rence of nitric acid in rain water. The reaction is now used on a large scale for the production of nitric acid which is then converted into nitrates and used for fertilizers and other purposes. The most important nitrogenous organic materials that are used for fertilizers are dried blood and tankage obtained from the packing houses, also fish scraps, guano and ordinary stable manure. Dried blood is a very valuable fertilizer because of the large percent of nitrogen which it contains (12 to 14 per- cent) and because it is readily available for assimilation by plants. Tankage consists of scraps of refuse meat, skin, etc., from which the oil has been removed by steaming and pressing. Fish scrap, guano and farm manures are valuable as nitrogenous fertilizers but are limited in quantity. Certain other nitrogenous materials that are sometimes added to mixed fertilizers because they are rich in nitrogen are useless on account of the fact that their nitrogen is only very slowly available for assimilation by plants. Such materials are hair, horns, hoofs, leather scrap, and peat. When these are finely ground and mixed with other fertilizing material they yield rela- tively high percents of nitrogen in the analytical process but are of little use to the plant life. Their detection is often possible only by means of the microscope and the determination of their quantity in a mixed fertilizer is a very difficult matter and for these reasons their admixture with commercial fertilizers is forbidden by many state laws. Fertilizers may contain nitrogen in only one form or in a mixture of two or more classes of compounds, as ammonium salts, nitrates, or organic compounds. The determination of the AGRICULTURAL MATERIALS 513 percent of nitrogen in each form is occasionally demanded but usually the determination of total nitrogen is all that is required, it being understood that hair, leather scraps and other such mate- rials are excluded. The method of Dumas or the soda lime" method may be used for the determination of total nitrogen but the method of Kjeldahl, or one of the modifications of this method, is better suited to this class of work. Kjeldahl's Method. KjeldahFs original process 1 consists in digesting the organic material with boiling concentrated sul- phuric acid until complete decomposition has been effected. The exact course of the reactions cannot be traced but the carbon and hydrogen are completely oxidized and nitrogen is converted into ammonia, which immediately combines with sulphuric acid and remains as ammonium sulphate. The completion of decom- position is insured by the final addition of a small amount of potassium permanganate. The solution is then diluted with water, an excess of sodium hydroxide is added and the resultant ammonia is distilled into a measured quantity of standard acid solution. To dete mine the excess of standard acid, potassium iodate and potassium iodide are added and the liberated iodine is titrated by a standard solution of sodium thiosulphate. Iodine is liberated according to the following equation: KIO 3 +5KI+3H 2 SO 4 -*3K 2 S04+6I+3H 2 O. Instead of this method of determining the excess of standard acid it is now customary to titrate the excess by means of a standard basic solution. Modification to Include Nitrates. The method is inapplicable to the determination of nitrogen of nitrates because of the loss of nitric acid which occurs as soon as the material is treated with sulphuric acid. Modifications of the method to suit the analysis of nitrates will presently be discussed. The digestion with sulphuric acid is best accomplished in a pear-shaped flask with a long.necky like that shown in Fig. 116. The concentrated sulphuric acid of commerce boils at tempera- tures ranging from 210 to 340, depending upon the percent of water contained in it. Such a temperature is high enough to *Z. anal. Chem., 22, 366 (1883). 33 514 QUANTITATIVE ANALYSIS permit condensation of nearly all of the vapor without the use of a water condenser, the long neck of the digestion flask serving for this purpose. If the solution is to be transferred to a special distilling flask the capacity of the digestion flask need not be greater than 200 cc. It is more convenient, however, to distill from the flask in which digestion is accomplished, in which case the capacity of the flask should be 500 cc. The digestion must be performed under a hood or some other provision must be made Fio. 116. Kjeldahl flask, stand and lead pipe ventilator. for carrying away the fumes. An excellent arrangement for this purpose, is a lead pipe, 6 inches in diameter and with holes in the side so that the necks of a number of digestion flasks may be inserted with the flask in an inclined position. The end of the lead pipe leads to a chimney. Catalytic Agents. Wilfarth showed 1 that the addition of mercuric oxide, - copper oxide or ferric oxide to the mixture of the organic material and sulphuric acid considerably accelerates the reactions that occur during digestion. The action is of a catalytic nature and depends upon the capability of the metal of existing in more than one state of oxidation. The metal is t *Z. anal Chem., 24, 455 (1885); Chem. Zentr., [3] 16, 17 and 113 (1885) AGRICULTURAL MATERIALS 515 thus alternately reduced by organic matter and oxidized by sulphuric acid, somewhat as follows: 2HgS04-*Hg 2 S0 4 +S0 3 +0, Hg 2 S0 4 +2H 2 S0 4 -+2HgS0 4 +2H 2 0+S0 2 . The nascent oxygen thus formed attacks the organic matter. Of the three metals named, mercury serves best because its salts are colorless and do not obscure the end point of the oxida- tion. It is necessary in this case to precipitate the mercury by the addition of potassium sulphide, before distillation, in order to prevent the formation of mercurammonium compounds which are not readily decomposed by sodium hydroxide. Copper sulphate as a catalyst is often preferred because it serves as an indicator when sodium hydroxide is added later, a deep blue solution being formed when the solution becomes basic. Prevention of Bumping. During the distillation of ammonia, after the addition of excess of sodium hydroxide, there is usually a tendency toward bumping. In order to prevent this the "official" method of the Association of Official Agricultural Chemists directs the addition of granulated zinc or pumice stone to the contents of the flask before distillation. The reaction of zinc with sodium hydroxide produces a continuous evolution of hydrogen and this effectually prevents bumping. There is, however, a disadvantage connected with the use of zinc which is sometimes serious, in that the sodium zincate that is formed by the reaction so increases the surface tension of the solution that troublesome frothing occurs. An excellent substitute for both zinc and pumice is a small amount (0.5 gm) of crushed porcelain from which the dust has been removed by sifting. Blank. Sulphuric acid nearly always contains a small amount of ammonium sulphate. Distilled water may also contain a small quantity of ammonium hydroxide. In order to make the proper correction for the ammonia that will be derived from the reagents a "blank" determination must be made, omitting the sample of fertilizer but carrying out the operations exactly as in the real determination. In this case cane sugar is added to reduce possible traces of nitrates existing in the reagents, as they would be reduced by the organic matter of the fertilizer. 516 QUANTITATIVE ANALYSIS Detection of Nitrates. Mix 5 gm of the fertilizer with 25 cc of hot water and filter. To a portion of this solution add 2 volumes of con- centrated sulphuric acid, free from nitric acid and oxides of nitrogen, and allow the mixture to cool. Add cautiously a few drops of a con- centrated solution of ferrous sulphate so that the fluids do not mix. If nitrates are present the junction shows at first a purple, afterward a brown, color or if only a minute quantity is present, a reddish color. To another portion of the solution add 1 cc of a 1 percent solution of sodium nitrate and test as before to determine whether enough sulphuric acid were added in the first test. Determination. Organic and Ammoniacal Nitrogen only. Kjeldahl Method. Prepare the following reagents. (a) Hydrochloric or Sulphuric Acid Solution, Half-Normal. Stand- ardize against pure sodium carbonate as directed on page 224, making the necessary changes in weight of carbonate to account for tjie different normality of the acid. The official method directs the standardization of these acids by weighing silver chloride or barium sulphate precipitated from measured volumes. That this method may lead to serious errors is explained on page 222. The method is not to be recommended. (6) Sodium Hydroxide or Potassium Hydroxide Solution, Tenth- Normal. Standardize by titration against exactly 10 cc of the acid, using methyl red as indicator. (c) Sulphuric Acid. The concentrated acid of the laboratory, specific gravity 1.84, as nearly as possible free from nitrates and ammo- nium salts. (d) Metallic Mercury or Mercuric Oxide. Mercuric oxide should be that prepared in the wet way but not from mercuric nitrate. (e) Potassium Sulphide Solution. Dissolve at the rate of 40 gm for each liter of solution. Commercial potassium sulphide is used. (/) Sodium Hydroxide Solution. A saturated solution, free from nitrates. (g) Methyl Red Solution. Dissolve 1 gm of methyl red (dimethylam- inoazobenzeneorthocarboxylic acid) in 100 cc of 95 percent alcohol. If the approximate percent of nitrogen in the sample is known, cal- culate the weight that will yield ammonia equivalent to about 35 cc of the standard acid. If nothing is known of the nitrogen content use about 1 gm of sample. The sample must contain no nitrates, nitrites or nitro-compounds. Place two weighed samples in 500 cc Kjeldahl digestion flasks, holding the latter in a vertical position to prevent the sample from sticking to the sides of the neck, which should be dry. Weigh 1 gm of sugar into another flask and treat the same as the fertilizer sample. Add about 0.7 gm of mercuric oxide or of mercury, or 0.3 gm of AGRICULTURAL MATERIALS 517 copper sulphate, also 25 cc of concentrated sulphuric acid. Incline the flask in a hood or with the neck inserted into a lead-pipe ventilator and heat gently until the violence of the reactions has moderated, then grad- ually raise the temperature until the acid is boiling. The flask may be heated without protection by a gauze if it is of Pyrex glass or a similar resistance glass and if it is placed over a hole in a stand of sheet iron in such a manner that the flame cannot come into contact with the sides of the flask above the liquid. (See Fig. 116.) Digest by gently boiling until the solution is nearly colorless. This may occur after a short time or the digestion may require several hours. Remove the flame and at once drop into the flask small quantities of powdered potassium permanganate until the solution acquires a green or purple tint which persists after shaking. Allow the flask to stand until cool. (Do not cool under a tap.) Carefully add 200 cc of distilled water and mix by rotating the flask. Add about 0. 5 gm of crushed porce- lain and 25 cc of 4 percent potassium sulphide solution, shaking as the latter is added. Have the connections with a tin condenser ready and have 50 cc of standard acid measured into a 400-cc flask into which the delivery tube (of glass) dips. Most laboratories in which much work of this kind is done will be equipped with a special form of apparatus for carrying on several distillations at once. The flask should be in a vertical position and some kind of trap should be used to prevent spray from being carried over by the steam. The delivery tube should be capable of being detached from the condenser for the purpose of cleaning and rinsing it. The entire condenser must be thoroughly rinsed before each distillation, to insure freedom from basic solutions. Pour 50 cc of saturated sodium hydroxide solution (which should contain but little carbonate) down the inclined flask in such a way that mixing does not occur. Immediately connect with the condenser, care- fully mix the contents of the flask by shaking, then distill until about 150 cc of distillate has been collected. It sometimes happens that too much sulphuric acid has been added to hasten a difficult digestion or that the sodium hydroxide solution is not saturated. The consequence is that the solution still contains an excess of acid when ready for distillation. This will not be the case if the directions have been carefully followed but the addition of a drop of phenolphthalein to the solution will serve to indicate the fact. It should be remembered, however, that a con- centrated solution of a base soon decolorizes phenolphthalein and this action may be mistaken for an indication of an excess of acid. If copper sulphate has been used as an accelerator a deep blue color will indicate the presence of sufficient sodium hydroxide. In this case the addition of potassium sulphide should be omitted. When the distillation is finished lower the receiving flask until the 518 QUANTITATIVE ANALYSIS delivery tube is above the liquid, then remove the flame. Disconnect the delivery tube from the condenser and rinse inside and outside, allow- ing the rinsings to run into the flask. Add enough methyl red to tint the solution, then titrate with standard base. Subtract the excess of acid thus indicated and calculate the percent of nitrogen in the sample, making proper correction for any nitrogen found in the reagents by the blank determination with sugar. Modifications to Include the Nitrogen of Nitrates. It has already been noted that most of the nitrogen of nitrates is lost by directly heating with sulphuric acid. Asboth 1 modified the Kjeldahl method by adding benzoic acid, nitrobenzoic acid being formed and later oxidized by potassium permanganate. Jodl- bauer 2 substituted phenolsulphonic acid for benzoic acid and reduced the resultant nitrophenolsulphonic acids to aminophenol- sulphonic acids by zinc dust. The amino compound was then oxidized by heating with sulphuric acid. The addition of phos- phoric acid was also found to hasten the oxidation. Both ben- zoic acid and phenolsulphonic acid are now generally substituted by salicylic acid and the reducing agent for the nitro compound is either zinc dust or sodium thiosulphate. The reactions may be represented as follows: 2KN0 3 +H 2 S0 4 ^K 2 S0 4 +2HN0 3 , , /OH HN0 3 +C 6 H 4 <( ->C 6 H 3 -COOH+H 2 0. COOH \N0 2 The nitro compound is then reduced by nascent hydrogen: /OH /OH 6H+C 6 H 3 -CbOH->C 6 H 3 --COOH+2H 2 0, \N0 2 \NH 2 or by sodium thiosulphate: Na 2 S 2 3 +H 2 S0 4 ->Na 2 S0 4 +H 2 S0 3 +S, /OH /OH 6H 2 SO 3 +2C 6 H 3 -COOH+'2H 2 O-6H 2 SO4+2C6H 3 -COOH. \N0 2 \NH 2 iChem. Zentr., [3] 17, 161 (1886). 2 Ibid., [3] 17, 433 (1886); Z. anal. Chem., 26, 92 (1887). AGRICULTURAL MATERIALS 519 The oxidation of the amino-salicylic acid by sulphuric acid is not well enough understood to be represented by an equation. Determination by the Kjeldahl Method, Modified to Include Nitro- gen of Nitrates. Weigh the sample of fertilizer and place in a digestion flask. The quantity to be used is determined as in the plain Kjeldahl method, (a) Add 30 cc of concentrated sulphuric acid to which has been added 1 gm of salicylic acid and mix by shaking. After 30 minutes add 5 gm of sodium thiosulphate or (b) add to the substance 30 cc of concentrated sulphuric acid containing 2 gm of salicylic acid, allow to stand 30 minutes and add 2 gm of zinc dust, shaking. Heat gently until frothing has ceased then boil until white fumes are no longer found. Add about 0.7 gm of mercury or of mercuric oxide and continue the digestion, distillation and titration as in the Kjeldahl method. Make blank determinations of nitrogen in the reagents. Calculate the per- cent of nitrogen in the fertilizer. Gunning Method. It was observed by Gunning 1 that in the ordinary Kjeldahl process the water produced by the oxidation of organic matter dilutes the sulphuric acid and retards its action. Gunning proposed the addition of potassium sul- phate which forms acid sulphates which lose water much more readily than the hydrates of sulphuric acid so that the solution does not become diluted. A mixture of one part of potassium sulphate and two parts of sulphuric acid is heated together and finally allowed to cool. This mixture is measured into the diges- tion flask, where the digestion is performed as in the Kjeldahl process except that no mercury is added and, consequently, no potassium sulphide is needed before the distillation. In the method as now carried out the required amounts of potassium sulphate and sulphuric acid are added directly to the flask without preliminary heating. Determination. Calculate the weight of sample required, as in the Kjeldahl method, and weigh this amount into digestion flasks. Add to the sample in the digestion flask 10 gm of powdered potassium sulphate and 15 to 25 cc of concentrated sulphuric acid. Digest as in the Kjeldahl process except that 0.3 gm of copper sulphate is used in- stead of mercury, mercuric oxide or potassium permanganate. When the solution is colorless, cool, dilute and conduct the distillation as in the Kjeldahl process, omitting, however, the potassium sulphide solution. iZ. anal. Chem., 28, 188 U889). 520 QUANTITATIVE ANALYSIS Make a blank determination as in the Kjeldahl process. Calculate the percent of nitrogen in the sample. Determination by the Gunning Method, Modified to Include Nitrogen of Nitrates. To the weighed sample in a digestion flask add 30 cc of concentrated sulphuric acid containing 1 gm of salicylic acid, mix and allow to stand for 10 minutes. Add 5 gm of sodium thiosulphate and heat for 5 minutes. Cool, add 10 gm of potassium sulphate and heat until nearly or quite colorless. Dilute, distill and titrate as in the plain Gunning method. Make a blank determination of nitrogen in the reagents. Calculate the percent of nitrogen in the fertilizer. Kjeldahl-Gunning- Arnold Method. This method of digestion combines the accelerating action of mercury salts, potassium sulphate and cupric sulphate. Otherwise the method is not essentially different from those already described. *It is not applicable to fertilizers containing nitrates. Determination. Digest the usual amount of sample with 15 to 18 gm of potassium sulphate, 1 gm of cupric sulphate, 1 gm of mercury or mercuric oxide and 25 cc of concentrated sulphuric acid. Heat gently until frothing ceases, then boil the mixture briskly and continue the digestion until the solution is colorless or nearly so or until oxidation is complete. Cool, dilute with about 200 cc of water, add 50 cc of potassium sulphide solution and make basic and distill as in the Kjel- dahl method. Nitrogen of Ammonia. The various methods already de- scribed for the determination of nitrogen give only the total, making no distinction between nitrogen in different forms. It is sometimes desirable to know the relative amounts of this element existing as ammonium salts, nitrates and organic com- pounds. The official magnesium oxide method for determining nitrogen of ammonium salts follows : Determination. Place the weighed sample in a 400 or 500 cc distilling flask (the Kjeldahl digestion flask is a good substitute) with about 200 cc of water and 5 gm or more of magnesium oxide, free from carbonates. Connect with a tin condenser and distill 100 cc of the liquid into a measured volume of standard acid (50 cc of half-normal acid is usually suitable). Titrate the excess of acid with standard base and calculate the percent of ammoniacal nitrogen in the sample. For the determination of nitrates and ammonium salts to- gether the Ulsch-Street method is made official. AGRICULTURAL MATERIALS 521 Determination. Nitric and Ammoniacal Nitrogen. Place 1 gm of the sample in a 500-cc flask, add about 30 cc of water and 2 to 3 gm of iron reduced by hydrogen and, after standing sufficiently long to insure solution of the soluble nitrates and ammonium salts, add 10 cc of a mix- ture of strong sulphuric acid with an equal volume of water; shake thor- oughly, place a long funnel in the neck of the flask to prevent mechan- ical loss and allow to stand for a short time until the violence of the reaction has somewhat moderated. Heat the solution slowly, boil for five minutes and cool. Add about 100 cc of water, a little paraffin to prevent foaming and 7 to 10 gm of magnesium oxide, free or nearly so from carbonates. Connect with the tin condenser and boil for 40 minutes, or nearly to dryness, collecting the ammonia in 50 cc of half- normal acid. Titrate the excess of acid and calculate nitrogen of nitrates and ammonia. If the sample consists of nitrates alone, proceed as above, except that 25 cc of the nitrate solution, equivalent to 0.25 gm of the sample, is used and 5 gm of reduced iron. After boiling add 75 cc of water and an excess of saturated sodium hydroxide solution (instead of magnesium oxide) and distill as above described. Availability of Nitrogen. Mention has already been made of the low fertilizing value of certain nitrogenous materials because of the slow decomposition that results when the fertilizer is added to the soil. Nitrogen is probably directly assimilated by plants only in the most highly oxidized form, i.e., that of nitrates. Ammonium salts and certain organic materials, such as dried blood, have almost as great value because they readily decom- pose and oxidize in the soil, forming nitrates. Hoofs, hair, leather and hide do not so decompose, except very slowly and a method of differentiating between available and non-available forms of nitrogen is desirable. The microscope will detect ground hair and other similar materials but it can give only qualitative results. Fortunately qualitative results are all that are necessary in states where the addition of such materials is contrary to law, but for scientific purposes a quantitative dis- tinction between available and non-available nitrogen may be of great practical use. An exact analytical method for such a pur- pose seems to be impossible because there is no sharp distinction to be made between the classes of fertilizer materials. Great reliance is placed upon culture experiments, comparing the effect of using different fertilizers with plants under otherwise identical 522 QUANTITATIVE ANALYSIS conditions. Such experiments are slow and have no value what- ever for analytical purposes. An approximate distinction can be made by the use of potassium permanganate in either neutral or basic solution. Readily decomposable materials are oxidized and the nitrogen is converted into ammonia. It has not yet been determined how much reliance is to be placed upon these methods but they have been adopted as official methods by the Associa- tion of Official Agricultural Chemists. 