QUANTITATIVE CHEMICAL ANALYSIS. EXERCISES QUANTITATIVE CHEMICAL ANALYSIS: - WITH A SHORT TREATISE ON GAS ANALYSIS. BY W. DITTMAR, LL.D. (Edin.), F.R.S.; F.R.S.E. if PROFESSOR OF CHEMISTRY IX THE GLASGOW AND WEST OF SCOTLAND TECHNICAL COLLEGE. UHITTEESITr LONDON: \VILLIAMS & NORGATE, 14 HENRIETTA STREET, COVENT GARDEN; AND 20 SOUTH FREDERICK STREET, EDINBURGH. GLASGOW: WILLIAM HODGE & CO., 123 HOPE STREET. 1887. [All Rights Reserved.] GLASGOW : PRINTED BY WM. HODGE & Co., 123 HOPE STREET. PREFACE. A PRELIMINARY edition of this book, as many of my friends are aware, was issued a little over a year ago for the benefit of chiefly my own students. That edition, however, was little more than a reproduction of what had long had currency in my laboratory as a trypographed book. Hence this volume may be introduced as having already been used and I hope not without success in a largely-attended teaching laboratory for a series of years. However it may stand with analysis generally, quantitative analysis can be taught only by examples ; and in the earlier stages of the course the technicalities of the subject are the principal things to be taught. Hence our exercises on "analy- tical methods" (section 2) are arranged, not according to any scientific system, but so that, at any given point of his progress, the student has become familiar with as many different opera- tions as could have been learned during the time. To some of my readers the tone of the earlier exercises more especially may savour a little of mechanical drilling. Why tell the student so minutely what he has to do and hinder him from exercising his own ingenuity ? Some ten years ago, if a book like the present had been placed before me, I should have asked this question myself. But I have since come to modify my views. The technicalities of quantitative analysis are the very things which the student is not likely to find out by himself. He had better be drilled into doing them correctly. What is the good, for instance, of letting him spoil a series of ammonia determinations by mismanaging his chloro- platinate precipitates ? It surely is better to show him quite directly what he has to do ; and if it is, why should not the book tell him, and thus save the time of the teacher? No fear of any talented student being spoiled by a course of judi- cious drilling. It is just he that must be made alive to the fact that no amount of scientific knowledge will enable a man to get through a quantitative analysis successfully unless he has the "canning" as well as the knowing, and unless he attends to all those little practical details which to him at first A vi PREFACE. sight may appear to be irrelevant. Cheerful and conscientious devotion to all the protracted drudgery that may be involved in one's duty is certainly a lesson worth learning, and it is one of the educational functions of quantitative analysis to inculcate the lesson. And as to the talented student's weaker brother? Why, he must be drilled, or else he may learn nothing at all. Of course, here, as everywhere, we must beware of extremes, and take care not to disgust the student with his work. It is as well, even at the earlier stages, to occasionally break the monotony of analytical work by the interpolation of an exercise in preparative chemistry. With students who have already been " broken in," a capital plan is to give them unnamed substances and let them find the exercise or set of exercises that they are meant to work, taking care not to give them any help except where it may become necessary to prevent sheer waste of time. The section on Gas Analysis will, I hope, be welcome to many, because there is no treatise on the subject in any language, as far as I know, excepting, of course, Bunsen's Gasometrische Methoden; but it treats only of the great master's own methods. In conclusion, I have much pleasure in acknowledging my indebtedness to my assistants, Messrs. John M'Arthur and Archibald Kling, for the valuable aid they have given me in preparing the book for and passing it through the press ; to the same gentlemen and Messrs. Frank Lyall, James Robson, William Cullen, and Andrew Hodge, for the careful execution of test-analyses ; and last, not least, to my publisher, for having spared no expense in bringing the book into a pre- sentable form. W. D. CHEMICAL LABORATORY, ANDERSON'S COLLEGE BUILDINGS, GLASGOW, September, 1887. CONTENTS. PAGE Table of Atomic Weights, - / xii EXERCISES IN EXACT WEIGHING AND MEASURING. 1. Practice in the Routine Method of Relative Weighing, - 1 2. The Method of Vibration, 3 3. Absolute Weighing, - 5 4. Reduction to the Vacuum, 8 5. Weighing of Pre-determined Quantities of Liquids, 10 6. Making of Apparatus for the Exact Measurement of Liquids, - 10 7. Graduation of a Measuring Flask, - - 1 1 8. Graduation of a Pipette for "Delivering " Pre-determined Volumes, 1 2 9. Determination of Specific Gravities of Liquids, -. 12 10. Preparation and Adjusting of Standard Solutions, - 13 EXERCISES IN ANALYTICAL METHODS. 11. Determination of Filter Ashes, - - 16 12. Analysis of Chloride of Barium, - - 17 Rules regarding the Booking and Reporting of Analyses, - 19 13. Analysis of a Silver Coin, 20 14. Sulphate of Iron and Potash, - 22 15. Analysis of Sulphate of Magnesia and Potash, - 24 Tatlock's Method, - - 25 Finkener's Process, - 26 16. Analysis of Phosphate of Lime, - 30 17. Separation of Iron and Alumina, - - 33 18. Titrimetric Determination of Iron, - 34 The Bichrome Method, - - 37 The Stannous Chloride Method, 39 viii CONTENTS. PAGE 19. Analysis of Sulphate of Copper and Ammonia, - 42 Supplementary Methods, - 46 20. Analysis of Carbonates, - 47 21. Separation of Iron, Manganese, and Calcium, 52 22. Spathic Iron Ore,- 55 23. Preparation of a Standard Solution of Iodine, 56 24. Analysis of Bleaching Powder, - 60 25. Gas- Volumetric Analysis, - 64 26. Native Oxide of Manganese, - 67 27. De Haen's Method for the Determination of Copper in Alloys, Ores, &c., 75 28. Determination of Water by the " Direct " Method, - 76 29. Determination of Nitrogen by Varrentrapp and Will's Method, 79 Notes regarding Special Classes of Substances, 30. Kjeldahl's Method of Nitrogen Determination, 83 31. Schloesing's Method of Nitric and Nitrous Acid Determination, 85 31a. Walter Crum's Method, 89 32. Quantitative Electrolysis, 91 32a. Electrolytic Separation of Copper and Nickel, 100 33. Analysis of Brass, 101 34. Analysis of German Silver, - 104 Separation of Nickel and Cobalt, - 106 35. Analysis of Iron or Copper Pyrites, 111 Assay for Sulphur, 111 Determination of the Metals, - 114 Assaying for Copper, 116 36. Analysis of Galena, 119 37 Silicate Analysis, - 122 38. Analysis of Chrome Iron Ore, 128 Assay for Chromic Oxide, 128 Complete Analysis, 130 39. Separation of Lead and Antimony, 135 By Means of Alkaline Sulphide, - - 135 By Means of Chlorine, - - 136 40. Analysis of Gold and Silver Alloys by Cupellation, 140 CONTENTS. ix PAOE ELEMENTARY ANALYSIS OF COMBUSTIBLE CARBON COMPOUNDS. 41. Analysis of Ferrous Oxalate,- 143 42. Analysis of Cane Sugar or Mannite, 146 43. Analysis of Benzoic Acid or Naphthalene, - - 147 44. The Old Method of Organic Analysis, 147 45. Analysis of Liquids, - - 1 49 46. Analysis of Nitrogenous Substances, - 151 47. Determination of Nitrogen in Organic Substances by Dumas' Method, 152 GAS ANALYSIS. Generalities and Theory, - - 155 Proximate Analysis of Gas Mixtures, - - 166 The Law of Gas Absorption and its Applications, - - 168 Ultimate Analysis of Gases generally, - - 173 Apparatus and Working Methods, - - 179 EXERCISES 1. Making of a Bunsen Eudiometer, - - 208 2. Calibration of the Eudiometer, - 211 3. Preparation of a Reduced Volume Scale for the Eudio- meter, - - 212 4. Making and Calibration of a Bunsen Absorption Tube, 212 5. Calibration of the Measurer in the Author's Apparatus, 212 6. Analysis of a Mixture of Air and Carbonic Acid, - 212 7. Analysis of a Mixture of Carbonic Oxide and Carbonic Acid, - 213 8. Analysis of Marsh Gas, - 213 9. Analysis of Ethylene, 213 10. Analysis of a Mixture o/H 2 , CO, CH 4 , and N 2 , - 215 11. Extraction of the Absorbed Gases from a Water, and their Determination, - 215 PROMISCUOUS EXERCISES IN APPLIED ANALYSIS. 1. Analysis of a Sea Water, - 219 Chlorine, 219 Sulphuric Acid, Note (8), 313 Lime and, Magnesia, - 224 CONTENTS. PAGE The Potash, - -225 Total (Bases as) Sulphates, - - 226 The Carbonic Add, - - 227 Alkalinity, - - - 230 Total Salts (Tornoe's Method), - - 231 The Bromine, - . - - 232 Average Composition of Ocean- Water Salts, 234 2. Stassfurth Potash Salts, . - - ; - 235 Tatlock's Method, - - 236 Our own Form of Finkener's Method, - '-236 3. Cast-iron, - . ."' 238 The Total Carbon, ..... 238 The Sulphur, - - . .' - - - 241 The Phosphorus, - ; - ' -.,. % .- - 242 The Manganese, - .- - - 244 The Silicon,- - - - - - 246 The Iron and Foreign Metals, '- - 246 Determination of the Silica, &c., present as Slag, - - 247 Steel and Wrought-iron, , .' . . * . ... 247 4. Superphosphate of Lime, - ... _ 248 5. Guano, - . . . - - - 252 6. Milk, - . 255 Albumenoids and Sugar, - 256 The Sugar, - '. 257 7. Soap, - . . 259 Fatty Acid, - . 260 The Alkali Salts generally, 261 The Glycerine, - . . 262 Resin, - - 263 8. Butter, _ _ 265 Foreign Fat, . ! . 265 Rancidity, - - - - 267 9. Assaying of Tanning Materials, . 267 Assaying of Sumach, - - - - -'.. 267 10. Partial Analysis of Tea, - - - - 271 11. Analysis of Wood Spirit, - . 272 Determination of the Methyl-Alcohol, - - - - 276 CONTENTS. XI PAGE Determination of the Acetone, - 279 Dittmar and Fawsitt's Table of the Specific Gravities of Aqueous Methyl-Alcohols, - 281 12. Determination of Ethyl- Alcohol, 284 Table giving the relation in Aqueous Ethyl-Alcohol between Specific Gravity and Percentage, - 286 Separation of Eihyl and Methyl- Alcohol, - 292 NOTES. (1) Theory of the Balance, - 302 (2) Specific Gravities of Aqueous Hydrochloric Acids. The Author's Differential Method of Specific Gravity Deter- mination, - - 305 (3) Double Salts (Mg or Fe)S0 4 .K 2 S0 4 + 6H 2 0, - 308 (4) To Ex. 24, - 309 (5) To Ex. 26, - 309 (6) Determination of Cobalt as Phosphate, - 309 (7) On Platinum Solution and Platinum Residues, 310 (8) Determination of the Sulphuric Acid in Sea Water, 313 (9) Greiner and Friedrich's new Stopcocks, - 313 INDEX, - 315 ATOMIC WEIGHTS. = 16 After Lothar Meyer and Seubert's calculations. Exceptions : according to authorities quoted in the notes. Name of Element. Atomic Weight. Name of Element. Atomic Weight. Symbol. Value. Symbol. Value. Aluminium, - Al 27*10 Molybdenum Mo 96*2 Antimony, - Sb 119-9 Nickel, - - Ni 587 Arsenic, - - As 75'9 Niobium, Nb 93'9 Barium, - - Ba 137-20 Nitrogen, N 14-046 Beryllium, - Be 9-1 Osmium, - - Os i95' Bismuth, 1 - Bi 208*0 Oxygen, - - 16- Boron, - - B io'9 Palladium, - Pd io6"6 Bromine, Br 79'95 2 Phosphorus, P 3 I- 4 Cadmium, - Cd II2'0 Platinum, 6 - Pt !95'5 Caesium, - - Cs I 33' Q Potassium, - K 39'!3 6 Calcium, - Ca 40*02 Rhodium, Rh 104-3 Carbon, - - C I 2 '00 Rubidium, - Rb - 85-4 Cerium, - - Ce I41'5 Ruthenium, - Ru 103-8 Chlorine, Cl 35'454 Scandium, - Sc 44-1 Chromium, 3 - Cr 52-13 Selenium, - Se 79-07 Cobalt, Co 587 Silicon, 7 - Si 28-399 Copper, - - Cu 6 3'34 Silver, Ag 107-93 Didymium, - Di i45'4 Sodium, - Na 23'053 Erbium, - E 166*4 Strontium, - Sr 87'5 2 Fluorine, F 19-1 Sulphur, - S 32-06 Gold, 4 An 197-324 Tantalum, - Ta 182-7 Hydrogen, - H i '0024 Tellurium, - Te 128-0 Gallium, - Ga 70-1 Thallium, Tl 204*2 Indium, - In H37 Thorium, Th 232-5 Iridium, - Ir 193-0 Tin, Sn 117-6 Iodine, - - I 126-85 Titanium, 8 - Ti 48-08 Iron, - - - Fe 56*02 Tungsten, - W 184*0 Lanthanum, La 138-9 Uranium, U 240-5 Lead,- - - Pb 206*9 Vanadium, - V 5 1 ' 2 Lithium, Li 7-02 Yttrium, - Y 89-8 Magnesium, 2 Mg 2 4'37 Ytterbium, - Yb J 73" Manganese, 5 - Mn 55' Zinc, 9 - Zn 6 5'37 Mercury, - - Hg 200-3 Zirconium, - Zr 90*6 2 j Marignac; Fres. Zeitschrift, 1884; 118. 3 Siewert, as calculated by Clarke. 4 Thorpe and Laurie ; Chem. Soc. Trans., 1887, p. 565. 6 Dewar and Scott ; Marignac. 6 Dittmar and M' Arthur ; Memoir communicated to the Roy. Soc. of Ediii. in summer, 1887. 7 Thorpe and Young; Chem. Soc. Trans., 1887, p. 576. 8 Thorpe; Roy. Soc. Proc., 1883, p. 43. 9 Mean of determinations by Marignac and by Baubigny (1884). [UHI7BESITY] EXERCISES IN EXACT WEIGHING AND MEASURING* '. Ex. 1. Practice in the Routine Method of Relative Weighing. As a necessary preliminary to the actual practice of quantitative analysis, begin by learning to execute exact weighings by means of a precision balance. But a mere knowledge of the method is - not sufficient : you must practise the art for hours and hours until you have it at your fingers' ends. To enable you to do this, and to check the results, we keep a set of pieces of metal marked A, B, C, &c., of which the exact weights in grammes are noted down in a book. Procure this set, and then, in order to determine the several weights, First, set your balance in order; i.e., clean the case outside and inside, and remove every particle of dust from the instrument itself by means of a camel's-hair brush. Second, " level " the instrument by working the screws at the bottom of the case until the spirit-levels inside stand at zero, which shows that the (plane) central support of the beam is exactly horizontal, and the zero of the scale lies vertically below the axis of rotation. Third, see if your balance is in equilibrium. For this purpose shut the case, and, by letting down the arrestment, enable the beam to seek its position of rest. As a rule, it will vibrate about this position. Should it stand still, cause it to oscillate by sending a current of air down on one pan (using your hand as a fan), or else by placing the rider on the beam for an instant and then removing it. Follow the excursions of the needle, and note the * See Note (1) at end of volume. 2 EXERCISES IN EXACT WEIGHING AND MEASURING. successive turning points. If in two consecutive excursions the needle goes as far to the one side of the zero as to the other, the position of rest is at zero; it agrees with that position in which the centre of gravity of the empty beam (by intention at least) is vertically below the axis of rotation ; the balance is in equilibrium/ -Should it .turn put that it is not, shift the little knob at the* one end"-t)f: tfie ^beam along the horizontal screw to which it is j attached,: ^f onwards \ and backwards, until perfect equilibrium is established'.' In order now to determine the relative weight of an object, place it on the left pan, and then try greater or lesser combina- tions of standards ("weights") on the right until the beam's position of rest lies within its angle of free play ; that is, until the beam, when the arrestment is down, oscillates freely. It stands to reason that these trials must be executed in a syste- matic manner. Supposing we had found that P grammes is greater, while p grammes is less, than the weight sought for, the correct rule is to add to p that one weight which comes nearest to the value J (P p). Take care not to make any alteration in the charge on either pan without having first arrested the beam ; and in letting down the arrestment, do it slowly so that the beam (as long as it is to any considerable extent out of equili- brium) in its descent keeps pace with the arrestment. Supposing now equilibrium to be established to the extent above referred to, shut the case, allow the beam to oscillate, and note two consecutive excursions of the needle (neglecting the first, which is liable to be infected by irregularities), mark with the plus sign ( + ) what lies to the left, and with the minus sign ( - ) what lies to the right, of zero. The algebraic sum* of the two readings, representing as it does the distance from the zero, in half-degrees, of that point which the needle would point to if the beam were in its position of rest, is proportional to the number of milligrammes (A) which must be added to the lighter side to establish equilibrium. This distance we call the "deviation," and are in the habit of * Supposing the needle to move (a) from -4 to + 6 ; (b) from + 0'5 to + 4 '5; (c) from - 1 to -17, the deviation would be respectively (a) + (6 - 4) = 4- 2 ; (b) + (0-5 + 4-5) = + 5 ; (c) - (1 + 17) = - 27 half -degrees, THE METHOD OF VIBRATION. 3 designating it by the symbol a. Determine approximately the relation between a and A by seeing to what extent the devia- tion is diminished by the addition of, say, 2 milligrammes to the lighter side. Supposing a had been = + 5 (half-degrees), and by taking away 2 mgms. been reduced to +1, i.e., by 4, every 2 of deviation means approximately a plus or minus of 1 mgm. Take away 5 -=- 2 = 2'5 mgms., take another couple of readings, correct the remaining error, and so on until equilibrium is established to within as little as the balance is capable of indicating. Tabulate your results, and ask the Demonstrator to check them by comparison with the standard tables. If your results are unsatisfactory, do the whole work over again (with- out referring to your previous notes), and so on until you are able, without any effort, to effect a weighing with rapidity, and yet with unerring certainty. Ex. 2. The Method of Vibration. A REFINEMENT on the ordinary routine method, which consists in this, that after having established equilibrium to within a few milligrammes, we determine the deviation with precision, and from it calculate the exact weight (of A m gs.) which is required to establish absolute equilibrium. To learn the method, deter- mine the relation in your balance between charge P, overweight A, and deviation a, as follows: Bring the empty instrument into approximate equilibrium, and then, in order to determine the exact deviation, let down the arrestment, and take, say, four or six consecutive excursions of the needle (neglecting the first one) so that you have 3, 5, . . . , or a greater odd number of read- ings to calculate from : we leave it to you to find out why an odd number is in practice preferred to an even one. Supposing the needle to turn successively at (*) W (4) (5) (6) + 4-2 -3-0 +4-0 -2-8 + 3'8, which results are conveniently taken down as follows: + 4.2 4-0 3-8 - 3-0 - 2-8, 4 EXERCISES IN EXACT WEIGHING AND MEASURING. we have from (2) and (S) a= + T2 half-degrees. (3) (4) a =+l'0 (4) (5)a=+l-2 (5) (0)a=+rO Mean =+ 1-10 = a . Now, place successively 1, 2, 4 mgms. on one of the pans, again determine the deviation, and in each case calculate the over- weight corresponding to 1 half-degree of deviation. Supposing we found, for the increase which a suffers by 124 mgms., (a -ao) = 3-0 6'2 117 half -degrees, we have = 0'33 O32 O34 mean = '33. a If, for simplicity's sake, we substitute a for a a , , as the example illustrates, is constant ; we will call it c, a and consequently put down that for the empty instrument, c = 1/3 (0-33 + 0-32 + 0-34) = 0-33 mgs. This being done, charge your balance, say, with 20 grms. on each side, establish equilibrium by adding small bits of thin wire, and determine the value which c now possesses. Do the same for charges of 40, 60, 80, 100 grms., and tabulate your results. The several values for c, in a good balance, will not differ much from one another, so that the cs corresponding to intermediate values can be obtained by interpolation. Draw a straight line, and lay down on it a scale of convenient units of length ; at the points 0, 20, 40, 60, 80, 100, erect perpendiculars, and (choosing a convenient unit of length) make the lengths of them equal numerically to the c's corresponding to the charges of 0, 20, 40, &c., grms. Draw a continuous curve so that it passes, as nearly as possible, through the end-points of the perpendiculars, and take the length of the ordinates from the points 0, 10, 20, &c.', of the base line. These lengths represent the A's corresponding to 1, at the charges 0, 10, 20, &c., grms. respectively. How such a table can be utilized for exact weighings need not be explained. In ordinary practice, however, it is best to strike a compromise between the strict method of vibration and the routine method explained in Ex. 1, ABSOLUTE WEIGHING. 5 For the same charge, the value c depends on the distance of the centre of gravity of the empty instrument from the axis of rotation. Every complete balance has an arrangement (gravity bob) for varying that distance. The best method is so to adjust the bob that, for some medium charge, the weight-value of 1 half-degree is, say, exactly 1, J, or i mgm., and to use this mean value of c for the approximate translation of deviations into differences of weight. Ex. 3. Absolute Weighing. IN Chemical Analysis, we have to do only with weight-ratios, not with absolute weights ; hence we need not trouble ourselves about the absolute correctness of our set of gramme weights; and as we always, in weighing, place the substance on the left, and the weight standards on the right, pan, it does not matter if our balance is not exactly equal-armed. But we should not use a set of weights without having tested it for its relative correct- ness, and this involves absolute weighings, though the unit, even here, as a matter of principle, is arbitrary. EXERCISE. Procure a set of only approximately adjusted gramme weights such as are sold in Germany and France for commercial purposes, take out the pieces representing nominally 1, 1, 1, 2, 5, 10 grammes, and in order to determine their relative errors, adopt provisionally one of the grammes as your standard, and with it compare the rest, as follows : Place the standard 1 gramme on the right pan of a fine balance, and, by placing metallic objects (wire, &c.), on the left, establish approximate equilibrium. Allow the balance to vibrate, take down, say, four vibrations, and find the deviation a, as shown in the preceding exercise. Calculate the number of milligrammes A which correspond to the a, and book your results thus (we assume that A = 0'5 a) : In pan, a say, -2-3 1*15 mgs. Now, take away that one gramme (1), and substitute another which we will call (1)^ add the requisite number of milligrammes 6 EXERCISES IN EXACT WEIGHING AND MEASURING. to establish approximate equilibrium, and again book your results. Supposing we find (In pan) (1), + 1 m a = -4-0. A = - 2-0 mgs., obviously perfect equilibrium would have been established, or, in other words, the constant tare would have been exactly counterpoised in the case of (1) by adding to it 1*15 mgs., (1), 3-00 Hence, (IX + 3 mgs. = (1) + 115 mgs., or (IX = (1)- (3 -I'lo) = (l)-l-85 mgs., (1) being = unity, by assumption. In a similar manner, determine the error in the third gramme piece (1)3. Say you find it = (1) + T2 mgs. Now, clearly, any two of the three weights (1), (1) 15 (1) 2 , when taken together, represent a known weight equal approximately to the 2-gramme piece (2), and the difference between, for instance, (1) + (l)j and (2) can be ascertained as you determined that between (1) on the one hand and (l) x or (1) 2 on the other. Suppose you had found the constant tare equivalent to (1) + (l)i + 1'3 mgs., (2) -37 mgs., then, as we know that (l) x = (1) T85 mgs., we have (1) + (1)! + 1-3 mgs. = 2 x (1)-1'85 + 1-3 mgs. = (2) -37 mgs. Hence, (2) = 2 units - T85 + T3 + 37 mgs., or (2) = 2 units + 3*15 mgs. In the same way, determine the values of the (5) and (10) in terms of the assumed unit (1) and of milligrammes, and all that we then need to be able to draw up a correction table for our set of weights is the exact relation between the milligrammes (i.e., the rider) used and our unit (1). But this, if it had to be done by experiments independent of any guaranteed set of weights, would involve more work than the result, in this case, would be worth. Hence, you had better now assume the 10 grammes out ABSOLUTE WEIGHING. 7 of the fine set to be equal exactly to 10 grammes, and equal also to 10,000 times 1 nig. as counted by the rider, and, after having compared that " true " 10-gramme piece with the (10) in the bad set, calculate the values of your several pieces in terms of the new unit. Supposing we found (10) = (10 x "1 gramme ")*+ 67 mgs, and before that (10) = 10 x (l)-9'8 mgs., then (10) x (1) = 10 grammes + 16'5 mgs., and (1) = 1 gramme + T65 mgs., which value must be substituted for every one (1) in the pro- visional table of errors. To make sure of your results, compare your several weights, (1), (IJj, (2), &c., with the respective pieces in the set of precision weights, and compare the thus directly determined errors with the ones entered in your table. In order to learn something new, effect these comparisons as follows : After having brought the balance into equilibrium, f place the object on the left pan, and weigh it with precision grammes placed on the right pan, using the method of vibration for the final adjustment. Then reverse the position of object and weight proper, and again note down the result. Supposing we found that the unknown weight x of the object was balanced in the first case by 10*0126 grammes, second by 10'0132 then, taking I' as being the length of the left arm, and I" as being the length of the right arm of our balance, x X I' = 10-0126 grammes X I" Eq. (1), and x x I" = 10'0132 x I' Eq. (2) : hence x H' I" = 1Q-Q126 x IQ'0132 I' I" Eq. (3) : hence x = ^10-0126 x 10'0132, or practically, as is easily proved, x = J (10-0126 + 10-0132). * i.e., the value of the standard 10-gramme piece or ten " true grammes." f Absolutely exact equilibrium is not needed. Assuming, for instance, the left pan were, in itself, by e grammes too heavy (e being something like, say, O'Ol grms. or less), then we have in lieu of eq. (3), x x (x + e) = (10'0126) x (10 '01 32 + e), and we may well cancel (x x c) against 100126 x c without real error. 8 EXERCISES IN EXACT WEIGHING AND MEASURING. The method illustrated in this exercise is the same as the one which we use for determining the errors in a set of precision weights, except that what we ultimately adopt as our standard is an imaginary quantity, so chosen that the average error of the single piece of the set, when its value is expressed in terms of the standard, is reduced to a minimum. Observe that, even in the best sets of gramme weights, as made by the most eminent artists, the " unit," in passing from one set to another, is, strictly speaking, a variable quantity. What that quantity is, in the absolute sense, it is of no use to the chemist to know ; but obviously, where two or more sets are used to supplement one another, it is necessary to determine the relations of the several units (" grammes ") to one another. Ex. 4. Reduction to the Vacuum. IN chemistry, when we " weigh " a thing, what we want to measure is, not its weight in the strict sense of the word, but its "mass," i.e., the quantity of matter which it contains. Our method of weighing really is a method of mass-measurement, founded upon the proposition that two things which (at a given place and time) are equal in weight are equal also in mass. If a thing weigh fifty times as much as our 1 -gramme piece, its mass also is fifty times the mass of that 1 -gramme piece ; and this is what we mean by saying that it " weighs 50 grammes." But the proposition, in strictness, holds only w T hen the weighings are made in vacuo, because when a body is plunged in air its weight is partly compensated for by the upward pressure of the air, which is equal to the weight of air displaced. Hence, when we have weighed an object, we should always by rights ascertain the difference between the volume V of the object and the volume v of the weights, and (if V > v) add to the weight registered the weight of V - v unit- volumes of air. To give an idea of the magnitude of the correction as applied to ordinary cases, assume a quantity of water has been weighed by means of brass weights, and been found = 10 grms. Taking the volume of REDUCTION TO THE VACUUM. 