1 Determination. Organic Nitrogen Soluble in Neutral Permanga- nate. Make a preliminary test as follows : Place 1 gm of the material upon an 11 cm filter paper and wash with water at room temperature until the filtrate measures 250 cc. Dry and determine nitrogen in the residue by the Kjeldahl, Gunning or Kjeldahl-Gunning-Arnold method, making a correction for the nitrogen of the filter paper if necessary. Place a weighed quantity of the fertilizer, equivalent to 50 mg of the water-insoluble organic nitrogen as determined above, on a moistened 11 cm filter paper and wash with water at room temperature until the filtrate measures 250 cc. Transfer the insoluble residue with 25 cc of tepid water (at about 30) to a 300 cc low-form beaker, add 1 gm of sodium carbonate, mix and add 100 cc of 2 percent potassium permanga- nate solution. Cover with a glass and immerse for 30 minutes in a water or steam bath so that the level of the liquid in the beaker is below that of the heating medium. Keep at 100, stirring twice at intervals of 10 minutes each. At the end of this time remove from the bath, add immediately 100 cc of cold water and filter through a heavy 15 cm folded filter. Wash with small quantities of cold water until the filtrate measures about 400 cc. Determine nitrogen in the residue and filter by either of the three official methods already described, correcting for the nitrogen contained in the filter. The nitrogen thus obtained is the inactive water-insoluble organic nitrogen. Subtract this percent from that obtained in the preliminary test. The remainder is the percent of organic nitrogen soluble in neutral permanganate. As already explained, this is an approximate measure of nitrogen easily available for plant food. Determination. Organic Nitrogen Soluble in Basic Permanganate. This method is not applicable to fertilizers containing cottonseed meal or castor pomace. Prepare the sample as follows: (a) Mixed Fertilizers. Make a preliminary test as directed in the neutral permanganate method. Place an amount of material, equiva- i J. Assoc. Off. Agr. Chem., Vol. I, No. 4, Pt. II, p. 11. AGRICULTURAL MATERIALS 523 lent to 50 mg of water-insoluble organic nitrogen, on a filter paper and wash with water at room temperature until the filtrate measures 250 cc. (b) Raw Materials. Make the determination of water-insoluble organic nitrogen as with mixed fertilizers. Place an amount of material equivalent to 50 mg of this nitrogen, in a small mortar, add about 2 gm of powdered rock phosphate (to facilitate the washing process), mix thoroughly and transfer to a filter paper. Wash with water at room temperature until the filtrate measures 250 cc. When much oil or fat is present it is well to wash with ether and allow to stand until the odor of the latter has disappeared before washing with water. Dry the residue from either class of materials at a temperature not exceeding 80 and transfer from the filter to a 500 cc Kjeldahl digestion flask. Add 20 cc of water, about 1 gm of powdered porcelain to prevent bumping and about 1 gm of paraffin to prevent frothing. A solution of potassium permanganate is made by dissolving 25 gm in about 100 cc of water. Also dissolve 150 gm of sodium hydroxide in 500 cc of water and, after this has cooled, mix with the potassium permanganate solu- tion and dilute the whole to 1000 cc. 100 cc of this basic permanganate solution is added to the flask containing the fertilizer and this is then connected with the tin condenser, the lower end of which dips into 50 cc of half-normal acid. Digest slowly for at least 30 minutes, below distillation point, with a very low flame, using coarse wire gauze and asbestos paper between the flask and flame. Gradually raise the temperature and, after any danger of frothing has passed, distill until 95 cc of the distillate is obtained and titrate as usual. When a tendency to froth is noticed lengthen the digestion period and no trouble will be experienced when the distillation is begun. During the digestion gently rotate the flask occasionally, particularly if the material shows a tendency to adhere to the sides of the flask. The nitrogen thus obtained is the active water-insoluble organic nitrogen. Phosphorus. The most important sources of phosphorus for fertilizing purposes are mineral phosphates (chiefly apatite, which is calcium orthophosphate, Ca 3 (PO 4 )2, but contains some mag- nesium phosphate) ground raw and steamed bone, slag from basic Bessemer steel furnaces (known as " Thomas slag") and, to a less extent, fish scrap, oil cake and tankage. The normal calcium orthophosphate of mineral deposits has a very small solubility in water or in any salt or acid solutions commonly occurring in soils. It is, in consequence, generally 524 QUANTITATIVE ANALYSIS considered as a source of non-available phosphorus. In order to change it to a soluble form so that it may be used as a fer- tilizer, the mineral phosphate is treated with sulphuric acid, there being formed calcium acid phosphates and sometimes free phosphoric acid if an excess of sulphuric acid has been used: Ca 3 (P0 4 )2+H 2 S04->CaS0 4 +2CaHP04, Ca 3 (P0 4 )2+2H 2 SO 4 -2CaS04+Ca(H 2 P0 4 )2, Ca 3 (P04)2+3H 2 S04-*3CaS0 4 +2H 3 P04. Normal calcium phosphate (tricalcium phosphate) is almost insoluble in water. Di calcium phosphate, CaHP0 4 , dissolves in water to the extent of only 0.136 gm in 1000 cc at 25, but dis- solves quite easily in certain salt solutions, as ammoniiim citrate. Monocalcium phosphate, Ca(H 2 P04) 2 , is easily soluble in water. Salt solutions existing in soils dissolve dicalcium phosphate in a manner similar to that shown by ammonium citrate solution. On this account the analyst speaks of " water soluble," " citrate soluble,'' and " insoluble" phosphate. Water soluble and citrate soluble forms are taken together as "available" phosphate so that a distinction between the two forms composing this class is now seldom required. When a mixture of the compounds formed by " acidulating" phosphate rock is allowed to stand a reaction occurs between normal calcium phosphate and monocalcium phosphate: Ca 3 (P0 4 )2+Ca(H 2 P0 4 ) 2 ^4CaHP04. The dicalcium phosphate formed in this way is known as " re- verted" phosphate. This term is gradually falling into disuse because it is not possible to distinguish between truly reverted phosphate and dicalcium phosphate formed by the action of sulphuric acid upon the normal calcium phosphate. The phosphorus of bones is in the form of normal calcium phosphate but it is in a condition which makes it possible for soil acids to readily convert it into acid phosphates. It also dissolves in ammonium citrate solution for the same reason and it is therefore properly classed as citrate soluble and as available. In reporting the analysis of fertilizers phosphorus is calculated as phosphorus pentoxide, which is often improperly called "phos- AGRICULTURAL MATERIALS 525 phoric acid." It is therefore customary to speak of total "phos- phoric acid," and of water soluble, citrate soluble and available "phosphoric acid," meaning by these terms phosphorus in the various forms already described, but calculated as the pentoxide. These terms will not, however, be used in the following paragraphs. Total Phosphorus. The various methods for the determina- tion of phosphorus have already been discussed in connection with steel analysis, pages 452 to 457. The same methods are used in the analysis of fertilizers but the preliminary treatment will, of course, be quite different from that of steel. This treat- ment must include (a) destruction of organic matter and (b) solution of the phosphate. The details of the following methods are essentially those of the "official" methods of the A. 0. A. C. The choice of method for dissolving the sample will depend upon the nature of the latter. Preparation of Solution. Treat 2.5 gm of the sample by one of the following methods : (a) Ignite in a crucible until organic matter is removed and dissolve in hydrochloric acid. (6) Evaporate with 5 cc of magnesium nitrate solution made as follows: Dissolve 320 gm of calcined magnesium oxide in nitric acid, avoiding an excess of the latter; add a little calcined magnesium oxide in excess, boil, filter from the residue and dilute to 2000 cc. After evaporating the fertilizer and magnesium nitrate solution, ignite until organic matter is removed and dissolve in hydrochloric acid. (c) Boil with 20 to 30 cc of concentrated sulphuric acid in a Kjeldahl flask, adding 2 to 4 gm of sodium nitrate at the beginning of the diges- tion and a small quantity after the solution has become nearly colorless, or adding the nitrate in small portions from time to time during the digestion. After the solution is colorless add 150 cc of water and boil for a few minutes. (d) Digest in a Kjeldahl flask with concentrated sulphuric acid and such other reagents as are used in either the plain or modified Kjeldahl or Gunning method for the determination of nitrogen. Do not add any potassium permanganate but, after the solution has become colorless, add about 100 cc of water and boil for a few minutes. (e) Dissolve in 30 cc of concentrated nitric acid and 5 cc of concen- trated hydrochloric acid and boil until organic matter is destroyed. (/) Add 30 cc of concentrated hydrochloric acid, heat and add cau- 526 QUANTITATIVE ANALYSIS tiously, in small quantities at a time, about 0.5 gm of finely pulverized potassium or sodium chlorate to destroy organic matter. (g) Dissolve in 15 to 30 cc of concentrated hydrochloric acid and 3 to 10 cc of concentrated nitric acid. This method is recommended for fertilizers containing much iron or aluminium phosphate. After the sample of fertilizer has been brought into solution by any of the methods described above cool, dilute to 250 cc, mix and pour into a dry filter, discarding the first 10 cc of the filtrate and allowing the remainder to run into a dry flask which can be stoppered. Gravimetric Determination. The following special reagents are necessary: Ammonium molybdate solution, ammonium nitrate solution and "magnesia mixture," all to be prepared as directed for the determi- nation of phosphorus in steel, page 456. Fill a dry 100 cc volumetric flask with the phosphate solution and rinse into a 250 cc flask of resistance glass; or measure 50 cc or 25 cc, according to the percent of phosphorus present, by means of a pipette. Neutralize with ammonium hydroxide and clear with a few drops of nitric acid, thus dissolving the small amount of precipitated hydroxides of iron and aluminium. In case hydrochloric or sulphuric acid has been used as a solvent for the fertilizer material add also 15 gm of dry ammonium nitrate. To the hot solution add ammonium molybdate solution, about 70 cc for each decigram of phosphorus pentoxide that is thought to be present. Immerse in water and digest at 65 for an hour and determine whether the phosphorus has been completely precipitated by adding more moly- bdate solution to the clear, supernatant liquid. If more precipitate forms continue the digestion, followed by testing as before. Filter on paper and wash with cold water or, preferably, ammonium nitrate solution. During the washing the precipitate that adheres to the flask need not be completely removed but it must be washed. Place the flask in which precipitation was made under the filter and dissolve the precipitate on the filter in concentrated ammonium hydrox- ide (using as little as possible) followed by hot water, allowing the solu- tion to run into the flask, thus dissolving the adhering precipitate. Wash the paper very thoroughly with hot water. Transfer the entire solution and washings to a 250 cc beaker of resistance glass. The total volume of the solution should not be greater than 100 cc. Nearly neutralize with hydrochloric acid, the reformation of the yellow pre- cipitate serving as indicator. Redissolve the precipitate that finally forms, by the addition of a few drops of dilute ammonium hydroxide. Cool and add, very slowly and with vigorous stirring, 25 cc of magnesia mixture. After 15 minutes add ammonium hydroxide (specific gravity 0.90) equal to one-ninth of the total volume of the solution, stirring as AGRICULTURAL MATERIALS 527 this is added. Cover and allow to stand for two hours. Filter and wash with dilute ammonium hydroxide (the concentrated solution diluted to ten times its original volume) until practically free from chlorides, as shown by acidifying with nitric acid and adding silver nitrate solution. Dry the filter and precipitate and transfer the latter to a porcelain crucible, previously ignited and weighed. Ignite the filter separately and trans- fer its ash, when white, to the crucible containing the main precipitate. Ignite to whiteness or grayish white over the blast lamp or Me*ker burner, weigh and calculate the percent of phosphorus pentoxide. Volumetric Determination Have the following solutions ready: (a) Ammonium Molybdate. To 100 cc of the molybdate solution that was prepared for the gravimetric determination of phosphorus add 5 cc of concentrated nitric acid. The solution should be filtered imme- diately before using. (6) Fifth-normal Sodium Hydroxide. Prepared from boiled water. (c) Standard Hydrochloric or Nitric Acid. This solution should be equivalent in strength to the standard base. It should be made from previously boiled and cooled water and should be standardized by titra- tion against sodium carbonate, using methyl orange as indicator. The fertilizer is dissolved by either of methods (6), (e), (/) or (0). Method (e) is to be preferred if the material will yield to this treatment. The solution is to be diluted and filtered as already directed. In the case of fertilizers containing less than 5 percent of phosphorus pentoxide, use an aliquot corresponding to 0.5 gm of substance. If the percentage is between 5 and 20 use an aliquot corresponding to 0.1 gm of substance. Add 5 to 10 cc of concentrated nitric acid, the amount depending upon whether this acid has been used in making the solution; or add ammon- ium nitrate equivalent to this amount of nitric acid. Nearly neutralize with ammonium hydroxide, precipitation of hydroxide of iron or alu- minium serving as indicator. Clear with a drop of nitric acid, dilute to about 100 cc and heat by immersing in water to 60 to 65. For phos- phorus pentoxide percents below 5 add 20 to 25 cc of freshly filtered molybdate solution; for percentages between 5 and 20 add 30 to 35 cc of molybdate solution. For percentages greater than 20 add sufficient molybdate solution to insure complete precipitation of the phosphorus. Stir, allow to stand in the bath for fifteen minutes and filter at once. Wash twice with water by decantation, using 25 to 30 nc each time and agitating and settling each time before decanting. Transfer the precip- itate to the filter as thoroughly as can be done without the use of a policeman and wash the flask and filter with cold water until the filtrate from two fillings of the filter yields a pink color upon the addition of phenolphthalein and one drop of the standard base. 528 QUANTITATIVE ANALYSIS Return the filter paper and precipitate to the flask in which pre- cipitation was made. Add a measured, small excess of the standard base to dissolve the yellow precipitate, add a few drops of phenolphthalein and titrate the unused excess of base with standard acid. Calculate the percent of phosphorus pentoxide in the sample. The following changes in the method just described are made optional: (a) Heat the solution to only 45 to 50 and allow to stand in the bath, after the addition of the molybdate solution for 30 minutes. (6) Cool to room temperature before adding the molybdate solution. Add the latter at the rate of 75 cc for each decigram of phosphorus pentoxide present, place the stoppered flask containing the solution in a shaking apparatus and shake for 30 minutes at room temperature. Fil- ter at once and proceed as already directed. Water Soluble Phosphorus. The phosphorus of untreated phosphate rock and of bone and other organic sources is in- soluble in water and this determination is omitted. In "acidu- lated " samples it is sometimes required although, as has already been stated, citrate insoluble phosphorus subtracted from total phosphorus gives available phosphorus and there is little object in making the determination in any case. Following is the "official" method., Gravimetric Determination. Place 2 gm of the sample on a 9-cm filter, wash with successive small portions of water, allowing each portion to pass through before adding more, until the volume of the filtrate is about 250 cc. Preserve the residue on the filter for the determination of citrate insoluble phosphorus. If the filtrate is turbid add nitric acid until clear. Dilute to 500 cc in a volumetric flask, mix and determine the phosphorus in 50 or 100 cc portions by the method above described for total phosphorus. Calculate the percent of water soluble phos- phorus, expressed as phosphorus pentoxide. Volumetric Determination. Wash a 2-gm or 4-gm sample as directed for the gravimetric method and dilute the filtrate to 500 cc. To 50 cc portions add 10 cc of concentrated nitric acid, then ammonium hydrox- ide until a slight permanent precipitate is formed. The volume should now be between 60 cc and 75 cc. Proceed from this point as in the volu- metric determination of total phosphorus. Citrate Insoluble Phosphorus. Phosphates that are insoluble in neutral ammonium citrate solution, and therefore in soil solutions, are so slowly assimilated by plants that they are often considered to be of small value as fertilizers. The greatest diffi- AGRICULTURAL MATERIALS 529 culty exists, however, in making an accurate determination of insoluble phosphate because of the difficulty that is encountered in the preparation of a neutral ammonium citrate solution. In the discussion of indicators it was shown that no indicator is sufficiently sensitive to both acids and bases as to indicate accu- rately the point of neutralization if both acid and base are weakly ionized. It is therefore extremely difficult to neutralize exactly the weak citric acid by the weak base, ammonium hydroxide, with the aid of any organic indicator. The action of ammonium citrate upon dicalcium phosphate is due to the presence of citric acid of hydrolysis: It is therefore highly important that the concentration of citric acid in the solution should be the same in all cases if the ana- lytical results are to possess any significance, as the action of the solution is, at best, but an arbitrary and approximate imitation of the action of solutions found in soils. Ammonium Citrate Solution. Two methods for the prepara- tion of neutral ammonium citrate solution are approved by the A. O. A. C. The first method is that of neutralizing a stated quantity of citric acid in solution by ammonium hydroxide, the indicator being litmus or azolitmin paper. In the second method the solution is nearly neutralized and then a measured volume of a solution of calcium chloride in water and alcohol is added. Calcium citrate is at once precipitated. If the solution was neutral, ammonium chloride is the only other product of the reaction; if an excess of citric acid was present hydrochloric acid is also produced: If the solution was basic instead of neutral or acid, ammonium hydroxide remains after the precipitation. By testing the solu- tion with cochineal after .filtration and then adding either am- monium hydroxide or citric acid, as is indicated as being neces- sary by the reaction with cochineal, and by repeating this process as often as is necessary, the solution may finally be brought to a neutral condition. The advantage of this method over the first 34 530 QUANTITATIVE ANALYSIS official method is in the substitution of an equivalent amount of a strong acid (hydrochloric acid) for the weak citric acid, so that an indicator may now be chosen of sufficient sensibility toward ammonium hydroxide to give indication of real neu- trality. Even by this method, however, the ratio of ammonia to citric acid is found to vary. McCandless 1 found that this ratio for solutions made by nine different chemists varied between the limits 1 : 3. 775 and 1 :4.189. The calculated ratio for normal ammonium citrate is 1:3.766. The excess of acid indicated by the above ratios would give the solution a greater solvent power for calcium phosphate and would give rise to incorrect results in the determination of insoluble phosphorus. Patten and Marti have devised a method 2 which they call the "titration method" and have shown that strictly rreutral solu- tions can easily be made. The solution is first made approxi- mately neutral and then both ammonia and citric acid are deter- mined. The determination of ammonia is made by treating 5 cc of the solution with magnesium oxide and distilling the ammonia into standard acid: 2(NH 4 )3C 6 H50 7 +3Mg(OH) 2 ->Mg3(C 6 H50 7 )2+6NH 3 +6H 2 0. The excess of standard acid is titrated as in the Kjeldahl method for the determination of nitrogen. The determination of total citric acid in the solution depends upon the reaction of formalde- hyde with ammonia, either free or combined in salts, to form hexamethylenetetramine, a substance so weakly basic that it does not affect phenolphthalein. The reaction is represented by the following equations: 4NH 4 OH+6HCHO->N(CH 2 NCH 2 ) 3 +10H 2 0, 3(NH 4 )3C 6 H 5 07+18HCH0^3N(CH 2 KCH 2 ) 3 +18H 2 04- 4H 3 C 6 H 5 07. The free citric acid is then titrated by a standard solution of a strong base, as sodium hydroxide, in presence of phenolphthalein. This device is somewhat similar to that used in the calcium chloride method where the weak acid of the citrate solution is 1 U. S. Dept, Agr., Bur. Chem., Bull. 122, 147. 2 J. Ind. Eng. Chem., 5, 567 (1913). AGRICULTURAL MATERIALS 531 substituted by a strong acid and the question of neutrality is qualitatively decided by the use of an indicator that is sensitive -to weak bases. In the Patten and Marti method the weak base of the citrate solution is destroyed and the remaining acid is titrated by a strong base, using an indicator that is sensitive to weak acids. The ratio of ammonia to citric acid now having been quantita- tively determined the solution is exactly neutralized by the addi- tion of the calculated quantity of either ammonium hydroxide or citric acid. Determination. In addition to the solutions used for the determina- tion of total phosphorus, prepare a neutral solution of ammonium citrate by either of the following methods : (a) Dissolve 370 gm of commercial citric acid in 1500 cc of water and nearly neutralize with commercial ammonium hydroxide. Cool and add dilute ammonium hydroxide until exactly neutral, testing with litmus or azolitmin paper. Dilute the neutral solution until the specific gravity is 1.09 at 20 C. .(&) To 370 gm of commercial citric acid add commercial ammonium hydroxide until nearly neutral. Reduce the specific gravity by dilution until it is slightly greater than 1.09 at 20 and make exactly neutral, testing as follows : Prepare a solution of fused calcium chloride in water, 200 gm to each liter, and add 4000 cc of 95 percent alcohol. Make this solution neutral, using a freshly prepared corallin solution as preliminary indicator, and test finally by withdrawing a portion, diluting with an equal volume of water and testing with cochineal solution. 50 cc of this calcium chloride will precipitate the citric acid from 10 cc of the citrate solution. To 10 cc of the nearly neutral citrate solution add 50 cc of the alcoholic calcium chloride solution, stir well, filter at once through a folded filter, dilute the filtrate with an equal volume of water and test the reaction with a neutral solution of cochineal. If the citrate solution is shown not to be neutral, carefully add ammonium hydroxide or citric acid, as may be necessary, to the solution, mix and test again. Repeat this process until a neutral reaction is obtained. Add sufficient water to reduce the specific gravity to 1.09 at 20. (c) Method of Patten and Matfi. Dissolve 370 gm of commercial citric acid in 1500 cc of water; add 358.3 cc of ammonium hydroxide (sp. gr. 0.90) and allow to cool. Measure 50 cc of this solution into a 250-cc volumetric fl ask, dilute to the mark and mix thoroughly. From a burette measure 5 cc of the diluted solution into a beaker, add 4 cc of a neutral 40-percent solution of formaldehyde and titrate with tenth-normal 532 QUANTITATIVE ANALYSIS sodium hydroxide or potassium hydroxide solution in presence of phenolphthalein. The pink color should remain after the solution is boiled; if it does not the ammonia has not been entirely decomposed and another titration should be made, using more formaldehyde. Determine the total (free and combined) ammonia in the solution as follows: Carefully measure 5 cc of the diluted solution into a 500-cc Kjeldahl digestion flask, add 0.5 gm of magnesium oxide and at once distill into 50 cc of fifth-normal acid. Titrate the excess of acid using cochineal as indicator. From the titration of citric acid and of ammonia calculate the amount of citric acid or of a standard solution of ammonium hydroxide that must be added to 1450 cc of the stronger solution of ammonium citrate in order to make an exactly neutral solution. After neutralization dilute the solution to 2000 cc. The specific gravity of the solution as finally diluted should be 1.09. Acidulated Samples. Place 100 cc of ammonium citrate solution in a 250 cc flask which is immersed in warm water. Heat to 65 then drop into the solution the paper containing the washed residue from the determination of water-soluble phosphorus, close the flask with a rubber stopper and shake vigorously until the paper is reduced to a pulp. (If a determination of water soluble phosphorus has not been required the untreated sample may be used for the determination of citrate insoluble phosphorus.) Replace the flask in the water bath and keep at 65, shaking every 5 minutes. At the expiration of exactly 30 minutes from the time that the material was introduced remove the flask from the bath and filter the contents as rapidly as possible. Wash thoroughly with water which is at a temperature of 65 until the filtrate measures about 350 cc, allowing time for thorough draining before adding new portions of water; then determine nitrogen in the residue by either of the methods used for total phosphorus, using any of methods (a), (6), (c), (d) or (e) for dissolving the extracted material. From the percent thus obtained for citrate insoluble phosphorus and the percent of total phosphorus calculate the percent of available phosphorus. Also, if water soluble phosphorus has been determined, calculate the percent of citrate soluble phosphorus. Express all results as percents of phosphorus pentoxide. Non-acidulated Samples. Treat 2 gm of the sample, without pre- vious washing with water, as directed for acidulated samples, except when the material contains much animal matter, such as bone, fish, etc., in which case dissolve the residue which is insoluble in ammonium citrate by either of methods (6), (c) or (d). Determine phosphorus by gravimetric or volumetric methods, already described. AGRICULTURAL MATERIALS 533 Potassium. All soils contain certain minerals, one component of which is potassium, and these are continually undergoing decomposition yielding potassium in a soluble form. The most important of such minerals belong to the class of felspars. Soils are frequently deficient in potassium and require fertiliza- tion with some material containing this metal. It is interesting to* note that sodium, so nearly allied to potassium in chemical properties, is able to replace it in the plant organism very little, if at all. The various potassium fertilizers may be sold singly or they may be components of mixed fertilizers. Organic matter may or may not be present. If it is present it must be destroyed by oxidation and the earth and alkaline earth metals must be separated. The separation of organic matter and interfering metals and the determination of potassium are best accomplished by the Lindo-Gladding method, the principles of which have already been discussed (page 102). The reagents there described are necessary in this connection. Following are the official methods of the A. O. A. C. 1 Determination. Prepare the potassium solution from the fertilizer as follows: (a) Mixed Fertilizers, Wood Ashes and Cotton Hull Ashes. Place 2.5 gm of sample upon a 12.5 cm filter paper and wash with boiling water until the nitrate measures about 200 cc. Add to the filtrate 2 cc of concentra- ted hydrochloric acid, heat to boiling and transfer to a 250 cc volumetric flask. Add to the hot solution a slight excess of ammonium hydroxide and enough ammonium oxalate to precipitate all of the calcium. Cool to 20, dilute to the mark on the flask, mix and pass through a dry filter, rejecting the first 25 cc of the filtrate. (6) Potassium Salts, Potassium Magnesium Sulphate and Kainite. Dissolve 2.5 gm of sample in a 250 cc volumetric flask and dilute to the mark without the addition of ammonium hydroxide or ammonium oxalate. (c) Organic Compounds: Cottonseed Meal, Tobacco Stems, Etc. Satu- rate 10 gm of sample with concentrated sulphuric acid and evaporate and ignite at dull redness to destroy organic matter. A muffle furnace will be found to be convenient for this operation. Add a little concen- trated hydrochloric acid and warm slightly in order to loosen the mass 1 J. Assoc. Off. Agr. Chem., Vol. I, No. 4, Pt. II, p. 12. 534 QUANTITATIVE ANALYSIS from the dish. Dissolve in hot water, cool and dilute to 250 cc. Mix thoroughly in the flask. For each class of materials two methods for the determination are official. For both of these the following special reagents will be nec- essary. (a) Ammonium Chloride Solution. Dissolve 100 gm of ammonium chloride in 500 cc of water, add 10 gm of powdered potassium chlor- platinate, warm slightly and shake at intervals for 6 to 8 hours. Allow the mixture to settle for several hours and filter. The residue of potas- sium chlorplatinate may be used for the preparation of a fresh supply. (b) Chlorplatinic Acid Solution. This solution should contain 1 gm of platinum in each 10 cc. (c) Alcohol. An 80 percent solution, specific gravity 0.8645 at 15/15. Denatured alcohol, made according to formula 1, (U. S. Internal Rev., Reg. No. 30, Revised Aug. 22, 1911, p. 45) and diluted "to make 80 percent alcohol by volume, may also be used. Method I (Lindo-Gladding) (a) Mixed Fertilizers, Wood Ashes and Cotton Hull Ashes. Evaporate 50 cc of the prepared solution nearly to dryness in a dish, add 1 cc of sulphuric acid (1 to 1), evaporate to dryness and ignite at a full red heat until organic matter is removed and the residue is white. Dissolve the residue in hot water, using at least 20 cc for each decigram of potas- sium oxide present, add a few drops of concentrated hydrochloric acid and enough chlorplatinic acid to precipitate all of the potassium and to leave about 1 cc of platinum solution in excess. If the percent of potas- sium is approximately known the quantity of platinum solution that is necessary should be calculated. Contamination with ammonia vapor must be avoided. Evaporate the solution on a steam bath to a thick paste and add to the residue 25 cc of 80 percent alcohol. Stir thoroughly and allow to stand for* 15 minutes. Filter through a Gooch or paper filter. If the filtrate is not colored sufficient chlorplatinic acid solution is not present and the analysis must be begun again with another portion of the solution, in- creasing the amount of platinum solution. Wash the precipitate on the filter with 80 percent alcohol, continuing the washing after the filtrate has become colorless. Remove the filtrate and washings to the bottle which has been provided for waste platinum solutions and wash the precipitate five times with 10 cc portions of the ammonium chloride solution. Wash again thoroughly with 80 percent alcohol, exercising particular care to remove ammonium chloride from the upper part of the filter. Dry the precipitate and filter for 30 min- AGRICULTURAL MATERIALS 535 utes at 100. For the subsequent treatment in case a paper filter has been used, see page 102. For the Gooch filter the weight of potassium chlorplatinate is given without further treatment. (6) Commercial Potassium Chloride ("Muriate of Potash"). To 50 cc of the solution already prepared add a few drops of hydrochloric acid and 10 cc of chlorplatinic acid solution. Evaporate over a steam bath to a thick paste and treat the residue as in the case of mixed fertilizers. (c) Potassium Sulphate, Potassium Magnesium Sulphate and Kainite. Acidify 50 cc of the solution with a few drops of hydrochloric acid, add 15 cc of chlorplatinic acid solution and evaporate on the steam bath to a thick paste. From this, point proceed as with mixed fertilizers, except that 25 cc portions of the ammonium chloride solution should be used in the washing process. The potassium is reported as percent of potassium oxide (often called "potash") instead of as the element. Method II This method is not recommended for materials containing soluble sulphates, therefore it is practically restricted to certain mixed fertilizers and to potassium chloride and nitrate. The required reagents are the same as in method (/). The solution of potassium is prepared as already directed except that in all cases the addition of ammonium hydroxide and ammonium oxalate is omitted. Dilute 25 cc (50 cc if the percent of potassium oxide is less than 10) to 150 cc, heat to boiling and add, drop by drop and with constant stirring, a slight excess of barium chloride solution until no further precipitate of barium sulphate is produced. Without filtering add in the same manner barium hydroxide (saturated solution) in slight excess. Filter while hot and wash until the washings test free from chlorides. Add to the filtrate 1 cc of concentrated ammonium hydroxide and then a saturated solution of ammonium carbonate (prepared without heating) until the excess of barium is precipitated. Heat and add, in fine powder, 0.5 gm of pure oxalic acid or 0.75 gm of ammonium oxalate. This completes the precipitation of barium. Filter, wash free from chlorides, evaporate the filtrate to dry ness in a platinum dish and ignite care- fully below redness until all volatile ammonium salts are driven off. Digest the residue with hot water, filter through a small filter and dilute the filtrate, if necessary, so that for each decigram of potassium oxide there will be at least 20 cc of liquid. Acidify with a few drops of hydrochloric acid and add an excess of chlorplatinic acid. Evaporate on a steam bath to a thick paste, cool and add 25 cc of 80 percent alcohol. Filter on a Gooch or paper filter and wash with 80 percent alcohol 536 QUANTITATIVE ANALYSIS several times after the filtrate is colorless. Dry for 30 minutes at 100 and weigh. If there is an appearance of any white salts in the precipi- tate, washing with ammonium chloride solution, followed by 80 percent alcohol, may be necessary, as described for method (/). MILK The milk of most mammals has been analyzed and its compo- sition determined but, for practical purposes, the analyst rarely has to do with any other than cow's milk and human milk. The analysis of cow's milk may be made for purely scientific purposes as, for instance, the determination of the relation between the composition of milk and the breed of animal, the season of the year or the rations upon which the animal is fed, or the determination of the changes that occur in composition during the period of storage, and other similar questions. The analysis may also be made for purposes of legal control to detect sophistication. The analysis of woman's milk is usually made for hygienic purposes, in order to provide a basis for modification of the mother's diet, etc., in cases where the infant is not thriving. The percentage composition of milk varies rather widely although the same substances are found in practically all milk from a given species of animal. It is therefore not possible to fix, by legal enactment, the exact composition of milk that is to become an article of commerce, but certain minimum figures are usually established by law and any milk containing a con- stituent in quantity below the legal minimum is considered to be adulterated. The average composition of cow's milk is given by Babcock in the table on the following page. Adulteration of Milk. Adulteration is practised by skimming or watering, or by both methods. There is a rather widely disseminated popular belief that the only part of milk that has much value is the cream. Many municipalities control the com- position of milk only by specifying a minimum for fat and the milk inspection will often include little else than a determina- tion of the percent of fat. Milk containing a high percent of butter fat, such as that from Jersey cows, could then be watered in such a way as to leave the legal minimum of fat but the solids not fat would thereby fle lowered. AGRICULTURAL MATERIALS 537 ?o CO CO O5 ^ 00 3 1 8 1 J CO co oo ic i" H 1> CO d co rt d o* Jl trs"" "o g :3 03 ij 03 *i tc | o | '| ?. o 3 c . 1 2 - s g -3 a 'gsS ll'| tj o ^ ^ "^ ^ 2 PK OQ fi 03 TK QQ ^ r-j ^ *" 1 ^^ CO C^ co d o 1000 t^^Ht^ 00 l^t^-Tti r i O C^ I>O 1 t O > 1 OOO 1-H I-H - d d o" d o d d d 'aT'oT ' ^ : .o ^ : a -^ ' JJ : 'S i d fl O PH *H O G O +* "^ * ^ r* a a .S .S .S !2 .S ^ a fS '3 c3 QJ O co ^rtxsg 'TJ-csa) ^So^ 52 ^ fH 3 d '43 ^^flS^ O >H ^ x p .s a 8 538 QUANTITATIVE ANALYSIS Proper control of adulteration can be secured by a considera- tion of the relations between total solids, solids not fat and fat. Watering has the effect of lowering the percent of total solids, solids not fat and fat, proportionately. Skimming lowers the percent of fat and total solids and slightly raises the percent of solids not fat. If both watering and skimming are practised the ratio of solids not fat to total solids is slightly increased, while the ratio of fats to total solids is abnormally low, all three per- cents being lowered. The methods for the analysis of dairy products as used in most laboratories are the official and provisional methods of the A. O. A. C. 1 These are given, with certain modifications, in the following pages. It should be noted that the fat of milk quickly rises when the milk is quiet and the sample must be mixed thoroughly by pour- ing from one vessel to another immediately before the removal of any sample for a determination. Violent agitation must be avoided as this will result in coalescence of the fat globules. The analysis should be completed within the shortest possible time after the sample is received, in order to avoid the inter- ference of fermentative or putrefactive changes. If it is impos- sible to make all of the determinations within a short time the sample for each determination should be removed and weighed, after which the work may be interrupted for a reasonable period at the following stages in the various determinations: Determination j Stage after which work may be interrupted Specific gravity Total solids Ash Total nitrogen Casein and albumin. . Lactose Fat (centrifugal) Fat (gravimetric) Formaldehyde Must be completed at once Evaporation of sample Weighing the sample Addition cf sulphuric acid Addition of sulphuric acid to filtered proteids Addition of mercuric salt solution Addition of sulphuric acid Drying of paper coil with milk Test must be made at once. In case it is impossible to begin any of the determinations while the sample is fresh, add formaldehyde (1 part of 40 percent solu- tion to 2500 parts of milk) and place on ice. !J. Assoc. Off. Agr. Chem., Vol. II, No. 3, Pt. II, p. 287. AGRICULTURAL MATERIALS 539 Specific Gravity. The specific gravity of milk may be deter- mined by means of a hydrometer, a Westphal balance or a pyc- nometer, or by any method used for other liquids. It is most convenient, and sufficiently accurate, to use a hydrometer, which should be standardized on account of the large errors that are frequently made in graduating these instruments. It is unfortu- nate that arbitrary scales have come into use for the expression of the density of milk. The instrument called the "lactometer" is simply a hydrometer having an arbitrary scale of its own, sometimes also reading directly in specific gravity. Quevenne's lactometer is graduated in degrees from 15 to 40, correspond- ing to specific gravity 1.015 to 1.040. The New York Board of Health lactometer is graduated also in arbitrary degrees, corresponding to a specific gravity of 1 and 100 to a specific gravity of 1.029, the latter figure being considered as the specific gravity of pure average milk. Degrees on the New York lac- tometer would thus roughly indicate the percent of whole milk in a milk and water mixture. Such an indication can have but slight value in milk testing. Determination. Mix the sample of milk by pouring from one vessel to another several times, avoiding violent agitation. Immediately determine the specific gravity at 15.5, using a lactometer. Record both specific gravity and lactometer degrees. Total Solids. A statement of the chemical nature of the solids of milk has already been given in the table on page 537. The fatty constituents are present in the form of an emulsion of minute globules which can be made to coalesce by agitation, such as occurs during churning. The proteids are chiefly in the form of sols from which they are flocculated by such means as the addition of acids, boiling, etc. The well-known " curdling" of milk when it "sours" is chiefly due to the development of lactic acid by fermentation of lactose. Lactose itself is present as a true solution, as are also the . inorganic constituents, although calcium phosphate combines in some manner with casein. The determination of total solids must be made at a tempera- ture not higher than 100 in order to avoid chemical changes. Great difficulty is often experienced in the attempt to remove completely the water, on account of the formation of a skin of 540 QUANTITATIVE ANALYSIS flocculated proteids. This effect may be minimized by placing a small amount of sand in the dish and stirring occasionally. The drying surface is, in this way, very much increased. Aluminium dishes are suitable for this determination. They should be not less than 5 cm in diameter and should be flat on the bottom. Determination. Clean, dry and weigh a dish containing 15 to 20 gm of clean sand and a short glass rod, then add about 5 gm of milk. The sample may be weighed in the dish or it may be accurately measured after the specific gravity has been determined. In either case it must be well mixed as directed for the determination of specific gravity. Dry at 100 until the weight is constant, stirring occasionally with the glass rod, which is left in the dish. Cool in a desiccator and weigh rapidly, in order to avoid absorption of moisture. Calculate the loss in weight as percent of water. Ash. The ash of milk contains the inorganic constituents in the form in which they are left after burning the solids and it does not represent the original combinations of these constituents. Determination. Weigh a platinum dish, add about 20 gm of milk and quickly reweigh. Add 6 cc of concentrated nitric acid, evaporate to dryness over a steam bath and ignite at a temperature just below redness until free from carbon. Weigh and calculate the percent of ash. Total Nitrogen. The nitrogenous constituents of milk are usually calculated as proteids, although traces of other nitroge- nous compounds are present. While the molecular constitu- tion of the proteids is not known the percentage composition is established as varying within fairly narrow limits for the different proteids. The following table expresses the approximate composition. Element Percent Carbon 50 to 55 Hydrogen 6 9 to 73 Nitrogen 15 Oto 19 Oxygen 19 to 24 Sulphur 0.3 to 2.4 Both casein and albumin contain slightly less than 15.7 percent of nitrogen and, since these compounds make up more than 95 percent of the total proteids of milk, the percent of total proteids is found with sufficient accuracy by multiplying the percent of total nitrogen by 6.38 ( = AGRICULTURAL MATERIALS 541 Nitrogen is determined by the Kjeldahl or Gunning method, as already discussed in connection with fertilizers, the sample of milk being digested without previous evaporation. Determination. Having accurately determined the specific gravity of the sample, mix well and measure 5 cc into a Kjeldahl digestion flask, using a calibrated pipette, and calculate the weight. Without evaporating the milk proceed to determine nitrogen by either the Kjeldahl or Gunning method, described on pages 516 and 519. Mul- tiply the percent of nitrogen by 6.38 and record the result as percent of total proteids. Casein. Casein is present in milk as a sol and it has a high degree of molecular association. It is flocculated by dilute acids and will not thereafter redissolve, so that it is to be classed as an irreversible colloid. It dissolves in concentrated acids but its chemical nature is thereby changed. Casein is approximately separated from the other proteids by warming to 40 after the addition of water and acetic acid. Casein flocculates and is separated by filtration. Nitrogen in this residue is determined and casein is calculated. Casein may also be precipitated by heating to 40 with a solution of potassium aluminium sulphate. Determination. Method I. Weigh a covered 125 cc beaker to milli- grams, add 10 cc of milk and quickly reweigh. Add 90 cc o f water which is at a temperature of 40 to 42 and add, at once, 1.5 cc of a 10 percent solution of acetic acid. Stir with a glass rod and allow to stand for 5 minutes longer. Filter and wash three or four times with cold water, saving the filtrate and washings, the total volume of which should not be greater than 125 cc. The filtrate should be quite clear. Place the paper and casein in a Kjeldahl digestion flask and determine nitrogen by the Kjeldahl or the Gunning method. Multiply the percent of nitrogen by 6.38 and record the result as the percent of casein. Method II. Weigh a covered 125 cc beaker to milligrams, add 10 cc of milk and quickly reweigh. Add 50 cc of water at 40 then add 2 cc of alum solution, saturated above 40. Stir then allow the precipitate to settle, transfer to a filter and wash with cold water Transfer the filter and casein to a Kjeldahl digestion flask and determine nitrogen by the Kjeldahl or Gunning method. Multiply by 6.38 to obtain the percent of casein. Albumin. Albumin is flocculated and separated from the re- maining proteids by the use of a more dilute solution of acetic 542 QUANTITATIVE ANALYSIS acid but at a higher temperature. The determination of nitrogen in the precipitate gives a basis. for the calculation of the percent of albumin. Determination. Method I. Add a drop of phenolphthalein to the nitrate from casein, obtained by method (I), and neutralize with an approximately tenth-normal solution of sodium hydroxide or potassium hydroxide. Add 0.3 cc of 10 percent acetic acid solution and heat on the steam bath until the albumin is completely flocculated. Filter, wash and determine nitrogen in the precipitate as in the determination of casein. The percent of nitrogen multiplied by 6.38 gives the percent of albumin. Method II. To the nitrate from the casein obtained by method (II) add 0.3 cc of 10 percent acetic acid and boil until albumin is completely flocculated. Filter and wash with cold water. Transfer the paper and albumin to a digestion flask and determine nitrogen by the Kjel- dahl or Gunning method. Multiply by 6.38 and report the percent of albumin. Lactose. The only carbohydrate occurring in milk in suffi- cient quantity to be of any importance is lactose. It may be de- termined by either optical methods, depending upon its power to rotate the plane of polarization of plane-polarized light, or by reduction methods, based upon its power to reduce a cupric salt in a basic solution containing a tartrate. Optical Methods. With the exception of the proteids lac- tose is the only optically active substance in milk and it is only necessary to remove the proteids in order to apply the polari- scope to the determination of the concentration of lactose. It is assumed that the student is familiar with the physical prin- ciples upon which the construction of the various types of polari- scopes is based. The best-known instruments used for this pur- pose, are provided with scales reading in angular degrees and also with special sugar scales, upon which each degree indicates 1 percent of sugar if a certain specified weight of material is con- tained in 100 cc of solution and polarized in a 200 mm tube. This specified weight is the "normal" weight of a given instru- ment. The normal weight of cane sugar for the Laurent scale is 16.19 gm. For the Ventzke scale the normal weight is 26.048 gm. These weights are, however, based upon AGRICULTURAL MATERIALS 543 dilution to 100 Mohr cubic centimeters. (See page 172.) Since volumetric apparatus is now generally calibrated upon a basis of the true cubic centimeter it is necessary to correct these weights by multiplying by the ratio of the volume of the true cubic cen- timeter to that of the Mohr cubic centimeter. This ratio is: Density of water at 17.5 _ r . Density of water at 20 XWt * f true CC > m air at 20 or gg| X 0.9972 = 0.9977. The Laurent normal weight for cane sugar is therefore 0.9977 X 16.19=16.153 gm and the Ventzke normal weight is 0.9977X 26.048 = 25.988 gm. If lactose is to be determined by use of the sugar scale the nor- mal weight for lactose is calculated by multiplying that for sucrose by the ratio of the specific rotatory power of sucrose to that of lactose. [a] D for sucrose = 66. 5 and [a] D for lactose = 52.53. Therefore the Laurent normal weight for lactose is f\(\ f\ P2~KoX 16. 153 = 20.449 and the Ventzke normal weight for lac- f*f r tose is ^-53 X 25.988 = 32.899 gm. Both of these finally ob- tained weights are based upon dilution to 100 true cubic centi- meters. These weights of milk are too small for convenience and it is customary to use three times the normal weight for the Laurent, and twice the normal weight of milk for the Ventzke scale. These weights then become, finally, 61.347 and 65.798 gm for the Laurent and Ventzke scales respectively. In order to avoid weighing each sample of milk the table on page 545 is given for measuring milk having a predetermined specific gravity. For making these measurements a burette is not suitable because of the length of time that is thereby required, segrega- tion of fat taking place. Neither is the ordinary measuring pipette with a cylindrical tube 4 satisfactory as such a pipette of sufficient capacity is too long to be used conveniently. A special pipette having a bulb in the upper part is better. The zero mark is above the bulb and the first mark below is 55. The lower part is a narrow tube graduated from 55 to 65 cc. 544 QUANTITATIVE ANALYSIS Precipitation of Proteids. The milk solution may be clari- fied, without heating, by Wiley's method, 1 adding a solution of mercuric nitrate containing free nitric acid, or by adding a solu- tion of mercuric iodide containing free acetic acid. This is prob- ably a purely colloidal flocculation, the composition of the pro- teids being unchanged by the action. If this precipitation is carried out in a 100 cc volumetric flask and the solution is after- ward diluted to the mark the volume of the solution will be less than 100 cc, on account of the volume occupied by the precipi- tated proteids. The concentration of lactose is consequently greater than would be calculated unless allowance were made for the volume of the precipitate. The official method directs that the volume of the precipitate should be considered as 2.4 cc for the quantity of milk that is used for the Laurent polari- scope or 2.6 cc for the quantity that is used for the Ventzke polariscope. This is only a close approximation but it is suffi- ciently accurate for ordinary work. A more accurate method for making allowance for the volume of the precipitate is that of Wiley and Ewell. 