9 1 grm. of water as our unit of volume, the weights of unit- volume of brass, water, air, are about 8, 1, 1/800 respectively. Hence, 10 grms. of water are 10 volumes by definition, and 10 grms. of brass are 10 -r 8 = T25 volumes ; and Vv = 10 T25 = 875 ; so many volumes of air weigh 8*75 -r 800 = 0'0109 grms. Hence the true weight (" mass ") of the water is 10'0109 grms., and not lO'OOOO. Most of the things which we weigh in the ordinary practice of analysis are nearer in specific gravity to water than they are to brass ; hence the error which we commit when we neglect the correction is by no means insignificant. But its effect on the final result is in general far less than would appear from our example. A single weighing in chemistry is nothing more than one term of a ratio, and as both terms are affected by the error similarly (though not equally), the quotient is nearer the truth than the student may be inclined to think. Sup- posing, for instance, in addition to our mass of water we had weighed the same volume of sulphuric acid in order to determine the specific gravity, and found the apparent weight to be = 12 grms., then the corrected weight of the acid would be 12'0109, and the true specific gravity = 12'0109 -=- 10'0109 = 1-1998, which differs only by 0*0002 from the apparent specific gravity, 12 -r 10 = T2000. In practice, the correction can be neglected, unless we have to weigh a gas or other substance of very small density, and even then we may always neglect the weight of the air displaced by the weights, because obviously we have the right to assume as our unit of weight the weight which our 1 -gram me piece has in air. It would even be a mistake not to do so, because our weights are adjusted by comparison against each other in air, and we would introduce an error by allowing, for instance, for the difference in density between the platinum of the decigrammes and the brass of the grammes and gramme- multiples. Our unit of weight, then, is variable with the density of the air we weigh in, but this is a matter of no moment practically. To calculate the weight of air displaced by the object, we divide its apparent weight by its specific gravity, and thus obtain a sufficient approximation to its volume. Taking the volume of 1 gramme of water (the cubic-centimetre) as unit volume, 1 cc. of air of t and P mms. pressure weighs 10 EXERCISES IN EXACT WEIGHING AND MEASURING. S = 0-4648 x P -r (273 + t) mgs. Hence, supposing a body in air of 15 and 750 mms. weighs apparently 44 grms., and its specific gravity is 4'0, its volume is 44 -f- 4 = 11 cc., and its real weight equals 44 grms. + (11 X 0'4648 x 750) -=- 288 = 44-0133 grms. Ex. 5. Weighing 1 of Pre-determined Quantities of Liquids. To practise this operation, and at the same time to procure a useful piece of apparatus, take a cylindrical test-glass of 4 to 6 ounces capacity ; weigh into it successively 10, 20, . . 100 grms. of water ; mark the several levels, first with pencil on a strip of paper fixed along the glass, and finally with a diamond. In reading the levels, to obtain constant results, we must adopt a definite modus operandi. The one adopted by us is this Place the measure so that its axis is vertical, and the level of the liquid in the same horizontal plane with the observing eye, so that the meniscus appears as a band. Take the lower boundary of this band as defining the " level of the liquid." Ex. 6. Making of Apparatus for the Exact Measurement of Liquids. THE most convenient unit for this purpose is the volume at the ordinary temperature of that quantity of water the apparent weight of which in air is equal to 1 gramme (of our set of weights). We will adopt this unit, and call it " fluidgramme," symbol "fgr." The value of our fluidgramme, it is true, is, strictly speaking, a variable quantity, as it depends on the temperature of the water, and the density of the air in which the weight is taken ; but the effect of these variations rarely goes beyond the unavoidable uncertainty of the measurement. For the sake of precision, however, we will adopt standard conditions, and herewith define the fluidgramme as being " the volume at 15C. of that quantity of water which, at 15C. and GRADUATION OF VOLUMETRIC APPARATUS. 11 760 mms. barometric pressure, weighs apparently 1 gramme." The fluidgrainme, as is seen from our definition, is almost iden- tical with the cubic centimetre, or milli-litre of the French system. What, in commercial volumetric apparatus, figures as " cubic centimetre " or " litre," usually comes nearer to our fluidgrainme and 1000 fluidgrammes respectively than to the standard units nominally adopted. Ex. 7. Graduation of a Measuring" Flask. To graduate a measuring flask, take a long-necked flask of. say, -1 litre's capacity, and after having cleaned and dried it, weigh into it the nearest convenient multiple of 10 grammes of water of 15C., mark the level of the water with the diamond, and reckon the flask as holding " so many fluidgrammes." In- stead of bringing the water to 15C., we may, of course, use the stock of water we have at our disposal as it is, determine its actual temperature, and, by the following table, calculate the weight of water corresponding to the required volume. 1000 grammes of water (weighed in air of t and 760 mms., brass standards, unconnected for displaced air) occupy at this temperature of t the volume of (1000 + x) fluidgrammes = (1000 + y) cubic centimetres. y + 1-89 2-04 2-20 2-37 2-55 274 2-95 3-17 3-39 3-63 Note. In using this table you may always say, without appreciable error, (1000 x) : 1000 = 1000 : (1000 T x). t x y t x + -0-64 + 1-25 + 15 +0-00 4 078 112 16 015 8 0-68 1-21 17 0-30 9 0-63 1-27 18 0-47 10 0-56 1-34 19 0-66 11 0-47 1-43 20 0-85 12 0-38 1-52 21 1-06 13 0-27 1-63 22 1-28 14 0-14 176 23 1-51 15 o-oo 1-89 24 175 12 EXERCISES IN EXACT WEIGHING AND MEASURING. Ex 8. Graduation of a Pipette for "Delivering" Pre-determined Volumes. EXAMPLE chosen: A pipette graduated to deliver |, 1, 1|, 2 ... up to 5 fgrs. To make such an instrument, select a glass tube as nearly as possible cylindrical, and make it into a pipette. Paste along it a narrow strip of paper, then fill it by suction with water to near the top and let the water run out again, taking care not to remove more of the last drop than goes out by itself when the end of the pipette is held against the wet side of the vessel used to receive the liquid. (This method must subse- quently be strictly adhered to in the use of the instrument.) In order now to find the approximate position of the marks, tare a crucible with a little more than 5 grms. of water in it on a balance, and suck out water into the moist pipette until the crucible has lost exactly 5 grms. of its weight. Mark the level of the water in the pipette with a pencil. In a similar manner find the position of the \ fgr. mark. The intermediate marks may be determined by measurement, which will give incidentally the length of tube corresponding to every centigramme of water. Now determine the exact actual capacities corresponding to the several marks by means of an exact balance, and correct the errors by linear measurement. Supposing the pipette when filled up to the 5 mark delivers 4'94 grms. instead of 5 grms., and the distance from the 4 to the 5 mark is 30 mms., we see that the mark must be shifted upwards through 0'06x30=l - 8 mms. to assume its right position. Finally, mark the several points with the diamond. Ex. 9. Determination of Specific Gravities of Liquids. EXAMPLE chosen: Pure muriatic acid (which in this exercise means aqueous acid ; by " hydrochloric acid " we mean the real substance HC1). Take a quantity of pure, fuming, muriatic acid, determine its approximate density by a Twaddell's hydrometer, and then dilute it with water so as to bring it down to 20 Twad. at 15C. Supposing the acid showed 35 Twad. (which corresponds, by definition, to the specific gravity (5 x 35 + 1000) : 1000 = 1175), SPECIFIC GRAVITIES OF LIQUIDS. 13 the excess of weight per 1 fgr. over that of the same volume of water is 35/20 of what it is meant to be. Hence, every 20 volumes of the acid must be diluted to 35 -^ 20 x 20 = 35 volumes, which, if the two liquids when mixed did not contract, would give exactly what is wanted. [The acid being intended to be used in Ex. 10, work on such a scale as to produce about 300 cc.] After having made sure that the dilute acid really is of the intended density according to the hydrometer, determine the specific gravity more exactly by the successive application of the following two methods : 1. Take a narrow-necked flask of about 50 or 100 fgrs. capacity, make a diamond mark on the neck, and determine its capacity when filled up to this mark with water of 15C. Repeat the experiment with acid, taking particular care in this case to make sure of the temperature, as hydrochloric acid expands far more largely by heat than water does. Supposing the flask to hold 50 grms. of water and 5 5 '21 of acid, the specific gravity of the latter, by definition, is 55*21 -=- 50 = 11042. 2. Make a " plunger " (Fig. 1 ) which displaces about 20 grms. of water, charge it with mercury so that it weighs FIG. 1. in all about 40 grms., attach to its upper end a very thin platinum wire, and then determine the loss in apparent weight which the plunger suffers when immersed in (a) water and (b) the acid at 15C. These losses, according to a well-known hydrostatic proposition, equal the weights of liquid displaced, from which the specific gravity can easily be cal- culated. Half actual size. Ex. 10. Preparation and Adjusting* of Standard Solutions. EXAMPLE chosen: Preparation of a standard solution of hydrochloric acid containing exactly 1 gramme-equivalent, i.e., HC1 grms. = 36 '45 grms. of real hydrochloric acid per litre. For a first approximation, deduce from the specific gravity of your muriatic acid (Ex. 9) its percentage of HC1 by means of the table given in Note (2) at the end of this volume, and from that 14 EXERCISES IN EXACT WEIGHING AND MEASURING. calculate how much has to be diluted to, say, 1000 fgrs. to pro- duce the solution required. Supposing for the specific gravity found the table gives the percentage as 20*25 ; hence, 36*45 -7- 20*25 x 100 = 180 grms. may be taken as representing approxi- mately 36 '45 grms. of chloride of hydrogen. Weigh out the calculated quantity of your acid, pour it into a litre flask, dilute to about 4/5 of a litre, mix the liquids by giving them a rotatory motion ; then fill up exactly to the mark, shake up thoroughly (liquids don't mix by themselves), and preserve the mixture in a glass stoppered bottle, until, by Exs. 12 and 13, you have learned to determine chlorine quantitatively. You may then apply these methods to, say, 5 cc. of the approximately standardized dilute acid to ascertain its exact strength, and then correct it either by addition of water or of the mother-acid, as the case may be. As 5 cc. cannot be measured off with a suffi- cient degree of exactitude, it is better to determine the weight in grammes of, say, 100 cc. of your acid, to weigh the small quantity required for the chlorine determination, and to calculate the exact volume. To show how the correcting of the crude " standard " acid is effected, let us assume 1. The actual quantity of HC1 found in 1000 fluidgrammes, instead of being 36*45, were 36'7 grms. In this case all that is required is to dilute every 3645 volumes of acid with water to 3670 volumes ; or, what is more exact in practice, to add to every 3650 volumes (which need not be measured with any high degree of exactitude) exactly 3670 - 3645 = 25 volumes of water. 2. Supposing now the acid only contained 36'1 instead- of 36*45 grms. per 1000 fluidgrammes, to be able to easily rectify matters in this case, it is best to keep some of the original acid (of 20*25 per cent.) in reserve. Obviously what was weighed out of this stock for 1000 fluidgrammes of solution was only 36*1/36*45 of what ought to have been taken ; hence, since 180 grms. had been used, 180 x 36*45 -f- 36*1 = 181*75 grms. is what ought to have been taken. If we simply added the 175 grms. of stock acid, we should have to remove 1*75 -|- 1*10 = 1*6 fluidgrammes of water by evaporation to bring down the volume to its proper limit. But clearly there is no need for this troublesome operation. We can calculate the stock acid STANDARD SOLUTIONS. 15 required to convert 1000 volumes of the stock acid as it is into, say, 1003 volumes of what it ought to be, thus: needed for 1003 fgrs., 1-003 x 18175 = 182*29 grms. ; 182'29 - 180 = 2'29 grms. = 2*08 fgrs. of stock to be added to 1000 fgrs. of the solution to form 1002'08 fgrs. of a mixture, which obviously requires to have added to it 3'00 2 - 08 = 0'92 fgrs. of water to be made all right. In this case, as in the former, only the small quantities of stock acid and of water added to the large volume of solution need be accurately weighed or measured. Finally, in order to make sure that everything was done correctly, again analyse the adjusted acid. The result should not differ more from the intended value of 36 '45 grms. than by O'Oo grm. (or O'l grm. at the outside) ; the mean between the intended value, 36'45, and the one resulting from the final analysis is the most probable value of the actual " litre." The student will notice that we neglect the unknown contrac- tions involved in the mixings ; in a case like the present we may do so without going wrong practically. If the highest precision is demanded, the surest method is to ascertain the specific gravity of the acid to be produced, and to reduce all the quantities involved to weight. The contractions, then, are out of court. In obedience to a general habit, we shall from now use the terms " cubic centimetre " and " litre " as designating the unit of volume used in graduating the respective measuring instrument and its thousand-fold respectively. EXERCISES IN ANALYTICAL METHODS. Ex. 11. Determination of Filter Ashes. IN quantitative analysis, it is best to make a rule of using only Swedish paper, as this gives less ash per square unit than any other filter paper in commerce (except the ash-free filters lately introduced by Schleicher and Schiill). But, in any case, before using a filter paper, the exact amount of ash which it contains must be determined as follows : Cut out a stock of filters of the several sizes required in analysis. We are in the habit of naming the sizes by the length of their radii in centimetres, and speak of a No. 2, 3, 4, 4J, 5 filter when we mean a filter of 2, 3, 4, 4J, or 5 centimetres radius respectively. Take, say, three No. 8 filters, fold and roll them up tightly into a small parcel, and, after having wound the end of a platinum wire round it, kindle it, and let it burn itself out completely, over a porcelain plate placed beneath it to catch any particles. Put the ash into a small, clean, platinum crucible, burn off the charcoal completely, allow to cool, weigh the whole, remove the ash (by means of a feather if necessary), and weigh the empty crucible. The difference gives the weight of the ash. Supposing it amounts to n ings., we have, for the ash yielded by a filter of r centimetres radius, the formula Ash in mgrs. = n xr 2 = kr 2 . Calculate k. o X o Multiply k successively by the squares of the radii of the several sizes, and enter the results in a table. As treatment with acids removes a large proportion of the ash of a paper, repeat the determination with three No. 8 filters which have been previously CHLORIDE OF BARIUM. 17 washed with hot 10 per cent, hydrochloric acid, and then with water, and add a second column to your table as giving the ashes of the niters when " exhausted by acids." Ex. 12. Analysis of Chloride of Barium. SUBSTANCE. Pure chloride of barium occurs in commerce. To make sure of the crystal water, recrystallize it from hot water, allow the crystals to drain, and dry them on filter paper in the air. Take about 3 grms. of the air-dry crystals, grind them up in a mortar, and keep the now absolutely homogeneous sub- stance in a small corked preparation-tube. Determination of the Water. Weigh out, say, 1 grm. of salt in a tared porcelain crucible, and expel the water by very gradually raising the temperature to a dull-red heat, which maintain for about ten minutes. Allow to cool, and weigh. Now, again heat the crucible for about five minutes, allow to cool, and weigh again, and so on until the weight is constant to within *5 mgr. at most. From the final loss of weight calculate the percentage of water. To Determine the Chlorine. Dissolve a known weight of substance (about 1 grm.) in a beaker in about 50 cc. of water, acidulate with nitric acid, and then add nitrate of silver, at last drop by drop, and stirring up diligently after each addition until the chlorine is completely precipitated. Avoid application of heat, and don't use an immoderate excess of reagent. Stir up the mixture until it has quite lost its milky appearance, allow it to settle in the dark (or in gaslight), and carefully decant the liquor through a 4| filter. Stir up the precipitate with cold water, acidulated with nitric acid, again allow to settle, and decant through the filter. Repeat this operation once more, then wash the precipitate on to the filter, and wash it with water (cold) until the filtrate is free from acid and from barium. Dry the precipitate in the funnel, detach it as completely as possible from the paper, and incinerate the latter in a porcelain crucible. To remove the charcoal from the lid, place it, inside up, on a triangle over a Bunsen, and cover it with a bit of platinum foil; the C 18 EXERCISES IN ANALYTICAL METHODS. charcoal will soon vanish. In a similar way the charcoal inside the crucible is removed by placing the crucible slantingly over the flame, and successively heating the several portions of the char- coal under a small reverberator of platinum foil placed over the edge and stretching down into, but not touching the inside of, the crucible. The small particles of chloride of silver which stuck to the filter are now converted into metal. To chlorinate them, dissolve them in one drop of nitric acid, add a drop of hydrochloric,, and carefully evaporate to dryness. Now, add the bulk of the chloride of silver, and heat the crucible over a flame so as to just fuse the chloride at the lowest temperature. Allow to cool, weigh, deduct the weights of empty crucible and ash (as "exhausted by acid"), and, from the weight of the chloride, calculate the percentage of chlorine. AgCl : Cl = 143*38 : 35*45 = 1 : 0*24727. To Determine the Barium. First boil down the wash -waters from the chloride of silver so far that, when united with the filtrate, the whole amounts to 100-150 cc. Place this liquid in a beaker, or better, in a Florence flask of 200-250 cc. capacity, heat to boiling, and then add excess of sulphuric acid so as to pre- cipitate the barium. Keep up a boiling heat until the precipitate becomes compact and readily settles down. Allow the precipitate to settle down completely, decant the clear liquor through a 4| filter, wash the precipitate twice by decantation with hot water acidulated with sulphuric acid ; then collect it on the filter, con- tinue washing with pure hot water until the last washings are free from sulphate and free acid. Dry the precipitate, detach it from the filter, incinerate the latter in a platinum wire spiral, add the ash to the precipitate, and ignite the whole for about a quarter of an hour in a tared platinum crucible, allow to cool in an exsic- cator, and weigh. From the weight of the sulphate of baryta calculate the percentage of barium. BaS0 4 X 0*5882 = Ba. During the evaporations, &c., involved in the above operations, practise the following two methods for the determination of the barium : 1. Dissolve, say, 1 grm. of the salt in about 100 cc. of water, heat to boiling, precipitate the barium by sulphuric acid, and collect and weigh the precipitate as just explained. REPORTING OF ANALYSES. 19 2. Weigh out about 1 grm. into a platinum crucible, dissolve, over a water bath, in the least quantity of water, add a measured volume of a standardized sulphuric acid, presumably not quite sufficient to decompose the salt, evaporate to as near dryness as possible on a water bath ; then place the crucible slantingly on a triangle, and complete the evaporation by heating the lid and the upper portions of the crucible by means of a Bunsen. Finally, heat the crucible to redness from below, allow to cool, and weigh. Moisten the residue with a few drops of sulphuric acid, evaporate, again ignite and weigh, and so go on until the weight of the residue is constant. Each of the above analytical methods must be practised again and again if necessary, until you have them at your fingers' ends. Observe, the methods in this book are not selected on the prin- ciple of simplicity, but with the view of teaching you as much as possible by each exercise. RULES KEGARDING THE BOOKING AND REPORTING OF ANALYSES. 1. Keep one book for the data of the analyses ; another for the calculations. Both to be paged, and every item in either book to be marked with the page corresponding to it in the other. 2. No need of a scroll book, as a rule. The weighings should go directly into the report book meant to be submitted to the teacher. Write distinctly. 3. All results to be given in, whether successes or no. 4. At end of report, summarize in the following style : 1. Percentage of Water. Analysis, 1. 2. 3. Mean. Results, 14-65 14-92 (16-33) 1478 " ( ) " means excluded in calculating the mean. The numbers 1, 2, &c., refer to sections of the detailed report. 2. Percentage of Barium. Analysis, 1. 3. 5. Mean. Results, 5610 56'33 56'20 56-21 3. Percentage of Chlorine. Similarly. EXERCISES IN ANALYTICAL METHODS. Mean Results. Theory. Found. Barium, Ba = 137'20 56'20 Chlorine, C1 2 = 70*91 29'05 Water, 2H 2 = 36-01 1475 24412 100-00 Two decimals are plenty. Learn the methods of shortened decimal multiplication and division, and of logarithmic calcula- tion. The four-place table in the author's "Tables to facili- tate," &c., will be found convenient. It is so constructed that the uncertainty in even an interpolated logarithm is less than 1 J unit of the fourth place, and consequently corresponds to less than 0-00026, or 1/3850 of the number. The uncertainty in a number furnished for a given logarithm, even by interpolation, falls short of 0*000145, or 1/6900 of the value of the number. This is more than can be attained with an ordinary four-place table, and suffices in almost all cases. Ex. 13. -Analysis of a Silver Coin. DISSOLVE, say, 1 grm. of the alloy by heating it with 10 cc. of 1*2 nitric acid in a slanting Florence flask. If gold is present it remains as a black powder, which collect on a very small filter, ignite precipitate and filter in a small porcelain crucible, and weigh. To the filtrate (which should be in a beaker), add hydrochloric acid drop by drop, stirring up diligently after each addition, so that the chloride of silver coagulates, and you are able to see if the next drop produces an additional precipitate. In this way continue until the silver is completely precipitated by the least practicable quantity of reagent. Manipulate the pre- cipitate as shown in Ex. 12, and from the weight of the chloride of silver calculate the silver. AgCl x 07527 = Ag. While the AgCl is drying, practise the following " titrimetric " method of silver determination: Volhard's Method. By dissolving 1/10 NCS.NH 4 = 7'6 grms. of sulphocyanate of ammonium in water, and diluting to 1 litre, SILVER COIN. 21 prepare a solution which contains approximately 1/10 NCS.NH 4 mgs., and consequently precipitates about 1/10 x Ag. nigs. = 10793 mgs. silver per cc. To standardize this solution exactly, weigh out from 4 to 5 decigrms. of pure silver, dissolve it in 10 cc. of pure nitric acid, of 1'2 specific gravity, in a slanting flask, dilute with water to about 100 cc., add about 5 cc. of saturated solution of iron-alum, and then run in sulphocyanate from a burette until the red colour of ferric sulphocyanate, which makes its appearance locally from the first, through the addition of the last drop of reagent, has become permanent on stirring. Supposing 450 mgs. of silver have been taken, and 42'1 cc. of standard solution been required to effect a complete precipitation, 1 cc. of the latter may be put down as corresponding to 450 -r 42'1 = 10*689 mgs. of silver. This determination must be done twice at least (the two results should agree to within 1/1000, or, at most, 1/500 of their value), and the mean of the results be adopted. To determine the silver in the coin, just treat \ grm. of it as you did the pure silver in determining the " titre " of the reagent. The rest requires no explanation. The copper may be determined by difference.* You may now resume Ex. 10, and determine the exact weight of hydrochloric acid contained in every cc. of your standard muriatic acid by means of the following two methods : 1st, the one explained in Ex. 12; and 2nd, as follows: Weigh out (i.e., indirectly measure off) a convenient quantity of your muriatic acid, and, from the result of the gravimetric analysis, calculate the weight of silver which that quantity of acid should be capable of precipitating. Supposing it is 1*106 grms., weigh out a little more than that, say 11 2 grms. of pure silver, dissolve it in nitric acid, dilute to 100 cc. or so, add the weighed-out sample of your muriatic acid, stir up violently, and allow to settle in the cold, and in the dark. Then pour off the clear liquor as completely as possible, add iron alum, and, by titrating with sulphocyanate, determine the weight of silver which is present in the solution. From the result you will see * By intention, British coins contain 92 '5 per cent., German and French coins 90*0 per cent., of real silver. The rest, apart from an occasional trace of gold, is copper. 22 EXERCISES IN ANALYTICAL METHODS. whether or not it is worth while to recover and estimate the small remnant of dissolved silver contained in the moist precipi- tate. Deduct the silver found by titration from the silver originally taken, and, from the difference, calculate the hydro- chloric acid, remembering that every Ag = 107'93 parts of silver corresponds to HC1 = 36*45 parts of real hydrochloric acid. Note. In either case, you had better state your result in multiples of HC1 mgs. per cc. Supposing, for instance, you obtained 145*3 mgs. of chloride of silver per cc., report the hydrochloric acid as being 145*3/143*38 x (HC1 = 36*45) mgs., reducing the fraction 145*3/143*38 to a decimal. And, similarly, if with the second method you found 1 cc. of your acid precipi- tated exactly 109*1 of silver, report thus 109*1 /107'93 X (HC1 = 36*45) mgs. of chloride of hydrogen. Ex. 14. Sulphate of Iron and Potash. FeS0 4 K 2 S0 4 + 6H 2 0. SUBSTANCE. Dissolve 64 grms. of pure ferrous sulphate crystals and 29 grms. of pure powdered sulphate of potash* in 110 cc. of hot water and 1 cc. of 20 per cent, sulphuric acid ; filter hot into a basin, and allow to crystallize. Throw away the mother liquor. Dry the crystals on filter paper by exposure to ordinary air. Do not re-crystallize your salt, or else it changes composition. Determination of the Sulphuric Acid. Dissolve 1-1*5 grms. of salt in about 100 cc. of water, with addition of 2-3 cc. of 20 per cent. HC1, heat to boiling, then add a slight excess of chloride of barium. Keep the mixture at a boiling heat until the precipi- tate of sulphate of baryta has settled. Manipulate the precipitate as shown in Ex. 12, except that you acidulate the wash- water with hydrochloric, instead of sulphuric, acid. From its weight calculate the percentage of sulphuric acid. BaOS0 3 x 0*3432 = S0 3 . Keep filtrate and wash-waters for the determination of the iron and potash. * See Note (3) at end of volume. SULPHATE OF IRON AND POTASH. 