2 Two flasks are used, of 100 cc and 200 cc capacity, respectively. The same quantity of milk is placed in the two flasks, the proteids are pre- cipitated by acid mercuric nitrate solution and each solution is diluted to the mark on the flask. The polarization of the solu- tion in the 200 cc flask is slightly less than half of that of the more concentrated solution. The product of the two polarizations divided by their difference gives the corrected reading. Determination. Prepare a solution of either mercuric nitrate with nitric acid or mercuric iodide with acetic acid, as follows: Acid Mercuric Nitrate. Dissolve 25 gm of mercury in 50 cc of con- centrated nitric acid, then add 35 cc of water and mix. Acid Mercuric Iodide. Dissolve 33.2 gm of potassium iodide, 13.5 gm of mercuric chloride and 20 cc of glacial acetic acid in 640 cc of water. Determine the specific gravity of the milk by means of a sensitive hydrometer or a picnometer. Refer to the table on page 545 and measure, at the temperature at which the specific gravity was taken, the quantity of milk indicated, the sample having been mixed thoroughly 1 Am. Chem. J., 6, 289 (1889). J. Am. Chem. Soc., 18, 428 (1896). AGRICULTURAL MATERIALS 545 immediately before making both measurements. A pipette which is capable of delivering the required amount in one portion is desirable. The milk is run into a volumetric flask, graduated at 102.4 cc for the Laurent or 102.6 cc for the Ventzke type of polarimeter. Add 1 cc of acid mercuric nitrate solution or 30 cc of acid mercuric iodide solution, dilute to the mark on the flask, mix well and allow the precipitate to settle. Filter through a dry filter, rejecting the first 25 cc of the filtrate, receiving the remainder in a dry flask. Polarize in a 200 mm or 400 mm tube. If the 200 mm tube is used the reading on the sugar scale is to be divided by 3 for the Laurent or by 2 for the Ventzke scale. In case the 400 mm tube has been used these numbers become 6 and 4, respectively. The quotient in either case is the percent of lactose in the milk. Specific gravity Volume of milk to be used Laurent scale Ventzke scale 1.024 59.90 64.25 1.025 59.85 64.20 1.026 59.80 64.15 1.027 59.75 64.05 1.028 59.70 64.00 1.029 59.60 63.95 1.030 59.55 63.90 1.031 59.50 63.80 1.032 59.45 63.75 1.033 59.40 63.70 1.034 59.35 63.65 1.035 59.30 63.55 1.036 59.20 63.50 Reduction Methods. Fehling first perfected 1 a method for the determination of reducing sugars by measuring the amount of reduction of a cupric salt that takes place in the presence of a strong base and a tartrate, cuprous oxide being formed. Soxh- let found, 2 however, that the reaction is not to be expressed by any single equation and that the quantity of cuprous oxide pro- duced varies somewhat with the conditions of the experiment with regard to the concentration of the sugar, time of heating, amount of excess of cupric solution and kind of sugar. Thus, 1 Ann. Chem. Pharm., 72, 106 (1849); 106, 75 (1858). 2 Chem. Zentr., [3] 9, 218 and 236 (1878). 35 546 QUANTITATIVE ANALYSIS sugars having the same molecular formulas (as dextrose and invert sugar, maltose and lactose) do not reduce the same weight of cuprous oxide when the conditions of the experiment are other- wise the same. The method has been so modified and standard- ized that reducing sugars may now be determined with a consider- able degree of accuracy by either measuring the volume of copper solution required to react with the sugar, or by weighing, or other- wise determining, the weight of cuprous oxide that is produced by the reaction. If the latter class of methods is used the weight of sugar cannot be directly calculated from the weight of cuprous oxide, on account of variation in the products formed by the reaction. The weight of sugar is found by the use of a table that has been constructed from the results of direct experi- ments. Also it is not practicable to make a direct weighing of cuprous oxide because of the oxidation that takes place during the drying process. It is therefore dissolved in nitric acid and the copper is determined by one of the volumetric or electrolytic processes, or else the cuprous oxide is reduced to metallic copper by heating in an atmosphere of hydrogen, the copper being then weighed. For the determination of lactose in milk it is necessary first to remove other substances that would reduce cupric salts in basic solution. The milk proteids are the only substances of this nature to be considered and these may be removed by pre- cipitating cupric hydroxide in their presence. For this purpose cupric sulphate and sodium hydroxide are added to the milk. The colloidal cupric hydroxide, which then precipitates, com- pletely removes the colloidal proteids. A slight error is occasioned by the fact that a small amount of lactose is also adsorbed by the precipitate but this is usually negligible. Determination. Prepare 500 cc of a half-normal sodium hydroxide solution, also Fehling-Soxhlet solutions, as follows: (a) Cupric Sulphate Solution. Dissolve 34.639 gm of dry, crystal- lized cupric sulphate in water and dilute to 500 cc. If not clear, filter through washed asbestos. (6) Tartrate Solution. Dissolve 173 gm of potassium sodium tar- trate ("Rochelle salts") and 50 gm of sodium hydroxide in water and dilute to 500 cc. Allow to stand for 2 days and filter through washed asbestos. AGRICULTURAL MATERIALS 547 Measure 25 cc of milk into a 500 cc volumetric flask and calculate its weight. Add 400 cc of water and 10 cc of copper sulphate solution (a), mix and then add 8.8 cc of half-normal sodium hydroxide solution. After the copper hydroxide has precipitated the solution should still contain a slight excess of copper. Dilute to the mark on the flask, mix and filter through a dry paper, rejecting the first 25 cc of the filtrate. Place 25 cc of solution (a) and 25 cc of solution (b) in a 250 cc beaker of resistance glass and add 50 cc of the filtrate from proteids, already obtained. Heat over a burner which is regulated so that the boiling begins in 4 minutes and continue boiling for exactly 2 minutes longer. Keep the beaker covered during the heating. Filter immediately through a Gooch filter and wash with hot water but making no attempt to transfer all of the cuprous oxide to the filter. Determine the copper in this precipitate by one of the following methods : (a) Electrolytic Method. Transfer the asbestos and most of the cuprous oxide to the beaker in which precipitation was made. Dissolve the oxide remaining in the crucible by means of 2 cc of concentrated nitric acid, adding the latter with a pipette and receiving the solution in the beaker containing the asbestos. Rinse the crucible with a jet of hot water, allowing the rinsings to flow into the beaker. Heat until all of the cuprous oxide is dissolved, then filter through paper and wash the paper thoroughly with hot water to remove all copper nitrate, re- ceiving the solution in the beaker that is to be used for the electrolysis. Electrolyze and determine the copper, observing all of the precautions mentioned in the discussion on page 156. (6) Proceed as in method (a) until the copper solution is obtained and filtered. Determine the copper in the solution by the volumetric iodide method described on page 267, beginning with "Boil until all bromine is removed . . . " in the third paragraph of page 269. Conduct a blank experiment, using 50 cc of water instead of the milk solution. If more than 0.5 mg of cuprous oxide is obtained make a correction in the amount of cuprous oxide found in the analytical experi- ment. The alkaline tartrate solution deteriorates on standing and the amount of cuprous oxide obtained in the blank increases with time. Find the weight of lactose corresponding to the copper obtained, using the table on page 548. Calculate the percent of lactose in the milk. Fat. The composition of butter fat has already been discussed in connection with edible oils and fats. The fat of milk is in a state of emulsification and the microscopic globules are largely responsible for the white appearance of milk. The determina- tion of fat is made by a method belonging to any one of three 548 QUANTITATIVE ANALYSIS TABLB FOR THE DETERMINATION OP LACTOSE (SOXHLET-WEIN) Milli- grams of cop- per Milli- grams of lac- tose Milli- grams of cop- per Milli- grams of lac- tose Milli- grams of cop- per Milli- grams of lac- tose Milli- grams of cop- per Milli- grams of lac- tose Milli- grams of cop- per Milli- grams of lac- tose Milli- grams of cop- per Milli- grams of lac- tose 100 71.6 150 108.8 200 146.9 250 184.8 300 224 . 4 350 263.9 101 72.4 151 109.6 201 147.7 251 185.5 301 225.2 351 264.7 102 73.1 152 110.3 202 148.5 252 186.3 302 225.9 352 265.5 103 73.8 153 111.1 203 149.2 253 187.1 303 226.7 353 266.3 104 74.6 154 111.9 204 150.0 254 187.9 304 227.5 354 267.2 105 75.3 155 112.6 205 150.7 255 188.7 305 228.3 355 268.0 106 76.1 156 113.4 206 151.5 256 189.4 306 229.1 356 268.8 107 76.8 157 114.1 207 152.2 257 190.2 307 229.8 357 269.6 108 77.6 158 114.9 208 153.0 258 191.0 308 230.6 358 270.4 109 78.3 159 115.6 209 153.7 259 191.8 309 231.4 359 271.2 110 79.0 160 116.4 210 154.5 260 192.5 310 232.2 360 272.1 111 79.8 161 117.1 211 155.2 261 193.3 311 232.9 361 272.9 112 80.5 162 117.9 212 156.0 262 194.1 312 233.7 362 273.7 113 81.3 163 118.6 213 156.7 263 194.9 313 234.5 363 274.5 114 82.0 164 119.4 214 157.5 264 195.7 314 325. 3^ 364 275.3 115 82.7 165 120.2 215 158.2 265 196.4 315 236.1 365 276.2 116 83.5 166 120.9 216 159.0 266 197.2 316 236.8 366 277.1 117 84.2 167 121.7 217 159.7 . 267 198.0 317 237.6 367 277.9 118 85.0 168 122.4 218 160.4 268 198.8 318 238.4 368 278.8 119 85.7 169 123.2 219 161.2 269 199.5 319 239.2 369 279.6 120 86.4 170 123.9 220 161.9 270 200.3 320 240.0 370 280.5 121 87.2 171 124.7 221 162.7 271 201.1 321 240.7 371 281.4 122 87.9 172 125.5 222 163.4 272 201.9 322 241.5 372 282.2 123 88.7 173 126.2 223 164.2 273 202.7 323 242.3 373 283.1 124 89.4 174 127.0 224 164.9 274 203.5 324 243.1 374 283.9 125 90.1 175 127.8 225 165.7 275 204.3 ' 325 243.9 375 284.8 126 90.9 176 128.5 226 166.4 276 205.1 326 244.6 376 285.7 127 91.6 177 129.3 227 167.2 277 205.9 327 245.4 377 286.5 128 92.4 178 130.1 228 167.9 278 206.7 328 246.2 378 287.4 129 93.1 179 130.8 229 168.6 279 207.5 329 247.0 379 288.2 130 93.8 180 131.6 230 169.4 280 208.3 330 247.7 380 289.1 131 94.6 181 132.4 231 170.1 281 209.1 331 248.5 381 289.9 132 95.3 182 133.1 232 170.9 282 209.9 332 249.2 382 290.8 133 96.1 183 133.9 233 171.6 283 210.7 333 250.0 383 291.7 134 96.9 184 134.7 234 172.4 284 211.5 334 250.8 384 292.5 135 97.6 185 135.4 235 173.1 285 212.3 335 251.6 385 293.4 136 98.3 186 136.2 236 173.9 286 213.1 336 252.5 386 294.2 137 99.1 187 137.0 237 174.6 287 213.9 337 253.3 387 295.1 138 99.8 188 137.7 238 175.4 288 214.7 338 254.1 388 296.0 139 100.5 189 138.5 239 176.2 289 215.5 339 254.9 389 296.8 140 101.3 190 139.3 240 176.9 290 216.3 340 255.7 390 297.7 141 102.0 191 140.0 241 177.7 291 217.1 341 256.5 391 298.5 142 102.8 192 140.8 242 178.5 292 217.9 342 257.4 392 299.4 143 103.5 193 141.6 243 179.3 293 218.7 343 258.2 393 300.3 144 104.3 194 142.3 244 180.1 294 219.5 344 259.0 394 301.1 145 105 1 195 143.1 245 180.8 295 220.3 345 259.8 395 302.0 146 105.8 196 143.9 246 181.6 296 221.1 346 260.6 396 302.8 147 106.6 197 144.6 247 182.4 297 221.9 347 261.4 397 303.7 148 107.3 198 145.4 248 183.2 298 222.7 348 262.3 398 304.6 149 108.1 199 146.2 249 184.0 299 223.5 349 263.1 399 305.4 AGRICULTURAL MATERIALS 549 general classes, there being many modifications of the methods belonging to each class. These classes are (1) extraction meth- ods, the fat being dissolved by a volatile solvent which is later evaporated, leaving the fat to be weighed, (2) volumetric meth- ods, the proteids being dissolved by strong acids, the fat being collected and its volume measured, and (3) refractometric meth- ods, based upon the variation of the index of refraction of milk with varia- tion in the concentration of suspended fat particles. Paper-coil Method. The best-known extraction method is that of Adams. 1 In this method the weighed sample of milk is absorbed by a roll of porous paper, which is then dried and the fat is extracted by ether or volatile petro- leum spirit. The removal of water is essential as otherwise other substances, such as lactose and proteids, would also be extracted. For the same reason the ether must be entirely free from water and alcohol. These substances are re- moved by allowing the ether to stand over metallic sodium until hydrogen is no longer evolved, then distilling. Paper may be prepared for fat determinations by extracting strips of filter paper with ether and drying, but it is now possible to obtain extracted paper of special quality which is suitable for this purpose. The apparatus for continuous extraction is that devised by Soxhlet or some other apparatus, similar in principle. Soxhlet's extractor is shown in Fig. 117. The volatile solvent is boiled in the lower flask which has previously been weighed. The vapor is condensed above, drops into the extractor and there remains in contact with the paper and milk solids until it has filled the apparatus to the height of the bend in the siphon tube (s) when it siphons back into the flask to be again vaporized. At the last, the ether 1 Analyst, 10, 46 (1885). FIG. 117. Soxhlet ex- tractor with fat flask and Hopkins condenser. 550 QUANTITATIVE ANALYSIS is evaporated from the flask and the latter is then weighed with the extracted fat. Determination by the Paper-coil Method. Prepare anhydrous ether by allowing ordinary ether to stand over sodium, the bottle being loosely stoppered, until hydrogen is no longer evolved. The sodium is cut into small pieces or, better, is pressed into wire by means of a sodium press, and is immediately placed in the ether. When gas ceases to be evolved distill the ether from an electrically heated apparatus or from a water bath from which the flame has been removed, receiving the distill- ate in a bottle which is protected by a drying tube filled with calcium chloride. All flames must be removed from the vicinity when ether is being handled or distilled. Roll a strip of fat-free porous paper into a coil and bind with a plati- num or iron wire, avoiding contact with the hands as far as possible. Special extracted paper is obtainable in strips 6.5 cm by 60 cm, already thoroughly extracted with ether. If such paper is not available cut heavy filter paper into strips 6.5 cm wide and roll into a coil which will easily slip into the continuous extractor. Mix the sample of milk, immediately withdraw 5 cc by means of a pipette and allow the milk to run into the end of the coil, by which it is absorbed. The weight of milk is calculated from the volume and specific gravity. Dry the coil at 100 for 3 hours, then place it in the extraction apparatus. Clean, dry and weigh a low, wide flask of the form shown in Fig. 117 and having a capacity of 100 cc. Place 60 cc of anhydrous ether in the flask and connect as shown, using pressed corks that have already been extracted with ether. Rubber stoppers must not be used. Heat the flask containing the ether until the latter boils, using a water bath or an elec- trically heated stove, all flames being removed. Keep the ether boiling and extract for 2 hours, the extractor emptying approximately every 10 minutes. Finally mterrupt the extraction at a point just before the extractor is ready to empty by siphoning, the flask then containing but little ether. Connect the flask with a condenser in such a way that the ether may be distilled and recovered. When no more ether will distill remove the condenser and heat the open flask over the bath until the odor of ether nearly disappears, then place the flask on its side in an oven and heat at 100 until the ethereal odor has been completely removed. Explosions sometimes occur as a result of placing the flask in the oven when too much ether remains, the ether-air mixture becoming ignited in the oven. Cool the flask in a desiccator and weigh. The increase in weight is calculated as fat. AGRICULTURAL MATERIALS 551 The paper-coil method, while the most accurate of all, has now been displaced as an official method by the Rose-Gottlieb method 1 because of the greater simplicity of the latter. Here the proteids are dissolved by ammonium ~ hydroxide. The butter fat is then ex- r\ tracted by shaking with ether and petro- ^ ( leum ether in a cylinder which is provided with a side tap for drawing off the ethereal solution. This is shown in figure 118. The purpose of the petroleum ether is to diminish the solubility of lactose in the ether. The fat solution thus obtained is heated to vaporize the solvent and the fat is then weighed. Determination by the Official Rose-Gottlieb Method. Measure 10 cc of milk into a Rohrig tube or some similar apparatus and calculate the weight from the specific gravity. Or weigh, to milligrams, approximately 10 cc of milk in a small covered beaker, pour into the tube and reweigh the covered beaker. Add 1.25 cc of concentrated ammonium hydroxide and mix thoroughly. (If the sample is sour 2 cc of ammonium hydroxide should be used, although it should be remembered that accurate sampling of fermented milk is difficult on account of coclusion of fat by flocculated proteids.) Add 10 cc of alcohol, 95 percent by volume, and mix well, then add 25 cc of washed ether, stopper and shake vigorously for 30 seconds. Add 25 cc of petroleum ether (distilling below 60) and shake again for 30 seconds. Allow to stand for 20 minutes or until the upper liquid is practically clear. Draw off as much as possible of the ether-fat solution (usually 0.5 to 0.8 cc ^"^ ... . . . , ,. , FIG. 118. Rohrig tube will be left) through a small, quick-acting dry for fat determination. filter, into a dry 100 cc wide necked flask. Again extract the liquid remaining in the tube, using 15 cc of washed ether, shaking for 30 seconds, adding 15 cc of petroleum ether, shaking 1 Z. Nahr. Genussm., 9, 531 (1905). 552 QUANTITATIVE ANALYSIS again for 30 seconds and allowing to stand for 20 minutes for separation. Draw off the clear upper layer through the same filter as before and into the same flask. Wash the tip of the outlet tube, the funnel and the filter with a few cubic centimeters of a mixture of equal parts of the two ethers. For very accurate work a third extraction should be made. Usually this will yield less than 1 mg of fat if a close separation has been made on the previous extractions. As this represents less than 0.01 percent it may be ignored in most work. Evaporate the ethers slowly on the steam bath then dry the fat to constant weight at 100. Dissolve the fat in a little petroleum ether and wash the flask and any solid residue that may remain free from fat with petroleum ether, leaving the residue in the flask. Dry the flask as before and reweigh with the residue. This weight subtracted from the first will .give the weight of fat. Blank determinations should be made on the reagents, correcting for any soluble matter that may be obtained. From the corrected weight of fat calculate the percent of fat in the milk. Volumetric Methods. In order to separate the fat for a volume measurement it is necessary first to dissolve the pro- teids existing in a state of colloidal suspension. This can be done by adding sulphuric acid or both sulphuric and hydro- chloric acids, with or without amyl alcohol. The fat is then col- lected by whirling in a centrifugal machine. A special bottle is used, having a graduated neck in which the volume of fat can be read. If proper relations between the quantity of milk and the graduations on the bottle are observed the percent of fat may be read directly. These methods are sufficiently accurate for control purposes and they can be performed in a very short period of time. The chief causes of inaccuracies are incorrect graduations of milk bottles and pipettes and variation in the temperature at which the volume of fat is read. Babcock Method. The best known of the volumetric methods is that of Babcock. 1 In this method sulphuric acid is the only proteid solvent used, its specific gravity lying between 1.82 and 1.83. If acid of lower specific gravity is used solution of proteids is incomplete, while acid of higher specific gravity than 1.83 causes charring of the milk solids, resulting also in incomplete 1 Wis. Exp. Sta. Bull., 24 (1890); Milch. Zeit., 19, 746 (1890). AGRICULTURAL MATERIALS 553 separation of fat. In order to avoid all weighing of samples a special pipette is provided, graduated to deliver 17.6 cc. Upon the assumption that the specific gravity of milk is 1.032 this volume of milk would weigh 18.2 gm. This assumption is sufficiently close to the truth for ordinary testing purposes. The graduated bottle has the form shown in Fig. 119. The numbered divi- sions include a volume of 0.20 cc, each division having ten subdivisions. If the specific gravity of butter fat is taken as 0.91 (which is practically correct at 60, the temperature of measurement) each numbered division on the bottle will hold 0.18 gm of fat, which will be 1 percent of the weight of sample. It will thus be seen that the method involves several somewhat arbitrary assumptions but, in spite of this fact, it admirably fulfills the requirements of a rapid and fairly accurate method for the control of milk supply. Official Specifications for Test Bottles and Pipettes. 1 The standard Babcock test bottle must conform to the following specifications: The total percent graduation shall be 8. The total height of the bottle shall be 150 to 165 mm. The capacity of the bulb to the junction with the neck shall be not less than 45 (true) cc. (See page 172.) The graduated portion of the neck shall have a length of not less than 63.5 mm and the neck shall be cylindrical for not less than 9 mm on either side of the graduated portion. The gradua- tions shall represent whole percent, halves and tenths of a percent. The capacity of each percent on the scale shall be 0.20 true cc. Standard milk pipettes are graduated to deliver 17.6 true cc of water at 20 in 5 to 8 seconds. Graduated cylinders should be provided for the acid, capable of measuring 17.5 cc. The official method for calibrating Babcock test bottles is to 1 J. Assoc. Off. Agr. Chem., Vol. II, No. 3, Ft. II, p. 289. FIG. 119. Babcock bottle for fat determi- nations. 