23 To Determine the Iron. Boil down the filtrate* (with addition of a few granules of chlorate of baryta to peroxidize the iron) to about 200 cc. Then add excess of ammonia, and boil off that excess again (in order to minimize the precipitation of barium). Collect the precipitate on a No. 5 filter, and wash it with hot water until the last drop of wash-water leaves no residue on evaporation on platinum foil. As this precipitate is always contaminated with barium, scrape it from the filter (before drying) as completely as possible by means of a platinum spatula, dissolve what sticks to the filter in hydrochloric acid, and use the liquor for dissolving the bulk of the precipitate. Precipitate the barium with a few drops of sulphuric acid in the heat, allow to settle, filter and wash the precipitate, and in the filtrate repre- cipitate the iron with ammonia. Dry, detach the oxide as completely as possible from the paper, and place it in a tared platinum crucible. Incinerate the paper in a platinum spiral, add the ash to the precipitate, ignite, and weigh. From its weight calculate the percentage of ferrous oxide. Fe 2 3 x 0-9 - 2 FeO. The Potash is determined in the filtrate obtained in the first precipitation of the oxide of iron. Concentrate the wash- waters by evaporation in a slanting flask, add the filtrate, and next precipitate the barium by addition of ammonia and carbonate of ammonia. Keep the mixture on a water bath until the carbonate of baryta has settled, then filter it off and wash it with hot water, always adding a few drops of carbonate of ammonia to increase the insolubility of the precipitate. Boil down the filtrate in a flask, and ultimately evaporate to complete dryness in a Berlin basin. Scrape out the dry residue, transfer it to a platinum crucible, put on the lid, and, by applying a flame, gradually and slowly burn off the ammonia salt, taking care never to let the temperature rise beyond dull redness. To recover the small quantity of salt sticking to the basin, dissolve it in a few cc.'s of sal-ammoniac solution, evaporate to dryness in the basin, scrape out that residue, and treat it like the first. The non- volatile residue should, by theory, consist of pure chloride of potassium, * "Filtrate" always includes the washings as far as they contain anything of the substance to be determined in the nitrate. 24 EXERCISES IN ANALYTICAL METHODS. but is generally not quite pure. Weigh the crude chloride, then see if it is quite soluble in water, and if the solution remains clear on adding a drop of carbonate of ammonia. If the mixture is turbid, filter off the small precipitate formed, evaporate the filtrate to dryness in the crucible, ignite gently to drive off the chloride of ammonium, and weigh the residue as (now sufficiently pure) chloride of potassium. Reduce to percentage of K 2 by eq. 2KC1 x O6319 = K 2 0. The Water is determined by difference. Theoretical Composition. Ferrous oxide, 72'02 16'58 Sulphuric acid, 16012 36*86 Potash, 94-27 2170 Water .. . 108'03 . 24*87 434-44 100-01 Ex. 15. Analysis of Sulphate of Magnesia and Potash. SUBSTANCE. Dissolve 53 grms. of crystallized sulphate of magnesia and 29 grms. of sulphate of potash in 150 cc. of boiling water, filter, and allow to crystallize. Do not re-crystal- lize your double salt.* Place the crystals on filter paper, and expose them to the air to dry. The Water is determined as shown for chloride of barium in Ex. 12. Take care that the temperature, while fully up to, never rises beyond, dull redness, or else some of the SO 3 goes with the H 2 O. The Magnesia. Dissolve 07 1 grm. of the salt in 40 cc. of water, add a few cubic-centimetres of sal-ammoniac, then 10 cc. of ammonia, which should produce a clear liquid. If a precipitate is produced, take it away by cautious addition of more sal- ammoniac. Then add a slight excess of phosphate of ammonia, and allow to stand for twelve hours in a covered beaker. In stirring the mixture, avoid touching the beaker with the glass rod, or else the parts of the precipitate which form at the scratches produced * See Note (3) at end of volume. TATLOCK'S METHOD FOR POTASH. 25 will be very hard to remove. After having made sure that the precipitation is completely effected, collect the precipitate on a No. 5 filter, wash it with a mixture of 1 volume of 10 per cent, ammonia and 2 volumes of water until the last runnings are free from sulphate, dry it, separate it from the filter, incinerate the filter in a platinum spiral, and ignite precipitate and ash in a porcelain crucible until the weight remains constant, even at a bright red heat. The ignited precipitate is P 2 7 Mg 2 ; from its weight calculate the magnesia. P 2 O 7 Mg 2 x 0'3623 = 2MgO. The Sulphuric Acid is determined by precipitation with chloride of barium, as shown in Ex. 14. From the weight of the sulphate of baryta calculate the percentage of sulphuric acid as S0 3 . For the Determination of the Potash, we recommend the following two methods. Both should be practised by the student : 1. TATLOCK'S METHOD. Dissolve, say, 0'5 grm. of the double salt in 20 cc. of water and a few drops of hydrochloric acid, add a quantity of chloro- platinic acid solution, containing 1'3-1'4 grm. of metallic platinum, i.e., about 2*5 times the quantity which is required, by calcula- tion, to convert both the magnesium and the potassium into chloroplatinates, and evaporate over a water -bath until the residue, after cooling, forms a semi-solid magma. [If the substance to be analysed contains chlorides, add a little water to the residue, and re-evaporate to expel the HC1 eliminated from the PtCl 6 H 2 as far as conveniently practicable. In our case there is no hydrochloric acid produced, as PtCl 6 H 2 -f K 2 S0 4 = PtCl 6 K 2 + H 2 S0 4 .] After cooling, add 2 - 5 to 3 cc. of water,* allow to stand, with occasional stirring, for an hour; then throw the whole on a filter, No. 3, and use first the filtrate, and then about 10 drops of chloroplatinic acid solution, containing 5 grms. of metal per 100 cc. for rinsing the basin. Other 10 to 15 drops of reagent serve to bring the bulk, if not the whole, of what has remained of precipitate in the basin on the filter. After the platinic * The intention is to extract the soluble parts with a PtCleH 2 solution, con- taining at least 5 grms. of metal per 100 cc. 26 EXERCISES IN ANALYTICAL METHODS. liquor has drained off as completely as possible, use small succes- sive instalments of strong alcohol, to bring all the precipitate on the filter and wash it free of chloride. About 20 cc. of alcohol suffice for this purpose (Mr. Tatlock recommends alcohol of " 95 per cent."; but ordinary alcohol of 85 per cent, by weight no doubt is sufficiently strong). Dry the precipitate in the filter in a drying chamber below 100C., and transfer the bulk to a watch-glass. To recover what sticks to the filter, replace the filter in the funnel, dissolve off the adhering salt in hot water, and evaporate the solution in a tared platinum crucible. Now add the bulk from the watch-glass, dry at 100C. until constant, and weigh. PtCl 6 K 2 x 01940 = K 2 0. Another mode of manipulating the filter is to incinerate it in a small crucible (which converts the chloroplatinate into Pt + 2KC1), to wash the ash by decantation with hot water, and to weigh the (ignited) platinum. Pt X 2'4936 = PtCl 6 K 2 . Pt x 0-4839 = K 2 0. The chloroplatinate, if dried at 100C., retains an appreciable proportion of water ; it, besides, is liable to include traces of sulphates. On the other hand, however, a small quantity of the chloroplatinate remains dissolved in the mother liquor, and another is dissolved by the wash liquors ; and as a matter of experience the errors very nearly compensate one another. Mr. Tatlock's method, indeed, gives very satisfactory results with any kind of salt that fairly falls within the denomina- tion of (impure or pure) potash salt. It is not intended for the determination of small percentages of potassium in, for instance, what substantially consists of sodium salts. The prin- cipal advantages of the following method are that it is very widely applicable, and gives exact results, even if the potash to be determined is diffused through a large mass of foreign salts ; besides, it is less wasteful of platinum than any of the older forms of the platinum process. 2. FINKENEK'S PKOCESS. (As modified ly the AUTHOR and Mr. JOHN M 'ARTHUR). Dissolve, say, 1 grm. of salt in water, and add about T25 times the quantity of chloroplatinic acid required, by calculation, for FINKENER'S METHOD FOR POTASH. 27 the conversion of the potassium into chloroplatinate, i.e., about 0'6 grin, of metallic platinum for 1 grm. of our double salt. Heat the mixture, and, if necessary, add more water, so that the chloroplatinate formed almost or quite dissolves in the hot liquid. Then evaporate on a water-bath to a very small volume, so that what is left forms, after cooling, a solid magma soaked in, say, 1 cc. of mother liquor ; as soon as salt begins to separate out, you must stir up to prevent the formation of large crystals. After cooling, add, first, about 10 cc. of absolute alcohol, and then, after some time, 5 cc. of ether, mix well, and allow the basin to stand under a bell jar for one or two hours, or until the pre- cipitate (a mixture of chloroplatinate of potassium and sulphate of magnesia in the present case ; if Na 2 or CaO were present they would go down in similar forms) has settled completely. Decant off the liquid through a filter and wash with ether- alcohol, 1 volume of ether to 2 of alcohol. The precipitate as thus obtained usually contains a small quantity of foreign chloroplatinate. To remove this, rinse the precipitate with a few cc.'s of plain ether, allow it to dry up in the air or at a very gentle heat, and then heat it on a water-bath with a quantity of water, which almost, if not quite, suffices to dissolve the chloroplatinate of potassium. Then re-evaporate to a magma, allow to cool, and again apply alcohol and ether as in the first production of the mixture. As a rule, some 2 to 3 mgs. of platinum pass into solution. After this process of purification all that remains to be done is to determine the potassium present in the precipitate. This may be done in two ways. First method. Decant the clear liquor through a filter, No. 4, wash the precipitate by decantation with ether-alcohol as above, dry it (i.e., both the filter and the contents of the basin) at a gentle heat, scrape off what is on the filter and add it to what is in the basin. Incinerate the filter and add also the filter-ash. Then place a glass funnel over the salt in the basin (see Fig. 2), and, while sending a pretty brisk current of hydrogen gas through the stem over the substance, raise the temperature of the latter to about 300C. by means of an Argand or rose burner, and keep up this heat until the 28 EXERCISES IN ANALYTICAL METHODS. chlorine of the Pt01 4 part of the salt is completely expelled as HC1. The residue now consists of metallic platinum plus sul- phates (MgS0 4 in our case). Extract them, the bulk by means of water, the last remnant (and other impurities that may be pre- sent) by means of warm hydrochloric acid ; collect the platinum on a No. 3 filter, wash, ignite, and weigh it. From the weight calculate the potash. (Vide supra.) The filtrate from the metallic platinum must be tested by sulphuretted hydrogen, and any PtS 2 that may come down collected, washed, and ignited, to be recovered as metal. A more exact form of the method is to dissolve all the precipi- tate in hot water (about 100 cc. for 0*5 to 1 grm. of platinum in the precipitate), and precipitate the platinum as such by hydrogen gas. Requirements. (1.) A "Kipp" affording a con- tinuous supply of washed hydrogen. (2.) An Erlenmeyer flask, holding about twice the volume of the liquid, provided by means of a well-fitting india-rubber "cork" with an inlet tube Wet Way. Reduction of K 2 PtCl 6 in the Dry Way. One-tenth actual size. terminating 1 to 2 centms. above the liquid, and an outlet tube commencing just below the cork and terminating in a short bit of india-rubber tubing. (2a.) A pinch-cock to close the latter. (3.) A large beaker rigged up as a hot-water bath. (4.) A car- bonic acid apparatus. The flask is filled with hydrogen. When FINKENER'S METHOD FOR POTASH. 29 all the air is out, the outlet is closed with the pinch-cock ; but the communication with the Kipp maintained. The flask is immersed almost totally in the water bath, which is kept at 80 to 90C. until all the platinum is eliminated, which is seen by the disappearance of all colour from the solution. Before opening the flask the hydrogen must be displaced by air-free carbonic acid to avoid explosions. The rest explains itself. Second method. After having washed the mixture of PtCl 6 K 2 and MgS0 4 with ether-alcohol, digest it in a cold, saturated solution of sal-ammoniac, which readily dissolves the MgS0 4 (and Na. 2 SO 4 , if present), but does not change the chloroplatinate if not allowed to act for an unnecessarily long time. Collect the residual Pt01 6 K 2 on a small filter, wash it as quickly as possible with sal-ammoniac solution until the last runnings are free from S0 3 , run alcohol through the precipitate once to remove part of the sal-ammoniac, then dry it (not completely, but only so that the filter can be safely lifted out of the funnel), and, lastly, ignite precipitate and filter at a dull red heat in a porcelain crucible. Heat the residue in an atmosphere of hydrogen, and extract the chloride of potassium by means of water. The potassium now can be determined in three ways 1, by igniting and weighing the platinum (this is the safest plan) ; 2, by evaporating the KC1 solution to dryness, igniting gently to expel any (NH 4 ) Cl that may have remained, and weighing the KC1 ; 3, (if the (NH 4 ) Cl is sure to be away), by determining the chlorine in the KC1 solution. The Second (sal-ammoniac) method comes in useful chiefly for the analysis of substances poor in potash. In their case the Finkener process serves only for, so to say, concentrating the potassium so as to bring it within the range of the ordinary forms of the platinum process. The recrystallization process is dispensed with. The crude mixture of sulphates and potassium chloroplatinate, after having been washed and dried, is reduced dry in hydrogen as explained, the chloride of potassium extracted by water, and then determined as chloroplatinate, preferably as follows : The calculated amount of chloroplatinic acid, plus about a quantity representing 50100 mgs. of metallic platinum, is added to the dissolved chloride of potassium, the whole 30 EXERCISES IN ANALYTICAL METHODS. evaporated to a magma on a water-bath, as usual, and after cooling, washed judiciously, first with 0'5 cc. of water, then again with 0'5 cc. of water, or 0'5 cc. of 5 per cent, reagent should the excess of platinum not produce about this strength in 0"5 cc. of added water, then with small instalments of alcohol of about 74 per cent, by weight, and lastly, with nearly or perfectly absolute alcohol. The chloroplatinate is dried at 100, and weighed as such.* Theoretical Composition. Magnesia, MgO = 40'37 10-02 Potash, K 2 - 94-27 23'40 Sulphuric acid, 2S0 3 = 16012 3975 Water, 6H 2 = 108*03 26'82 40279 99-99 Ex. 16. Analysis of Phosphate of Lime. EXAMPLE chosen: Hydrated di-calcic phosphate, P 2 6 (2CaO.H 2 0) -f ccAq. Determination of the Water. Ignite 0"7 to 1*5 grms. in a platinum crucible repeatedly until the weight is constant, and from the loss calculate the percentage of water. The residue, being pyrophosphate, is not available for the following deter- minations. Determination of the Lime. Dissolve 1 to T5 grms. of sub- stance in the least quantity of hydrochloric acid, dilute with water to about 50 cc., and to the cold liquid add ammonia until alkaline ; then add acetic acid, drop by drop, until the precipitate is redissolved. If an insoluble precipitate remains, filter it off, ignite, and weigh it as P 2 5 (Fe 2 3 or A1 2 3 ). To precipitate the lime in the filtrate, add a measured volume of an approximately standardized solution of oxalate of ammonia, which is sure to be more than equivalent to the lime present. Allow the mixture to stand in the cold for about half-an-hour, then put it on a water * See Note (4) at end of volume. PHOSPHATE OF LIME. 31 bath until the oxalate of lime has completely settled. Filter off the oxalate on a No. 5 filter, wash it with hot water, dry, and heat it with the filter in a close platinum crucible until the paper is completely charred. In order now to burn the charcoal, shift the lid a little to one side so that the crucible is only three- fourths covered, and give it the full heat of a good Bunsen until the residue is white. Weigh the crucible, then heat for five minutes over the blow-pipe and weigh again, and so continue until the weight is constant, which shows that the residue has been converted into pure (C0 2 -free) lime. The filtrate serves for the Determination of the Phosphoric Acid. Evaporate filtrate and washings to about 100 cc., add 25 cc. of ammonia, then magnesia mixture in sufficient quantity, and allow to stand for twelve hour-s in a covered beaker. Collect and weigh the pre- cipitate as shown in Ex. 15 for magnesia. (P 2 5 2MgO) x 0*6377 = P 2 5 . While the above operations are progressing, start an independent determination of the phosphoric acid by means of The Molybdenum Process, which is of great importance on account of its almost absolutely general applicability. Dissolve O'l to 0*15 grm. of substance in nitric acid, add enough of a nitric solution of molybdate of ammonia to precipitate the whole of the P 2 5 (which means at least 40 parts of Mo0 3 for every 1 part of P 2 5 present),* and keep the mixture over a water bath, at about 40 C., for five or six hours, when the P 2 5 may be assumed to be completely precipitated. But to make sure of this, decant off some of the clear liquor into a large test-tube, add about one- third of its volume of reagent, heat gently for a while, and observe the effect. Supposing the precipitation to be complete, decant the liquor through a No. 4 filter as completely as possible, and wash the yellow precipitate four times by decantation with small quantities (of about 10 cc.) of reagent, or of a liquid prepared by dissolving 15 grms. of nitrate of ammonia in water acidulated by 1 cc. of nitric acid, and dilut- ing to about 100 cc. In order now to determine the P 2 5 in the precipitate, dissolve what is on the filter in the least quantity * In calculating the reagent, assume your preparation to contain 75 per cent, of P 2 5 . 32 EXERCISES IN ANALYTICAL METHODS. of ammonia, and use the resulting liquid for dissolving the bulk of the precipitate which was left in the precipitating vessel. To the solution (which should be perfectly clear) add hydrochloric acid, drop by drop, until the yellow precipitate locally formed is somewhat slow in disappearing ; then add magnesia mixture to precipitate the P 2 5 as phosphate of magnesia and ammonia, which is filtered off and washed with dilute ammonia. It is always contaminated with molybdenum. To remove this, dis- solve the precipitate in the least quantity of hydrochloric acid, pass H 2 S into the heated solution, allow to stand until the precipitate has settled, then filter and reprecipitate by ammonia after addition of a little magnesia mixture. The precipitate now is pure, and can be manipulated as usual. According to Finkener,* the phospho-molybdate of ammonia can be weighed directly. For this purpose, precipitate in the usual way, keeping the bulk of the precipitate in the beaker, and wash with an acidified 20 per cent, solution of nitrate of ammonia. Transfer the precipitate to a tared porcelain crucible (preferably a shallow, flat-bottomed one), evaporate to dryness, and heat on a piece of wire-gauze, over a very small flame, till vapours cease to be evolved. Most of the precipitate can be washed over mechanically ; what of it sticks to the filter or beaker is dissolved in ammonia ; the ammonia solution is added to the precipitate. But according to Finkener, an acid reaction must be established by HN0 3 before evaporating. Weigh and calculate the P 2 5 . The precipitate contains 3794 per cent, of P 2 5 . Although it may be slightly discoloured, this will not appreciably affect the final result. From the percentages found, calculate the formula of the salt that is, assuming the salt to be 1P 2 5 + #CaO + 2/H 2 O, find x and y ; x should be very nearly equal to 2 ; y, in a preparation which has not been specially dried under definite conditions, is subject to variation. Supposing you find x = 2'02, y = 4*2, calculate the percentages corresponding to P 2 5 (2CaO.H 2 0) + 3'2 H 2 0, and contrast them with the direct result of your analysis. * Ber. Deutsch. Chem. Ges., 1878, p. 1639. SEPARATION OF IRON AND ALUMINA. 33 Ex. 17. Separation of Iron and Alumina. WEIGH out about 1 grm. of potash alum, and '5 grin, of stan- dardized sulphate of iron (see Ex. 18), dissolve in water, add hydrochloric acid, and a few drops of nitric, and boil, to per- oxidize the iron. Then dilute to about 20 cc., add enough of pure potash to produce a strong permanent precipitate, then place the mixture in a nickel or platinum basin.* Now add excess of strong caustic potash, and keep at a boiling heat, stirring up continually with a platinum rod to prevent actual boiling, until the A1 2 O 3 can be assumed to have passed into solution, and dilute, first as far as possible in the metal basin, then by instalments more largely in a Berlin basin provided with a good spout, and filter off the ferric hydrate. Wash the precipitate thoroughly with hot water; the result is that substantially the iron and alumina are separated from each other. But, in general at least, what should be ferric oxide retains a more or less considerable portion of the alumina. This alumina must be extracted by re-dissolving the precipitate in hydrochloric acid, and repeating the treatment with alkali. The crude ferric oxide obtained cannot be ignited and weighed as it is, because it contains com- bined fixed alkali. It must be dissolved in hydrochloric acid, reprecipitated by ammonia, and ignited as before explained. From the Alkaline Liquors precipitate the alumina by adding an excess of sal-ammoniac, and heating the mixture until the ammonia liberated is substantially expelled. The test for seeing whether the sal-ammoniac is in excess is, after expulsion of the ammonia, to add some more of the salt, and see if this causes an additional evolution of ammonia. If not, the decomposition of the aluminate is accomplished, and all that remains to be done is to collect the alumina on a filter, to wash it thoroughly with hot water, dry, ignite, and weigh it. The best way of manipulating the alumina is to allow it to dry up in the filter at a gentle heat, until the paper ceases to be actually wet, and the precipi- tate has shrunk considerably without having become hard and dry ; and in this condition to put it, wrapped up in the filter, * If a platinum basin is used, the acid liquor may go in directly. P 34 EXERCISES IN ANALYTICAL METHODS. into a platinum crucible, and ignite it, finally in the presence of air, until all the charcoal is burned away. For a check, determine the iron and alumina conjointly by precipitating with ammonia and sal-ammoniac. Compare your results with the numbers demanded by the synthesis. On this occasion, you should learn to work the Bunsen pump. Ask the Demonstrator to show vou how to use it. Ex. 18. Titrimetric Determination of Iron. THE method is founded upon the fact that strongly acid dilute solutions of ferrous salt are readily oxidized into ferric salt by addition of permanganate of potash. This reagent forming intensely pink or violet solutions, while dilute iron and man- ganous solutions are practically colourless, the end-point of the oxidation process is easily seen by the appearance of a per- manent red colour in the mixture. REQUIREMENTS. (1.) Standardized Ferrous Sulphate. Dissolve 100 grms. of the ordinary pure salt in 250 cc. of water and 10 cc. of 60 per cent, sulphuric acid, heat the solution for a while with a quantity of small iron nails, and then filter it through a double filter straight into half a litre of strong alcohol, with constant agitation of the latter. The greater part of the salt comes down as a fine crystal meal. Collect this on a funnel over a small perforated cone of vegetable parchment, and, with the help of the filter pump, suck off the mother-liquor, and wash off the adhering remnant of the latter by means of alcohol. After having dried the salt on filter paper (which should be renewed after the tangible liquor is gone), crush the lumps by gentle pressure, and expose the salt on a bed of fresh paper to the air in a dry place (not in artificially dried air) until it has lost every trace of stickiness and rolls over a sheet of paper like gunpowder. Lastly, sift off and reject the coarser parts, and preserve the powdery salt in a dry, well-stoppered bottle. To determine the iron, heat, say, 2 grms. in a platinum crucible, finally to bright or TITRIMETRIC DETERMINATION OF IRON. 35 redness in the presence of air, until the residue no longer changes weight, i.e., has become pure Fe 2 O 3 . Before accepting the result, make sure of the absence of FeO and of SO 3 by dissolving a little of the oxide in hydrochloric acid, and applying the well- known tests. The (unavoidable) ferricum amounts to very little; yet it had better be determined, in some 5 grms., by titration with stannous chloride (vide infra), and allowed for. (2.) A Solution of Permanganate of Potash, so adjusted that 1 cc. = 5*6 nigs, metallic iron.* Since Mn 2 O 7 contains 5x"O" available for oxidation of iron, and every 1"O" oxidizes 2FeSO 4 + 1H,SO 4 into Fe,(SO 4 )., + H 2 O; K,OMn 2 O 7 grms. = 10 Fe grms. 316 grms. = 560 grms.; c 3'16 grms. = 5'60 grms. As the salt is never absolutely pure, the solution cannot be standardized synthetically. Weigh out, say, 2 x 3*2 = 6'4 grms. of permanganate, dissolve in a mortar by successive instalments of water, and dilute to 2 litres. To determine the exact strength of the solution, weigh out 0'8 to TO grm. of the ferrous sulphate, dissolve in 200 to 300 cc. of water previously mixed with about 20 cc. of 20 per cent. H 2 S0 4 , and drop in permanganate solution from a Gay-Lussac burette until the last drop added produces a permanent red colour. Calculate the iron corresponding to 1 cc. of permanganate in multiples of 5'6 mgs. Supposing you find 1 cc. = T012 x 5'6 mgs. of iron, then obviously every 1 lit. of the reagent requires the addition of 012 lit., or 12 cc. of water. Add the calculated quantity of water, and repeat the standardization. All these standard titra- tions ought to be done in duplicate at least, and not to be accepted unless the results agree to O'l or 0'2 cc. per about 30 cc. of reagent used. The application of the process to solutions containing iron as ferrous *tilj_>]n:ite requires no explanation. The following points, however, must be attended to : * We often use the symbol = for " equivalent to " (in the sense of the reaction considered). 36 EXERCISES IN ANALYTICAL METHODS. 1. The iron solution must be so dilute that 1 lit. contains at most 1 grm. of metal. 2. The solution must be cold. 3. There must be a large excess of H 2 S0 4 , far more than the chemical equation demands. 4. The solution must be free from chlorides, or else the HC1 takes up part of the reagent, forming C1 2 and water : Mn 2 T + 14HC1 = 7H 2 + 2MnCl 2 + 5C1 2 . This, however, can be remedied by addition of a sufficiency of sulphate of manganese (Zimmer- mann). Prepare a solution of 20 grms. ordinary crystallized MnS0 4 4H 2 to 100 cc. In titrating, about 20 cc. of this solution should be added per 60 cc. of 20 per cent. HC1 used in dissolving the substance to be analysed. When the iron is given as ferric salt, its reduction to ferrous salt may be effected by the following methods: (1.) By Sulphuretted Hydrogen. Pass H 2 S into the solution until it is completely reduced, boil off the H 2 S, filter quickly, pass in a few extra bells of the gas to correct any adven- titious oxidation, boil off the now present excess of H 2 S in a narrow -necked flask, cool quickly, and titrate. (2.) By Sulphite of Soda. The ferric chloride solution, which should contain about 5-10 cc. of HC1, and about 0*1 grm. of metallic iron, and amount to 100 cc., is placed in a 300 cc. flask with a long neck, diluted to 200 cc., and heated nearly to boiling. The heat is then removed, and a saturated solu- tion of sulphite of soda is added in small quantities till 15-20 cc. in all have been added. As soon as the solution of the sulphite has been all added, an india-rubber stopper provided with a FIG. 3. TITKIMETRIC DETERMINATION OF IRON. 37 syphon-shaped outlet-tube, as shown by Fig. 3, is inserted, and the liquid boiled briskly. The boiling is continued until the steam, when passed through a dilute solution of permanganate containing a little H 2 SO 4 , no longer destroys the colour of the reagent. As soon as this test shows that all the S0 2 has been boiled away, the tube is dipped into a beaker containing pure water, the boiling continued for a short time, and the clip then closed, the lamp being, of course, removed at the same time. After a minute or two, the clip is opened and the flask allowed to fill itself up to about the beginning of the neck. The con- tents are then cooled and the titration is proceeded with, after addition of the requisite quantity of sulphate of manganese. [P. T. Austen and G. P. Hurff ; Chemical News, vol. xlvi. (1882), p. 288.] To practise the method of permanganate titration, apply it (1) To a known weight of pianoforte wire dissolved in dilute sulphuric acid. Good wire contains from 99'6 to 100 per cent, of metal. Calculate for 99*8, and the titration should correspond with that to 0'2 or 0'3 per cent. (2) To the case of iron and alumina in Ex. 17. Produce a mixed precipitate of Fe 2 O 3 and A1 2 O 3 , containing an exactly known weight of iron ; dissolve in hydrochloric acid, reduce by means of sulphurous acid, add sulphate of manganese, and titrate. APPENDIX. The following two additional methods are given here for future reference. The student is not meant to take them up now ; he makes better use of his time by practising gravimetric work. THE BICHROME METHOD was invented at about the same time by Penny and by Schabus ; and it is worth noting that in the latter 's hands it virtually led to the discovery of an error in the then atomic weight of chromium. In principle it is similar to the permanganate method, only bichromate of potash is used as an oxidant, and the end point is sought for by taking out a drop of the titration mixture from time to time, and examining it with ferricyanide of potassium. 