554 QUANTITATIVE ANALYSIS fill the dry bottle to the zero mark with pure mercury at 20, weigh, fill to the highest mark and re weigh, calculating the bulb and stem capacities on the basis of 13.5471 gm of dry mercury for each cubic centimeter at 20. It is difficult to see what ad- vantage this possesses over the method of calibrating by weighing water at 20, discussed in the first part of Chapter V, especially since the Babcock bottle filled with mercury must weigh more than 600 gm. Accurate weighing of such a quantity would require a special balance, as sensitive as the analytical balance and having large capacity. Milk pipettes and graduates are calibrated according to the official method by measuring in a burette the quantity of water delivered by the instrument at 20. Unless care has been exercised in wetting the inner surface of the burette,, using the standard method by which the burette was calibrated, this method will be subject to considerable error since all burettes are graduated for delivery and not for capacity. A better method for calibrating pipettes is described on page 189. Determination. For the determination measure 17.6 cc of carefully mixed milk into each of at least two test bottles. If the centrifuge carries more than two pockets for bottles more than two determinations may be run at a time but in any case an even number of bottles must be used and they must be placed opposite each other in the machine. Add to each bottle 17.5 cc of sulphuric acid which has a specific gravity of 1.82 to 1.83. Mix and when the proteids are dissolved whirl the bottles in the centrifuge for 4 minutes at the rate prescribed for the machine used . Add boiling water, filling to the bottom of the necks of the bottles and whirl for 1 minute. Again add boiling water to bring the fat layer to about the middle of the bottle necks and whirl again for 1 minute. Immediately place the bottles in a glass beaker which is filled with water at a temperature between 57 and 60. The water should surround the bottle necks to a point above the fat layer. After 1 minute read the length of the fat column, taking the lowest point of each meniscus. Record directly as the percent of fat. Relation between Fat, Total Solids and Specific Gravity. On page 536, the relations of fat, solids not fat and total solids to each other were discussed, also the effect of watering and skim- ming upon the relations existing among these numbers. There is also a nearly constant relation between the specific gravity and AGRICULTURAL MATERIALS 555 the percents of fat and total solids, so that if two of these quanti- ties are known the third can be closely approximated by the use of a formula. In control work the determination of fat by Babcock's method is rapid and easy and the specific gravity may be quickly obtained by means of a lactometer. Neither determination requires a balance or other expensive apparatus. On the other hand the determination of total solids requires the use of an analytical balance and the necessary evaporation, cooling and weighing may extend over a period of several hours. A formula for calculating this property is therefore often desir- able. Richmond's formula 1 is as follows: T = 0.25S + 1.2F + 0.14, where T and F represent the percents of total solids and fat, respectively, and S is the Quevenne lactometer reading. Added Water. Watering may be detected even more readily by observing certain properties, such as specific gravity, index of refraction or percent of ash, of the milk serum from which the curd has been removed. The soluble solids in the serum do not vary as much as do the total solids. Proteids and fat may be separated by the addition of acetic acid or copper sulphate or by allowing the milk to "sour" spontaneously. In the latter case the constants of the serum are changed somewhat by the conversion of lactose into lactic acid. The refractive index of the serum may be determined by any of the standard instruments but the immersion refractometer is most conveniently employed. In this instrument the position of the prism system in the tube is fixed, as in the butyrorefracto- meter, mentioned on page 361. The chief distinguishing feature is the immersion of the prism in the liquid to be tested. The scale is an arbitrary one and reads from 5 to +105, corre- sponding to indices of refraction from 1.32539 to 1.36640. Examination of Acetic Serum, (a) Zeiss Immersion Refractometer Reading. To 100 cc of milk at a temperature of about 20 add 2 co of 25 percent acetic acid (specific -gravity 1.035) in a beaker and heat the mixture, covered by a watch glass, immersed in a water bath at 70. Place the beaker in ice water for 10 minutes and separate the curd by filtering through a 12.5 cm folded filter. Transfer about 35 cc Analyst, 20, 57 (1859). 556 QUANTITATIVE ANALYSIS of the serum to one of the beakers that accompanies the temperature control bath used in connection with the Zeiss immersion ref ractometer ; take the refractometer reading at 20, using a thermometer graduated to tenths of degrees. A reading below 39 indicates added water. If the reading is between 39 and 40 the addition of water is not certain but is to be suspected. (6) Ash. Transfer 25 cc of the serum to a flat bottomed platinum dish and evaporate to dryness on a steam bath, then heat over a low flame until the solids are thoroughly charred. Place the dish in an electric muffle furnace and ignite to a white ash at a temperature not higher then 500. Cool and weigh. Express the result as grams per 100 cc. Ash content below 0.715 gm per 100 cc indicates added water. Multiply by the factor 1.021 to correct for the addition of acetic acid. The result is the ash on the undiluted sour serum. Examination of Sour Serum, (a) Zeiss Immersion Refractometer Reading. Allow the milk to sour spontaneously, filter and determine the immersion refractometer reading of the clear serum at 20. A read- ing below 38.3 indicates added water. (6) Ash. Determine the ash in 25 cc of the serum obtained from the soured milk, using the method described above for acetic serum. Re- sults below 0.730 gm per 100 cc indicate added water. Examination of Copper Serum. Zeiss Immersion Refractometer Reading. Use a solution of copper sulphate containing 72.5 gm per liter, adjusted if necessary to read 36 at 20 on the scale of the immersion refractometer. To one volume of this solution add four volumes of milk. Shake well and filter. Determine the immersion refractometer reading of the clear serum at 20. A reading below 36 indicates added water. Detection of Formaldehyde. Formaldehyde is sometimes added to milk in order to prevent souring, which is due to bac- terial action. Such preservatives as this are forbid'den in most cities and states. Several qualitative methods will serve to detect formaldehyde. The following is Hehner's method. 1 Half fill a test tube with concentrated sulphuric acid, then carefully add milk, pouring down the side of the tube in such a way as to prevent mixing of the milk and acid. If formaldehyde is present a violet or blue color will be developed at the junction of the two liquids. This test depends upon reactions that are not understood. A trace of ferric chloride or other oxidizing agent was formerly 1 Analyst, 20, 155 (1895). AGRICULTURAL MATERIALS 557 thought to be necessary but milk containing formaldehyde will respond to the test without the addition of any reagent other than sulphuric acid. CREAM For the examination of cream the methods described for milk are used, the following modifications being made in two determina- tions, to provide for the differences in composition. Ash. Use 2 to 3 gm of sample instead of 3 to 5 gm. Fat. Rose-Gottlieb Method. Weigh 4 to 5 gm of the sample into the Rohrig tube and dilute with water to 10.5 cc. Proceed as directed for milk. Babcock Method. Two standard cream test bottles are described in the official methods. One of these is 6 inches and the other 9 inches long. The 6-inch bottle is described as follows: The total percent graduation shall be 50. The total height of the bottle shall be 150 to 165 mm. The capacity of the bulb up to the neck shall be not less than 45 cc. The graduated portion of the neck shall have a length of not less than 63.5 mm and the neck shall be cylindrical for at least 9 mm below the lowest and above the highest graduation marks. The graduations shall represent 5 percent, whole and halves of a percent. Weigh 9 of 18 gm, depending upon the fat content of the sample, into a standard Babcock cream test bottle and proceed as directed for the determination of fat in milk. If a 9-gm sample was used multiply the reading by 2. CONDENSED MILK Condensed milk is a product obtained by evaporating either whole milk or skimmed milk, with or without the addition of cane sugar as a preservative. The use of other preservatives is forbidden by the U. S. food and drugs act of 1906. Unsweetened condensed milk ("evaporated milk") must contain not less than 34.3 percent of total solids, including fat, and not less than 7.8 percent of fat. It must contain no added butter or butter oil. 1 Sweetened condensed milk is usually evaporated skimmed milk. In the original package the solids usually separate to some extent and careful maceration is necessary in order to obtain a uniform mixture for analysis. X U. S. Dept. Agr., Food Insp. Decision 131 (1911). 558 QUANTITATIVE ANALYSIS The following methods for the preparation of samples and for their analysis are official. Unsweetened Condensed Milk. Dilute 40 gm of the homogeneous sample with 60 gm of water and proceed as directed for the analysis of milk, correcting the results for the dilution. Sweetened Condensed Milk. Prepare the sample as follows: If cold, place the can in water at 30 to 35 until warm. Open, scrape out all of the milk adhering to the interior of the can and mix in a dish sufficiently large to permit thorough stirring so as to make the mass homogeneous. Weigh 100 gm into a 500 cc flask and make up to the mark with water. If the milk will not dissolve completely weigh out each portion of the mixed sample for analysis, without dilution. Total Solids. Use 10 cc of the sample, prepared as directed above, and proceed as with milk, using either dry sand or dry asbestos fiber for keeping the mass open. Ash. Evaporate 10 cc of the solution to dryness on the steam bath and ignite the residue at a temperature just below redness until the ash is free from carbon. Total Nitrogen. Use 10 cc of the solution and proceed as for the deter- mination of total nitrogen in milk. Multiply the result by 6.38 and report as protein. Lactose. On account of the presence of sucrose, lactose cannot be determined in sweetened condensed milk by a single polarization. It is therefore more convenient to use a gravimetric method. Measure 100 cc of the milk solution into a 250 cc volumetric flask. Dilute to about 200 cc and mix. Add 6 cc of copper sulphate solution ((a), page 546), dilute to the mark on the flask and mix. Filter through a dry filter and determine lactose in the filtrate by the gravimetric method, described for the analysis of milk. Fat or Ether Extract. Weigh 4 to 5 gm of the homogeneous, undiluted sample into a Rohrig tube, dilute to about 10.5 cc and proceed as directed for milk. Sucrose (Tentative Method}. From the percent of total solids sub- tract the sum of the percents of proteids, lactose, fat and ash. Report the difference as the percent of sucrose. BUTTER AND SUBSTITUTES The examination of butter is made in order to determine its quality, if known to be genuine, or to detect the partial or total substitution of other oils or of process butter. The examination of butter substitutes is usually made for the purpose of determin- AGRICULTURAL MATERIALS 559 ing their quality as edible fats and also for the detection and determination of the various oils and fats composing them. Process butter is a product obtained by steaming rancid butter, cooling and churning with a small amount of sweet milk. It is distinguished from fresh butter by the appearance of fat crystals. Butter fat that has not been melted is not crystallized. Butter and butter substitutes contain water, curd, salt and small amounts of milk solids, in addition to the fats. These substances are, for the most part, undissolved and the sample is therefore non-homogeneous, so that sampling must be performed with unusual care. Preparation of Sample. If large quantities of butter are to be sampled, use a butter "trier" or sampler. Withdraw portions of about 100 to 500 gm in a closed vessel at as low a temperature as possible. When softened, cool and at the same time shake the mass violently until it is homogeneous and sufficiently solid to prevent the separation of water and fat, then pour a portion into a small beaker, from which it is to be weighed for analysis. The sample should nearly or completely fill the beaker and should be kept in a cold place until analyzed. Moisture. After butter is churned it is washed to remove most of the buttermilk and it is then worked to remove the water. A determination of moisture will serve to show the efficiency of the working process. Moisture will vary between the approxi- mate limits of 4 percent and 35 percent. About 15 percent should be regarded as the normal water content. Determination. Place 1.5 to 2.5 gm of butter in a weighed dish with a flat bottom which has a surface of at least 20 sq cm and dry over a water or steam bath until the weight is constant, each drying to be for one hour. The use of clean dry sand is admissible but must be omitted if this dried sample is to be used for the indirect determination of fat. Calculate the percent of moisture in the sample. Fat or Ether Extract. Fat is determined indirectly, weighing the solids left after extraction with ether or gasoline, or directly by the methods already described for the determination of fat in milk. There is little to be gained by making the determina- tion, since fat necessarily comprises the major portion of the butter. 560 QUANTITATIVE ANALYSIS Determination by the Indirect Method. Dissolve in the dish with anhydrous, alcohol-free ether or gasoline the dry butter obtained in the moisture determination in which sand was not used, transfer to a weighed Gooch crucible by the acid of a wash bottle filled with the solvent and wash until the residue is free from fat. Dry the crucible and contents at 100 for 1-hour periods until the weight is constant and calculate the percent of fat. Preserve the contents of the crucible for the determina- tion of casein. Determination by the Direct Method. From the dry butter obtained in the determination of moisture, either with or without the use of sand, wash the fat with anhydrous, alcohol-free ether, receiving the solution in a weighed flask. Distill the ether and dry the extract at 100 until it ceases to lose weight, the dryings not to exceed one hour in duration. Calculate the percent of fat in the sample. Casein. Buttermilk which is not removed by washing adds a flavor to the butter and this is considered desirable by many users. It also injures the keeping qualities of the butter because of its tendency toward the formation of substances of unpleasant taste and odor by subsequent fermentation. The presence of buttermilk is made evident by proteids, which are usually re- ported as casein. The percent of proteids in good butter is less than 1. Determination. Cover the -crucible containing the residue from the determination of fat by the indirect method and heat, gently at first, gradually raising the temperature to just below redness. The cover may then be removed and the ignition continued until the residue is white. Weigh and calculate the loss in weight as casein. Salt. Good, marketable butter usually contains between 1 and 6 percent of added salt. Its determination is made by ex- traction with water, followed by titration or precipitating and weighing as silver chloride. Determination. Weigh 5 to 10 gm of butter in a counterpoised beaker, using portions of about 1 gm from different parts of the sample. Add about 20 cc of hot water and, after the butter is melted, transfer the whole to a 50 cc separatory funnel. Insert the stopper and shake for a few moments. Let stand until the fat has all collected on the top of the water, then draw off the latter into a flask, being careful to let none of the fat globules pass. Again add hot water to the beaker and repeat the extraction from ten to fifteen times, using each time from 10 to 15 AGRICULTURAL MATERIALS 561 cc of water. The washings will contain all but a trace of the sodium chloride originally present in the butter. Determine its amount in the whole or in an aliquot part of the liquid by titration with standard silver nitrate solution, using potassium chromate as indicator. This method is described on page 397. Examination of the Fat. The question as to the nature of the oils and fats composing butter and butter substitutes can be answered only by a more thorough examination than that already described. The composition of butter fat and of oleomargerine has already been discussed in connection with the analysis of edible fats and oils. The examination of the fat will be made by the same methods as are there described. It is sufficient to recall the following approximate figures as the most character- istic differences between butter fat and the mixture known as "oleomargerine." Butter fat Oleomargerine Butyro-refractometer. . . 42 48 Iodine value 36 53 Hehner value 87 93 Soluble acids 4 5 7 Reichert-Meissl value Jean'a eolubilitv test. . 24 63 1 30 If cocoanut oil is used in oleomargerine the Reichert-Meissl value is raised but this oil is easily detected by noting the Polenske value. (See page 375.) For the examination of the fat the water, curd and salt must be removed and the clear fat obtained. Preparation of Sample. Melt the butter and keep it in a dry place for two or three hours, the temperature being held at about 60, until the water and curd have entirely separated. Pour off the clear, super- natant fat and filter through a dry paper in a funnel heated by boiling water, or in an oven at about 60. Should the filtered liquid fat not be perfectly clear it must be filtered again. Determination. Determine the .constants mentioned in the above table, using the methods earlier described for the analysis of all edible fats and oils. Qualitative Tests : Microscopic Examination. Place a small portion of the fresh, unmelted sample, taken from the interior of the mass, on a glass slide, add a drop of pure olive oil, gently press over it a cover glass 36 562 QUANTITATIVE ANALYSIS and examine for crystals of lard, etc., using a magnification of 120 to 150 diameters. Examine the same specimen with polarized light and a selenite plate without the use of olive oil. Pure fresh butter will show neither crystals nor a parti-colored field with selenite. Other fats, melted and cooled and mixed with butter, will usually show crystals and variegated colors with the selenite plate. For further microscopic study dissolve about 4 cc of the fat in 15 cc of ether in a test tube. Close the tube with a loose plug of cotton and allow to stand 12 hours at about 20. When crystals form at the bottom of the tube remove them with a pipette, glass rod or tube, place on a slide, cover and examine. The crystals formed by later deposits may be examined in a similar way. Foam Test. Heat 2 or 3 gm of the sample in a spoon or dish over a free flame. Genuine, fresh butter will foam abundantly, whereas process butter will bump and sputter without foaming. Oleomargerine behaves like process butter but chemical tests will determine whether the sample is oleomargerine or process butter. Detection of Annatto and Saffron. Dissolve 5 gm of the fat in 50 cc of ether in a wide tube and shake the solution vigorously with 12 to 15 cc of a very dilute solution of potassium hydroxide, which must still be basic after it separates from the ether solution. Allow to stand for a few hours, then draw off the aqueous layer, evaporate to dryness and test with sulphuric acid which in the presence of annatto gives first a blue or violet blue, changing quickly to green and finally to brown. Saffron, which would be extracted at the same time, acts differently when treated with sulphuric acid, not giving the green coloration. The aqueous solutions, if not clear enough for the tests, must not be filtered, as the filter paper will take up large quantities of the coloring matter. They can be clarified by shaking with fresh portions of ether or carbon disulphide. Uncolored butter, treated in this way, gives only a very slight trace of coloring matter. Geissler's Test for Azo Colors. Spread a few drops of the clarified fat upon a porcelain surface and add a pinch of fuller's earth. In the presence of the various azo coloring matters a pink to violet red colora- tion will appear in a few minutes. Some varieties of fuller's earth react much more readily with azo colors than do others. The earth should therefore be previously tested with dilute solutions of known azo dyes. Low's Test for Azo Colors. Melt a small amount of the fat in a test tube, add an equal volume of a mixture of one part of concentrated sul- phuric acid and four parts of glacial acetic acid and heat nearly to the boiling point, the liquids being thoroughly mixed by shaking. Set aside and after the acid solution has settled out it will be colored wine AGRICULTURAL MATERIALS 563 red in the presence of azo colors, while with pure butter fat little or no color will be produced. Acid and Base Tests for Added Colors. Pour into each of two test tubes about 2 gm of the filtered fat dissolved in ether. Into one of the tubes pour 1 or 2 cc of hydrochloric acid and into the other about the same volume of 2 percent potassium hydroxide solution. Shake the tubes and allow to stand. In the presence of azo dye the test tube to which the acid has been added will show a pink to wine red coloration, while the basic solution in the other tube will show no color. If, on the other hand, annatto or other vegetable color has been used the basic solution will be colored yellow while no color will be apparent in the acid solution. CHAPTER XVII THE FIRE ASSAY GOLD AND SILVER ORES The determination of gold and silver in ores is not conveniently made by ordinary gravimetric or volumetric methods. This is because of (1) the relatively small quantity of these metals often contained in ores that are commercially valuable, (2) the diffi- culties involved in dissolving the relatively large quantities of ore that must therefore be used for analysis and (3) the commer- cial demand for methods that will not involve a large expenditure of time. The method that is commonly used for this purpose involves the fusion of the ore with relatively large quantities of the various fluxes and other reagents, the gold and silver being finally obtained in the elementary condition, in which form they are weighed. Dependence is placed upon the "noble" character of gold and silver, oxidation by heating in air being impossible, upon the readiness with which these metals alloy with lead and upon the "base" character of the latter metal, separation by oxidation being therefore possible. Lead or a lead compound is made a part of the fusion mixture and it completely removes the gold and silver from the fusion. From this alloy lead is later oxidized and removed, leaving gold and silver free from other metals and from the gangue of the ore. The term "assay" was formerly used to designate all such work as is now included by the term "analysis." In order to distinguish the analysis of ores by fusion processes such an analy- sis was called a "fire assay." "Assay" now usually applies only to the latter class of methods. Sampling. Gold commonly occurs in the elementary condi- tion in ores. It is sometimes in the form of nuggets of varying size but the greater part of the world's supply of gold comes from ores in which the metal is not visible, it being in very small flattened plates or scales which are brown instead of showing the 564 THE FIRE ASSAY 565 brassy yellow color of massive gold. These scales may be found associated with almost any other mineral. Quartz, hematite, pyrite, chalcopyrite, slate, syenite and sand are common gold- bearing minerals but the metal is found in certain localities in a great many other minerals. Gold also occurs in silver ores and in combination with tellurium as sylvanite, calaverite, and petzite, tellurides of gold and silver. Silver also occurs native, but more commonly combined with other elements, common ores being argentite, Ag 2 S, pyrargerite, AgsSbSs, and cerargerite or horn silver, AgCl. Silver is also nearly always found in lead ores, so that both lead and silver (and frequently gold) can be extracted from a single ore with profit. The general discussion of sampling, found in an earlier section of this book, will apply to the sampling of gold and silver ores, with the additional note that all operations must be performed with even greater care than in most other cases. The compara- tively large variation in the commercial value of a gold ore, as indicated by a small variation in analytical results, makes a high degree of accuracy absolutely essential. The finally ob- tained sample should, if possible, be at least as large as 100 gm and it may be 500 gm. It must be ground to pass a sieve having 100 meshes to the linear inch and it is often ground to pass a 200-mesh sieve. Thorough mixing and careful quartering are necessary for each stage in the division of the large sample. Weighing. A quantity of ore varying from about 3 gm to about 60 gm will be used for the assay, the weight taken depend- ing upon the richness of the ore. For weighing the ground ore or "pulp" the analytical balance may be used but a balance of somewhat less sensibility is convenient on account of the greater speed possible in weighing the relatively large samples. The ordinary "pulp balance" is enclosed in a glass case, as are more sensitive balances. Its sensibility need not be greater than 1, this making a somewhat stronger and heavier balance possible. The pans should be detachable so that they may be lifted by forceps when brushing off the ore, and they should be about 8 cm in diameter. On account of the confusion of systems of weights in use in the mining industries it is not convenient to use the metric 566 QUANTITATIVE ANALYSIS weights for weighing the sample of gold or silver ore. Ores are always estimated in avoirdupois tons of 2000 pounds. On the other hand, gold and silver are weighed in the Troy system so that the content of the ore must be stated in Troy ounces per avoirdu- pois ton. In the work of assaying, the metals as finally obtained from the ores are weighed on a very sensitive balance, using milligram weights and fractions. The "assay ton" (abbre- viated to A. T.) has been invented as a unit for weighing the sample for assaying, this unit bearing the same relation to the milligram as that of the avoirdupois ton to the Troy ounce. 1 avoirdupois ton =2000X7000=14,000,000 grains; 1 Troy ounce = 480 grains, therefore 1 avoirdupois ton = ~' L ~AQQ =29,166+ Troy ounces. The assay ton is therefore made to weigh 29166 mg or 29.166 gm. If an assay ton of ore is taken for assaying, the number of milli- grams and fractions of gold and silver obtained will indicate directly the "ounces -per ton" of these metals in the ore. Simple fractions or multiples of the assay ton may be used with equal convenience. The value of gold is fixed by statute at $20.67 per ounce Troy and the value, per ton, of a gold ore may be calculated from this. The commercial value of silver fluctuates, although the legal coinage ratio of silver to gold in the United States is -fa, so that 1 ounce Troy of silver has a coinage value of $1.29. For the final weighing of the gold and silver a balance of high sensibility and precision is necessary. Differences as small as 0.01 mg should be perceptible. If the sample used were as small as 0.1 A. T., as is frequently the case, even this amount of error would involve a difference of 0.1 ounce per ton and this, in a gold ore, would involve an uncertainty in value of $2.06 per ton of ore. In order to produce a balance of such sensibility the knife edges must be very finely and accurately ground and the moving parts must be very light. The pans are only 2 to 3 cm in diameter and instead of the usual supporting bows the pans are suspended from a single wire. This type of assayer's balance is called a "button balance" because it is used only for weighing the "buttons" of metal. THE FIRE ASSAY 567 Decomposition of the ore and extraction of the gold and silver may be accomplished by either the crucible process or the scorifi- cation process. Crucible Process. In this process the finely ground ore is intimately mixed with fluxes and lead oxide and a small amount of an organic reducing agent is also added unless the ore contains enough material of a reducing character. Three essential changes take place when the mix- ture is heated in a crucible: (1) The ore combines with the fluxes and a fused slag is produced, (2) the lead oxide is reduced to finely divided lead, and (3) the particles of reduced lead, settling in a fine spray through the whole mass of slag, alloy with the parti- cles of mechanically freed silver and gold, later collecting at the bottom of the crucible. For the fusion of the mixture of ore and fluxes there is required a large crucible of fire clay, of the form shown in Fig. 120. This is placed in a furnace large enough to heat two, four, six or even more crucibles at a time. Such a furnace is found in all assaying laboratories. It may be heated by coke, gas, gasoline, kerosene or crude oil. A detailed description of such furnaces will be found in any of the special books dealing with assaying. The reagents that are used in the crucible process may be classified as fluxes, reducing agents, oxidizing agents and lead compounds, although some reagents fall in more than one of these classes. Fluxes. Fluxes must be chosen with regard to the nature of the ore that is being decomposed. The formation of a fusible slag is due to a chemical combination of acid and basic materials to form a salt or mixture of salts. If the gangue of the ore is of an FIG. 120. Assay crucible. 568 QUANTITATIVE ANALYSIS acid nature, as is the case with quartz and many silicates, a basic flux will be required. If, on the other hand, the ore contains much iron oxide, limestone or similar metallic oxides or carbonates it will require a flux having an acid character. Many materials, either ore refractories or fluxes, are classed as acid or basic when they have the composition of salts. This is because they have the power of combining with other substances to form salts of lower fusing points, or because they change in such a way as to become acid or basic when heated. Thus clay, a silicate of aluminium, is classed as an acid refractory because it will combine with calcium oxide, sodium oxide, etc., (added as car- bonates and changed by heating) forming double silicates, fusing at lower temperatures. Carbonates of the alkali and alkaline earth metals, while they have the composition of normal salts, lose carbon dioxide when heated with acid materials and thus act the same as though the oxides had been originally added. They are therefore classed as basic fluxes. The principal fluxes that are used in assaying are as follows: Acid Fluxes. Silica, borax and borax glass are the common members of this class although a great many other compounds might be used. Silica combines readily with metallic oxides or carbonates forming silicates of various melting points, the latter depending upon the proportions of flux and metallic oxide. Borax or sodium pyroborate, Na 2 B407.10H 2 O, and borax glass, which is the anhydrous salt, Na 2 B407, act as acid fluxes because of their power of combining with metallic oxides to form ortho- borates and metaborates: Na 2 B 4 07+5CaO->2NaCaB03+Ca 3 (B03)2. This action may be more easily understood if oxide formulas are used. Na 2 O.2B 2 03+5CaO^Na 2 0.2CaO.B 2 O3+3CaO.B203. The reactions of crystallized and anhydrous sodium pyroborate are identical but the swelling of borax which accompanies the loss of water of crystallization often causes a loss of a part of the fusion mixture. On this account the borax should first be fused, cooled and powdered. This gives the product known as borax glass. THE FIRE ASSAY 569 Basic Fluxes. Fluxes which are basic or which become basic when heated are sodium carbonate, sodium bicarbonate, potas- sium carbonate, potassium bicarbonate, calcium carbonate, calcium oxide, lead oxide, ferric oxide and argols. The alkali carbonates and bicarbonates become oxides when heated, espe- cially when an acid substance is present to combine with the oxides as formed. The bicarbonates are somewhat cheaper than the normal carbonates but they evolve twice as much carbon dioxide for a given amount of metallic oxide formed and the normal salts are therefore preferable. This difference is shown by the following equations: Na 2 C03+Si0 2 ->Na 2 Si03+CO 2 ; 2NaHC0 3 +SiO2-Na 2 Si03+2C0 2 +H 2 O. For similar reasons calcium oxide is preferable to calcium carbon- ate, although neither of these fluxes is extensively used in assay- ing. Lead oxide is an excellent flux for silica and silicates as most lead silicates fuse easily and become more fluid than many other slags. On account of its relatively high cost it is substi- tuted, as far as possible, by some of the cheaper fluxes already named. It is, however, always added to the mixture to serve as a source of finely divided lead and it therefore always acts to some extent as a basic flux. The amount which will actually enter the slag is diminished by increasing the proportion of other basic fluxes, such as sodium carbonate. Argols or crude potassium bitartrate, KHCiH^Oe, is not added to serve as a flux but to reduce lead oxide. However, it acts as a basic flux by virtue of the potassium which it contains, heating causing decomposition: 2KHC 4 H 4 6 ->K 2 0+5H 2 0+6CO+2C. Carbon monoxide and carbon thus formed reduce lead oxide. Reducing Agents. Metallic lead must be very intimately mixed with the ore after fusion has taken place, in order that it may alloy with all particles of gold and silver. It is not practica- ble to mix elementary lead with the ore before fusion because of the difficulty involved in reducing the metal to a sufficiently fine state of division and because lead so mixed would settle out of the mixture before complete decomposition and fusion of the ore. Instead, it is better to mix finely powdered lead oxide with the 570 QUANTITATIVE ANALYSIS charge, providing a reducing agent that will act upon the oxide somewhat slowly, in this way providing intimate contact of the minute particles of lead with every portion of the ore. Reducing agents may constitute a part or all of the gangue of the ore. The most important of such reducing agents occurring in ores are sulphides and arsenides. Reactions like the following may occur : 5PbO+FeS 2 ->5Pb+FeO+2S0 2 . 2PbO+PbS^3Pb+SO 2 . 3PbO+ZnS-ZnO-fSOs+3Pb. Many ores, on the other hand, contain no reducing agents or even contain oxidizing agents, such as ferric oxide or cupric carbonate. In such cases its becomes necessary to add a reducing agent to the crucible charge. Reducing agents used for this purpose are argols, charcoal, flour and starch. Many other organic compounds might be used with equal success but the cost would generally be higher and no advantage would be gained. The exact reducing power of none of the common reducing agents can be calculated because they are not usually uniform in com- position. A preliminary fusion with lead oxide and suitable fluxes will establish this value. 1 gm of argols of average purity will reduce about 10 gm of lead, under the average conditions that obtain in the crucible fusion. Oxidizing Agents. It has already been stated that some ores, particularly those containing sulphides, reduce lead oxide with the production of metallic lead. They may even reduce more oxide than is desirable, in which case the quantity of reduced lead must be diminished by the addition of an oxidizing agent. Whether this acts upon the sulphides or upon the lead after reduction is immaterial. Both actions occur to some extent but the ultimate effect of either is to cause less lead to be finally obtained. Instead of adding an oxidizing agent the ore may be subjected to a preliminary roasting or heating in air, oxidation of sulphides and arsenic leaving the ore almost entirely free from reducing power, but this is an operation that requires additional time and it is usually simpler to add an oxidizing agent to the mixture which is to be heated in the crucible. The only oxidizing agent that is commonly used in assaying THE FIRE ASSAY 571 is potassium nitrate. Sodium nitrate would serve equally well but its tendency to deliquesce in moist air hinders proper mixing with the other constituents of the crucible charge. As a substi- tute for an oxidizing agent metallic iron is often used, usually in the form of nails that are long enough to reach through the entire mixture in the crucible and to protrude above. Iron acts as a "desulphurizer," by forming ferrous sulphide, a compound which does not reduce lead oxide: PbS+Fe-*Pb+FeS. This last reaction produces lead but not as much as would be formed if lead sulphide (of galena) were allowed to react with lead oxide: PbS+2PbO->3Pb+S0 2 . Lead Compounds. Lead oxide has already been mentioned in connection with fluxes. In the form of litharge this is the only lead compound that is ever used for this purpose in the crucible process for decomposing ores. On account of the almost uni- versal association of lead with silver in nature it is difficult to obtain litharge that is entirely free from silver. It is often furnished by manufacturers under the label "silver free" but this is, by no means, to be taken as an assurance that it contains no silver. A preliminary assay of litharge should be made with each lot purchased and the proper correction made in the crucible assays of ores. This corresponds to the "blank" determinations often made in connection with other analytical processes. Crucible Charge. The correct charge for the crucible fusion can be made only upon the basis of a knowledge of the composi- tion of the ore. The experienced assayer can usually decide from the general appearance of the ore as to the approximate mixture that is necessary for the proper fusion and extraction of the gold and silver. This is true, however, only in case the unground ore can be inspected, since powdering destroys the characteristic appearance of most minerals. Fortunately a certain latitude in the proportion of the ingredients is permissible. Small varia- tions from the composition of the best possible mixture will often work no more serious harm than production of a slag 572 QUANTITATIVE ANALYSIS requiring a somewhat higher temperature for complete lique- faction or the production of a lead button that is somewhat larger than is desirable. The fact that some variation is per- missible and also that experienced assayers do not often make pre- liminary analyses of ores often leads the student to the conclusion that the mixing of charges is largely a matter of guess work in any case. This is very far from being true. To enter into a detailed discussion of all classes of ores is not possible in this brief treatment of the subject of assaying. Cer- tain typical ores will be considered and the approximate crucible charge for each class stated. A knowledge of the fundamental principles of assaying will thereby be gained and the proper treatment of other ores will be known as a result of more extensive work in the assaying laboratory. Preliminary Assay. In order to determine the exact propor- tions of the various components of the crucible charge necessary for the best results, a preliminary assay may be made, using smaller quantities of the materials than will be used in the final assay. This serves the purpose of a preliminary analysis and the degree of fluidity of the slag and the size of the lead button obtained will indicate the necessary modification of the charge for the final assay. The preliminary assay is usually omitted unless the available sample of ore is small or unless the ore is already powdered so that its chemical nature cannot be deter- mined by inspection. It is usually preferable to use the full quantities of ore and reagents. If the assay is successful it need not be repeated. If it is unsuccessful it serves all of the purposes of a preliminary and a new assay may then be made. The slag should be perfectly fluid and the lead button obtained should weigh 25 to 30 gm. Silicious Ores Containing No Reducing or Oxidizing Com- pounds. The chief gangue of this class of ores is quartz or sili- cates such as clay or the felspars. The crucible charge must contain the proper fluxes for the production of a liquid slag, also litharge and a reducing agent. The only fluxes required are sodium carbonate and litharge. Argols is probably the best reducing agent. After the materials are thoroughly mixed they are placed in the crucible and covered with a layer of sodium chloride about 1/4 inch thick. The salt fuses at a comparatively THE FIRE ASSAY 573 low temperature (776) and the liquid cover thus formed prevents the loss of powdered ore by action of escaping gases. Silicious Ores. A typical charge for a silicious ore would be as follows: Ore 1 A. T. Litharge 50 gm Sodium carbonate 50 gm Borax glass 2 gm Argols 2 . 5 gm Salt cover Borax glass is here added to increase the fluidity of the slag by the formation of lead borate. If crystallized borax is used, nearly twice as much will be required, as will be seen from the molecular weights of the two substances (202 and 382). In this, as in all other charges suggested in the following pages, the stated proportions of the components are merely average proportions for ores of the general characters indicated. Varia- tion in the charges will often be necessary but these will generally be made upon the advice of the instructor. No condensed state- ment of this kind could meet all of the conditions brought about by small variations in the composition of ores. It is well to remember that increasing the proportion of sodium carbonate somewhat increases the size of the lead button, also that too much borax gives a thick and viscous slag. The variation in the size of the lead button is due to the fact that with small quantities of sodium carbonate present more litharge enters the slag as a basic flux, leaving the argols to be oxidized to some extent by the air. Silicious and Carbonate Ores. For an ore in which. limestone occurs with the silicious gangue the following mixture will serve: Ore 1 A. T. Litharge 35 gm Sodium carbonate 35 gm Borax glass 5 gm Silica 5 gm Argols 2.5 gm Salt cover Oxidizing Ores. Oxidizing ores contain certain reducible oxides or carbonates such as those of iron or manganese, ferric 574 QUANTITATIVE ANALYSIS oxide being the most common. If the main portion of the gangue is still silicious the charge will be similar to that stated above for a silicious ore, but modified by increasing the amount of argols and by the addition of more borax glass, borax or silica to serve as a flux for the metallic oxides. The following charge may be used for a silicious ore containing about 50 percent of hematite: Ore 1 A. T. Litharge 40 gm Sodium carbonate 30 gm Borax glass 10 gm Silica 5 gm Argols 7 gm Salt cover If the slag is sticky instead of fluid increase the proportion of sodium carbonate. If the lead button is too small increase the amount of argols, calculating the amount to be added from the previously determined reducing power of the argols. If the ore contains more than 50 percent of hematite increase the borax glass, silica and argols accordingly. If it contains less than this quantity decrease the amount of these substances. Reducing Ores. Ores containing sulphides or other reducing agents capable of reducing lead oxide must first be roasted or else an oxidizing agent or a desulphurizer must be added. The addition of potassium nitrate is recommended. The production of nitrogen oxides and sulphur dioxide causes some disturbance and the preliminary heating must be moderated accordingly. The addition of iron as a desulphurizer is not suitable for ores containing arsenic on account of the formation of a "speiss" or arsenide of iron. This speiss separates from both slag and lead and it usually carries gold if this metal is present in the ore. If iron nails are used they must be removed from the crucible before pouring and they sometimes cause trouble on account of the adherence of small globules of lead, this causing a loss of gold and silver. For a pyritic ore the following charges are suggested for trial assays. Unless the reducing power of the ore is exactly known it is impossible to predict the weight of lead that will be obtained but the charge may be modified, if necessary, after the first trial. THE FIRE ASSAY 575 Nitrate Method Iron Method Ore.. 0.5 A. T. 70 gm 25 gm Borax glass 2.5 gm Litharge Sodium carbonate. Silica Potassium nitrate. . Nails, cut, 20d .... Salt. . . 2 gm 5 gm 0.5 A. T. 30 gm 50 gm 5 gm 2 gm 4 cover If the lead button obtained by the nitrate method is too small decrease the amount of potassium nitrate; if it is too large increase the potassium nitrate. If the button from the iron method is too small add a calculated weight of argols. Ores Containing Copper, Arsenic or Antimony. If ores con- taining compounds of copper, arsenic or antimony are treated by the methods already described, without modification, these metals will be reduced and will enter the lead button as constitu- ents of the alloy. The button will thereby be rendered brittle and difficult to free from slag. This action must not be prevented by the addition of potassium nitrate because this will prevent also the formation of a lead button of sufficient size. It can be prevented without this interference by largely increasing the proportion of litharge, which keeps the interfering elements in the form of their oxides, these entering the slag by combination with fluxes. Copper oxide will, of course, combine with acid fluxes and the proportion of borax glass or silica is, on this account, increased. Oxides of arsenic or antimony will require basic fluxes. The following charge may be used for an ore which is largely chalcopyrite, Cu 2 S.Fe 2 S 3 . Ore 0.5 A. T. Litharge 125 gm Sodium carbonate 30 gm Borax glass 10 gm Silica 5 gm Salt cover It is here assumed that sufficient lead will be reduced by the sulphides present without the necessity for the addition of another reducing agent. If this is not the case add the required amount of argols in the next fusion. 576 QUANTITATIVE ANALYSIS Copper in a slag is indicated by a red or green color, the former being due to cuprous silicate and borate, the latter to cupric salts of the same acids. Cupellation. After the ore has been decomposed and the alloy of lead, gold and silver has been obtained the button must be treated in such manner as to completely remove the lead. This is easily accomplished because of the readiness with which lead oxidizes when heated in contact with air. The alloy is placed in a small vessel called a "cupel," which is made of bone ash and is shaped as shown in Fig. 121. Bone ash is chiefly composed of calcium phosphate and this has the very valuable property of being able to absorb lead oxide at high temperatures. In making the cupel the ash is moistened and pressed together. This produces a mass which is quite porous so that the lead oxide is readily 12 i. cupel, absorbed. Cupels thus made are very fragile and must be handled carefully. Laboratories in which large numbers of assays are made usually make their own cupels by means of an inexpensive hand or power press. Manufacturers of cupels for shipping usually add to the water a small amount of glue or molasses which serves as a binding material. This chars and blackens when the cupel is heated but it soon burns out. The cupel must be large enough to easily contain the lead alloy after the button is melted. It should weigh at least as much as the button in order to efficiently absorb the necessary quantity of lead oxide. The furnace in which the cupel is heated is of the muffle type and it must have a good draught in order that the lead may be quickly oxidized. The muffle is an arched chamber of fire clay, varying from 4 to 12 inches wide, 8 to 20 inches long and 3 to 6 inches high. It is heated by a furnace in which the fuel may be any of those used for the crucible furnace. As cupellation proceeds the lead is oxidized, a part is volatilized and drawn into the chimney and the remainder is absorbed by the cupel. When this process is finished the button of gold and silver is weighed and it is then prepared for the process of parting. Inquartation. The gold and silver of the button are to be sepa- THE FIRE ASSAY 577 rated by dissolving the silver in nitric acid. If gold constitutes' more than about one-fourth of the weight of the button the silver will dissolve very slowly. In this case it is necessary to increase the proportion of silver after the button has been weighed. This process is known as "inquartation." The button is wrapped in the necessary quantity of pure silver foil and is then placed in a clean cupel and melted by means of a blowpipe. It is kept in a fused condition until the added silver has thoroughly dissolved and it is then allowed to cool. With some experience it is not difficult to determine whether it will be necessary to inquart the button before parting, the depth of yellow indicating the approxi- mate percent of gold. In the beginning it is better to inquart if the button is appreciably yellow. Parting. The addition of nitric acid (specific gravity 1.2) to a button which contains not more than about 25 percent of gold will cause the solution of all of the silver, leaving the gold in the form of a brown skeleton which later usually falls to a coarse powder. The nitric acid must be free from chlorine as otherwise gold will be dissolved to some extent. Annealing. Gold left from the parting process has not the characteristic yellow color of massive gold but acquires it upon being heated. The mere change of color is of little or no impor- tance but the heating that is necessary for complete drying causes the change. This process is called " annealing." After annealing the gold is brushed into the pan of the button balance and weighed. Determination. Sample the ore according to the usual plan but exercise extraordinary care in the operations of mixing and dividing. The last portion obtained should weigh at least 100 gm and it must be ground to pass a 100-mesh sieve without forcing by the brush. Mix the sifted sample by rolling and leave in a flattened pile on the paper. Take out the sample for weighing by means of a spatula, dipping from various parts of the pile and taking the entire depth of the pile at each dipping. Weigh in the pulp balance the amount o ground ore that is required for the fusion, using one of the charges already suggested or a modification made by the instructor according to known variations in the character of the ore. (If the ore is known to be a very rich one the quantity of sample used will be less than 1 A. T. and the other com- ponents of the charge will be reduced accordingly.) 