38 EXERCISES IN ANALYTICAL METHODS. The Ferricyanide is used as a J per cent, solution, which is best made extemp. from a compact crystal previously washed with water to remove any superficial ferrocyanide. The solu- tion does not keep long, especially not in the light. Before use it must be tested with pure ferric chloride, with which it should strike a brown colour, free of all shade of green. For imme- diate use drops of the solution are distributed on a flat porcelain plate or in the several cavities of a porcelain slab as used for water-colour painting. The Bichrome volution can be standardized synthetically. As OA + 6FeO = 3Fe 2 3 + Cr A 1/60 of 294-54 grms. = 4'909 grms. of the pure reagent dissolved to 1 lit. give a solution of which every cc. peroxidizes 1/10 Fe = 5'6 mgs. of ferrosum. The principal impurity in the salt to guard against is sulphate. To test for it dissolve 1 grin, in 10 per cent, hydrochloric acid, and heat with a few drops of alcohol to reduce the Cr0 3 to Cr 2 Cl . Dilute the dark green solution, add chloride of barium, and allow to stand for twelve hours. Then decant off all except a few drops into another vessel, and dilute the small residue with water. The least quantity of sul- phate of baryta becomes visible. To dehydrate the salt it is powdered, and, after having been kept near its fusing point for about ten minutes, fused at the lowest sufficient temperature. On freezing it breaks up into numberless little fragments, and thus spontaneously assumes a convenient form for being weighed out. The titre, as calculated from the synthesis, ought to be perfectly correct ; but unfortunately there is no method for proving the absence of surplus chromic acid in the salt, and besides, as a matter of principle, it is better to determine the iron-titre experimentally by means of standardized ferrous sul- phate. For this purpose about 1 grm. of the iron salt is weighed out exactly, dissolved in 70-100 cc. of water, previously mixed with enough of sulphuric acid to prevent the formation of basic salt in the oxidation. There is no need of any such large excess of acid as is required in the permanganate process. About 95 per cent, of the calculated volume of the bichrome solution are now added out of a burette, and a drop of the mixture TITRIMETRIC DETERMINATION OF IROX. 39 is taken out and tested with the i'erricyanide solution to make sure that it still contains unoxidized ferrosum, and strikes a blue colour with the reagent. More bichrome is now added in less and less instalments, until a drop, when taken out, no longer gives a greenish colour with ferricyanide. Supposing the oxidant at the end to have been added in instalments of 0*2 cc., and the ferrosum reaction to have ceased after addition of 32'2 cc., 32'litO'l is put down as the volume equivalent to the weight of ferrosum employed. Should the end point have been over- stepped, this is easily set right by adding a known, small, additional weight of iron salt ; the final titre is calculated from the mean of 3-4 well-agreeing titrations. In the determination of an unknoivn weight of dissolved fer- rosum, the finding of the end point is, of course, not so easy as in the case of the standardization ; but one soon learns to interpret the shades of blue or green observed in the drop tests, and to finish an analysis without wasting any appreciable quantity of the ferrosum over them. Iron given as ferric am must, of course, be reduced to ferrosum by means of sulphuretted hydrogen or sulphite of soda, as above explained. Penny recommends stannous chloride as a reducing agent (vide infra), but we do not like to use it, because every drop of stannous chloride that is added over and above the exact quantity needed of course vitiates the result. To obtain exact results we must work on only moderately dilute solutions, l>ecause the delicacy of even the ferricyanide test is limited, and, strictly speaking, the negative end-reaction comes out, not when the ferrosum is all oxidized, but when it is reduced to a small weight r per unit volume of mixture. If the iron solution is very dilute and not susceptible of concentration, this r must be ascertained synthetically and allowed for. Unlike the permanganate process, the bichrome one is not affected injuriously by the presence of chlorides. This is a great advantage, because hydrochloric acid is, so to say, the natural solvent for iron ores, while sulphuric acid, as a rule, attacks them only very sluggishly. THE STANNOUS CHLORIDE METHOD is founded upon the fact that strongly acid solutions of ferric 40 EXERCISES IN ANALYTICAL METHODS. chloride are most intensely yellow in the heat, and in the heat readily reduced (and decolorized) by stannous chloride. Fe 2 Cl 6 + SnCl 2 = SnCl 4 + 2FeCl 2 . Fresenius has established the conditions under which the reaction takes its normal course, and thus translated it into an exact method for the determination of ferricum. Requirements: (1.) A Standard Solution of Ferric Chloride. Pure ferric oxide is made by heating pure ferrous oxalate in a basin, and then in a platinum crucible in the presence of air until the weight is constant. (Make sure, by direct testing, that ferrous oxide is absent.) 1/20 Fe 2 3 = 8'00 grms. of such oxide are dissolved by prolonged digestion in strong, pure hydro- chloric acid, and the solution is diluted to 1 litre. (2.) A Standard Solution of Stannous Chloride. A weighed quantity of pure granulated tin is boiled with pure hydrochloric acid until a sufficiency of the metal is dissolved. The solution is decanted off, and the residual metal washed, dried, and weighed roughly, to see what weight of it has dissolved. For every 3 grms. of dissolved tin the solution is diluted (with boiled out water) to 1 litre to obtain a liquid which decolorizes about half its volume of standard iron solution. (3.) As a useful, though not indispensable, auxiliary, a solution of iodine in iodide of potassium, containing 3 to 5 grins, of free iodine per litre. It is standardized empirically against the tin solution. The ordinary standard solution (see Ex. 23) will do. Standardization of the Tin Solution. The tin solution readily takes up oxygen from the air, and its titre therefore is eminently variable. It is best kept in a tall bottle, provided with a narrow (clipped) syphon for letting out liquid, and an air inlet tube drawn out to a fine point. The titre then remains constant for at least the space of 2-3 hours. Even for a few analyses a large supply (say one Winchester quartful of solution) had better be made. In the course of time the solution becomes absolutely useless through formation of metastannic acid. Mulder recommends to connect the air space in the bottle permanently with a Kipp (or similar contrivance) yielding carbonic acid on demand ; but this complication is worth TITRIMETRIC DETERMINATION OF IRON. 41 employing only when the reagent is being used habitually day after day. To determine the titre, a known volume (say 20 cc.) of the iron solution is run into an Erlenmeyer standing on a flat sand-bath charged with perfectly white sand. By means of a bent glass tube connected with a Kipp discharging carbonic acid by means of a stiff joint, an atmosphere of carbonic acid is established in the flask, and (let us at once add) maintained to the end. The liquid is mixed with some pure hydrochloric acid, heated to incipient ebullition, and the tin solution is run in from a Mohr's burette, until the last drop, by converting an almost colourless into a perfectly colourless liquid, produces a peculiar transitory blue. The volume of the tin solution used is noted down, and the volume of the iron solution equivalent to one volume of tin calculated (from, of course, a number of agreeing titrations). The outlet tube of the tin burette should be made of a long narrow tube, and be bent thus, ' so that, while the outlet tube hangs in the titration flask, the burette is not being heated by the vapours. But it is not always (and not with all persons equally) easy to catch the end point. Hence, in general, it is an improvement to intentionally add a slight excess of tin, to allow to cool (in carbonic acid), to add starch solution, and then iodine solution, until the appearance of a permanent blue colour. To ascertain the saturation-ratio between the tin and the iodine solutions, under the circumstances of the analysis, undo the blue colour by adding, say, 2-3 cc. of tin, and then run in iodine again until the blue colour reappears. As the strength of the tin is variable, while that of the iodine solution is practically constant, the titre of the latter had better be referred to the iron solution. Supposing, for instance, we find 1 cc. of tin solution = 0*480 cc. of iron solution, 1 cc. =2*1 cc. of iodine solution, we note down that every 1 cc. of iodine is equivalent to 0'48 -f 21 = 0-229 cc. of iron. To determine an unknoivn weight x mgs. of ferricum, add pure, strong hydrochloric acid to convert it virtually into Fe 2 Cl c , 42 EXERCISES IN ANALYTICAL METHODS. and establish a strongly acid reaction, and then proceed as in the standardization of the tin. Supposing t cc. of tin and i cc. of iodine to have been used, 1 cc. of tm = k cc. of iron, and 1 cc. of iodine n cc. of iron solution, then we have the chemical equa- tion, t cc. of tin = x mgs. of f erricum + i cc. of iodine = x mgs. of f erricum -f ni cc. of f erricum solution = Id cc. of f erricum solution. Hence, arithmetically, x = (Id ni} x 8'0 mgs. of ferric oxide, or (kt ni) x 5*6 mgs. of f erricum. If the iron solution is given as ferrosum, it must be converted into f erricum, which is done best by means of a current of chlorine, followed, of course, by expulsion of the surplus chlorine by prolonged gentle heating. It stands to reason that the method may be reversed for the determination of an unknown weight of stannosum by means of ferric chloride ; but it is worth pointing out that it will not do in such a case to run the ferric solution into the stannosum solu- tion until a yellow colour begins to appear, because experience has shown that in this latter procedure the free oxygen (0. 2 ) dis- solved in the reagent-solutions takes part in the oxidation of the tin, while it does not in the normal process as described. Ex. 19. Analysis of Sulphate of Copper and Ammonia. CuS0 4 (NH 4 ) 2 S0 4 + 6H 2 0. DISSOLVE 250 grins, of pure blue vitriol and 132 grins, of pure sulphate of ammonia in 800 cc. of water at a boiling heat, filter while hot, and allow to crystallize. Dry the crystals in the air on filter paper. Take a sample of 3-4 grms., grind it up to make absolutely sure of its being homogeneous, and preserve it in a corked preparation-tube for the following exercise : Determination of the Copper. Dissolve, say, 1 grm. of the salt in 100 cc. of water, add 2 cc. of 20 per cent, sulphuric acid, h,eat to boiling, and pass sulphuretted hydrogen through the hot liquid until all the copper is precipitated. Collect the sulphide on a No. 5 filter, wash it with hot water, and dry it. Detach the precipitate from the filter, and incinerate the latter in a platinum spiral ; place precipitate and ash in a porcelain crucible provided SULPHATE OF COPPER AND AMMONIA, FIG. 4. with a perforated lid (see Fig. 4), and heat it over a Bunsen, while a pretty lively current of hydrogen gas is being passed through the crucible by means of a a porcelain (or hard glass) tube inserted into the perforation. The precipitate is thereby con- verted into a mixture of metal and sulphide. To produce a definite sulphide, allow to cool in hydrogen, add a pinch of pure sulphur and again heat in hydrogen, first gently, then more strongly, and lastly, for about twenty minutes, with the full heat of a good Bunsen. Now allow to cool in hydrogen, and weigh. The precipitate now is sure to at least substantially consist of Cu 2 S ; but it is better to repeat the last operation (i.e., addition of sulphur and ignition in hydrogen) until the weight is constant to within, at most, 1 mg. Cu 2 S x 0'9996 = 2CuO. To Determine the Ammonia. Construct the distillation appa- ratus represented in Fig. 5.* a is a Florence flask of 100 cc.'s capacity, communicating with a U-tube b, with inflated bulbs immersed in a cold-water bath. The U-tube, when filled up to the upper edge of the bulbs, should hold at least 120 cc. of water, c is a tower filled with glass beads or fragments of porcelain ; it communicates with an aspirator for reducing the pressure within the apparatus. Before using the apparatus make fcB sure of all the joints being air- F tight, by attaching a small 11 mercury manometer (Fig. 6) to the end of c, sucking out air so as to reduce the pressure by about J or 1 inch of mercury, closing the suction- tube, and * Another apparatus for the same purpose is described in Ex. 30 011 Kjeldahl's method. 44 EXERCISES IN ANALYTICAL METHODS. FIG. 6. allowing to stand for, say, ten minutes. If the mercury main- tains its level, the apparatus is sufficiently tight. If not, discover the leaky joints and replace them by better ones ; the application of sealing-wax, &c., cannot be toler- ated in a quantitative apparatus. Supposing the apparatus to be in a sound condition, place a weighed quantity (about 1 grm.) of the salt in the flask, and, on the other hand, run a mixture of 1 cc. of 20 per cent, hydrochloric acid and 5 cc. of water into the U-tube through the tower. Now pour on the salt about 40 cc. of 10 per cent, caustic soda (not potash), attach the U-tube and the aspirator to the tower, and, by sucking a little air out of the apparatus, reduce the pressure by about 2 inches of water, so that, if there should be a small unobserved leak left, only air can get in, but nothing can get out of the apparatus. Now apply heat to the flask, and distil until about three-fourths of the contents are driven over, when the ammonia can be assumed to have passed over completely. To determine its weight, trans- fer the contents of the U-tube and tower to a Berlin basin, add an excess of chloroplatinic acid (i.e., at least six parts of platinum for every one part of ammonia), and evaporate to dry ness on a water-bath. Allow the residue (PtCl 6 (NH 4 ) 2 + a;PtCl G H 2 ) to cool, treat it with ether-alcohol (a mixture of equal volumes of abso- lute alcohol and ether) until the excess of PtCl H 2 is dissolved, and then, by means of a wash-bottle charged with the same liquid, sweep the precipitate on a No. 4 filter, and wash it with ether-alcohol until the last runnings are colourless. Then dry the precipitate (at first outside the drying chamber, at a very gentle heat on account of the ether), fold it up in the filter, place it in this condition in a porcelain crucible, put on the lid, and next apply a very gentle heat until the charring of the paper is completed. Then gradually raise the temperature to ultimately a red heat. The precipitate is now reduced to spongy SULPHATE OF COPPER AND AMMONIA. 45 platinum, accompanied by charcoal, which coats the lid and the inside of the crucible. Burn away this charcoal as shown for chloride of silver in Ex. 12, and weigh the now pure metal. From its weight calculate the percentage of ammonia. Pt X 17508 = 2NH 3 . Instead of isolating and weighing the platinum of the chloroplatinate, we may, of course, weigh the precipitate itself, which may be done in two ways (1.) After having collected and dried the precipitate, we transfer the bulk of it to a tared crucible, and to recover the particles sticking to the paper replace the filter in the funnel, dissolve them by means of hot water, and evaporate the solution to dryness over the bulk of the salt by means of a water-bath. The residue is dried at 100 in an air chamber until constant. (2.) We collect the precipitate on a filter previously dried (until constant in weight) at 100 (and weighed), and at the end weigh filter and precipitate conjointly. As filter paper is very hygroscopic, both the filter and the filter plus precipitate must be weighed within two well-fitting watch-glasses held together by means of a suitable clip. The most exact modus operandi in this case is the following : Prepare two equal sized filters, put one on the left, the other on the right pan of the balance, and by means of a pair of scissors clip off bits from the apex of the heavier filter until it just balances the other. At this stage of the work the filters should not be handled directly, but by means of an ivory forceps. Put the entire filter inside the clipped one on the funnel, and collect the precipitate in this double filter. After washing and preliminary drying, dry the inner filter with the precipitate on one of the couple of watch- glasses, and meanwhile keep the tare filter in the same drying chamber in, say, a small porcelain dish. Immediately after the last weighing of the precipitate, take it and its filter out of the watch-glasses, substitute the tare filter as it comes out of the chamber, and weigh it along with the watch-glass couple to obtain the correct tare to be subtracted. PtCUNH^ x 0-07688 = 2NH 3 . The Sulphuric Acid is determined by precipitation with chloride of barium as usual, 46 EXERCISES IN ANALYTICAL METHODS. The Water is determined by difference. Theoretical Composition. Oxide of copper, CuO, - 79'34 1 9'85 Ammonia, 2NH 35 - 34'11 8'54 Sulphuric acid, 2SO S , = 16012 40'07 Water, 7H 2 0, - 126-03 31-54 399-60 100-00 SUPPLEMENTARY METHODS. Requirements: The standard muriatic acid of Ex. 10 and t standard solution of caustic alkali. Dissolve 30 grms. of pun caustic potash (purified by alcohol) in distilled water, and dilut< to 1 lit. To determine the strength of the solution, charge tw< burettes, one with the acid, the other with the alkali. Run int< a beaker about 20 cc. of the acid, colour it with a few drops o neutral litmus solution, and run in alkali until, by addition of th< last drop, the mixture, after stirring, is blue throughout its entin mass. That the mixture on standing gradually turns violet ii owing to the unavoidable presence of carbonic acid, whicl gradually acts on the litmus, and therefore must not be taker into consideration. Mark down the volumes of acid and alkal used as being equivalent to each other. Now undo the neutrality of the liquid by running in an additional small volume of acid again neutralize with alkali, and note the volumes of the reagent; used. So go on 5-6 times. From each couple of readings acid volume . calculate the ratio, ^ a ^ vo i um e :=: ' a mean of th< well-agreeing numbers. In order to determine the ammonia distil about 2 grms. of the salt with excess of caustic alkali collect the ammonia in a properly adjusted volume of youi standard acid, and, by titration, determine the acid left un saturated, taking care to determine the point of saturatior several times, as in the standardization of the alkali, to obtair as close an approximation to the truth as possible. Supposing the acid used was Ace and the alkali used was Pec, ther (A kP) cc. is the volume of acid equivalent to the ammonia CARBONATES. 47 and if 1 cc. of the standard acid was = to exactly 36*45 mgs. of real HC1, the ammonia amounts to (A kP)x 17'05 mgs.* Detei^niination of the Sulphuric Acid. Dissolve 23 grms. of salt in 200 cc. of water, pass H 2 S into the hot solution until all the copper is down, filter off the precipitate, boil down the nitrate to 100-200 cc., and, after cooling, determine the free acid by titration. This (assuming the salt to be a compound of CuSO 4 and (NH 4 ) 2 SO 4 ), gives the acid combined with the oxide of copper (by theory 20'03 per cent.). To determine the rest, add to the mixture a volume of additional standard alkali equal to a little more than was required for the first titration; boil down the mixture in a Berlin basin until the vapours no longer smell of ammonia, or colour turmeric paper. Then add standard acid out of the burette to produce a decidedly acid reaction, and lastly, titrate back with alkali until neutral. The additional alkali used, when corrected for the standard acid added, measures the sulphuric acid combined with the ammonia (again 20'03 per cent, by theory). Ex. 20. Analysis of Carbonates. EXAMPLE chosen: Iceland spar ; or pure precipitated carbonate of lime as used in Lawrence Smith's method of Silicate Analysis. (See Ex. 37). " The carbonic acid in this as in most other carbonates can be determined in two ways, namely, by decomposing a known weight by means of a suitable acid, and weighing the carbonic acid which goes off, either indirectly by identifying it with the loss of weight which apparatus, substance, and reagents con- jointly suffer in consequence of the withdrawal of the carbonic acid, or else directly by collecting it in an apparatus charged with potash or other chemical absorbent for carbonic acid, and * Xote to more advanced student*. To obtain the highest attainable precision, the ratio k must be determined in a solution as dilute, and containing about as much (NH 4 ) 2 SO 4 as the titration mixture hi the analysis did. Neutral (NH 4 ) 2 SO 4 is made by precipitating an ammoniacal solution of ordinary pure salt with strong alcohol, washing the precipitated crystal meal with alcohol, and drying in the air, 48 EXERCISES IN ANALYTICAL METHODS. ascertaining the increase of weight suffered by the absorption apparatus. The second method is susceptible of a higher degree of exactitude than the first ; hence, in the instruction concerning it, we introduce certain refinements omitted in the first, although equally needful theoretically on both sides. (1.) The Indirect Method. Construct an apparatus according to Fig. 7. a is a flask of about 100 cc.'s capacity; e, a U-tube charged with granulated porous chloride of calcium as used in organic analysis; b, a pipette of 15-20 cc.'s capacity, which must be provided either with a stopcock at b, or else with a small india-rubber cap at the top, and in any case with a small " thimble " below, which is tied to a contracted portion of the stem by means of a thin platinum wire fused to the edge of the thimble ; / is a very carefully fitted in and perfor- ated cork or india-rubber stopper. To execute an analysis, place, say, 1 grm. of substance in the flask, and add about 5 cc. of water. Charge the pipette with about 15 cc. of 10 per cent, hydrochloric acid, and put the apparatus together as shown by the figure, taking care that none of the acid gets in contact with the substance before it is wanted. Now, in order to cause the apparatus to quickly assume a constant weight, plunge it into a mass of water of the temperature of the balance-room up to almost the rim (without wetting the cork), wipe off the water, suspend the apparatus at the balance, and determine its weight repeatedly at intervals of about five minutes until two consecu- tive weighings agree to within 1 mg. While the apparatus is being left to itself, both the top end of the pipette and the exit end of the U-tube are closed with india-rubber caps ; the latter cap, however, must be removed before the weight is determined. Decompose the carbonate by gradual addition of acid from out of the pipette, then gently heat the contents, finally to near boiling, and, by gently shaking the flask round, expel the FIG. 7. CARBONATES. 49 dissolved carbonic acid ; lastly, remove the cap of the pipette and suck at the g end, by means of an india-rubber tube, until all the liberated carbonic acid can be assumed to be displaced by air. Now put on the caps, allow the apparatus to cool, and, in order to bring its outside as nearly as possible into the same condition as it was at first, keep it for a sufficient time immersed in a water-bath of the temperature of the balance-room. Then wipe off the water, remove the outlet cap, and determine what the weight now amounts to. The rest requires no explanation. (2.) For executing the direct method rig up the apparatus (Fig. 8). The flask a should be of 100-150 cc.'s capacity. The stem of the funnel should extend to near the bottom of the FIG. & flask ; the outside portion of it is shaped so that, supposing the stem to have once been filled with liquid, this liquid will, under all circumstances, be partly retained by the tube so as to prevent egress of gas. The flask, through an ascending, rather wide tube, communicates with a small U-tube 6, charged with pumice impregnated with dehydrated sulphate of copper* to absorb HC1 vapours. A thimble placed in the U-tube b (Fig. 9) below the entrance end of the tube from the flask serves to collect that part of the water and hydrochloric acid which has condensed * Soak fragments of pumice in solution of sulphate of copper, and heat them in a platinum dish until white. E 50 EXERCISES IN ANALYTICAL METHODS. during the descent of the gas through that tube. From the sulphate of copper tube (supposing carbonic acid to be evolved in the flask) the gas passes next through a chloride of calcium U-tube c to be dried, and thence through the Liebig bulbs d charged with liquid caustic potash (1 part of sticks to l^ parts of water), where the carbonic acid is almost completely absorbed ; but a small quantity escapes absorption, and besides, some of the water of the ley is always carried away as vapour by the un- absorbed part of the gas (the air). To condense both is the function of the little U-tube e, of which the limb nearer to the potash-bulb is charged with granulated soda-lime, while the other is filled with chloride of calcium of the same kind as c. From the soda-lime tube the gas passes through a pro- tection tube / to an aspirator adjusted so that when the flask communicates freely with the atmosphere a slow current of air bubbles through the potash- bulbs, supposing these to be filled and placed pro- perly. The gas must come in by the larger to go FIG. 9. ou jjy ^g smaller of the two terminal bulbs, and the three-bulbed basis of the apparatus slant slightly upwards so that the gas ascends in going through the three bulbs. To fill the bulbs, put a sufficiency of the ley into a crucible, dip in the entrance (i.e., the major bulb) end, and, while taking care to let the base slant upwards in the sense explained, suck at the outlet end by means of a long enough india-rubber tube to enable you to see what you are doing, until a slow current of air, sucked through after the ley, sends a little of the latter up into the outlet bulb (see Fig. 10). The inlet end which dipped into the ley must be carefully cleaned, the outside in an obvious manner, the inside by means of filter paper rolled round a thin knitting wire. To make sure that no ley hangs about the outside of the apparatus, finish up by dipping it as far as safe CAR BON A IKS. 51 into a mass of water, and then wiping it dry by means of a soft towel or linen handkerchief. Supposing the water to be of the temperature of the balance-room, the immersion process is always useful as bringing the bulbs more quickly to a constant weight. The aspirator arrangement is only meant to make assurance doubly sure : don't trust to its compensating for actual leaks, but test your apparatus before use by means of the mercury manometer represented by Fig. 6, p. 44- Supposing this point to be satisfactorily settled (the potash-bulb and soda-lime tube are supposed to have been care- fully tared before), the analysis itself is easy. Decompose your carbonate by gradual addition of 10 per cent, hydro- chloric acid, expel the dissolved carbonic acid by cautious heating, and then, after having made sure that the pressure in the flask is less than one atmo- sphere, suck air through the apparatus by means of the aspirator until all the carbonic acid can be assumed to have been swept into the potash-bulb. A volume of air equal to, say, three times the empty space in the apparatus up to the potash-bulb is sufficient, which, in the case of your apparatus, if in accordance with the figure, means half a litre of air. Half a litre of air, even in a laboratory, is not likely to contain more than half a mg. of carbonic acid : yet it is Letter to purify the air before allow- ing it to enter the apparatus by means of a soda-lime tube attached to the funnel by means 52 EXERCISES IN ANALYTICAL METHODS. % of a well-fitting cork. The sulphate of copper tube in the present case, in the hands of a careful operator, is almost orna- mental ; but in an apparatus meant to be preserved, it is as well to have it so as to be at liberty, where necessary, to use stronger acid. Whenever the bases in the carbonate form soluble sulphates, dilute sulphuric acid should be employed instead of hydrochloric acid, on account of its absolute non- volatility from aqueous solu- tions. The sulphate of copper tube then, of course, can be dispensed with. An improvement upon the apparatus for the direct method (introduced by Classen) is to insert an inverted condenser between the flask and the first U-tube, as shown by Fig. 11. The condenser enables one to dispense with the sulphate of copper tube, 'because it condenses all the hydrochloric acid along with the water as long as the acid used does not contain more than 20 per cent, of real HC1. It comes in particularly useful in the analysis of dilute solutions or of difficultly decomposible carbonates. Theoretically, Iceland spar contains 44'00 per cent, of carbonic acid (CO,). Ex. 21. Separation of Iron, Manganese, and Calcium. SUBSTANCE. To obtain a convenient substance-solution of known composition, dissolve, say, 300 mgs. of pianoforte wire in hydro- chloric acid, and drop in, say, 50 cc. of standard permanganate of potash as used for iron titrations. On the other hand, dis- solve, say, 100 mgs. of precipitated carbonate of lime in hydro- chloric acid, and mix the solution with the first. Everything, of course, to be measured or weighed exactly, so that the absolute weights of the components can be calculated* conveniently in terms of FeO, MnO, and CaO. The solution is meant to contain a remnant of ferrosum. Separation. Begin by fully peroxidizing the iron by heating * Constants needed for the calculations : Mn 2 Or K-2 = 31 6 '3; 10 -Fe = 560 '0; 2 Mn = 110 ; CaO CO 2 = 100 ; CaO = 56. SEPARATION OF IRON, MANGANESE, AND CALCIUM. 53 with a few granules of chlorate of potash. Expel the chlorine oxides by gentle heating, and next boil for some time, so that any manganic chloride that may be present assumes, through the action of the hydrochloric acid the state of manganous salt. Then allow to cool, and dilute to about half a litre ; add sal- ammoniac if necessary, and then ammonia until it predominates. From the moment when a permanent precipitate begins to form, the ammonia should be added by drops and with constant stirring. This and, more still, the presence of a considerable quantity of sal-ammoniac, is an essential condition of success. As soon as an alkaline reaction is established, at once heat up the mixture, and boil it until the vapours no longer smell of ammonia. Then filter hot, and wash with hot water. The pre- cipitate, in addition to all the iron, contains part of the man- ganese and calcium. To eliminate them, re-dissolve it in hydro- chloric acid, and again apply the process of separation just explained ; or else, in order to learn something new, use The Acetate Method, which is as follows: After diluting to about half a litre, cautiously add ammonia in the cold until the locally-formed precipitate is slow in re-dissolving, and the liquid has become dark brown. Then add (neutral) acetate of ammonia (not too little), heat to boiling, and filter hot* Wash the precipi- tate without intermission with hot water containing a little of the acetate until it is pure. Ignite the precipitate, and weigh it as Fe 2 O 3 , but do not forget to test it for manganese with the blow-pipe. (If manganese be present, the precipitate must be re-dissolved, and the remnant of manganese recovered, weighed, and brought into account). To determine the Manganese and Calcium, concentrate the filtrate by evapora- tion to about half a litre, neutralize (exactly) with ammonia, and add a slight excess of sulphide of ammonium to precipitate the manganese as sulphide. The manipulation of this precipitate presents certain difficul ties, which, however, according to Fresenius, are overcome by taking care to have a sufficiency not too much, though of sal-ammoniac in the mixture, to use colourless, or * Prolonged boiling brings the precipitate into a condition in which it refuses to " filter " properly. 54 EXERCISES IN ANALYTICAL METHODS. almost colourless, sulphide of ammonium (instead of the ordinary yellow reagent), and otherwise operating as follows : The pre- cipitation is effected at a gentle heat in a flask (conveniently an Erlenmeyer), which should not be much larger than necessary ; boiled-out warm water is then added so as to almost fill the flask, which is corked and allowed to stand in a warm place for 24 hours or longer until the sulphide has settled completely. The clear supernatant liquor is now syphoned off into another flask and kept there, meanwhile, protected against access of air. The precipitate is mixed with warm air-free water containing a little sal-ammoniac and sulphide of ammonium, the flask corked up again and put aside to allow the contents to clear up. This second liquor is syphoned off and preserved like the first. With a large precipitate the process is repeated once more (with plain sulphide of ammonium this time) ; but in our case one washing by decantation will do. Supposing it to be effected, the wash- waters are filtered first (beginning with the more dilute one), and only then the precipitate is collected on the filter, and washed with warm water containing a little sulphide of ammonium. According to Finkener, the precipitation of the manganese can be effected promptly and successfully by adding the sulphide of ammonium to the neutralized liquid at a boiling heat, boiling for about ten minutes, and filtering off through a double filter. The precipitate is washed with hot water containing sulphide of ammonium. According to the Author's experience (see his "Manual of Analysis," pp. 96 and 97), precipitated manganese sulphide is decomposed by boiling sal-ammoniac ; hence, the volatilized sulphide of ammonium must be replaced from time to time by fresh reagent. Of our various methods for bringing precipitated sulphide of manganese into a weighable form, no one is quite satisfactory. Perhaps the best gravimetric* method is Rose's, who, after incineration of the filter, mixes precipitate and filter ash with sulphur, and ignites in hydrogen, &c., as explained for copper, Ex. 19. The thus ignited precipitate is MnS. MnS x O8155 = MnO, and MnS x (V6318 - Mn. * More advanced students may consult the section on manganese under Cast-iron. " SPATHIC IRON ORE. 55 In the filtrate from the sulphide of manganese, the calcium, after due concentration, is determined by precipitation with oxalate of ammonia in the presence of free ammonia in the heat. The mixture must be allowed to stand hot until the oxalate has settled completely ; it is then filtered off, washed hot, ignited, and weighed as CaO. (See Ex. 16). The student, after having learned this method of separation with solutions of known composition, may apply it in the com- plete analysis of a (manganiferous) spathic iron ore. Ex. 22. Spathic Iron Ore. Impure MOCK, where M = chiefly Fe, but in general includes Mn, Ca, Mg. The quantitative should be preceded by an exhaus- tive qualitative analysis. Powder a supply of the ore, dry it at 100C., and analyse it in this condition. For the determination of the metals, prepare a standardized SUBSTANCE-SOLUTION as follows : Dissolve 5 grins, of sub- stance in hydrochloric acid, and next evaporate to dryness to render the silica insoluble ; then separate out the silica (plus insoluble silicates) as shown in the Exercise on Silicate Analysis. The impure silica is filtered off, ignited, weighed, and calculated as "gangue-." The filtrate is diluted to 500 cc. or other con- venient volume, and aliquot parts are used for the determina- tions. We are in the habit of doing these fractionations both by volume and by weight, and of calculating from the recorded weights (see Exercise on Sea Water) ; but in this case volume- measurement will suffice. The Metals are determined as explained in the preceding Exercise, it being remembered, however, that the sesquioxide precipitate may contain alumina, which, if present, must be eliminated and determined as explained in Ex. 17, and that the oxide of iron may contain silica. This silica remains when the ignited precipitate is dissolved in strong hydrochloric acid, and thus becomes weighable. 56 EXERCISES IN ANALYTICAL METHODS. As the presence of alumina involves a second precipitation of the oxide of iron, in its presence one precipitation of the mixed sesquioxides will do ; but the nitrate from the (re-precipitated) alumina-free ferric oxide must, of course, be worked up along with the first filtrate from the sesquioxides (for manganese, lime, and magnesia). The magnesia is determined in the filtrate from the oxalate of lime by means of phosphate of ammonia, as explained in Ex. 15. For the Carbonic Acid, see Ex. 20. To check the results, determine the iron titrimetrically in an aliquot part of the substance solution, and ascertain by direct experiment what percentage of fixed residue the ore leaves when ignited finally in the presence of air until the weight is con- stant. This percentage must agree with that calculated from the separate determinations, assuming the metals to remain as Fe 2 Oo, Mn 3 4 , CaO, MgO respectively. Note. The united ammoniacal filtrates (containing the man- ganese, &c.) may contain such masses of ammonia salt that it is necessary to remove these by evaporation to dryness and cal- cination of the residue before the determination of the protoxides is proceeded with. The calcined residue is dissolved in water, with addition of hydrochloric acid. Ex. 23. -Preparation of a Standard Solution of Iodine. A SOLUTION of iodine in iodide of potassium which contains an exactly known weight of free iodine per cc. is susceptible of manifold applications as a titrimetric agent. For its preparation the most exact method is a Direct Quantitative Synthesis. The necessary pure iodine is best prepared by Stas' method. A quantity (say 50 grins.) of commercial re-mblimed iodine is dissolved by grinding it up in a mortar with 75 grms. of water, and the least sufficient quantity about 35 grms. will do of iodide of potassium. As the solution is almost opaque, allow to stand for a time, decant off' the liquor, and, if any undissolved iodine is left, bring it into solution by means of the least sufficient quantities of STANDARD SOLUTION OF IODINE. 57 iodide of potassium and water, filter the solution through a bunch of glass wool stuck into the throat of a funnel, and dilute it with ten times its volume (or if necessary, more) of water to precipitate as large a proportion as possible of the dissolved iodine. The precipitated iodine is quite free from bromine and chlorine ; these remain in the mother -liquor. Wash it by decantation with pure cold water, collect it on glass wool, allow to drain, and next dry it further on a plate of unglazed porous stoneware. To dry the preparation further, keep it in a shallow basin over perfectly anhydrous nitrate of calcium under a bell- jar for a number of days. For its final purification distil it from out of a very short-necked retort heated by means of an air-bath which quite encloses it. The first instalment of vapours often contains a little water, and therefore is collected separately. As soon as the water is gone, the rest of the vapours are allowed to fall into a large wide-necked Erlenmeyer, where they con- dense, assuming, as a rule, the form of a semi-fused compact mass. This is taken out, ground up coarsely, and dried finally over fresh nitrate of calcium. By re-subliming (the compact mass) in instalments from out of a watch glass (which is being heated very slowly on a piece of sheet-iron or asbestos paste- board) into a beaker placed over the watch glass, the iodine is obtained in isolated crystals, which are more easily dehydrated finally than a powder prepared in a mortar. Small quantities of precipitated iodine are best sublimed at once in this way, the first instalment of vapour being allowed to go off with the moisture. After having produced a supply of pure anhydrous iodine, it is expedient to at once weigh out, say, 3-4 portions exactly in glass tubes and seal them up for future use. If there is a sufficient quantity, prepare from it a quantity of exactly standardized solution by weighing out some 5 grms. exactly, dissolving them in strong solution of 10 grms. of iodide of potassium, and diluting to 1000 cc. Such synthetically-produced iodine solution, however, is expen- sive ; hence it had better be used only for the exact standardiza- tion of a larger supply of roughly standardized solution made either from ordinary iodine weighed out approximately, or from 58 EXERCISES IN ANALYTICAL METHODS. the mother-liquor obtained in the Stas process of purification by adding some more iodide of potassium to it to preclude future precipitation of iodine, standardizing it roughly in the first instance, and diluting with the calculated volume of water. For an exact standardization, prepare first an auxiliary solution of thiosulphate of soda by dissolving 10 grms. of the pure salt Na 2 S 2 O 3 5H 2 in water, and diluting to 1 litre. When this solu- tion is mixed with iodine solution, the salt is oxidized into tetrathionate in exact accordance with the equation 2Na 2 S 2 3 + I 2 = 2NaI + Na 2 S 4 O 6 . On titrating a given volume of the iodine solution by added thiosulphate solution over a white underground, the end point of the reaction can be seen very sharply, without the use of starch solution, by the disappearance of the last trace of yellow colour. Of course, the addition of a little clear starch solution near the end of the reduction facilitates the recognition of the end point. Supposing unit volume of thiosulphate requires for its oxidation, according to one experiment, n volumes of the normal iodine solution, and, according to another, n' cc. of the iodine solution to be standardized, both contain the same weight of free iodine. It stands to reason that a directly weighed-out small quantity of dry iodine dissolved in iodide of potassium solution and con- veniently diluted may be substituted for the synthetically- standardized solution. The titre is noted by stating the free iodine per cc. as a multiple of I = 126'85 mgs. The iodine solu- tion, when kept in a good glass-stoppered bottle in a cool place and away from direct sunlight, retains its titre for a long time. It must not be measured in a Mohr's burette, because the vulcanized india-rubber of the pinch-cock reacts with the free iodine. The thiosulphate solution is not quite so stable as the iodine solution ; but if made from pure salt, it changes only very slowly. After long standing it deposits sulphur. As soon as this change shows itself, the solution must be thrown away and renewed. Examples of Applications. By means of standardized iodine solution we can determine (1.) Arsenious Acid given as dissolved alkaline arsenite. The solution must be free from caustic alkali, and should not contain STANDARD SOLUTION OF IODINE. 59 an immoderate quantity of carbonate. Dilute moderately, add starch solution, and titrate with iodine, stirring up constantly, until the liquid becomes permanently blue. Then add sesqui- carbonate of ammonia, which, in general, will cause the colour to disappear; if so, drop in more iodine solution until the mixture is blue and retains the colour on addition of a little more of the sesquicarbonate (Mohr). If enough of the latter is present to ensure to the mixture an alkaline re-action even at the end, the process goes on in accordance with the equation As 2 O 3 + 2H 2 O + 21, = 4HI + As 2 5 . (2.) Antimonious Oxide Sb,0 3 , given as dissolved tartar- emetic, or in similar conditions, behaves like arsenious acid. (3.) Dissolved Thiosulphate. Even in solutions acid by hydro- chloric acid, the reaction goes on, as above explained, if the solution is sufficiently diluted, and the thiosulphuric acid formed by the hydrochloric acid has had no time to suffer its character- istic spontaneous decomposition into sulphur and sulphurous acid. (4.) Dissolved Sulphurous Acid, H 2 S0 3 . The solvent must be air-free water, and the solution so dilute that it contains no more than about three parts of S0 2 per 10,000 parts (Bunsen), or else the sulphuric acid formed H 2 S0 3 + H 2 + I 2 = 2HI + H 2 S0 4 acts backwards on the 2HI produced in the sense of the same equation read from the right to the left. Hence, any free sul- phuric acid present from the first (in calculating the water to be added) must be taken into account as representing so much sulphurous acid. SO 2 per H 2 S0 4 . By the combined application of iodine and thiosulphate we can determine (1), and obviously, free iodine ; (2) free chlorine or bromine, because these, by mere addition of excess of iodide of potassium solution, furnish their equivalent of free iodine ; (3) hypochlorous acid. Every HC10, from an excess of dis- solved hydriodic acid (KI + enough of HC1), liberates I 2 ; hence, conversely, every I 2 = 2 x 126'85 mgs. of iodine liberated indicate HC1O mgs. of hypochlorous acid, or O = 16 mgs. of active oxygen, or Cl, = 70'9 mgs. of active chlorine. 60 EXERCISES IN ANALYTICAL METHODS. (5.) The loosely combined part of the oxygen in any peroxide which, either when brought into contact with iodide of potassium solution liberates a quantity of iodine equivalent to the loose oxygen, or which, when distilled with hydrochloric acid, liberates a similar equivalent of chlorine. Examples : Mn0 2 , Cr0 3 , &c. Ex. 24. Analysis of Bleaching" Powder. Exs. 24, 26, and 27 include cases of iodine titration. . The student may single these out and practise them before taking up the Exercises in their entirety. IN the analysis of a bleaching powder the most important item is the proportion of real Cl Ca OC1 ( = caCl + caOCl) which is contained in it. In practice, however, it is customary not to report the percentage of CaOCl 2 , but that of the chlorine equi- valent to the oxygen of the hypochlorite, which happens to be equal in weight to the chlorine in the CaOCl.,. Of the numerous methods for the Determination of the "Active" Chlorine, we select the following three. Whether you choose one or another, begin by weighing out 10 grins, of the "bleach," grind- ing them up in a porcelain mortar with a good spout with water so as to form a uniform paste, and diluting to 1000 cc. Aliquot parts of the turbid solution (to be referred to as the " liquor "), rendered uniform immediately before by turning the close flask upside down two or three times, are measured off for the several determinations. (1.) Bunsen's Method* Measure off 10 cc. of liquor = 100 ings, of substance, and add about 1 grin, of solid iodide of potassium, which dissolves readily with liberation of some iodine. To libe- rate all the iodine due (0 + 2HI = K 2 + I 2 ), acidify with hydro- chloric acid, dilute to about 50 cc., and titrate with thiosulphate solution previously standardized by means of standard iodine solution. The method is easy and exact ; but the imperfect * As modified by Rud. Wagner. In the original method very dilute sulphurous acid served as the reducing agent. BLEACHING POWDER. 61 stability of the thiosulphate is an inconvenience. Hence, when- ever bleaching powder tests have to be carried out frequently, (2.) Penot's method deserves the preference. This method measures more directly the oxygen of the hypochlorite by ascer- taining the weight of arsenious acid, As 2 O 3 , which it is capable of converting into arsenic acid, As. 2 5 . 1/40 x As 2 3 = 4'95 grms. of pure powdered white arsenic are placed in a litre flask along with about 11 grms. of. pure carbonate of soda, Na 2 CO 3 , or 30 grms. of the dekahydrated crystals, and some water, and the whole is heated over a water-bath until the arsenic is dissolved. A short- necked funnel suspended in the neck of the flask prevents loss by spirting. The (cooled down) solution is diluted to 1000 cc. Every 1 cc. requires 1/10 01 = 3*545 mgs. of chlorine for its exact oxidation,* a convenient litre for general purposes, f Where the solution is used only for " bleach " testing, it is, of course, better to adjust it so that 1 cc. = exactly (say) 2 mgs. of chlorine. 4 Cl X 1 "3974 = As 2 O 3 . Another requisite is iodide of potassium and starch paper, prepared by soaking strips of Swedish filter paper in a very dilute, clear starch solution con- taining some iodide of potassium, and allowing them to dry in a pure atmosphere. To make an analysis, measure of, say, 20 cc. of liquor = 200 mgs. of substance, and titrate with the arsenite until you have reached the exact point when a drop of the mix- ture no longer produces a blue stain (indicative of unchanged RC10) on a bit of the test paper placed on a porcelain slab. Should you have overstepped the end point, this can be set right by either adding another cc. of liquor, and resuming the addition of arsenite, or by adding sesquicarbonate of ammonia and some * The student, for his information, should titrate the arsenite, after addition of sesquicarbonate of ammonia, by means of his iodine solution, and compare the analytical with the synthetical result for the arsenious acid. The two results should agree fairly, but they will not agree absolutely, because the respective chemical equations, though right enough as such, as theories of the processes of I it ration are only approximately correct. This remark applies to the majority of titrimetric processes ; hence an obvious general rule for the standardization of a solution as a special reagent. For an example : iodine solution intended for determinations of arsenious acid had better be standardized with arsenious acid, although the iodine method is in itself more exact. f Standard arsenite has a standing as a general reducing agent in titrimetric analysis. 62 EXERCISES IN ANALYTICAL METHODS. starch solution, and titrating back with iodine solution.* But with some practice it is easy enough to catch the end point to within dbO'l cc., and even less. (3.) Dufios method, although antiquated now, affords an in- structive exercise to the student. It is founded upon the fact that hypochlorites in aqueous solutions are readily reduced by sulphurous acid with formation of sulphuric acid. CaCl 2 + S0 2 = CaCl. 2 + SO 3 . The sulphuric acid is determined gravimetrically by means of chloride of barium. Both it and the sulphurous acid are embodied in one reagent, prepared by dissolving 40 grins, of chloride of barium in 700 cc. of boiled-out water, adding 300 cc. of freshly made saturated sulphurous acid water, and allowing the mixture to stand in a well-closed bottle until the unavoidably formed sulphate of baryta has settled completely. By means of the syphon arrangement described for stannous chloride in Ex. 18, p. 40, it is easy to draw off portions of the clear liquid for immediate use. According to Sims, 1 cc. of sulphurous acid water saturated completely at 20C. contains 0104 grins, of S0 2 ; hence 10 cc. of our reagent should contain about 400 mgs. of chloride of barium = l'65xBaCl 2 mgs., and 312 mgs. of sulphurous acid = 5xS0 2 mgs., or amply enough of barium for, say, 71 mgs. of chlorine equal to about 200 mgs. of bleaching powder, and far more than enough of sulphurous acid ; but an ample excess should be employed in the analysis. To make a determination, run a sufficiency of clear reagent into an Erlenmeyer previously filled to overflowing with car- bonic acid gas, add the predetermined volume of bleach liquor, and allow the action to go on for a while in the cold, maintaining an atmosphere of air-free carbonic acid ; then dilute moderately with boiled-out water, and add some hydrochloric acid to prevent permanent precipitation of sulphate of lime, and at last boil off the surplus sulphurous acid. Collect the sulphate of baryta, and weigh it.f BaS0 4 x 0'3040 = C1 2 and BaS0 4 x 0'06859 - 0. * Which had better be standardized empirically by means of the arsenite solution. t Should the bleach contain sulphate, the BaSO4 corresponding to it must, of course, be determined, and deducted as a correction. ULEACHING POWDER. 63 The other Components. Tin- Chlorate. Most bleaching powders contains a little chlorate. For its exact determination, however, we have no really satisfactory method. Duflos', it is true, gives the total oxygen present as RC1O or RC10 3 , and, were we quite sure that Penot's gives only the hypochlorite oxygen, the chlorate could be obtained by the combined application of the two. But, apart from the uncertainty of this assumption, it is questionable whether the two methods conjointly afford a sufficient degree of precision for the determination by difference of the small quantity of chlorate under analysis. The student should try and see what he can do. The Total Chlorine. By boiling the liquor with excess of sulphurous acid solution under an inverted condenser, it is easy to reduce all the oxidized chlorine to chloride. The surplus sulphur dioxide is removed by heating the sufficiently -diluted mixture on a water-bath, and the chlorine determined as usual. By deducting the chlorines of the hypochlorite and chlorate, we obtain the chlorine which was present from the first as chloride. The Carbonic Acid is liberated by means of a mixed solution of hydrochloric acid and enough of stannous chloride to fix the active chlorine as SnCl 4 , and determined by the direct method. (See Ex. 20). The Sulphuric Acid, Silica, "Sand" A known weight of substance is decomposed by hydrochloric acid in the heat, the solution evaporated to dryness, and the silica (and "sand") elimi- nated as in a silicate analysis, care being taken not to expel any of the sulphuric acid. The impure silica is weighed as such ; a separation of it into silica proper and sand is not worth the trouble. In the solution the sulphuric acid is determined as usual. An aliquot part of the hydrochloric solution might serve for the determination of the Linie and Magnesia (and the traces of iron and alumina which are usually present) ; but on account of the great preponder- ance of the former, an exact separation can be effected only by means of sulphuric acid, which may as well be applied to an aliquot part of the liquor. A measured volume of it is reduced 64 EXERCISES IN ANALYTICAL METHODS. by means of sulphurous acid, which just produces the sulphuric acid needed for the CaCL 2 ; to satisfy the surplus lime a little more sulphuric acid is added, the whole reduced by evaporation to a small volume, and the sulphate of lime brought down, after cooling, by adding from a half to one volume of alcohol, and allowing to stand over night. The sulphate of lime is filtered off, washed first with aqueous, then with strong alcohol and ignited, to be weighed as sulphate of lime. CaS0 4 xO'4116 = CaO. From the alcoholic filtrates the alcohol is expelled by evaporation in the presence of enough of water to prevent etherification. The iron is then peroxidized with a little chlorine water, precipitated (with the alumina) by ammonia, the excess of the latter boiled off, and the precipitate filtered off to be weighed after ignition. From the filtrate any remnant of lime is elimi- nated by oxalate of ammonia, and the magnesia then determined as usual. The Water may be determined by means of the direct method, as given in Ex. 28, for ferrous sulphate. The column of hot chromate of lead retains the chlorine carried away by the steam ; but it is easier, of course, to determine it by difference. In reporting, give first the several components as determined ; then express each as a multiple of the respective formula value, e.g., the lime as x X 56, the total chlorine as y x 35*45, the active oxygen as 0x 16 parts ; then combine the chlorous and basilous radicals into multiples of CaOCl 2 , CaCL, CaS0 4 , &c., to obtain the percentages of chloro-hypochlorite, chloride, sulphate, &c. A novel method, invented by Lunge, for the determination of the active oxygen in bleaching powder is appended as affording an example of what is called " gas- volumetric analysis." Ex. 25. Gas- Volumetric Analysis, LUNGE'S method is founded upon the fact ascertained by him that the hypochlorite, when brought into contact with peroxide of hydrogen in aqueous solution, is reduced by and with the reagent thus : RC10 + HA = RC1 4- H 2 O + 0., The process is carried out in a flask whose atmosphere communicates directly with that GAS-VOLUMETRIC ANALYSIS. 