37 578 QUANTITATIVE ANALYSIS Before removing the weighed ore from the pulp balance weigh, on the ordinary laboratory balance, the other components of the charge with the exception of the salt, in the order named in the statement of the charge, beginning with the litharge. All of the reagents must be free from lumps. Place these substances in a flattened pile on a piece of mixing paper or oilcloth and finally brush the ore sample onto the top of the pile and mix well by rolling. Empty the charge into an assay crucible which is 6 to 8 inches in height, brush the paper to remove the last of the mixture and tap the crucible slightly to settle the charge. Lastly cover the mixture with a layer of salt about K inch thick and place the crucible in the furnace, which should not be hot enough to crack the crucible. Raise the temperature gradually, using a moderate temperature until violent effervescence has ceased. After the fusion is quiet heat to a temperature of bright redness for a period lasting from ten minutes to one hour, according to the difficulty experienced in obtain- ing complete decomposition. The pouring mould, which is made of iron and has conical depressions, should be warmed meanwhile to prevent sudden chilling of the button and slag when pouring. Lift the crucible from the furnace, tap lightly to settle globules of lead that may be suspended in the slag and pour quickly but steadily, into the mould. This mould is never made large enough to contain all of the slag, so that the latter will always run over. A mould which would contain the entire contents of the crucible would require an inconveniently long time for cooling. The attention should be fixed, not upon the slag but upon the lead alloy, which appears as a bright stream near the end of the pouring. This stream must be directed toward the center of the mould and it must be poured without splashing. The lead immediately sinks to the bottom where it later solidifies as an inverted cone under the slag. After pouring, the crucible must be free from masses of imperfectly fused slag and from particles of lead. Allow the mould to stand quietly until the slag and lead button are perfectly solid then invert the mould, when the contents will easily drop out. Carefully break the slag from the button by means of a hammer. Examine the slag in order to detect any particles of lead that may have been caught by it. Such particles may be saved but the assay is not reliable in such a case and it is better to begin again, changing the con- ditions as may be necessary to obtain perfect separation. The button should be quite malleable and it should separate easily without leaving a crust of lead on the slag. Carefully free the button from all adhering particles of slag by hammering on a small, clean anvil. This operation should be performed in such a manner as to leave the button in the form of a cube, finally truncating the corners to prevent later injury to the THE FIRE ASSAY 579 cupel If detached particles of lead have been recovered place these on the clean cube and weld into the latter by a stroke of the hammer. The lead button should weigh from 25 to 30 gm although a button as light as 20 gm will often contain all of the gold and silver of the ore, if the fusion has been normal. If a heavier button than 30 gm is obtained do not discard it but reduce its size by scorification, a process to be later described (page 580). If the button is too small or if the slag does not pour and separate well, make another charge, properly modified in accordance with the principles already discussed. Cupellation. Place a cupel in the already heated muffle furnace and bring to a temperature of bright redness, then carefully drop in the button by means of long tongs provided for the purpose. Close the muffle door and raise the temperature to about 700 (bright redness) when the black crust of lead suboxide, Pb 2 0, changes to the yellow monoxide, PbO, and this begins to volatilize. Remove the top of the muffle door, the lower half being left in to shield the cupel from the entering current of cold air. If the door is in one piece remove and place two or three empty cupels in front of the one that contains the alloy. Much of the lead oxide is vaporized and is drawn into the chimney while the remainder is absorbed by the cupel. If the temperature is too high the lead will boil, with consequent mechanical loss. If the temperature is too low the button will freeze and again loss will occur through "sprouting." The latter action is due to the contraction of the cooled and solidified skin of lead, the liquid alloy from the interior breaking through and often being thrown out of the cupel. A sprouted button should be at once discarded as the results obtained from it will be unreliable. The correct temperature for cupellation is indicated by the presence of a ring of litharge crystals upon the inner surface of the cupel, just above the liquid alloy. As cupellation proceeds the button becomes smaller. Toward the end of the process the temperature must be raised somewhat, on account of the rise in the melting point of the alloy which is now richer in gold and silver, metals of higher melting point than that of lead. So long as lead remains the surface of the alloy is covered by a thin, iridescent layer of oxide which is in continuous motion over the surface. As the last of the lead is oxidized the iridescence suddenly disappears and the surface brightens or "blicks." To insure the removal of the last trace of lead close the door of the muffle and raise the temperature for about one minute. Remove the cupel from the muffle and cover the former in order to prevent too rapid cooling and consequent sprouting of the button. When the button is cold take it out by means of a pair of 580 QUANTITATIVE ANALYSIS strong, pointed pliers and brush with a stiff brush to remove particles of oxide or cupel material. Weigh on the button balance and record the weight in milligrams. Inquartation. If the button is silver white or only faintly yellow it may be parted without previous inquartation. If the intensity of yellow indicates more than about 25 percent of gold cut a piece of pure silver foil weighing from one to three times as much as the button, according -to the indicated composition of the button. Wrap the button in this foil and place in a new, clean cupel. Carefully fuse by means of a blowpipe and keep in the fused condition for one minute, then allow to cool. Parting. Place the button in a No. 1 porcelain crucible and nearly fill the latter with nitric acid whose specific gravity is 1.2. Warm gently until action begins. The silver should dissolve rapidly enough to cause a moderate evolution of gas. If it does not do so. it contains too much gold and parting will be imperfect. In this case remove the button, wash, dry and fuse with more silver. When all action of the acid has ceased and only a brown skeleton of gold remains carefully decant the acid solution of silver nitrate into a porcelain dish, allowing no particles of gold to escape. Wash by decantation, using distilled water which has been tested and found to be free from chlorides. The washing process must be performed with extreme care as small particles of gold are easily detached and lost. A white porcelain dish is used for receiving the washings because of the consequent ease in detecting lost gold particles. Wash until all silver is removed, as shown by test- ing with hydrochloric acid. After the washing is completed dry the crucible on the hot plate. Annealing. Heat the crucible over the ordinary burner until the gold changes from brown to yellow, then allow to cool, brush into the pan of the button balance and weigh. Calculate the ounces per ton of gold and silver upon the basis of milligrams obtained per assay ton of ore, making the proper correction in the weight of the silver for any silver that has been found in the litharge. Scorification. For the treatment of high-grade silver ores, and especially those containing copper, arsenic, antimony or zinc, the scorification process is simpler than the crucible process. In the scorification. process the ore is mixed with granulated lead, usually with the addition of a small amount of borax glass or silica, and is heated in an oxidizing atmosphere in a shallow vessel of fire clay called a "scorifier." This is shown in Fig. 122. THE FIRE ASSAY 581 Its size varies between % mcn an d 4-J^ inches, inside diameter, but the size ordinarily used is about 2 inches in diameter. In the crucible process the reactions are between the ore and added fluxes, litharge being one of these, and enough lead is reduced to form a button of the correct weight. Air plays little or no part in this process. In the scorification process the chief flux is lead oxide but it is formed by the oxidation of lead which is added in relatively large quantities, the unused excess being vaporized as oxide. Because of this possibility of expelling the unused excess of flux there is a rather large permissible latitude in the quantity of lead that may be taken. Because the process FIG. 122. Scorifier. may be continued until the required quantity of lead is left for the button, no reducing agent being used, no preliminary calculation concerning this matter is necessary. Because the process is an oxidizing one copper, arsenic, antimony and zinc are easily driven into the slag and do not contaminate the button. In other words, the adjustment of the charge involves little more than the addition of a certain minimum amount of lead, with a small quantity of borax glass or silica to aid in the formation of a liquid slag if the ore contains little acid gangue. These features give the scorification process a decided advantage over the crucible process with ores to which it will apply. It is not suitable for low-grade silver ores because of the small amount of ore that must necessarily be used. Not more than 0.2 A. T. can conveniently be scorified and 0.1 A.T. will usually give better results if a scorifier of ordinary dimensions is used. The process does not work well with gold ores because of appreciable losses of gold in the slag. The following charges are suitable for the scorification of typ- ical ores of the classes named. 582 QUANTITATIVE ANALYSIS Silicious ores: Ore 0. 1 A. T. Lead 40 gm Borax glass 2 gm Ores containing arsenic and antimony: Ore 0.1 A. T. Lead 50 gm Borax glass 5 gm Ores containing copper: Ore 0. 1 A. T. Lead 65 gm Borax glass 1 gm Silica 1 gm Ores containing iron (pyritic ores) : Ore .'.... 0.1 A. T. Lead 50 gm Borax glass 3 gm The lead that is used for this purpose (known as "test lead") is a finely granular form and it should be as nearly as possible free from silver. The same difficulty is encountered in obtaining silver-free lead as was noted in the case of litharge and for the same reason. It is therefore necessary to make a preliminary assay of each lot of test lead and to correct the results of the assay of ores according to the amount of silver found in the lead. Determination. Sample and weigh the ore according to the direc- tions given for the assay by the crucible process. Weigh the materials for the scorifier charge according to one of the statements given above. Place one-half of the lead in the scorifier, add the weighed sample of ore and mix well with the lead by means of a platinum wire, then add the rest of the lead. Place the borax glass and silica on the top and then place the scorifier in the muffle, which should be hot. Close the door of the muffle and raise the temperature to the point where the lead is melted and the black suboxide changes to the yellow, more volatile monoxide. Open the door of the muffle and admit a full supply of air. The lead now rapidly oxidizes and a part of the oxide vaporizes but most of it attacks the ore and decomposes it with the formation of a liquid slag. This process of oxidation and slag formation will now continue until the ore is completely decomposed, a perfectly fluid slag forming a THE FIRE ASSAY 583 ring around the circumference of the scorifier leaving the lead exposed over a large circle in the center. As lead is thus used in slag formation and through vaporization of the oxide the exposed circle of the metal (the "bull's eye") becomes smaller on account of the descent of the sur-~ face toward the narrower part of the scorifier. When the process is finished the slag entirely covers the lead. After the "bull's eye" has disappeared close the muffle door and raise the temperature for a short time in order that the slag may become so thoroughly liquefied that it will not become viscous during pouring, then pour into the mould. The mould that was used in the crucible process may be used here also but a more shallow mould which has hemispherical depressions is preferred. When the slag and button are cold, remove from the mould and free from slag exactly as was done with the button obtained from the crucible fusion. Cupel, part and weigh as already directed. TABLE OF LOGARITHMS 586 QUANTITATIVE ANALYSIS LOGARITHMS Natural Numbers 1 2 3 4 5 6 7 8 9 Proportional Parts 1 2|3|4 5 6 7 8 9 10 0000 0043 0086 0128 0170 0212 0253 0294 0334 0374 4 8 12 17 21 25 29 33 37 11 0414 0453 0492 0531 0569 0607 0645 0682 0719 0755 4 8 11 15 19 23 26 30 34 12 0792 0828 0864 0899 0934 0969 1004 1038 1072 1106 3 7 10 14 17 21 24 28 31 13 1139 1173 1206 1239 1271 1303 1335 1367 1399 1430 3 6 10 13 16 19 23 26 29 14 1461 1492 1523 1553 1584 1614 1644 1673 1703 1732 3 6 9 12 15 18 21 24 27 15 1761 1790 1818 1847 1875 1903 1931 1959 1987 2014 3 6 8 11 14 17 20 22 25 16 2041 2068(2095 2122 2148 2175 2201 2227 2253 2279 3 5 8 11 13 16 18 21 24 17 2304 2330 2355 2380 2405 2430 2455 2480 2504 2529 2 5 7 10 12 15 17 20 22 18 2553 2577 2601 2625 2648 2672 2695 2718 2742 2765 2 5 7 9 12 14 16 19 21 19 2788 2810 2833 2856 2878 2900 2923 2945 2967 2989 2 4 7 9 11 13 16 18 20 20 3010 3032 3054 3075 3096 3118 3139 3160 3181 3201 2 4 6 8 11 13 15 17 19 21 3222 3243 3263 3284 3304 3324 3345 3365 3385 3404 2 4 6 8 10 12 14 16 18 22 3424 3444 3464 3483 3502 3522 3541 3560 3579 3598 2 4 6 8 10 12 14 15 17 23 3617 3636 3655 3674 3692 3711 3729 3747 3766 3784 2 4 6 7. 9 11 13 15 17 24 3802 3820 3838 3856 3874 3892 3909 3927 3945 3962 2 4 5 7 9 11 12 14 16 25 3979 3997 4014 4031 4048 4065 4082 4099 4116 4133 2 3 5 7 9 10 12 14 15 26 4150 4166 4183 4200 4216 4232 4249 4265 4281 4298 2 3 5 7 8 10 11 13 15 27 28 4314 4472 4330 4487 4346 4502 4362 4518 4378 4533 4393 4409 4548 4564 4425 4579 4440 4594 4456 4609 2 2 3 3 5 5 6 6 8 8 9 9 11 11 13 12 14 14 29 4624 4639 4654 4669 4683 4698 4713 4728 4742 4757 1 3 4 6 7 9 10 12 13 30 4771 4786 4800 4814 4829 4843 4857 4871 4886 4900 1 3 4 6 7 9 10 11 13 31 4914 4928 4942 4955 4969 4983 4997 5011 5024 5038 1 3 4 6 7 8 10 11 12 32 5051 5065 5079 5092 5105 5119 5132 5145 5159 5172 1 3 4 5 7 8 9 11 12 33 5185 5198 5211 5224 5237 5250 5263 5276 5289 5302 1 3 4 5 6 8 9 10 12 34 5315 5328 5340 5353 5366 5378 5391 5403 5416 5428 1 3 4 5 6 8 9 10 11 35 5441 5453 5465 5478 5490 5502 5514 5527 5539 5551 1 2 4 5 6 7 9 10 11 36 5563 5575 5587 5599 5611 5623 5635 5647 5658 5670 1 2 4 5 6 7 8 10 11 37 5682 5694 5705 5717 5729 5740 5752 5763 5775 5786 1 2 3 5 6 7 8 9 10 38 5798 5809 5821 5832 5843 5855 5866 5877 5888 5899 1 2 3 5 6 7 8 9 10 39 5911 5922 5933 5944 5955 5966 5977 5988 5999 6010 1 2 3 4 5 7 8 9 10 40 6021 6031 6042 6053 6064 6075 6085 6096 6107 6117 1 2 3 4 5 6 8 9 10 41 6128 6138 6149 6160 6170 6180 6191 6201 6212 6222 1 2 3 4 5 6 7 8 9 42 6232 6243 6253 6263 6274 6284 6294 6304 6314 6325 1 2 3 4 5 6 7 8 9 43 6335 6345 6355 6365 6375 6385 6395 6405 6415 6425 1 2 3 4 5 6 7 8 9 44 6435 6444 6454 6464 6474 6484 6493 6503 6513 6522 1 2 3 4 5 6 7 8 9 45 6532 6542 6551 6561 6571 6580 6590 6599 6609 6618 1 2 3 4 5 6 7 8 9 46 6628 6637 6646 6656 6665 6675 6684 6693 6702 6712 1 2 3 4 5 6 7 7 8 47 6721 6730 6739 6749 6758 6767 6776 6785 6794 6803 1 2 3 4 5 5 6 7 8 48 6812 6821 68306839 6848 6857 6866 6875 6884 6893 1 2 3 4 4 5 6 7 8 49 6902 6911 6920 6928 6937 6946 6955 6964 6972 6981 1 2 3 4 4 5 6 7 8 50 6990 6998 7007 7016 7024 7033 7042 7050 7059 7067 1 2 3 3 4 5 6 7 8 51 7076 7084 7093 7101 7110 7118 7126 7135 7143 7152 1 2 3 3 4 5 6 7 8 52 7160 7168 7177 7185 7193 7202 7210 7218 7226 7235 1 2 2 3 4 5 6 7 7 53 7243 7251 7259 7267 7275 7284 7292 7300 7308 7316 1 2 2 3 4 5 6 6 7 54 7324 7332 7340 7348 7356 7364 7372 7380 7388 7396 1 2 2 3 4 5 6 6 7 TABLES 587 LOGARITHMS Natural Numbers 1 2 3 4 5 6 7 8 9 Proportional Parts 1 2 3 4J5 6|7|8 | 55 7404 7412 7419 7427 7435 7443 7451 7459 7466 7474 2 2 3 4 5 5 6 56 7482 7490 7497 7505 7513 7520 7528 7536 7543 7551 2 2 8 4 5 5 6 57 7559 7566 7574 7582 7589 7597 7604 7612 7619 7627 2 2 a 4 5 5 6 58 7634 7642 7649 7657 7664 7672 7679 7686 7694 7701 1 2 a 4 4 5 6 59 7709 7716 7723 7731 7738 7745 7752 7760 7767 7774 1 2 a 4 4 5 6 60 7782 7789 7796 7803 7810 7818 7825 7832 7839 7846. 1 1 2 3 4 4 5 G 61 7853 7860 7868 7875 7882 7889 7896 7903 7910 7917 1 1 2 s 4 4 5 6 62 7924 7931 7938 7945 7952 7959 7966 7973 7980 7987 1 1 2 3 3 4 5 6 63 7993 8000 8007 8014 8021 8028 8035 8041 8048 8055 1 1 2 3 3 4 5 5 64 8062 8069 8075 8082 8089 8096 8102 8109 8116 8122 1 1 2 3 3 4 5 5 65 8129 8136 8142 8149 8156 8162 8169 8176 8182 8189 1 1 2 3 3 4 5 5 66 8195 8202 8209 8215 8222 8228 8235 8241 8248 8254 1 1 2 3 3 4 5 5 67 8261 8267 8274 8280 8287 8293 8299 8306 8312 8319 1 1 2 3 3 4 5 5 68 8325 8331 8338 8344 8351 8357 8363 8370 8376 8382 1 1 2 3 3 4 4 5 69 8388 8395 8401 8407 8414 8420 8426 8432 8439 8445 1 1 2 2 3 4 4 5 70 8451 8457 8463 8470 8476 8482 8488 8494 8500 8506 1 1 2 2 3 4 4 5 71 8513 8519 8525 8531 8537 8543 8549 8555 8561 8567 1 1 2 2 3 4 4 5 72 8573 8579 8585 8591 8597 8603 8609 8615 8621 8627 1 2 2 3 4 4 5 73 8633 8639 8645 8651 8657 8663 8669 8675 8681 8686 1 2 2 3 4 4 5 74 8692 8698 8704 8710 8716 8722 8727 8733 8739 8745 1 2 2 3 4 4 5 75 8751 8756 8762 8768 8774 8779 8785 8791 8797 8802 1 2 2 3 3 4 5 76 8808 8814 8820 8825 8831 8837 8842 8848 8854 8859 1 2 2 3 3 4 5 77 8865 8871 8876 8882 8887 8893 8899 8904 8910 8915 1 2 2 3 3 4 4 78 8921 8927 8932 8938 8943 8949 8954 8960 8965 8971 1 2 2 3 3 4 4 79 8976 8982 8987 8993 8998 9004 9009 9015 9020 9025 1 2 2 3 3 4 4 80 9031 9036 9042 9047 9053 9058 9063 9069 9074 9079 1 2 2 3 3 4 4 81 9085 9090 9096 9101 9106 9112 9117 9122 9128 9133 1 2 2 3 3 4 4 82 9138 9143 9149 9154 9159 9165 9170 9175 9180 9186 1 2 2 3 3 4 4 83 9191 9196 9201 9206 9212 9217 9222 9227 9232 9238 1 2 2 3 3 4 4 84 9243 9248 9253 9258 9263 9269 9274 9279 9284 9289 1 2 2 3 3 4 4 85 9294 9299 9304 9309 9315 9320 9325 9330 9335 9340 1 2 2 3 3 4 4 86 9345 9350 9355 9360 9365 9370 9375 9380 9385 9390 1 2 2 3 3 4 4 87 9395 9400 9405 9410 9415 9420 9425 9430 9435 9440 1 1 2 2 3 3 4 88 9445 9450 9455 9460 9465 9469 9474 9479 9484 9489 1 1 2 2 3 3 4 89 9494 9499 9504 9509 9513 9518 9523 9528 9533 9538 1 1 2 2 3 3 4 90 9542 9547 9552 9557 9562 9566 9571 9576 9581 9586 1 1 2 2 3 3 4 91 9590 9595 9600 9605 9609 9614 9619 9624 9628 9633 1 2 2 3 3 4 92 9638 9643 9647 9652 9657 9661 9666 9671 9675 9680 1 2 2 3 3 4 93 9685 9689 9694 9699 9703 9708 9713 9717 9722 9727 1 2 2 3 3 4 94 9731 9736 9741 9745 9750 9754. 9759 9763 9768 9773 1 2 2 3 3 4 95 9777 9782 9786 9791 9795 9800 9805 9809 9814 9818 1 2 2 3 3 4 96 9823 9827 9832 9836 9841 9845 9850 9854 9859 9863 1 2 2 3 3 4 97 9868 9872*9877 9881 9886 9890 9894 9899 9903 9908 1 2 2 3 3 4 98 9912 9917 9921 9926 9930 9934 9939 9943 9948 9952 1 2 2 3 3 4 QQ 9956 9961 9965 9969 9974 9978 9983 9987 9991 9996 1 2 2 3 3 3 TABLE OF ANTILOGARITHMS 590 QUANTITATIVE ANALYSIS ANTILOGAEITHMS Logarithms 1 2 3 4 5 6 7 8 9 Proportional Parts 1 2 3| 4 5 6|7|8 .00 1000 1002 1005 1007 1009 1012 1014 1016 1019 1021 1 1 1 1 2 2! .01 1023 1026 1028 1030 1033 1035 1038 1040 1042 1045 1 1 1 1 2 2 .02 1047 1050 1052 1054 1057 1059 1062 1064 1067 1069 1 1 1 1 2 2 .03 1072 1074 1076 1079 1081 1084 1086 1089 1091 1094 1 1 1 1 2 2 .04 1096 1099 1102 1104 1107 1109 1112 1114 1117 1119 1 1 1 1 2 2 2 .05 1122 1125 1127 1130 1132 1135 1138 1140 1143 1146 1 1 1 1 2 2 2 .06 1148 1151 1153 1156 1159 1161 1164 1167 1169 1172 1 1 1 1 2 2 2 .07 1175 1178 1180 1183 1186 1189 1191 1194 1197 1199 1 1 1 1 2 2 2 .08 1202 1205 1208 1211 1213 1216 1219 1222 1225 1227 1 1 1 1 2 2 2 .09 1230 1233 1236 1239 1242 1245 1247 1250 1253 1256 1 1 1 1 2 2 2 .10 1259 1262 1265 1268 1271 1274 1276 1279 1282 1285 1 1 1 1 2 2 2 .11 1288 1291 1294 1297 1300 1303 1306 1309 1312 1315 1 1 1 2 2 2 2 .12 1318 1321 1324 1327 1330 1334 1337 1340 1343 1346 1 1 1 2 2 2 2 .13 1349 1352 1355 1358 1361 1365 1368 1371 1374 1377 1 1 1 2 2 2 3 .14 1380 1384 1387 1390 1393 1396 1400 1403 1406 1409 1 1 1 ? 2 2 3 .15 1413 1416 1419 1422 1426 1429 1432 1435 1439 1442 1 1 1 2 2 2 3 .16 1445 1449 1452 1455 1459 1462 1466 1469 1472 1476 1 1 1 2 2 2 3 .17 1479 1483 1486 1489 1493 1496 1500 1503 1507 1510 1 1 1 2 2 2 3 .18 1514 1517 1521 1524 1528 1531 1535 1538 1542 1545 1 1 1 2 2 2 3 .19 1549 1552 1556 1560 1563 1567 1570 1574 1578 1581 1 1 1 2 2 3 3 .20 1585 1589 1592 1596 1600 1603 1607 1611 1614 1618 1 1 1 2 2 3 3 .21 1622 1626 1629 1633 1637 1641 1644 1648 1652 1656 1 1 2 2 2 3 3 .22 1660 1663 1667 1671 1675 1679 1683 1687 1690 1694 1 1 2 2 2 3 3 .23 1698 1702 1706 1710 1714 1718 1722 1726 1730 1734 1 1 2 2 2 3 3 .24 1738 1742 1746 1750 1754 1758 1762 1766 1770 1774 1 1 2 2 2 3 3 .25 1778 1782 1786 1791 1795 1799 1803 1807 1811 1816 1 1 2 2 2 3 3 .26 1820 1824 1828 1832 1837 1841 1845 1849 1854 1858 1 1 2 2 3 3 3 .27 1862 1866 1871 1875 1879 1884 1888 1892 1897 1901 1 1 2 2 3 3 3 .28 1905 1910 1914 1919 1923 1928 1932 1936 1941 1945 1 1 2 2 3 3 4 .29 1950 1954 1959 1963 1968 1972 1977 1982 1986 1991 1 1 2 2 3 3 4 .30 1995 2000 2004 2009 2014 2018 2023 2028 2032 2037 1 1 2 2 3 3 4 .31 2042 2046 2051 2056 2061 2065 2070 2075 2080 2084 1 1 2 2 3 3 4 .32 2089 2094 2099 2104 2109 2113 2118 2123 2128 2133 1 1 2 2 3 3 4 .33 2138 2143 2148 2153 2158 2163 2168 2173 2178 2183 1 2 2 3 3 4 .34 2188 2193 2198 2203 2208 2213 2218 2223 2228 2234 1 2 2 3 3 4 4 .35 2239 2244 2249 2254 2259 2265 2270 2275 2280 2286 1 2 2 3 3 4 4 .36 2291 2296 2301 2307 2312 2317 2323 2328 2333 2339 1 2 2 3 3 4 4 .37 2344 2350 2355 2360 2366 2371 2377 2382 2388 2393 1 2 2 3 3 4 4 .38 2399 2404 2410 2415 2421 2427 2432 2438 2443 2449 1 1 2 2 3 3 4 4 .39 2455 2460 2466 2472 2477 2483 2489 2495 2500 2506 1 1 2 2 3 3 4 5 .40 2512 2518 2523 2529 2535 2541 2547 2553 2559 2564 1 1 2 2 3 4 4 5 .41 2570 2576 2582 2588 2594 2600 2606 2612 2618 2624 1 1 2 2 3 4 4 5 .42 2630 2636 2642 2649 2655 2661 2667 2673 2679 2685 1 1 2 2 3 4 4 5 .43 2692 2698 2704 2710 2716 2723 2729 2735 2742 2748 1 1 2 3 3 4 4 5 .44 2754 2761 2767 2773 2780 2786 2793 2799 2805 2812 1 1 2 3 3 4 4 5 .45 2818 2825 2831 2838 2844 2851 2858 2864 2871 2877 1 1 2 3 3 4 5 5 .46 2884 2891 2897 2904 2911 2917 2924 2931 2938 2944 1 1 2 3 3 4 5 5 .47 2951 2958 2965 2972 2979 2985 2992 2999 3006 3013 1 1 2 3 3 4 5 5 .48 3020 3027 3034 3041 3048 3055 3062 3069 3076 3083 1 1 2 3 4 4 5 6 .49 3090 3097 3105 3112 3119 3126 3133 3141 3148 3155 1 1 2 3 4 4 5 6 TABLES ANTILOGARITHMS 591 Logarithms 1 2 3 4 5 6 7 8 9 Proportional Farts 1| 2 3| 4[ b 6| 7| 8| .50 3162 3170 3177 3184 3192 3199 3206 3214 3221 3228 1 1 2 3 4 4 5 6 7 .51 3236 3243 3251 3258 3266 3273 3281 3289 3296 3304 2 2 3 4 5 5 6 7 .52 3311 3319 3327 3334 3342 3350 3357 3365 3373 3381 2 2 3 4 5 5 6 7 .53 3388 3396 3404 3412 3420 3428 3436 3443 3451 3459 2 2 3 4 5 6 6 7 .54 3467 3475 3483 3491 3499 3508 3516 3524 3532 3540 2 2 3 4 5 6 6 7 .55 3548 3556 3565 3573 3581 3589 3597 3606 3614 3622 2 2 3 4 5 6 7 7 .56 3631 3639 3648 3656 3664 3673 3681 3690 3698 3707 2 3 3 4 5 6 7 8 .57 3715 3724 3733 3741 3750 3758 3767 3776 3784 3793 2 3 3 4 5 6 7 8 .58 3802 3811 3819 3828 3837 3846 3855 3864 3873 3882 2 3 4 4 5 6 7 8 .59 3890 3899 3908 3917 3926 3936 3945 3954 3963 3972 1 2 3 4 5 5 6 7 8 .60 3981 3990 3999 4009 4018 4027 4036 4046 4055 4064 1 2 3 4 5 6 6 7 8 .61 4074 4083 4093 4102 4111 4121 4130 4140 4150 4159 1 2 3 4 5 6 7 8 9 .62 4169 4178 4188 4198 4207 4217 4227 4236 4246 4256 1 2 3 4 5 6 7 8 9 .63 4266 4276 4285 4295 4305 4315 4325 4335 4345 4355 1 2 3 4 5 6 7 8 9 .64 4365 4375 4385 4395 4406 4416 4426 4436 4446 4457 1 2 3 4 5 6 7 8 9 .65 4467 4477 4487 4498 4508 4519 4529 4539 4550 4560 1 2 3 4 5 6 7 8 9 .66 4571 4581 4592 4603 4613 4624 4634 4645 4656 4667 1 2 3 4 5 6 7 9 10 .67 4677 4688 4699 4710 4721 4732 4742 4753 4764 4775 1 2 3 4 5 7 8 9 10 .68 4786 4797 48084819 4831 4842 4853 4864 4875 4887 1 2 3 4 6 7 8 9 10 .69 4898 4909 4920 4932 4943 4955 4966 4977 4989 5000 1 2 3 5 6 7 8 9 10 .70 5012 5023 5035 5047 5058 5070 5082 5093 5105 5117 1 2 4 5 6 7 8 9 11 .71 5129 5140 5152 5164 5176 5188 5200 5212 5224 5236 1 2 4 5 6 7 8 10 11 .72 5248 5260 5272 5284 5297 5309 5321 5333 5346 5358 1 2 4 5 6 8 9 10 11 .73 5370 5383 5395 5408 5420 5433 5445 5458 5470 5483 1 3 4 5 6 8 9 10 11 .74 5495 5508 5521 5534 5546 5559 5572 5585 5598 5610 1 3 4 5 6 8 9 10 12 .75 5623 5636 5649 5662 5675 5689 5702 5715 5728 5741 1 3 4 5 7 8 9 10 12 .76 5754 5768 5781 5794 5808 5821 5834 5848 5861 5875 1 3 4 5 7 8 9 11 12 .77 5888 5902 5916 5929 5943 5957 5970 5984 5998 6012 1 3 4 5 7 8 10 11 12 .78 6026 6039 6053 6067 6081 6095 6109 6124 6138 6152 1 3 4 6 7 8 10 11 13 .79 6166 6180 6194 6209 6223 6237 6252 6266 6281 6295 1 3 4 G 7 9 10 11 13 .80 6310 6324 6339 6353 6368 6383 6397 6412 6427 6442 1 3 4 6 7 Q 10 12 13 .81 6457 6471 64866501 6516 6531 6546 6561 6577 6592 2 3 5 6 8 9 11 12 14 .82 6607 6622 6637 6653 6668 6683 6699 6714 6730 6745 2 3 5 6 8 9 11 13 14 .83 6761 6776 6792 6808 6823 6839 6855 6871 6887 6902 2 3 5 G 8 9 11 13 14 .84 6918 6934 6950 6966 6982 . 