65 of a gas-measuring tube, and the oxygen measured as an increase of volume in the atmosphere common to both (" gas-volumetri- cally "). Lunge recommends to mix 10 grms. of bleach with enough of water to produce 250 cc. of (turbid) liquor, and to use 5 cc. of the latter, representing 200 mgs. of substance. Supposing these to contain 71 = C1. 2 mgs., the active oxygen amounts to J O 2 mg. = 11 cc. about. The same quantity of oxygen must be con- tributed by the reagent, which in its customary form contains ten times its volume of loose oxygen. Hence, 2 cc. of reagent (instead of I'l) are ample. The apparatus (Fig. 12) consists of a flat-bottomed flask of, say, 50 cc.'s capacity, wide enough in the neck to enable one conveniently to slide in a flat-bottomed thimble and let it stand upright in the bulb of the flask. By means of a well-fitting india-rubber cork the flask can be made to communicate with an inverted Mohr's burette, the lower end of which communi- cates through an attached narrow india-rubber tube with a reservoir. The bleach liquor is run into the flask ; the reagent placed in the thimble within the former. The burette is charged with water (or mercury) from out of the reser- voir, and things are so arranged that, the burette (i.e., cock k) being open, the two liquid levels coincide nearly with the top mark of the graduation. The flask is now joined on air-tight after having been immersed in a large beaker containing water of the temperature of the laboratory, the reservoir placed so that the pressure inside is exactly at that of the atmosphere, and the level in the gas-measurer noted down. F 66 EXERCISES IN ANALYTICAL METHODS. Now the cock h is closed, the flask tilted over, so that liquor and reagent mix and the reaction allowed to take its course, care being- taken to always keep the pressure at that of the atmosphere approximately at least. After 1-2 minutes the reaction is com- pleted. The pressure is now adjusted to exactly that of the atmosphere, and the increase of gas volume ascertained to obtain the volume of the oxygen. Lunge recommends to collect the gas over mercury; water, we think, will do equally well. If water be used, it is an improvement to provide the Mohr burette with a good Erdmann float d (we mean one which fits close, and yet does not jam) ; the loss by gas-absorption, then, is reduced to very little. A further improvement is to enclose the burette in a water jacket kept at the same temperature as that in the beaker. After establishing equilibrium of temperature, we read the temperature t, and the barometer (let its height, reduced to 0C., be = B mm.), and then we have all the data for calculating out the result. Sup- posing the tension of vapour of water at t to be =p ; the gas volume reduced to T = P ; say to T = 1 and P = 1 mm., is -* f B ~,^ = VQ cc. ; the weight of the oxygen = v x 0*03212 x 16 mgs. ; and consequently as 2 of oxygen evolved corresponds to C1 2 of active chlorine, the weight of the latter is v X 0'03212 x 35'45 mgs., and the percentage x of the latter (as 200 mgs. of sub- stance were used) half as much, i.e., x = 0'5693 x v . Supposing it were proved that the peroxide of hydrogen does not touch the chlorate, the residue from the Lunge test might serve for the direct determination of the oxygen in the latter by Duflos' method. All that would be necessary would be to destroy the surplus peroxide by means of platinum black, or (after diluting and acidifying) by means of the exact quantity of permanganate, and then to apply the Duflos reagent in an atmosphere of car- bonic acid, as above explained. Perhaps a better, and certainly a simpler, method would be, after (or perhaps without) destruction of the surplus peroxide, to next eliminate the chloride chlorine by means of nitrate of silver in the cold, and in the filtrate determine the chlorate chlorine by reducing the C10 3 with sulphurous acid. The Cl of the C10 3 assumes the form of chloride of silver, which is easily NATIVE OXIDE OF MANGANESE. 67 freed from the admixed sulphite by dilute nitric acid, and thus made fit for the balance. Ex. 26. Native Oxide of Manganese. THE variety of minerals which our heading includes, from the standpoint of the analyst, can all be viewed as compounds Mn(X + x MnO + 2/RO, where R may include Fe, Co, Ni, Ba, Ca, Mg, H 2 , but be absent ; the value of x varies from 2 (in Hausmannite, Mn 3 O 4 ) to nil (in pure pyrolusite, MnO 2 ). METHODS OF ASSAYING. " Manganese," as it is called in commerce, is used chiefly as a material for generating chlorine, and its value as such depends substantially, if not entirely, on its percentage of actual or potential Mn0. 2 . But the most exact determination of this per- centage, in reference to a given cargo or consignment of ore, is of little value unless accompanied by two determinations of the moisture, namely, one with a large average sample of the ore, pounded up quickly though not very finely, and another with the finely-powdered analyst's sample as it goes to the chemical balance. These two determinations for commercial purposes must be carried out by drying both samples in a conventionally- fixed (though arbitrary) manner, and determining the loss of weight as representing the moisture. The reason for this is, that even the air-dry ore always contains a greater or less pro- portion of hygroscopic water, which is subject to considerable spontaneous variation. According to Fresenius (an acknowledged authority on the matter), all the moisture goes away only at 120 C. ; yet the customary drying temperature is 100 C C. If the ore is meant to be used as a material for making ferro- manganese or manganese preparations, such as sulphate, &c., the percentage of manganese metal is the most important item in the assay ; but the nature and quantity of the impurities in this case form a more important element in the valuation than they do in the case which we considered first. Methods for the mere assay- ing for manganese (Mn) are given in the Exercise on Cast-iron ; we here confine ourselves to the determination of t or active oxygen. 0* ^ 68 EXERCISES IN ANALYTICAL METHODS. Of the many methods proposed, Fresenius and Will's unquestionably is the one best adapted to the requirements of the technical assayer. It is founded upon the fact that binoxide of manganese, when brought into contact with aqueous sulphuric and oxalic acids, is readily reduced to manganous sulphate with evolution of 2 x C0 2 of carbonic acid for every one Mn0 2 of pure binoxide. MnO 2 + H 2 S0 4 + H 2 C 2 4 The carbonic acid is easily determined by one of the two methods given in Ex. 20. 2CO, x O9886 - Mn0 2 ; 2CO 2 X 01818 = O. The sulphuric acid must not be too dilute if the reaction is to progress at a convenient rate ; on the other hand, it must not be allowed in its concentration to come too near the condition of undiluted vitriol, or else some of the oxalic acid may be converted into H 2 O + CO -f CO 2 . This condition, however, almost takes care of itself. It is more important to point out that, according to Luck, ferric oxalate, which is always liable to be present, at temperatures about 70C. breaks up with formation of carbonic acid and ferrous salt. If such decomposition is allowed to happen in an ore containing iron in only the ferric form, so much Fe 2 3 is mis-determined as MnO 2 . The ore to be analysed must be very finely powdered, or else the decomposition of the last particles proceeds very sluggishly. But finely-powdered manganese ore is very hygroscopic. Perhaps the best method is to preserve a sufficient supply of the fine powder in its air-dry condition in preparation tubes, and to weigh out best at the same time one portion for the deter- mination of the moisture, and another for the assay proper. The oxalic acid is best used in the form of neutral potash salt, K 2 C 2 O 4 H 2 O = 184 ; hence 1 grm. of real binoxide requires 212 grms. of this salt. Fresenius and Will prescribe 2 '5 grms. for 1 grm. of high-class ore. For the execution of the assay, the inventors recommend the apparatus represented in Fig. 13. Flask A should hold about 120 cc. ; flask B, about 100 cc. when filled up to the neck. B is charged with oil of vitriol up to about the middle of its belly. The weighed ore sample (3-5 grms.) goes into A, along with the requisite quantity of oxalate, and about 40 cc. of water. ASSAYING OF MANGANESE ORE. 69 The apparatus is then put together and tared. To start the reaction, some of the vitriol in B is made to flow over into A by sucking at <1 with a piece of india- rubber tubing, while b is closed with an india-rubber cap, and then allowing the atmosphere to press the acid over. This operation is repeated until all the binoxide is dissolved, which, as a rule, requires 5-10 minutes. In order now to expel the dissolved carbonic acid, a somewhat large quantity of vitriol is allowed to run into A so as to heat the mixture to near 70C. The FlG - carbonic acid stagnating in the apparatus is then sucked out by means of an india-rubber tube attached to d the cap at b being, of course, removed the apparatus allowed to cool completely, and then weighed again. Supposing C grms. of carbonic acid to have been evolved from P grms. of ore, the percentage of Mn0 2 is C x 0-9886 x 100 -P- Where such tests are made habitually, it is obviously expedient to weigh out exactly 3, 4, or 5 times 0'9886 grms. of ore, to facilitate the computation. In a completely equipped chemical laboratory the best modus operands is to decompose, say, 1 grm. of ore (even less will do), and weigh the carbonic acid directly. If the ore contains carbonates, these must be decomposed, and their carbonic acid expelled before the oxalic acid is added. In this case, the inverted-condenser form of the apparatus comes in particularly useful. The Permanganate Method*. The Mn0 2 is allowed to act on a weighed sufficiency of (a) standardized ferrous sulphate, or (b) oxalic acid in dilute sulphuric acid, and the surplus reducing agent left at the end determined by titration with permanganate. If form (a) is preferred, the reaction must be carried out in an atmosphere of carbonic acid to prevent oxidation of ferrosum by * Taken from Fresenius' "Quantitative Analyse." 70 EXERCISES IN ANALYTICAL METHODS. the atmosphere ; it offers the additional inconvenience that the ore sometimes is rather slow in dissolving. Hydrochloric, sub- stituted for sulphuric acid, works more promptly ; but this leads to other difficulties. The oxalic acid modus is, on the whole, preferable. For its execution, neutral oxalate of ammonia is a very convenient form of the reagent, on account of the greater definiteness in its composition as compared with, for instance, crystallized oxalic acid. The air-dry crystals are C 2 O 4 (NH 4 ) 2 -f H 2 O = 14212, which quantity, of course, requires = 16 parts of oxygen, or MnO 2 = 87*0 parts of binoxide of manganese for its oxidation. The permanganate had better be standardized ex- pressly with a weighed quantity of the oxalate. For this pur- pose, some 300 mgs. of the salt are dissolved in 10 per cent, sulphuric acid, conveniently, but not necessarily, along with some manganous sulphate, and permanganate dropped in until the liquid becomes permanently pink. If no manganous sulphate was added, the oxidation at first goes on very sluggishly, even on gentle heating, but it becomes more and more rapid as the amount of manganous salt increases. From the volume of permanganate added, the weight of surplus oxalate is calculated, to be deducted from the weight of salt taken. From the remainder the Mn0 2 is calculated. In making an analysis, it is expedient to work on a relatively large scale, so that aliquot parts of the mixture (of MnS0 4 and surplus C 2 4 H 2 ) produced can be used for repeated determination of the excess of oxalate. If the mixture, as usual, is turbid,* it should be filtered before applying the permanganate, or else the change of colour is not seen with sufficient distinctness. The process of fractional filtration may be employed, i.e., we may dilute the whole to, say, 500 cc., filter off, say, 100 cc. through a dry filter, and, neglecting the volume of the precipitate, look upon these 100 cc. as representing one-fifth of the whole solution. Bunsen's Method. A small known weight of the substance f is distilled from out of a small fiask with hydrochloric acid of 20 per cent. ; the chlorine evolved is passed into excess of solu- tion of iodide of potassium, and the liberated iodine determined * It is as well to make sure that the turbidity includes no oxalate of lime, f Calculate for 50-100 cc. of thiosulphate. ASSAYING OF MANGANESE ORE. 71 by titration with thiosulphate, i.e., comparison with the standard iodine solution. The distillation flask should not be larger than necessary ; it is best made before the blow-pipe out of a piece of stout glass tubing of 4-6 nuns, inner width. Another piece of the same tubing serves for a delivery tube. Both the entrance end of the latter and the edge of the flask are ground flat, and the two are so united by means of a piece of good, stout india- rubber tubing (previously de-sulphurized by treatment with hot, dilute caustic soda, and carefully washed) that the two ground edges touch each other inside. For the reception of the iodide of potassium solution, Bunsen uses a small untubulated retort. The solution is made so dilute that it just fills the belly of the retort, which when used is fixed to a clamp-stand in an inverted position, the neck slanting at about 45. In this arrangement the greater part of the gas-evolution tube must be as narrow as at all con- venient and permissible, and its end drawn out to a narrow point, so that, if the liquid threatens to go back, it does so slowly, and can be driven forward again by increasing the flame below the flask. The several safety contrivances which have been proposed to render the going back of the iodide impossible are all of no use. Another arrangement which generally works better in beginners' hands, is to put the iodide solution into a bulbed U-tube, provided with an attached bead-tube for catching any stray iodine, and kept cold by immersion in a water-bath. The connection of the flask with the U-tube is effected by means of a perforated paraffined cork. The end of the gas-evolution tube goes through this cork down to near the highest level which the liquid can possibly reach. Whichever modus operandi we may adopt, we must remem- ber that aqueous hydriodic acid absorbs oxygen from the air, especially at higher temperatures, with elimination of iodine. Hence, in the case of the retort-modus especially, the iodide solution should be prepared with boiled-out water ; and in the case of the U-tube method, the U-tube should be filled with car- bonic acid before starting the distillation. The substance is best weighed out direct in the tared dry flask, which is next filled with carbonic acid. Air-free, but cold, hydrochloric acid is then added, and the delivery tube joined on without delay. The 72 EXERCISES IN ANALYTICAL METHODS. iodine liquor must be cooled down (if necessary) by application of cold water,* and, after sufficient dilution with (cold) boiled-out water, be titrated with thiosulphate without delay. Every I = 126*85 mgs. of iodine liberated correspond to JO = 8 mgs. of oxygen, or JMn0 2 = 43'50 mgs. of binoxide. Note. If the ore contains a ferrous compound (such as mag- netic oxide of iron, Fe 3 4 ), the iron-permanganate and the Bunsen methods determine only that part of the active oxygen which is present over and above that needed for the peroxidation of the ferrosum. This, however, in the eyes of the assayer, is rather an advantage than a defect, because the ferrous oxide acts in the chlorine maker's still exactly as it does in the assay. In the case of the Fresenius and Will method the effect of the iron in the ore is not quite so definite. In general, only part of the ferrosum will be peroxidized at the expense of the Mn0 2 ; on the other hand, the ferric oxide produced or originally present may oxidize oxalic acid into carbonic acid, and thus acts in the opposite way. According to Luck, almost the whole of the ferrous oxide of the ore becomes ferric oxide if some 6 cc. of 10 per cent, acetate of soda solution are added to the (3 grms. of) ore in the decompo- sition flask ; and the method then becomes technically correct. In the oxalic acid and permanganate method, in which the decomposition usually proceeds slowly, and the ferric oxide has no chance as an oxidizing agent, the result should be exact in a technical sense. In a Complete Analysis, the determination of the moisture must be supplemented by that of the chemically combined water, which is effected by means of the "direct" method given in Ex. 28. The residue left in the boat should be weighed, because its weight affords a check on the rest of the determinations con- jointly. In it the metals are present as Mn 3 4 , Fe 2 3 , CaO, MgO respectively ; the nickel probably remains as NiO. About the cobalt we cannot venture upon any statement ; but both it and the nickel together always form only a small fraction of the whole. * The sensibility of the iodide of starch test decreases perceptibly as the tem- perature rises, from even ordinary low to ordinary high values. (Fresenius. ) COMPLETE ANALYSIS OF MANGANESE ORE. 73 A further check might be obtained by reducing the first residue in hydrogen, so as to obtain a mixture containing the respective elements as MnO, Fe, Co, Ni, CaO, MgO, SiO 2 . For the Determination of the Metals, the ore is dissolved in hot hydro- chloric acid, the solution evaporated to dryness, and the silica eliminated and weighed, as in a silicate analysis. This " silica," of course, includes any undecomposible silicates that may be present; but, as a rule, the analysis of the mixture offers no interest, and it is sufficient to report it in toto as " gangue." Only in the hands of a very negligent worker could the "gangue" include undecomposed free or combined MnO 2 ; but the residual silicate may be manganiferous, and its determination be required. The filtrate from the silica is diluted to a known volume, and aliquot parts of it serve for the several determinations. We will assume, in the first instance, that the qualitative analysis has proved the presence of only iron, manganese, calcium, and magnesium. In this case the first step is to separate The Iron, which, of course, is present as ferricum, from the divalent metals. If there is relatively much iron present (we mean much for a manganese ore), a good method for its elimina- tion is to neutralize the solution by cautious addition of carbonate of soda solution, i.e., to add such solution until a permanent pre- cipitate threatens to form, and then to add a sufficiency (but not more than necessary) of carbonate of baryta milk in the cold. Only the iron is precipitated ; the manganese, &c., remain dis- solved. After the precipitate has settled, it is filtered off and treated as a mixture of ferric hydrate and carbonate of baryta. From the filtrate the dissolved baryta is eliminated by the least sufficiency of dilute sulphuric acid as sulphate, and the manganese precipitated after neutralization with ammonia by means of sulphide of ammonium, as fully explained in Ex. 21. In the filtrate from the sulphide of manganese, the calcium and magnesium, after due concentration, are determined by the usual methods. If only little iron is present the use of carbonate of baryta can be dispensed with. It suffices to add carbonate of soda drop by drop until a small portion of the manganese has gone down visibly, and then to allow the mixture to settle. The precipitate 74 EXERCISES IN ANALYTICAL METHODS. (Fe 2 O 3 + ccMnCO 3 ) is dissolved in hydrochloric acid, the solution boiled to reduce the manganese to manganous chloride, and the two metals are then separated, as explained in Ex. 21. The ammoniacal manganiferous filtrates are, of course, added to the main portion and worked up along with it. We will now pass to the components not taken into account in our explanation. Alumina, if present, goes with the iron, and from it is sepa- rated by caustic potash, as shown in Ex. 17. Observe that neither the acetate of ammonia nor the carbonate of baryta process separates out alumina as completely as they do iron. A little reflection will show how small quantities of alumina which escaped precipitation can be recovered. Barium, if present, must be determined in a special portion of the substance solution by means of sulphuric acid. In the case of a manganese ore, and in the hands of a thinking analyst, there is no fear of its getting mixed up with the calcium. Cobalt and Nickel. Small quantities of these two metals are often present, but their exact determination is by no means easy. Assuming the oxide of iron to have been eliminated by frac- tional precipitation, the cobalt and nickel pass into the manganese liquors,* from which they can be recovered by fractional pre- cipitation with sulphide of ammonium. This reagent is added, drop by drop, until, instead of a black, a flesh-coloured precipi- tate is produced. The mixture is then boiled, which will cause some of the precipitated sulphide of manganese to re-dissolve, but will not affect the sulphides of nickel and cobalt. Collect the (black) precipitate on a filter, wash it with hot water, and extract the manganese by treatment with hot, dilute, acetic acid. A small quantity of the nickel and cobalt will probably pass into solution. If so, they must be recovered by a repetition of the process described. The cobalt-nickel sulphide is dissolved in * The iron (Fe20s) precipitate may include some cobalt or nickel. The best method for their extraction is Schwarzenbach's. Dissolve in hydrochloric acid and dilute to at least half a litre for every 1 grm. of iron present. Neutralize in the cold with carbonate of ammonia, i.e., add of this reagent drop by drop until the liquor is dark-brown and opalescent, though still free of tangible pre- cipitate. Then boil and filter hot. The cobalt and nickel are in the filtrate, from which they can be precipitated with sulphide of ammonium. DE HAEN'vS METHOD. 75 aqua regia, the metal is precipitated by dilute caustic potash (hot), the hydrate washed, ignited, and reduced in hydrogen. The spongy metal is washed to remove adhering alkali, re-ignited in hydrogen, and weighed. For the separation of cobalt and nickel, see Appendix to Ex. 34. Ex. 27. De Haen's Method for the Determination of Copper in Alloys, Ores, &c. (As modified by E. O. BROWN.) THE method is founded upon the fact that solutions of cupric salts when mixed with iodide of potassium are decomposed with precipitation of the copper as subiodide, Cu 2 L, and liberation of iodine. CuX 2 + 2KI = 2KX + Cul + I. The iodine is deter- mined by titration with thiosulphate. The thiosulphate had better be standardized empirically by means of a known weight of pure metallic copper. 0'3 to O5 grm. of pure electrotype copper is dissolved in nitric acid,* the solution boiled to expel nitrous fumes, and diluted with water. Carbonate of soda solu- tion is now added until a permanent precipitate has formed, and the precipitate re-dissolved in acetic acid. To the solution excess of iodide of potassium is added, and the liberated iodine titrated with thiosulphate, some starch solution being added towards the end to enable one to find the end point notwithstanding the presence of the cuprous iodide precipitate. To assay, for instance, a cupreous pyrites, 3-10 grms. are dis- solved in nitric acid or aqua regia, and the solution is evaporated to dryness with excess of sulphuric acid to convert the metals into sulphates.! The residue is treated with water, the solution filtered, and from the filtrate the copper re-precipitated as sulphide by thiosulphate of soda. The washed precipitate is dried, ignited, * In the method described for the assays the metal is always brought into the form of sulphate before the addition of the iodide ; hence it would no doubt be better to do the same in the standardization of the reagent, i.e., convert the dissolved metal into sulphate by evaporation with sulphuric acid before proceed- ing further. W. D. f For another method of disintegration, *ee Ex. 35 on Iron Pyrites, 76 EXERCISES IN ANALYTICAL METHODS. dissolved in nitric acid ; the solution again evaporated with sul- phuric acid to eliminate traces of lead, the residue re-dissolved, and the solution filtered. The filtrate is treated as the copper solution is in the standardization. The above is taken from a memoir by James W. Westmore- land in the " Soc. of Chem. Ind. Journal " for 1886, p. 48. Westmoreland has subjected the process to a long series of severe critical tests, and found it to give very exact results, even in the presence of any foreign metals likely to occur in a copper ore. In the absence of certain metals the precipitation of the copper as sulphide can no doubt be dispensed with, and the sulphate solution as obtained directly from the substance be neutralized. &c., and titrated. Unfortunately, it does not appear from West- moreland's memoir what metals one may allow to be present without falling into error ; all he says in this respect is, that lead must, and that iron had better be, removed. Ex. 28. Determination of Water by the "Direct" Method. IN many hydrated substances, although they may lose theii water readily enough on ignition, this component cannot be deduced from the loss of weight observed, because what really remains, through some cause or other, differs in its weight from the anhydride of the original substance. In such cases, what is usually done is to expel the water from a known weight of substance by heating it in a current of dry air, and aftei removal of what there may be of foreign vapours, to collect the water in a tared chloride of calcium tube, and weigh it elirectly To begin with an easy case (which is within the reach of the ordinary indirect method so that you can check your resuli by means of the latter), procure a supply of (1.) Pure Selenite (CaS0 4 . 2H 2 0). Grinel up a few grammes into a fine homogeneous powder, which preserve in a stoppered tube while constructing the necessary apparatus (Fig. 14). II WATER BY THE DIRECT METHOD. 77 consists of a hard Bohemian-glass tube a b, supported by the gutter of a combustion furnace about 35 ctms. long. The end a, which projects about 7 ctms. beyond the end of the gutter, communicates with a perspirator affording a current of dry \ FIG. 14. air, and provided with a stopcock to regulate the current. A little thimble into which the end of the perspirator's exit-tube projects (see Fig. 15) serves to prevent back currents. The end b of the combustion tube is drawn out, and serves to attach a U-tube, filled with chloride of calcium, by means of an india- rubber tube, as shown in the figure. The two glass tube ends must touch each other within the india-rubber. During the analysis this joint is enclosed in an air chamber c, kept at 105 110 by means of a special lamp to prevent condensation of the water. Such a chamber is easily made out of asbestos paste- board, thin brass wire serves to sew up the seams ; but one made of thin sheet-iron is better. Before starting the actual analysis, insert a Flu - 15 - plain glass tube in the india-rubber end at 6, and, while keeping up a tempera- ture of 105-110 there by means of the chamber, dry the combustion tube by heating it in the furnace and driving a rapid current of dry air through it until the moisture may be 78 EXERCISES IN ANALYTICAL METHODS. presumed to be gone ; then slacken the current of air, and while allowing the combustion tube (but not the air chamber) to cool, weigh the chloride of calcium tube, and attach it at b, in lieu of the plain tube. To enable one to make this exchange, either the perforations of the chamber must be so wide that the wide part of the combustion tube can pass through them, or the chamber must consist of two detachable parts. As soon as the joint is made, replace the chamber over it, and keep up its temperature to the prescribed value. Assuming the combustion tube to have cooled down sufficiently, all that remains to be done now is to weigh out, say, 1 grm. of the substance in a platinum boat, to place it in the tube, and to keep it therein at a dull-red heat, in a continuous slow current of air until the sublimate of water, which shows itself at first, has been away for about five minutes. Now allow the tube (but not the air chamber) to cool, and weigh the chloride of calcium tube. To make sure that the water is all away, resume the heating process (with the chloride of calcium tube replaced), and, after about ten minutes' heating, again stop the heat and weigh the U-tube. As a rule, the weight will be the same as before ; if not, the process must be repeated until the weight remains constant ; which shows that all the water has been expelled and collected in the U-tube. Finally weigh the boat ; the loss which it has suffered should be equal to the gain of the U-tube to within 1 or 2 mgs. After having by this exercise learnt to work the method, apply it to (2.) Sulphate of Iron and Potash, with this modification, how- ever, that about 15 ctms. of the front (b) part of the tube are filled with closely-packed granulated chromate of lead (as used for organic analysis), which, before the boat goes in, must be thoroughly dried at a red heat. To make sure that all the water is really gone, attach the tared chloride of calcium tube at 6, and keep it there, while a slow current of air is going over the heated chromate, for ten minutes. Then weigh the chloride of calcium tube, and see if it has gained anything. If it has, repeat the process until the weight of the U-tube is constant to 0*5 mg. During the analysis the chromate must be kept hot to absorb the sulphur dioxide which goes off along with the water. Notes. During the heating processes take care to protect the VARRENTRAPP AND WILL'S METHOD. 70 chloride of calcium tube against the radiant heat of the chamber by means of a screen of asbestos pasteboard. The column of chromate of lead is kept in its place by means of two plugs of asbestos. Observe that ordinary chloride of calcium gives up water to absolutely dry air, as produced by the action of vitriol. Hence, whatever method you may adopt for the preliminary drying of your current of air, before entering the combustion tube it must pass through a tube charged with the same chloride of calcium as is in the U-tube intended for the reception of the water. Theoretical percentage of water in selenite = 20*93. the iron salt = 24'87. Ex. 29. Determination of Nitrogen by Varrentrapp and Will's Method. PRINCIPLE. All nitrogen compounds which contain their nitrogen in direct combination with hydrogen, or metals, or carbon, when ignited with caustic soda, give off their nitrogen as ammonia. The H 2 O in the Na 2 O H 2 O, by its hydrogen, supplements the NEL or (NH), or N EE of the substance, into NH 3 , while the oxidizes the rest ; the C0 2 produced combining with the Na. 2 O of the reagent. In practice the caustic soda is used in the form of soda-lime, because the unmixed reagent attacks the glass too violently. REQUIREMENTS. 1. A combustion tube of the form shown in Fig. 16 ; con- venient width = 14 mms. inside ; length = 520 mms. 2. A nitrogen bulb ; see h in the figure. 3. A copper or brass mixing wire ; i.e., a stout wire about 600 mms. long, terminating in a kind of short corkscrew. 4. Soda-lime. Procure a caustic soda which is as free as possible from nitrates; dissolve 500 grms. of it with 5 grins, tliiosulphate of soda* in 1000 cc. of water in a cast-iron pot, add 750 grms. of quicklime in small fragments, cover the * To destroy traces of nitrates. 80 EXERCISES IN ANALYTICAL METHODS. pot with a tinned-iron plate, and allow the lime to slake. Then put the pot over a fire, and evaporate to dryness with constant agitation with an iron spatula. Transfer the dry mass to a Hessian or iron crucible, and expose it to a dull-red heat for one to two hours. After cooling, take out the mass, pound it up in an iron mortar, and pass the whole through a wire-gauze sieve of 7 meshes per linear inch. From the product separate out the powdery part by means of a sieve with 22 meshes to the inch. Preserve the granulated part and the powder separately in dry, well-corked phials. The phials A. FIG. 16. a is a plug of asbestos ; h, a mixture of zinc dust and fine soda-lime ; t 1 , a small quantity of fine soda-lime ; d, mixture of substance and fine soda-lime ; e, plug of asbestos ; /, granulated soda-lime ; f/, plug of asbestos. intended for immediate use should be of about 100 cc.'s capacity, and not have lips at the neck, so that the contents can be more conveniently poured into a combustion tube. 5. A supply of zinc dust. 6. Asbestos, to be ignited strongly before use. ANALYSIS. Good substances to practise upon are hippuric acid, oxamide, or urea. Powder up a gramme or two, dry it at 100C., and pre- serve in a preparation tube. For immediate use in the analysis prepare a substance tube, the outer width of which is no more than half at most of the inner width of the combustion tube. VARRENTRAPP AND WILL'S METHOD. 81 Mod an Operandi. Stop up the lower end a of the tail of the combustion tube with a loose stopper of asbestos, and introduce a layer of a mixture of one part of zinc dust and one part of tine soda-lime b (about 4 ctms.). Shove down a plug of asbestos c, and add about 8 ctms. of fine soda-lime. Now put in about the half of the substance to be analysed from out of the previously weighed substance tube, pour on a quantity of soda-lime powder, and mix the two intimately by means of the wire, taking care to keep the first 4 ctms. or so of soda-lime free from substance. Add now another supply of powdery soda-lime, introduce the rest of the substance, add more soda-lime, and mix first this upper portion by itself and then the two portions into one homo- geneous mass. Withdraw the wire, and, holding the tube ver- tically, gently tap it, so that the mixture falls down into its natural position, but take care to keep it pretty loose ; then add a small quantity of plain soda-lime and a plug of asbestos e. Fill up the rest of the tube with granulated soda-lime, and again insert an asbestos plug at g to keep this part of the reagent in its place. Close the tube with a plain cork, and tap it thoroughly on a table so as to form a canal all over the powdery part of the contents. The quantity of substance to be used for an analysis depends on its richness in nitrogen. With rich substances aim at producing 50 to 150 mgs. of ammonia; but even of the poorest substance more than 1 grm. should not be used. Charge the bulbs with about 15 cc. of 5 per cent, hydrochloric acid, by filling them from h like a pipette, and, by means of a first- class cork provided with a carefully made perforation, attach them to the end of the combustion tube. The next thing is to make sure that the apparatus is perfectly tight. For this purpose slip a narrow india-rubber tube over k, suck out some of the air, and then (using your fingers as a pinchcock to govern the backward flow of air) let in air so as to drive the liquor all into the pear- shaped bulb. If the liquor retains its level for about ten minutes the apparatus may be pronounced safe. Now place the tube in a combustion furnace, and next heat the granulated soda-lime to dull redness, beginning at the exit end. Then heat the mixture, progressing very gradually and slowly in the same way, until the G 82 EXERCISES IN ANALYTICAL METHODS. decomposition is complete, which is seen by the mass (which be- comes black at first) regaining a light-grey colour, and by the gas evolution coming practically to an end. Now heat the zinc dust to evolve hydrogen, and thereby drive the gas contents of the tube into the bulbs. At last detach the bulbs, empty their con- tents into a Berlin basin, rinse them twice with water and add the rinsings, and next evaporate on a water bath to about one- half or one-third of the original volume to expel volatile organic products. Now add excess of chloroplatinic acid, and proceed as shown in Ex. 19. Pt x 014421 - N 2 . Instead of weighing it as platinum, the ammonia may be titrated. In this case, however, we must work on a compara- tively large scale to obtain a sufficiently exact result. This method does not work in cases when the hydrochloric acid at the end is strongly coloured. Sometimes during the combustion the tube gets filled with relatively pure ammonia, while no fresh gas is being evolved; the hydrochloric acid then rushes back most violently, and you may lose your analysis. To be prepared for this, keep the india- rubber tube attached to h, and as soon as the liquid threatens to go back, clip it with your fingers, so as to slacken the speed of the backward flow of air. The possibility of this emergency is avoided, and the process rendered more easy of execution gene- rally, by connecting the tail end of the combustion tube with a Kipp's apparatus for hydrogen gas, and letting a slow current of hydrogen go through from beginning to end. The hydrogen had better be moist. NOTES REGARDING SPECIAL CLASSES OF SUBSTANCES. 1. Cyanides and other substances, which, by themselves, would give nearly pure ammonia, or ammonia and steam. Along with these add a small quantity of powdered sugar, to produce car- buretted hydrogens. 2. Substances consisting largely of cellulose or other carbo- hydrates. These, especially when they contain much water, cause the soda-lime to swell up so as sometimes to obstruct the tube. This can be avoided by substituting for the powdery soda-lime a mixture of caustic soda and magnesia, easily made KJELDAHL'S METHOD. 83 from calcined magnesia and caustic soda by a process similar to that used for soda-lime. Ex. 30. -Kjeldahl's Method of Nitrogen Determination. THIS method, though as unlike it as it could possibly be in its technical aspects, naturally joins on to Varrentrapp and Will's, because it is founded upon a similar general principle, and, as far as our present experience goes, has the same range of applicability. Requirements. 1. Strongest oil of vitriol, i.e., as good an approximation to real H 2 SO 4 as can conveniently be procured. Ordinary vitriol, distilled down in a current of air till it is reduced to Marignac's acid, 12SO 3 .13H 2 0, will do. 2. Sulphuric anhydride; or phosphoric anhydride, which, con- jointly with sulphuric acid, represents a certain quantity of S0 3 . We are in the habit of using a very strong crystallized acid, H 2 SO 4 + xS0 3 , which is to be had in commerce.* These reagents must be free from every trace of nitrous and nitric acid. In describing the method, we will assume that the substance to be analysed is an extract or syrup (for which class of sub- stances it was originally invented). A quantity of substance, which should not contain more than 1 grin, of dry organic matter, and which need not represent more than, say, 50 mgs. of ammonia, is introduced into a pear-shaped, long-necked flask of about 150 cc.'s capacity, and next dried there as far as possible by heating the flask in a steam or air bath, and blowing air through it. The residue then is dissolved in 10 cc. of strongest oil of vitriol (" 1 "), the flask placed slantingly over a flame on a wire-gauze support, and heated, ultimately to near the boiling point of sulphuric acid, until the organic matter is decomposed as far as possible with evolution of sulphurous acid, carbonic acid, &c. The flask is now allowed to cool, the product mixed with 5 cc. (or more) of crystallized acid (H 2 SO 4 + #S0 3 ), or an equivalent quantity of sulphuric or phosphoric anhydride,! and the heating * From Chapman, Messel & Co., 36 Mark Lane, London, E.G. f Kreusler recommends to use from the first a solution prepared by gradual addition of 200 grins, of P 2 5 to 1 lit. of distilled H 2 SO 4 . 20 cc. suffice for 1-1 '5 grms. of dry organic matter. 84 EXERCISES IN ANALYTICAL METHODS. resumed and continued until the resulting liquor is no longer darkly coloured. On long-continued heating it becomes almost colourless ; but the reaction need not, in general, be pushed so far. According to Kulisch (Fres. Zeitschr., year 1886, p. 150), the action of the vitriol is very greatly accelerated by the addi- tion of a little mercury. According to Kjeldahl, about two hours' heating, reckoning from the first addition of acid, suffices as a rule to effect a sufficient decomposition ; but Kreusler, working chiefly on albumenoids, finds that 4-6 hours' heating is required, for at least this class of substances. When the sulphuric acid has done its work, pow- FIG. 17. dered permanganate of potash is added, very cautiously, to the hot liquid, until after the disappearance of all brownish colour, the liquid assumes a greenish (or, if phosphoric anhydride was used, a bluish green) colour, from excess of reagent. Kjeldahl recommends to dust in the permanganate by means of a small sieve made out of a wide glass tube, slightly contracted at one end, and near it provided with a septum of ware-gauze. Only during the actual addition of an instalment of perman- ganate the lamp is removed. When the end-point has been reached, the flask is kept over a very much reduced flame for another 5-10 minutes, and then allowed to cool. The product SCHLCESING'S METHOD. 85 now contains all the nitrogen of the substance as sulphate of ammonia, and all that remains to be done is to dilute the product with water, to eliminate the ammonia by distillation with caustic soda, and determine it by platinum or otherwise. The caustic soda is employed conveniently in the form of a roughly standardized ley (2 parts of fused alkali, and 5 parts of water, Kreusler), so that the volume required can be determined beforehand. A convenient distillation apparatus (see Fig. 17) is a globular flask of 500-700 cc., connected with a Liebig's con- denser by means of an adapter, which serves to collect any projected drops of liquid, and return them to the flask. Accord- ing to Kreusler, the acid intended to receive the ammonia may be contained in an open beaker. If the outlet end of the con- denser tube dips into the acid, there is no fear of any ammonia being lost. A few pieces of zinc placed in the flask prevent bumping ; but before resorting to this expedient, we must make sure that the soda contains no nitrates or nitrites, which would give ammonia with the nascent hydrogen. The best final test for the absence of nitrogen oxides from the reagents is to make a blank experiment with, say, 0'3-0'5 grm. of cane sugar. The ammonia should not amount to more than a fraction of a mg., and if so, the quantity found may be subtracted from the ammonia obtained in the actual analysis as a correction. If it amounts to over 1 nig., the reagents must be rejected, and better preparations be procured. Ex. 31. Schloesing's Method of Nitric and Nitrous Acid Determination. Principle. If a nitrate or nitrite be heated with an excess of a solution of ferrous chloride in strong hydrochloric acid, the nitrogen acid is completely converted into nitric oxide, whose reduced volume measures the nitrogen present as nitric or nitrous acid. REAGENTS: 1. A strong solution of ferrous chloride; best prepared extemp. by dissolving fine iron wire (by theory 3*1, in practice, say, 10 parts of iron for 1 of N 2 5 ) in 20 per cent, hydrochloric acid. 2. Hydrochloric acid of 30 per cent, or more, 86 EXERCISES IN ANALYTICAL METHODS. Assuming the nitrate to be given as a solution, a quantity (containing from 50-100 mgs. of N 2 O 5 , if conveniently possible) is introduced into a pear-shaped flask of about 150 cc.'s capacity, which is provided with a narrow, drawn-out neck, so that a stout quill-sized india-rubber tube can be drawn over it. This serves to join on a narrow gas evolution tube, so that the evolved gas can be collected over water. The india-rubber tube should not be too thin in the body ; and yet sufficiently so to enable one to see by its shape whether the pressure within the flask is above or below that of the atmosphere. A Mohr's clip, attached to the india-rubber joint (but meanwhile kept open by a wedge of wood stuck in between its two sides), enables one at a moment's notice to compress or open the india-rubber joint. The modus operandi is as follows : The nitrate solution is heated and kept boiling until all the air can be assumed to be expelled, and the volume of the liquid is reduced to, at most, 5 cc. The clip is now closed, and the lamp withdrawn, the gas evolution tube having pre- viously been dipped into the ferrous chloride solution, contained in a beaker. By cautiously opening the clip, we now suck in the iron solution, and after it a sufficient volume of the strong hydrochloric acid, taking care to allow no air to enter the flask or even the gas evolution tube. The latter is now adjusted to its right position under a graduated glass tube filled with water and placed for gas collection in a trough. The contents of the flask are then cautiously heated until the india-rubber joint- i.e., the part nearest to the flask, which, on starting, was in a collapsed condition has assumed its normal shape, and begins to expand. The clip is now opened and slipped off, so that it hangs at the tube by its ring, and the boiling continued until it is seen by the changes of colour in the boiling liquid that the reaction is over. The boiling is continued for some five minutes, to make sure of all the nitric oxide having gone over, and the india-rubber tube closed again, the lamp having been withdrawn immediately before. The tube with the nitric oxide in it is allowed to cool, transferred to a deep cylinder full of water in such a way that the hydrochloric acid in the tube falls down into the cylinder, and its place is taken by pure water. Adjust the position of the tube so that about 1-2 ctms. of water are SCHLCESING'S METHOD: KREUSLER'S APPARATUS. 87 suspended in it, wait till temperature-equilibrium is established, read off the volume, temperature and barometric pressure, and measure the height of the suspended column of water to obtain the data for calculating the reduced volume of the nitric oxide, and from it the weight of N 2 6 (or N 2 O 3 ) present in the sub- stance analysed. The tension P mms. of the gas, supposing it to be dry and at its observed volume (V cc.) and temperature (), is : Barometer reduced to mimes [suspended column of water, -r 13'6 4- tension of water at t]. The weight of the nitric oxide is 2 and the corresponding weight of, say, nitrate of potash, {^T x - 321 }^ Nitric oxide being appreciably soluble in water, and besides characteristically prone to unite with oxygen, in the presence of water into N 2 O 3 and N 2 O 5 , the method as described is obviously infected with a number of errors, some tending to increase, others to lessen, the volume of gas obtained. The best mode of correct- ing for this is, after a preliminary analysis, to carry out two experiments side by side of one another, one with the substance to be analysed, and the other with a known weight of pure nitrate of potash, so adjusted, that by calculation it yields the same weight of nitric oxide. If the two gases are measured under precisely the same conditions, their volumes are proportional, vei*y nearly, to the respective quantities of nitrogen present. Supposing, for instance, the analysis yields 29 cc., the standard experiment 30 cc. of nitric oxide, the nitric acid in the substance is 29/30 of that in the standard nitrate. The following Modus operandi of Kreuslers eliminates all the errors alluded to very completely. A glance at his apparatus (Fig. 18) almost enables one to divine his mode of procedure. Flask T is a reservoir for previously boiled out hot water, which is con- stantly kept simmering by means of a small flame beneath it. The syphon n z is filled with water by blowing in at s, allowed t< > run for a little while, and then stopped in its action by closing clip x. The nitrate solution goes into R, and the first step is to 88 EXERCISES IN ANALYTICAL METHODS. boil it down (clips x and w closed ; clips g and m open), and utilize the steam for expelling the air from S. After a time m is closed, and the steam blown off* through w. When all the air may be presumed to be out, x is opened (to be kept open to the end) and w closed. A jet of lukewarm (relatively cold) water directed against S causes this tube to fill from T, and enables one to see any remaining air-bells, which are let out through m. Supposing by this time the liquid in R to have got sufficiently reduced, y is closed, the lamp withdrawn, and the flask charged through the funnel i, first with the ferrous chloride, then with FIG. 18. fuming hydrochloric acid, and lastly with some 20 per cent, acid, to prevent the formation of gas-bells in the funnel tube. A SchifFs nitrogen-measurer,* charged with air-free 7 per cent, caustic soda, is now joined on at m, but without meanwhile un- doing the clip there, and the nitric oxide evolved and driven over into 8, to be collected there over hot water. At last the clip m is opened, the reservoir of the Schiff having been lowered * As represented in the Exercise (47) on Dumas' Method of Nitrogen Deter- mination. WALTER CRUM'S METHOD. 89 before, so as to suck over the gas into the latter, where it is allowed to cool under, and at last measured at, the pressure of the atmosphere. Bunsen's Gasonieti'ische Methoclen (second edition) contains a table giving the tensions of vapour of water given out by such a caustic soda. If this table is not at hand, subtract from the tension of pure water the following correction : At 10 to 15C 1-1 mms. ) , v Atloto2oC 1-4 . f( Kreusler -) and the result is right to within 0*2 mms. of mercury. Ex. 31a. -Walter Crum's Method. BY way of appendix to Schloesing's, we will describe Walter Crum's method for the determination of nitric or nitrous acid, which is founded upon the fact that a nitrate or nitrite, when shaken with sufficiently strong oil of vitriol and mercury, is decomposed completely, even in the cold, so that all the nitrogen is eliminated as nitric oxide. Lunge's nitrometer, represented in Fig. 19, is a convenient apparatus for the execution of the method. It consists of a graduated glass tube a, the lower contracted end of which, by means of a stout narrow india-rubber tube, com- municates with a plain tube b serving as a mercury reservoir, and which terminates above into a funnel / the stem of which bears a peculiar kind of stopcock, the construction of which is seen from the supplementary figures I., II., III. To prepare the apparatus for use, it is charged with mercury through tube h (the cock being in position I.), so that the measurer a is full up to the top end of the straight boring of the cock. Care must be taken to see that there are no air- bells anywhere in a, or in the india-rubber tube, or in the lower portion of 6. There must be one continuous mass of mercury from the stopcock on to the meniscus in tube b, which, of course, at this stage stands at a level with the meniscus in a. The cock is now turned so that both its ways are stopped, and the reservoir tube b lowered, so that the pressure at the top of the measurer is decidedly less than that of the atmosphere. 90 EXERCISES IN ANALYTICAL METHODS. Supposing now we wish to analyse a sample of nitrous vitriol as produced in a Gay-Lussac tower, we measure off 0*5 to 5 cc. of the given vitriol into the funnel, and by cautiously turning the cock into (or rather towards) position I., cause the liquid to be sucked into a without allowing any air to follow. To recover what sticks to the funnel, this latter is rinsed, once with 3 cc. and then with other 2-3 cc. of pure vitriol, and the rinsings are FIG. 19.* sucked into a as explained, to be shut up, air-free, by turning the cock into position II. In order now to induce the reaction, tube a is taken out of its clamp, and its contents are agitated by inclining the tube so as to make it almost horizontal, and then suddenly bringing it back into a vertical position. This is done again and again for some two minutes, until it ceases to cause renewed gas evolution. * Taken from W inkier and Lunge's Technical Gas Analysis. London, Van Voorst, QUANTITATIVE ELECTROLYSIS. 91 This point having been reached, tube b is adjusted so that the pressure of the nitric oxide in a is equal to that of the atmo- sphere. To fulfil this condition the mercury meniscus b must be above that in a by the number x of millimetres, which exactly balances the h mm. of acid liquor in a. If this liquor has the specific gravity s, that of mercury being 13*6, we obviously have 3X13-6 = ^X8: or# = -^L 13'6 A solid nitrate or nitrite given for analysis is dissolved in the least sufficiency of water, the solution sucked into the measurer, and what sticks to the funnel washed in by means of pure concentrated oil of vitriol. At least three volumes of the latter must be used for every two of aqueous liquor introduced. Even in this case the nitric oxide, in the calculation, may be assumed to be dry. The final reading of the volume of gas produced must of course be postponed until the froth within a has subsided, and until the gas can be assumed to have the temperature of the sur- rounding atmosphere. The pressure is then finally readjusted, &c., &c. As shown by Walter Crum, his method applies to nitrates of cellulose (gun-cotton, &c.) as well as to inorganic nitrates. A known weight of the gun-cotton is dissolved in the funnel in oil of vitriol, the solution sucked in, and the analysis completed as usual. Ex. 32. Quantitative Electrolysis. WHEN a galvanic current is passed through the solution of a heavy-metallic salt, the general result is that the salt is decom- posed into metal which goes to the negative electrode as such, and into acid-rest which goes to the positive electrode (chlorine or bromine as such; the sulphate-rest SO 4 as S0 3 + JO 2 , &c.). Such a process is available for the quantitative determination of the respective metal, if it is possible to establish conditions under which the whole of the metal separates out (within a reasonable time) as a coherent deposit on the negative electrode, 92 EXERCISES IN ANALYTICAL METHODS. so that its weight can be determined exactly by weighing the electrode before and, along with the deposited metal, after the electrolysis. Of course the electrodes must con- sist of a solid, not in itself liable to change of weight during the pro- cess, which in practice means that they must consist of platinum. The decomposition cell and its electrodes may assume a variety of forms. In the so-called Mans/eld apparatus, which was introduced by Luckow, the nega- tive electrode consists of a platinum cone, represented in Fig. 20, while the positive is made of stout platinum wire, and shaped as shown by Fig. 21. Both our figures give about one-fourth of the actual size, j, i To effect an electrolysis, the solution is placed in a beaker somewhat higher but not much wider than necessary to accommodate the cone ; the positive electrode stands within the cone, as indicated in Fig. 22, which at the same time represents a water -bath, enabling one, if necessary, to raise the temperature of the contents. This apparatus is used very largely at the famous Mansfeld copper works in Germany in the assaying of ores for copper, and it works very well for this purpose. For general purposes, another appa- ratus of Luckow's invention is prefer- FKJ. 21. FIG. 22. QUANTITATIVE ELECTROLYSIS. 