6998 7015 7031 7047 7063 2 3 5 6 8 10 11 13 15 .85 7079 7096 7112 7129 7145 7161 7178 7194 7211 7228 2 3 5 7 8 10 12 13 15 .86 7244 7261 7278 7295 7311 7328 7345 7362 7379 7396 2 3 5 7 8 10 12 13 15 .87 7413 7430 7447 7464 7482 7499 7516 7534 7551 7568 2 3 5 7 9 10 12 14 16 .88 7586 7603 7621 7638 7656 7674 7691 7709 7727 7745 2 4 5 7 9 11 12 14 16 .89 7762 7780 7798 7816 7834 7852 7870 7889 7907 7925 2 4 5 7 9 11 13 14 16 .90 7943 7962 7980 7998 8017 8035 8054 8072 8091 8110 2 4 6 7 9 11 13 15 17 .91 8128 8147 8166 8185 8204 8222 8241 8260 8279 8299 2 4 6 8 9 11 13 15 17 .92 8318 8337 8356 8375 8395 8414 8433 8453 8472 8492 2 4 6 8 10 12 14 15 17 .93 8511 8531 8551 8570 8590 8610 8630 8650 8670 8690 2 4 6 8 10 12 14 16 18 .94 8710 8730 8750 8770 8790 8810 8831 S851 8872 8892 2 4 6 8 10 12 14 16 18 .95 8913 8933 8954 8974 8995 9016 9036 9057 9078 9099 2 4 b 8 10 12 15 17 19 .96 9120 9141 9162 9183 9204 9226 9247 9268 9290 9311 2 4 6 8 11 13 15 17 19 .97 9333 9354 9376 9397 9419 9441 9462 9484 9506 9528 2 4 7 9 11 13 15 17 20 .98 9550 9572 9594 9616 9638 9661 9683 9705 9727 9750 2 4 7 9 11 13 16 18 20 .99 9772 9795 9817 9840 9863 9886 9908 9931 9954 9977 2 5 7 9 11 14 16 18 20 INDEX Abbe" refractometer, 359 Absorbents for carbon dioxide, 133 for hydrochloric acid, 131 Accuracy, limit of, 2 Acetate method for manganese, 465 for phosphorus, 452 Acetic acid in vinegar, 236 Acetyl value, 376 Acid fluxes, 50, 568 hydrochloric, standard solution, 224 value for oils, 368 Acidity of oils, 352 Acids, gravimetric standardization of, 222 laboratory, concentration, 236 standard, 221 primary standards for, 217 Adjustment of balance, 58 of standard solutions, 205 Adsorption, 27 by aluminium hydroxide, 89 Adulteration of milk, 536 Agricultural limestone, 239 materials, 510 Albumenoid nitrogen, 419, 422 Albumin of milk, 541 Aliquot parts, 194 Alkali metals in silicates, 291, 294 Alkalinity of limestone, 239 of water, 229 Allotropism of iron, 481 Alloys, anti-friction metals, 508 antimony in, 508 brass and bronze, 157, 505 copper in, 157, 161, 506, 509 lead in, 506, 509 nickel-chromium, for triangles, 44 Alloys, nickel in, 163 silver in, 161, 276 tin in, 505, 508 zinc in, 507 Aluminium, 87 hydroxide, adsorption by 89 solubility, 88 in minerals, 288, 294 Alundum, 42 Ammonia in water, 417, 422 Ammonium, 101 citrate solution for phosphorus, 529 Analyzed chemicals, 70 Annealing of gold, 577, 580 of steel, 488, 504 Anti-friction metals, 508 Anti-logarithms, table, 590, 591 Antimony in anti-friction metals, 508 in the fire assay, 575 Apparatus for electro-analysis, 148 general, 77 Apparent valence, 241 Arachis oil, 383 Argols, 570 Arm ratio, 65 Arsenical insecticides, 272 Arsenic in fire assay, 575 in insecticides, 272 in Paris green, 273, 274 oxidation by iodine, 260 reduction by hydriodic acid, 261 Arsenous chloride, distillation, 274 Ash, fusing point, 302 of coal, 300, 305 of condensed milk, 558 of cream, 557 of milk, 540 Assay, fire, 564 preliminary, 572 38 593 594 INDEX Assay ton, 566 Atomic weights, 8 Austenite, 485 Availability of nitrogen, 521 Available chlorine in bleaching powder, 270 oxygen, Bunsen's method, 265 in peroxides, 255, 264 Averages, rule of, 4 B Babcock method for fat, 552, 557 Balance, 51 adjustment, 58 arm ratio, 65 arms, relative length, 61, 65 assay, 566 beam rests, 52 . chainomatic, 57 . pulp, 565 rests, 63 sensibility, 53, 60 to set in motion, 58 zero point, 59, 63 Barium, 91 hydroxide, standard solution, 438, 442 salts, solubility, 91, 92 sulphate, decomposition, 93 occlusion of salts, 93 Bases and carbonates, mixtures, 227 Bases, primary standards for, 235 standard solutions, 233 Basic fluxes, 50, 569 Basicity of platinum after ignition, 40 Baudouin test for sesame oil, 385 Baum6 scale, 345 Beam, balance, to set in motion, 58 Beeswax, 369 Bicarbonates, titration, 229 Bismuthate method for manganese, 461 Blast lamps, 47 Bleaching powder, 270 Blistering of platinum, 40 Boiler compounds, 403 Borax, 238, 568 Borda method for weighing, 65 Boric acid, 237 Brass, 157, 505 British thermal unit, 314, 327 Bromine, 115 Bronze, 505 Bulbs for standardizing, 185 Bumping, 515 Bunsen's method for available oxy- gen, 265 Burette, gas, 328 Schellbach, 171, 176 Burettes, 170 calibration, 188, 189 outflow time, 176 * Burners, 45 blast lamps, 47 E. and A., 46 Meker, 47 Teclu, 46 Burning oils, 345 point of oils, 348, 350 Butter, 558 casein in, 560 colors, 562 fat in, 559 moisture in, 559 salt in, 560 substitutes, 558 Butyrin, 369 Butyro-refractometer, 361 Cain method for vanadium, 475 Calcium, 78 chloride as drying agent, 33, 132 determination by permanga- nate, 252 in minerals, 288, 294 oxalate, decomposition, 79 solubility, 78 salts, solubility, 78 sulphate, 80 Calculations of volumetric analysis, 191 INDEX 595 Caldwell crucible, 25 Calibration by standard bulbs, 184, 187, 189 of burettes, 188, 189 of flasks, 187, 188 of pipettes, 189 of standard bulbs, 186 of volumetric apparatus by weighing, 179, 188 of weights, 66, 67 temperature for, 182 Calorie, 314, 327 Calorimeter, 317, 319 Calorimetry, radiation correction, 322, 326 Capacities, absolute and relative, 172 Caprin, 369 Caproin, 369 Caprylin, 369 Carbonate minerals, 284 Carbonates and bases, 227 and bicarbonates, 228 titration, 226, 228, 229 Carbon, combustion, in steel, 445 dioxide, 127 absorbents for, 133 absorption by standard bases, 234 in carbonate minerals, 287 in gas mixtures, 333, 341 monoxide, in gas mixtures, 334, 341 total, in coal, 311 Carbonic acid, 127 in water, 404 Care of platinum, 40 Carius method for halogens, 126 Case hardening, 497, 504 Casein in butter, 560 in milk, 541 Cathode, mercury, 166 Caustic potash, 228 soda, 228 Cementite, 483, 485 Chain rider, 57 Chainomatic balance, 57 Chemical glassware, 71 porcelain, 37 Chemicals, analyzed, 70 Chemists' slide rule, 7 Chill test, 353 Chimney gases, 343 Chlorides, 87 reduction of permanganates, 244 Chlorine, 115 in water, 415 available, in bleaching powder, 270 Chlorplatinate method for potas- sium, 97 Chlorplatinic acid, 103 Chromite, 259 Chromium, 259 and nickel alloys, 44 in steel, 468 Chromophors, 208 Circuit fo'r electro-analysis, 150 Circular action of beam rests, 52 Citrate insoluble phosphorus, 528 Citric acid, 236 Clark's method for hardness, 230 Classes of methods, 4 Classification of indicators, 209 Cleaning solution, 186 Cleanliness, 1 of balance, 58 Clinker of coal, 302 Clupanodonic acid, 364 Coal, 297 ash, 300, 305 clinker, 302 combustion apparatus, 312 fixed carbon, 299, 306 fuel value, 314, 318, 324 fusing point of ash, 302 hydrogen in, 311 moisture in, 299, 305 nitric acid formation, 321 oxygen in, 309 sampling, 298, 303 sulphur in, 307, 310 sulphuric acid formation, 321 596 INDEX Coal, total carbon in, 311 volatile combustible matter in, 299, 306 Cochineal, 211, 214 Coefficient of fineness, 413 Coke, 299 Cold test, 354 Colloids, 19 Color change of indicators, 207 of iron solutions, 245, 248 of water, 413 reactions for oils, 386 Colors in butter, 562 Combined carbon in steel, 445 Combustion apparatus, 312, 436 Condensed milk, 557 Copper, 156, 161, 163, 166, 167 determination by thiosulphate, 267 in alloys, 161, 269 in anti-friction metals, 509 in brass, 157, 506 in fire assay, 575 in ores, 269 in Paris green, 273 iodide method, 267 Corallin, 213 Corrosives in water, 392 Cotton seed oil, 383 Cracking of coal products, 300 Cream, 557 Critical points of steel, 485, 487 Crucible charge for fire assay, 571 Gooch, 24, 86 process for fire assay, 567 Crucibles, platinum, 38 porcelain, 37 Cupellation, 576, 579 Current density, 145 Cyanides, 278 Decimal system. 202 Decomposition in a crucible, 37 Decomposition voltage, 140 Density, current, 145 of water, 181 Desiccators, 30 Direct combustion of carbon, 436 Distilled water, 72 Double layer, 142 Drying agents, 33, 132 of oils, 363 precipitates, 29 E Edible fats and oils, 354 Elaidin test, 386 Electro-analysis, 138 apparatus for, 148 <, circuit for, 150 electrodes, 146, 164, 166 electrolytes for, 138 electromagnetic stirring, 165 laboratory, 150 records, 155 solvents, 140 temperature of electrolyte, 140 Electrodes for electro-analysis, 146, 164, 166 mercury cathode, 166 moving, 164 Electrolytes for electro-analysis, 139 Electromagnetic stirring, 165 Elliott tester, 349 End point, 192 with permanganate, 245 Enlargement of particles, 20 Equilibrium, in reactions, 16 Equivalent weight, 195 Erythrosine, 211, 214 Etching metals, 501 Ether method for iron separations, 472 Ethylene in gas mixtures, 335, 341 Ethyl orange, 211, 214 Evaporation of steel solutions, 434 Evolution method for sulphur, 450 Explosion of gas mixtures, 336, 342 Expression of results in water analysis, 393 INDEX 597 Factors, 6, 9 Factor weights, 6 Fatigue of steel, 498 Fat in butter, 559 in condensed milk, 558 in cream, 557 in milk, 547 Fats, melting point, 362 Felspars, 290 Ferrite, 482, 485 Ferrous ammonium sulphate as primary standard, 247 Fertilizers, 510 moisture in, 511 nitrates in, 513, 518 nitrogen in, 511 phosphorus in, 523 potassium in, 533 Filtering by Munro's method, 25 Filter paper, 22 Filters, inorganic, 24 Filtration, 22 Fire assay, 564 crucible process, 567 cupellation, 576, 579 fluxes, 567 inquartation, 576, 580 oxidizing agents, 570 parting, 577, 580 reducing agents, 569 scorification, 580 Fire test for oils, 348, 350 Fish oils, 385 Fixed carbon in coal, 299, 306 Flash point of oils, 346, 349 Flasks, volumetric, 169, 175 calibration, 187, 188 Fleming tube, 438 Fluxes, 49, 567 acid, 50, 568 basic, 50, 569 for fire assay, 567 Foaming of water, 393 Ford method for manganese, 463 Ford-Williams method for mang- anese, 463 Formaldehyde in milk, 556 Fractional distillation of oils, 350 Fuel gases, 333 Fuel value, 314, 318, 324 from proximate analysis, 316 from ultimate analysis, 315 Fuels, 297 Fusing point of ash, 302 Fusion, 48 G Gas burette, 328 illuminating, 333, 340 mixtures, 328 carbon dioxide in, 333, 341 ethylene in, 335, 341 hydrocarbon vapors in, 336, 341 hydrogen in, 336, 341, 342 methane in, 339, 341, 342 nitrogen in, 334, 341 oxygen in, 334, 341 pipette, 330, 335 explosion, 336, 342 Gases, chimney, 343 fuel, 333 solubility in reagents, 332 Gauss method for weighing, 65 Gay-Lussac method for silver, 276 General operations, 9 principles, 1 Gladding method for potassium, 100 Glass, solubility, 71 Glassware, chemical, 71 Glycerine, compound with boric acid, 237 Glyoxime method for nickel, 470 Gold for crucibles, 42 ores, 564 Gooch crucible, 24, 86 Graded oxidation, 118 Granulation of steel, 493, 499, 504 Graphitic carbon in steel, 445 Gravimetric analysis, 5 standardization of acids, 222 Gunning method for nitrogen, 519 598 INDEX Halogen compounds, organic, 125 oxy acids, 125 Halogens, Volhard method for, 278 free, 125 indirect method, 115 Hardened oils, 389 Hardening of steel, 488, 504 Hardness of water, 229 Clark method for, 230 non-carbonate, 232 permanent, 229 temporary, 229 Heat units, 314, 327 Hehner value for oils, 370, 372 High-speed steels, 467 Hydriodic acid, reduction of arsenic, 261 Hydrocarbon vapors in gas mixtures, 336, 341 Hydrochloric acid, absorbent for, 131 standard solution, 224 Hydrogenation of oils, 389 Hydrogen in coal, 311 in gas mixtures, 336, 341, 342 peroxide, with hypobromites, 123 sulphide, in water, 405 reduction of iron, 249 Hydrolysis of iron salts, 246 Hydrosols, 19 Hypobromites, with hydrogen per- oxide, 123 Hypothetical, compounds in water, 394 Ignition of precipitates, 33 Illuminating gas, 333, 340 Immersion refractometer, 555 Incrustants in water, 392 Index of refraction, 358 Indicators, 193, 207 chromophors in, 208 Indicators, classification, 209 cochineal, 211, 214 color change, 207 corallin, 213 description of, 212 erythrosine, 211, 214 ethyl orange, 211, 214 iodeosine, 214 lacmoid, 211, 214 litmus, 211, 213 methyl orange, 208, 209, 211, 214 red, 211, 215 phenolphthalein, 207, 208, 210, 211, 213 p-nitrophenol, 211, 213 potassium ferricyanide, 257 rosolic acid, 211, 213 starch, 261, 263 Indirect method, 115 Industrial analysis of water, 399 products, analysis, 283 Ingot iron, 482 Ingotism, 495 Inquartation, 576, 580 Insoluble acids, 370, 372 matter in minerals, 287 Interpretation of results in water analysis, 408 Inorganic filters, 24 Insecticides, 272 Iodeosine, 214 Iodide method for copper, 267 Iodine, 115 absorption number, 363 oxidation of arsenic, 260 standard solution, 260, 261, 271 Iron, 161, 163, 167 allotropism, 481 in minerals, 288, 294 in ores, 244, 250, 259 in water, 405 primary standard, 246 reduction of, 248 salts, hydrolysis, 246 solutions, color of, 245, 248 INDEX 599 K Kaolin, 290 Kerosene, 348 Kjeldahl method for nitrogen, 513 Kottstorfer number, 368, 371 Laboratory for electro-analysis, 150 Lacmoid, 211, 214 Lactose in condensed milk, 558 in milk, 542 Lead, 162 in anti-friction metals, 509 in brass, 506 Limestone, alkalinity, 239 Limit of accuracy, 2 Lindo-Gladding method for po- tassium, 102, 534 Lindo method for potassium, 99, 102 Linolenic acid, 364 Linolic acid, 364 Litmus, 211, 213 Logarithms, 7 table of, 586, 587 Low method for copper, 267 Lubricating oils, 350 M Magnesium, 106 ammonium phosphate, 107, 108, 111 decomposition, 108, 111 in minerals, 289, 294 salts, solubility, 106 Manganese, 114 by potassium permanganate, 253 in minerals, 288, 294 in steel, 460 Volhard method, 277 Manganous acid, 254 sulphate in iron titration, 244 Marine animal oils, 385 Martensite, 487 Mass law, 15 Maumene number, 381 Maximum size of particles, 12 Meke* burner, 47 Melting point of fats, 362 Mercury cathode, 166 Metallography, 479, 500 Methane in gas mixtures, 339, 341, 342 Methyl orange, 208, 209, 211, 214 red, 211, 215 Meyer bulbs, 439 Milk, 536 added water, 555 adulteration, 536 albumin in, 541 ash, 540 casein in, 541 composition, 537 condensed, 557 fat in, 547 formaldehyde in, 556 lactose in, 542 nitrogen in, 540 specific gravity, 539 total solids, 539 Mixing and dividing, 10 Mohr's salt as primary standard, 247 Moisture in butter, 559 in coal, 299, 305 in fertilizers, 511 in silicates, 293 Molybdate method for phosphorus, 453 Moments, principle of, 54 Morse and Blalock bulbs, 185 Moving electrodes, 164 Munroe's method for filtering, 25 N Nesslerization, 418 Nessler's reagent, 417, 419 Nickel, 163 alloys with chromium, 44 in steel, 468 Nitrates in fertilizers, 513, 518 in water, 425 600 INDEX Nitric acid from coal combustion, 321 Nitrites in water, 424 Nitrogen, availability of, 521 Gunning method, 519 in condensed milk, 558 in fertilizers, 511 in gas mixtures, 341, 342 in milk, 540 in water, 415 Kjeldahl method, 513 Non-carbonate hardness of water, 232 Normal solution, 199, 201 Occlusion by barium sulphate, 93 Odor of water, 413 Oils, acetyl value, 376 acidity of, 352 acid value, 368 burning, 345 point, 348, 350 chill test, 353 cold test, 354 color reactions, 386 constants, table, 388 "drying" of, 363 edible, 345 elaidin test, 386 fats and waxes, 345 fire test, 348, 350 fish, 385 flash point, 346, 349 fractional distillation, 350 hardened, 389 Hehner value, 370, 372 hydrogenation, 389 index of refraction, 358 insoluble acids in, 370, 372 iodine absorption number, 363 lubricating, 350 marine animal, 385 Maumene number, 381 mineral from saponifiable, sepa- ration, 352 Oils, Polenske value, 375 qualitative reactions, 382 Reichert-Meissl number, 373 Reichert number, 373 saponification number, 368, 371 soluble acids, 370, 372, 377 specific gravity, 345, 352, 356 temperature reaction, 381 titer test, 363 viscosity, 350 volatile acids, 377 Oleic acid, 363 Olein, 354, 369 Optical methods for lactose, 542 Ores, copper, 269 gold, 564 iron, 250, 259 silver, 564 zinc, 280 Organic nitrogen in water, 423 Orthoclase, 290 Outflow time for pipettes and burettes, 176 Ovens, 29 Overheating of steel, 494 Oxalic acid as primary standard, 248 Oxidation, graded, 118 in a crucible, 36 potential, 119 reactions of, 240 selective, 119 Oxidizing agents for fire assay, 570 Oxy acids of halogens, 125 Oxygen in coal, 309 in gas mixtures, 334, 341 in steel, 476 Palau, 41 Palladium tube, 338, 342 Palmitin, 354, 369 Paper coil method for fat, 549 filter, 22 reduction of precipitates, 22, 34, 85 Paris green, arsenic in, 273, 274 copper in, 273 INDEX 601 Parting, 577, 580 Peanut oil, 383 Pearlite, 484, 485 Pemberton method for phosphorus, 454 Percent by direct reading, 198 Perchlorate method for potassium, 104 Permanent hardness, 229 Permanganate, end point, 245 standard solution, 243, 250 Peroxides, available oxygen. in, 255, 264 Persulphate method for manganese, 465 Phenolphthalein, 207, 208, 210, 211, 213 Phenolsulphonic acid method for nitrates, 427 ^ Phosphate, magnesium ammonium, 107, 108, 111 Phosphorus in fertilizers, 523 in steel, 4'52 pentoxide, 33, 132 Physical examination of water, 410 tests of steel, 505 Pipette, gas, 330, 335, 336, 342 Pipettes, 169, 175 calibration, 189 outflow time, 176 Platinum, 102 basicity after heating, 40 blistering, 40 care of, 40 crucibles, 38 deterioration, 39 melting point, 38 recovery from waste, 103 substitutes, 41 p-nitrophenol, 211, 213 Polarimeter, 542 Polenske value, 375 Polishing machines, 500 Porcelain crucibles, 37 for chemical uses, 37 Potability of water, 407 Potassium, 96 Potassium, chlorplatinate, 98 dichromate, 242 standard solution, 257 ferricyanide, indicator, 257 in fertilizers, 533 in silicates, 291, 294 Lindo-Gladding method, 102, 534 perchlorate, 104 permanganate, 240 primary standards for, 246 reduction by chlorides, 244 standard solution, 243, 250 thiocyanate, standard solu- tion, 277 Potential differences, 143 oxidation, 119 Precipitates, 33 drying of, 29 enlargement of particles, 20 washing, 26 Precipitation, 15 in volumetric analysis, 276 Preliminary assay, 572 Preparation of samples, 9 Pressure reduction for filtration, 23 Primary standards, 217 for permanganate, 246 for standard acids, 217 for standard bases, 235 oxalic acid, 248 sodium carbonate, 222, 224 sodium oxalate, 247 Proximate analysis, 285, 297, 299, 305 fuel value from, 316 Pulp balance, 565 Pyrex glass, 72 Pyrolusite, 255, 265 Pyrosulphates, formation, 101 Qualitative reactions, oils, 382 Quartering, 11 Quenching of steel, 482, 485, 490, 504 602 INDEX R Radiation corrections, calorimeter, 322, 326 Rapid methods, 432 Reactions, velocity, 16 Reagents, 69 Records, 73, 155 Reducing agents for fire assay, 569 Reduction method for nitrates, 428 methods for lactose, 545 of iron, 248 of precipitates by paper, 22, 34, 85 reactions of, 240 Refractometer, 359 immersion, 555 Reichert-Meissl number, 373 Reichert number, 373 Relative length of balance arms, 61, 65 Renard test for arachis oil, 383 Required oxygen, 429 Resin oil, 383 Rests, balance, use of, 63 Reversible colloids, 19 Rhotanium, 41 Ricinoleic acid, 363 Rider, chain, 57 Riffle, 14 Rock analysis, 284 Rose-Gottlieb method for fat, 551 Rosolic acid, 211, 213 Rule of averages, 4 S Salt in butter, 560 Sampler, riffle, 14 Samples, preparation, 9 Sampling coal, 298, 303 gold and silver ores, 564 quartering, 11 size of particles, 12 steel, 434 water, 409 Sanitary examination of water, 404 Saponifiable oils, separation from mineral oils, 352 Saponification number, 368, 371 Schellbach burette, 171, 176 Scorification, 580 Seger cones, 303 Segregation of carbon in steel, 495 Selective oxidation, 119 Sensibility of balance, 53, 60 Sesame oil, 385 Silver in alloys, 276 in ores, 564 Silica in carbonate minerals, 287 in silicate minerals, 293 in steel, 446 ware, 42 Silicate minerals, 289, 292* Silicates, alkali metals in, 291, 294 moisture in, 293 potassium in, 291, 294 sodium in, 291, 294 Silicic acids, 290 Silver, 83, 159, 161, 167 Gay-Lussac method, 276 ores, 564 salts, solubility, 83 Volhard method, 277 volumetric determination, 276 Slag in steel, 500 Slide rules, chemists', 7 Snelling method for filtering, 25 Soda ash, 225 lime, 132 Sodium, 96 bicarbonate, 229 carbonate, primary standard, 222, 224 tiration, 226, 228, 229 in silicates, 291, 294 oxalate, primary standard, 247 thiosulphate, standard solution, 260, 262, 263, 268 Solid solutions, 486 Sols, 19 Solubility of aluminium hydroxide, 88 of barium salts, 91, 92 INDEX 603 Solubility of calcium salts, 78 of gases in reagents, 332 of glass, 71 of magnesium salts, 106 of silver salts, 83 product, 16 Soluble acids, 370, 372, 377 Solution, 15 of steel, 434 tension, 141 Solutions, solid, 486 standardization, 216 Solvents for electroanalysis, 140 Sorbite, 490 Specifications for apparatus, 174 Specific gravity, Baume scale, 345 of burning oils, 345 of edible oils and fats, 356 of lubricating oils, 352 of milk, 539 temperature reaction, 381 Specified weight, to obtain, 64 Spermaceti, 369 Standard acids, 221 bases, 233 absorption of carbon dioxide by, 234 bulbs, calibration of, 186 methods for steel analysis, 433 samples, 435 solutions, adjustment, 205 decimal system, 202 normal system, 199 standardization, 216 temperature corrections, 204 temperature for calibration, 182 Standardization by direct weighing, 223 of solutions, 216 Standards, primary, 217 use of two, 238 Stannous chloride, reduction of iron, 249 Starch indicator, 261, 263 Stearin, 354, 369 Steel, 432 annealing, 488, 504 Steel, carbon in, 435 case hardening, 497, 504 chromium in, 468 combined carbon in, 445 critical points, 485, 487 effect of work, 498 etching, 501 evaporation of solutions, 434 fatigue of, 498 granulation, 493, 499, 504 graphitic carbon in, 445 hardening, 488, 504 high-speed, 467 . manganese in, 460 metallography, 479, 500 overheating, 494 oxygen in, 476 physical tests, 505 phosphorus in, 452 quenching, 482, 485, 490, 504 rapid methods for analysis, 432 sampling, 434 segregation of carbon in, 495 silicon in, 446 slag in, 500 solution of, 434 standard methods for analysis, 433 streaks in, 495 sulphur in, 448 prints, 496 tempering, 491, 504 thermal changes, 480, 485, 487 titanium in, 457, 458 treatment, 478 tungsten in, 467 vanadium in, 475 Streaks in steel, 495 Strontium, 96 Substitutes for platinum, 41 Sucrose in condensed milk, 558 Sulphates, 95 Sulphur in coal, 307, 310 in steel, 448 prints, 496 Sulphuric acid from coal combus- tion, 321 604 INDEX Sulphurous acid, reduction of iron, 249 Supersaturation, 20 Suspended matter in water, 415 Systems of volumetric analysis, 198, 204 Teclu burner, 46 Temperature correction for standard solutions, 204 for electrolysis, 140 Tempering of steel, 491, 504 Temporary hardness, 229 Tension, solution, 141 Thermal changes in steel, 480, 485, 487 Thiocyanate, standard solution, 277 Time-temperature curves, 323 Tin in anti-friction metals, 508 in brass, 505 Titanium in steel, 457, 458 Titer test for oils, 363 Tolerance, 173, 178 Total solids in condensed milk, 558 in milk, 539 in water, 415 Transfer of liquids, 73 Treatment of steel, 478 of water, 401 Trial of weights, 63 Triangles, 42 Troostite, 489 Tungsten in steel, 467 Turbidity of water, 410 U Ultimate analysis, 285, 297, 307, 310 fuel value from, 315 Units of volume, 172 Use of logarithms, 7 Valence, apparent, 241 Vanadium in steel, 475 Velocity of reactions, 16 Vertical action of beam, 52 Villavecchia test for sesame oil, 385 Vinegar, 236 Viscosity of oils, 350 Volatile acids, 377 combustible matter, 299, 306 Volhard method for halogens, 278 for manganese, 253 for silver, 277 Voltage, decomposition, 140 Volumetric analysis, 168 apparatus, absolute and rela- tive capacity, 172 calibration, 179 specifications, 174 tolerance, 173, 178 flasks, 169, 175 calibration, 187, 188 Volume, units of, 172 W Wash bottles, 28 Washing precipitates, 26 Water, added, in milk, 555 alkalinity, 229 ammonia in, 417, 422 analysis, 391 carbonic acid in, 404 chlorine in, 415 color of, 413 corrosives in, 392 density, 181 distilled, 72 expression of results of analysis, 393 foaming of, 393 hydrogen sulphide in, 405 hypothetical compounds in, 394 incrustants in, 392 interpretation of results of analysis, 408 iron in, 405 nitrates in, 425 nitrites in, 424 nitrogen in, 415 INDEX 605 Water, non-carbonate hardness, 232 odor, 413 organic nitrogen in, 423 permanent hardness, 229 physical examination, 410 potability, 407 required oxygen in, 429 sampling, 409 sanitary examination, 404 suspended matter in, 415 temporary hardness, 229 total solids of, 415 treatment, 401 turbidity, 410 Waxes, 355 Weighing, 50, 61 Borda method, 65 Weighing, Gauss method, 65 gold and silver ores, 565 Weight, equivalent, 195 Weights, 56 calibration, 66, 67 specified, 64 trial of, 63 Working steel, 498 Zero point, 59 estimation, 63 Zimmermann-Reinhardt solution, 246 Zinc, ferrocyanide method, 279 in brass, 507 in ores, 280 THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW RENEWED BOOKS ARE SUBJECT TO IMMEDIATE RECALL RET. FEB 2 4 1956 REf/0 LIBRARY, UNIVERSITY OF CALIFORNIA, DAVIS Book Slip-70m-9,'ti. : J(F7151.s4)4.')S UNIVERSITY OF CALIFORNIA LIBRARY