93 able. In it the negative electrode consists of a platinum basin, shaped as shown by Fig. 23, which accommodates the solution to be operated upon. Classen recommends the following dimen- sions Diameter = 90 mm. ; depth = 42 mm. ; capacity = 225 cc. Such a basin needs not weigh more than some 35-38 grms. The positive electrode (Fig. 24) FIG. 23. consists of a stout circular disc of sheet-platinum, with a long stout platinum wire riveted vertically to its centre. When used, the basin stands on a metallic ring, forming part of a metallic stand (Fig. 25), while the stem of the disc is held by a metallic screw-clip, cemented on to the end of an arm of the same stand, of which arm the middle portion is made of glass or vulcanite. A second screw-clainp, forming part of the same mass of metal as the first, serves to attach the positive* wire of the battery; the nega- tive wire is attached to a screw-clip fixed to a lower point of the pillar of the stand. As all electrolyses which come into consideration for us involve gas evolutions from the positive electrode, the decomposition cell must be covered by a concave glass plate to catch projected droplets of liquid, or rather by two semi-circular concave plates pro- vided at their open sides with the necessary notches to let the wires of the electrodes pass through, and constituting conjointly FlG - * In the case of a Grove battery, the positive wire goes to the platinum, and the negative to the zinc pole of the battery, t Taken from Classen's book. FIG. 24. 94 EXERCISES IN ANALYTICAL METHODS. a watch-glass-shaped cover. In any given case, the success of the analysis depends chiefly on the proper adjustment of (1) the composition of the liquid to be electrolyzed, and (2) the density of the current, meaning the number of amperes (vide infra) per square unit in the section of the body of current passing through the electrolyte. Chlorides, at least plain chlorides, are unfit for our purpose, because the chlorine which goes to the positive electrode attacks platinum ; nitrates may work ; sulphates, if their solution is of the proper degree of acidity or alkalinity, work well in many cases. According to Classen, the soluble double oxalates, which are produced from almost all heavy metals by addition to their neutralized sulphate or chloride solu- tions of excess of alkaline oxalate, lend themselves character- istically well for electrolysis generally, because the oxygen which goes to the positive electrode is taken up by the oxalic acid with formation of carbonic acid, a bye-product which is sure not to cause inconvenient secondary reactions. In regard to the power of the battery to be used no general direction can be given, except by saying that four good Grove's cells, as a rule, afford an amply sufficient (if not too strong a) current. The circuit should include a galvanometer, to be able at any moment to see what the strength of the current is ; and, besides, a rheostat, to be able to reduce this strength to the most convenient value. This, of course, implies that the avail- able maximum current-strength is in excess of what is wanted. Current-strengths are customarily stated in amperes. Imagine a current, which passes through a voltameter, and causes the evolution of 10*436 cc. of Knallgas (H 2 + JO 2 ) (reduced to and 760 mm.) per minute of time ; such a current is said to be of "one ampere."* Hence, supposing the Knallgas to be reduced to and 793 mm., one " ampere " corresponds to exactly 10 cc. of Knallgas per minute. If the gas be measured moist at 15C. and 760 mm., every cc. of gas evolved per minute corresponds to 0-08933 ampere. The Author is in the habit of using a special kind of galvano- * The statement, of course, refers to the current-strength as it is in the given connection ; if the voltameter be taken out, the current-strength rises immensely. QUANTITATIVE ELECTROLYSIS. 95 meter, represented in Fig. 26, which Sir William Thomson, some time ago, had the kindness of designing for him. It is a small tangent galvanometer, constructed on the Helmholz principle, in so far as the centre of the needle is not in, but removed hori- zontally from, the centre of the ring which conveys the current by a distance equal to half the radius of the ring. This latter consists of a solid copper strip, and has a diameter of about 22 FIG. 26. centimetres. The two ends of the strip dip, each into a mass of mercury, to enable one to make the necessary connections without disturbing the adjustment of the galvanometer. The needle consists of a small circular magnetized disc of steel. Crossways to it there is fixed on a long, very light, hollow aluminium wire, which serves as an index. The needle, as shown by Fig. 27, is suspended from an elliptical spring, which keeps the thread from snapping. A little wooden peg k, termin- ating in a ball above, which passes through a perforation in the 96 EXERCISES IN ANALYTICAL METHODS. glass lid of the suspension arrangement, enables one to arrest the needle by pushing it down to the floor of the limb-box, which latter is covered with a glass plate. Each of the two ends of the index points to a graduation which gives tenth- amperes directly, in this sense, that the reading 10 x corresponds to (1 +a) x amperes, where (1 -fa) is a factor, which must be FIG. 27. determined experimentally with the instrument in its intended place. The graduations go from 50 to + 50, i.e., from 5 to + 5 amperes. To standardize the instrument, put it into the circuit of a sufficiently strong constant battery, along with a voltameter,* a rheostat, and a commutator (an arrangement for reversing the current), and, at each of a properly selected series of current- strengths, determine the corresponding values for the volume of gas evolved per minute (at t and B mm.) and the mean of the two readings of the needle, always once with the current going one way, and then with the current reversed, to eliminate the zero and orientation errors. If the first of such a couple of readings is + 3'84, the second - 4'00, then \ (3'84 + 4'00) = 3'92 * Bunsen's Knallyas apparatus (see section on " Gas Analysis ") will do. QUANTITATIVE ELECTROLYSIS. 97 is put down as the reading. Instead of calculating the above factor (1 + a), which cannot be expected to be quite a constant in practice, it is better to embody the relation between readings and amperes in a curve drawn within a system of rectangular co-ordinates, so as to do as full justice as possible to all the good observations. Instead of a voltameter properly so called, a copper-voltameter may be used, which measures the current by the weight of copper precipitated by it from a solution of sulphate per minute. As shown by Fig. 28,* the instrument con- sists of a cylindrical lipless beaker, within which two square pieces of sheet copper are suspended vertically in the blue-vitriol solution (three volumes of saturated solution mixed with two volumes of water). A square plate of sheet platinum is connected with a stand in such a way that it can be immersed at a moment's notice between the copper plates. These latter are con- nected electrically with each other and the positive pole. The platinum plate is the terminal of the negative pole of the battery. The platinum plate is weighed before and, along with the deposit of copper formed on it, after the experiment (vide supra). Kittler recom- mends one square-decimeter of platinum surface for currents up to three amperes. Every one ampere corresponds to 19*686 mgs. of copper precipitated per minute, or every 1 mg. of copper corresponds to 0'0508 ampere. FIG. 28. * Taken from Kittler's Elektrotechnik, vol. i., p. 170. (Stuttgart, Enke.) H 98 EXERCISES IN ANALYTICAL METHODS. My galvanometer was made by Mr. James White's successors, Sauchiehall Street, Glasgow. I can confidently recommend it for the purposes of quantitative electrolysis. The needle shows a degree of steadiness (I mean relative independence of the vibra- tions of the floor) which at first quite surprised me. Fig. 29 represents a rheostat, which Mr. White's successors supplied along with the galvanometer. It consists of a wooden cylinder, with two parallel screw -lines cut into its surface. The two grooves accommo- date two independent wires, A and B. Both these wires terminate in screw -clamps, situ- ated at the same end of the cylinder. A belt of brass, provided with a circular spring inside, slides along the cylinder from one end to the other, and, where it happens to stand, unites a turn of wire A with the neighbouring turn of wire B, so that the current, supposing it to have come via A, returns via B to the end of the cylinder whence it came. In my apparatus the wire is of platinoid,* its thickness is 1*37 mm., and the total length of A + B is 17'8 metres. The resistance, when at its maximum, amounts to 4'9 ohms. Fig. 30 represents a kind of rheostat which Professor Blyth has devised for electrolytic and similar work. It consists of a wooden framework about five feet high, which serves as a support for six grooved pulleys, arranged in two blocks according to the ordinary block and tackle system. The three pulleys in the lower block are of wood, while in the upper, two are of wood and the third (the outside one) is of brass. One end of a fine iron wire is attached to the upper part of the framework, and the wire, after passing round the pulleys, is coiled upon a wooden or metal drum, by means of which the lower block with its attached weight can be raised. The * Platinoid is practically German silver, with an addition of 1 or 2 per cent, of tungsten (Bottomley ; Roy. Soc. Proc., 1884-5, vol. xxxviii., p. 341). QUANTITATIVE ELECTROLYSIS. 99 axis of the drum is screwed and moves in a nut, so that while revolving it has, at the same time, a longitudinal motion in the direction of the axis. The object of this is to prevent overlapping of the coils while the wire is being wound on. When the apparatus is used as a rheochord, the current is led through the iron wire by means of two terminals, one attached to the fixed end of the wire, and the other to the upper block which contains the brass pulley. At this point electrical contact is made simply by the pressure of the wire on the pulley. As the weight is raised, more and more wire is taken out of the circuit. In the apparatus represented, the iron wire used is 0'3536 mm. = 0'014 inch thick, and the greatest distance between the centres of the two blocks is 1/2 metre = 4 feet. With it a differ- ence of resistance of from seven ohms to half an ohm can readily be obtained. A special book on quantitative electrolysis (founded largely on researches of his own) was pub- lished lately by Classen.* Referring to it for fuller information, we satisfy ourselves with giving the following Exercise. * Quantitative Analyse durch Electrolyse, von Alex. Classen. Berlin, 1886. 100 EXERCISES IN ANALYTICAL METHODS. Ex. 32a. Electrolytic Separation of Copper and Nickel. WE choose this example because the electrolytic method works better with these two metals than with any others that could be named. To fix ideas we assume that the Mansfeld apparatus is used for the electrolyses. A good subject to choose is a German nickel coin. Take 0'25 to 0'30 grms. of the alloy, place it in a basin under a funnel, add about 20 cc. of 20 per cent, sulphuric acid, and enough of nitric acid to convert the metals into oxides. Dissolve, and evaporate so as to expel the nitric acid.* Re-dissolve in water, dilute to 200 cc., place the solution in a beaker of 7-8 ctms. diameter, and subject it to electrolysis in the cold by means of the Mansfeld apparatus. The strength of the current should be equal to 0*3 to 0*6 amperes. Three ordinary size Groves with chromic acid solution instead of nitric acid inside will do. After some hours the copper will be all precipitated ; which, however, should be made sure of firstly, by dipping the cone a little further in (or adding some water), and seeing whether there is any fresh deposit; and, secondly, by taking out 1 cc. and testing with sulphuretted hydrogen. When the copper is all down, withdraw the beaker from below and substitute one with 200 cc. of water (the current remaining closed) to wash the electrodes. (See note at end of article.) Then undo the connections, give the cone a final wash by plunging it into strong alcohol, dry it in an air- bath at 100C., and weigh it. This gives the copper. To deter- mine the nickel, concentrate first the aqueous washings, then the mother-liquor, the whole to about 100 cc., add a slight excess of ammonia and 4 grms. of solid oxalate of ammonia, and electrolize again, this time, however, in the heat, the beaker being suspended in a steam -bath. The current (of 1*1 to T3 amperes) is passed through the solution until 1 cc. of the liquor, when tested with sulphide of ammonium, after expelling the excess of ammonia by heating, * The presence of a remnant of nitric acid rather promotes the precipitation of the copper, but we are not quite sure that it does not interfere with the subse- quent precipitation of the nickel. (See under " Assaying for Copper," Ex. 35). ANALYSIS OF BRASS. 101 shows that all the nickel is down. This~ (atcdrding ^to ^Classen, whom we follow here as far as tire nickel is-cciidgriMjdJ ^i)l be the case after about two hours. Wash" thV cone, as explained for the copper, dry, and weigh it. This gives the nickel. For the first trial take, preferably, known weights of the two separate metals. German nickel coins contain, by intention, 25 per cent, of nickel and 75 per cent, of copper. Note. Classen insists that, whenever an oxidizable metal, such as copper, has been precipitated from an acid liquid, it must be washed completely while within the current. This, of course, is easily done. All that is needed is to apply a narrow syphon and transfer the liquor from the beaker to a suitable vessel, and at the same time to pour pure water on the top of the (nickel) solution so as to gradually displace it. In this manner oxida- tion of the copper precipitate is completely avoided, but the quantity of wash-liquor produced becomes inconveniently large. With a compact deposit of copper our method, as given in the exercise, is safe enough. Ex. 33. Analysis of Brass. Alloy of copper and zinc, usually contaminated with traces of iron and lead. DISSOLVE about one gramme of alloy in dilute sulphuric, with the help of nitric, acid, and chase away the excess of the latter. Treat the residue of sulphates with water, filter off the sulphate of lead, wash it first with dilute sulphuric acid, then with alcohol, and weigh it, best on a tared filter dried at 120C. as PbS0 4 . The alcohol is applied only after the copper and zinc are removed, so that the alcoholic wash-liquors can be thrown away. In the filtrate which contains the copper and the zinc as sulphates, both these metals can be determined electrolytically, the copper as shown in Exs. 32 and 35 ; the zinc according to Classen, as follows : (Assuming about 100 mgs. of zinc to be present, i.e., that only about 0'5 grm. of alloy was taken originally), con- centrate the liquor by evaporation to about 100 cc., neutralize very nearly, but not quite, with ammonia ; dissolve in the liquid 102 EXERCISES IN ANALYTICAL METHODS. 2-3 grms. of oxalate of ammonia, dilute to 150-200 cc., and electroh^e at 70-80C, iii the basin apparatus, Figs. 23-25, using a current of 0'8 to 1 ampere. The metallic deposit is washed first with hot water (without interrupting the current), then with alcohol, and after having been dried on the electrode at 80-100, weighed. The zinc precipitate after drying ' : sticks so fast" to the electrode that it is difficult to dissolve it off with dilute acid. As a rule the acid leaves a dark coating, which can be removed only by igniting the electrode (basin), and then again applying acid. Hence it is expedient to provide the basin with a coating of silver, copper, or tin before taring it and using it for the precipitation of the zinc (Classen).* If iron be present it passes into the zinc precipitate. Its deter- mination is easily effected by dissolving the whole in dilute sulphuric acid and titrating the iron with permanganate. The ordinary method is to separate the copper and zinc by means of sulphuretted hydrogen in the presence of a suffi- ciency of free mineral acid. For this purpose the (acidified) sulphate solution is heated to near boiling, and sulphuretted hydrogen passed through it until all the copper is precipitated as sulphide, which latter is filtered off and washed with hot water until the last washings are proved to be free of zinc and free acid. The copper precipitate is free of zinc (ZnS) only if the solution contained a very considerable surplus of sulphuric acid, but it is not quite easy to hit the exact degree of acidity. Hence the best course is to make sure that there is a suffi- ciency of acid, even at the risk of leaving a little of the copper in solution, and, should any copper remain unprecipi- tated, to bring it down from the filtrate, after dilution, by applying the sulphuretted hydrogen in the cold. The pre- cipitate of sulphide of copper obtained in this second precipita- tion must be washed, first with very dilute sulphuretted hydrogen water acidified to about the same extent as the mother-liquor is with sulphuric acid, and then with unacidified * In other words, part of the zinc deposit is an alloy of zinc and platinum, which gives off its zinc to acid only after the base metal has been oxidized by ignition in air. I fear a platinum basin will not stand this ordeal many times without getting rotten. ANALYSIS OF BRASS. 103 (H 2 S) water. The two copper precipitates are ignited together, and after treatment with sulphur and hydrogen at a red heat, weighed together as Cu 2 S, as shown in Ex. 19. To make quite sure that the sulphide is free of zinc, dissolve the ignited precipitate in fuming nitric acid of 1*5 specific gravity (at a gentle heat to keep the sulphur from running together into a globule), add a little sulphuric acid, chase away the nitric by evaporation, dilute and precipitate the copper with sulphuretted hydrogen, to test the filtrate for zinc by means of sulphide of ammonium. If sulphide of zinc is produced it must, of course, be collected and weighed (vide infra) to be allowed for. If the quantity of zinc thus recovered is small, it may be presumed to have been present in the ignited precipitate in the form of ZnS (beside Cu 2 S). From the united filtrates from the copper, the zinc, after expulsion of the sulphuretted hydrogen by boiling, may be pre- cipitated by boiling with an excess of carbonate of soda in a platinum (or nickel) basin.* The precipitate of basic carbonate is filtered off, washed completely with hot water until all the alkali is proved to be away, then dried, ignited (in porcelain), and weighed as ZnO. The filter, of course, must be incinerated by itself, under circumstances which reduce the unavoidable loss of zinc (by volatilization as metal) to a minimum. A good method is to soak the filter in nitrate of ammonia, dry it, and then burn it in instalments in an open porcelain crucible. But a better method is to re-place the empty filter in its funnel, dis- solve away the adhering oxide of zinc in boiling acetic acid, mixed with a little nitric, to evaporate the solution to dryness, finally in a porcelain crucible, and ignite the residue. The bulk of the basic carbonate is now added, and the whole ignited until the weight is constant. ZnO x 0*8034 = Zn. The oxide of zinc includes the iron (if any is present) as Fe 3 O 3 . To determine the iron, dissolve the ignited precipitate in hot ammoniacal sal- ammoniac as far as conveniently possible, and filter off the pre- cipitate (Fe 2 O 3 -f 'ZnO). Then dissolve this in the least quantity of hydrochloric acid, add ammonia to neutralize the acid, then * Remember that nickel does not stand acid solutions (wee Ex. 17, 011 Separa- tion of Iron and Alumina) ; nor does it stand treatment with ammonia. 104 EXERCISES IN ANALYTICAL METHODS. acetate of ammonia, boil, filter off the basic acetate of ferricum, ignite and weigh it as Fe 2 3 . (See Ex. 21, p. 53.) Another method is to take an aliquot part of the filtrate from the sulphide of copper, boil off the sulphuretted hydrogen, and determine the iron by titration with specially diluted perman- ganate, taking care to determine, and allow for, the small volume of reagent required to produce the end-reaction. (See Ex. 18.) In either case the weight of the zinc precipitate must be cor- rected for the weight of ferric oxide contained in it. Another mode of separating out the copper is to precipitate it from the (lead free) sulphate solution, as sulphocyanate of cuprosum. The solution for this purpose should not be too dilute. It is neutralized very nearly with carbonate of soda, mixed with a sufficiency of sulphocyanate of potassium and sulphurous acid in excess, and allowed to stand cold over night. The precipitate is filtered off, washed, and either dried at 100 C C. to be weighed as Cu.NCS, or, what is better, ignited with sulphur in hydrogen, and weighed as cuprous sulphide. Cu(NCS) x O5216 = Cu. From the filtrate the zinc is precipitated as carbonate, as explained above. Another method for the determination of the zinc is given in the following exercise. Ex. 34. Analysis of German Silver, Alloy of (chiefly) copper, nickel, and zinc. Ordinarily present impurities : lead, iron, and cobalt. CONVERT about 1 grm. of the alloy into sulphates, eliminate the lead (as PbS0 4 ) and the copper (as CuS by sulphuretted hydrogren),* as shown in the Exercise (33) on Brass. To Analyse the Filtrate, neutralize it with ammonia exactly, then, for every 1 mg.-atom of zinc., i.e., 6 5 '4 mgs., add about 4 mg.-equivalents, i.e., 4 x C 2 H 3 C10 2 mgs. of chlor-acetic acid, previously neutralized with ammonia, and 10 mg.-equiva- lents, i.e., 10 X C 2 H 3 C10 2 mgs. of free chlor-acetic acid, keep at 50 60C. over a water-bath, and precipitate the zinc by a * Or electrolytically. See Exs. 32-35. ANALYSIS OF GERMAN SILVER. 105 current of sulphuretted hydrogen. Sulphide of zinc thus pro- duced filters and washes quite easily. It is quite different in this respect from the ordinary sulphide of ammonium pre- cipitate. It may, indeed, be filtered off without allowing it to settle completely. It is washed with dilute sulphuretted hydro- gen water, acidulated with a little chlor-acetic acid, and lastly with plain dilute sulphuretted hydrogen water. It is then dried, separated carefully from the filter, and, after addition of the filter ash and some sulphur, ignited in hydrogen over a good Bunsen, as shown for copper in Ex. 19, to be weighed ultimately as ZnS. ZnS x 0'6709 = Zn. Solid chlor-acetic acid can be had from Trommsdorff, Erfurt, or Kahlbaum, Berlin, as " purum," at 20 marks the kilogramme. 4x C 2 H 3 C10 2 = 378'0 decigrammes dissolved to a decilitre in water gives a convenient solution. The filtrate from the sulphide of zinc contains all the nickel, cobalt, and iron. To simplify explanations we will assume, in the first instance, that iron is absent. To eliminate the nickel and cobalt conjointly add ammonia in excess, then excess of colourless sulphide of ammonium, and boil until the precipitate has become compact. Filter it off, wash it with hot water, dry it, and dissolve it with the filter ash in nitric acid of 1'5, evapo- rate with sulphuric acid to produce sulphates (as explained in Exercise on Brass, for copper). From the sulphate solution the nickel and cobalt may be precipitated electrolytically, as ex- plained for nickel in Ex. 32. For most practical purposes the cobalt (which never amounts to much) needs not be separated from the nickel, because it is quite equivalent to the latter metallurgically. But the problem is important scientifically, and we therefore append to this exer- cise a few methods for its solution. We will, however, first dispose of the Iron, of which traces are, as a rule, present in the alloy. For its determination the filtrate, from the sulphide of zinc, is freed of its sulphuretted hydrogen by boiling, the iron peroxidized by addition of a small granule of chlorate of potash and boiling again, and the iron then precipitated by addition of excess of ammonia. The precipitate is a very impure hydroxide, which must be purified by re-dissolving it in hydrochloric acid and 106 EXERCISES IN ANALYTICAL METHODS. re-precipitating it by the Schwarzenberg process. (See foot-note on p. 74, Ex. 21.) This process may have to be repeated before the iron is fit to be ignited and weighed as Fe 2 O 3 . From the united filtrates the nickel and cobalt are precipitated as sulphides, as explained above. Note on the Chlor-acetic Acid Process. This method for the separation of zinc from nickel, cobalt, iron, &c., was worked out in Oswald's laboratory, Riga, by P. von Berg. It has been tested in our laboratory by Mr. Frank Lyall, who obtained the follow- ing very satisfactory results : Zinc Oxide taken.* (Mgs.) Admixtures. ZnS obtained. (Mgs.) ZnO corresponding. (Mgs.) 315-0 None. 377-5 315-3 315-0 0-515 grm. of NiS0 4 , 7H 2 0. 376-0 314-0 315-0 0-523 grm. of CoS0 4 , 7H 2 0. 376-0 314-0 315-0 1-5 grm. of Fe(NH 4 ) 2 , S 2 8 , 6H 2 0. 374-5 312-8 SEPARATION OF NICKEL AND COBALT. All our methods for the solution of this problem start with the presumption that the solution to be operated upon contains no other heavy metals than these two, or, in other words, that any iron, manganese, &c., that may have been present originally, have already been eliminated by suitable methods, which, in the case of manganese, as we may say in passing, is not an easy condition to fulfil. (1) Liebig's Process is of purely historical interest now, but the student ought to know it. It is based upon the fact that a solution of Co(NC) 2 + 4K(NC), which contains free hydrocyanic acid, if boiled, passes into one of cobalticyanide Co(NC) 3 3K(NC), with evolution of hydrogen from the extra (NC)H, while the corresponding nickel salt remains unchanged. The metals had * As a standard solution of ZnSO 4> analysed by evaporating a known volume to dryness, and igniting the residue, so as to convert it into ZnO. SKPARATION OF NICKEL AND COBALT. 107 bettor bo used as nitrates ; in this caso the precipitation of the nickel (present as Ni(NC)., + x KNC) can be effected by boiling the solution with oxide of mercury. The nickel goes down as NiO, and is recovered by igniting the precipitate (in a good draught place). From the filtrate the cobalt, present as (NC) 6 Co'"K 3 is precipitated, after cautious neutralization with nitric acid, by mercurous nitrate. The precipitate (NC) 6 Co /// Hg 3 is washed with water, dried, and next heated strongly in air (in a draught place) ; the residue, lastly, in hydrogen, which yields weighable metallic cobalt. (2) Ease's Process (also superseded now). The metals are made into chlorides, and the solution is diluted to about a litre per gramme of total metal present. The solution is saturated with chlorine in the cold, mixed with carbonate of baryta, and allowed to stand cold. The cobalt goes down as Co 2 O 3 , the nickel remains dissolved substantially, not exactly. By re-dissolving the cobalt precipitate in hydrochloric acid, and repeating the treatment with chlorine and carbonate of baryta, the nickel in it can to a great extent be eliminated, and a better separation be effected. (3) The Nitrite Process, as worked out by Stromeyer. (Succeeds best with nitrates or sulphates, but may be applied to chlorides). The reagent is a solution of nitrite of potas- sium, which, according to our experience, had better be freed from its silica, which is almost invariably present, by cau- tiously acidifying with acetic acid, and removing the precipitated silica. The mixed solution is neutralized by adding caustic potash until a small permanent precipitate is produced, which is then removed by addition of acetic acid. A very considerable excess of nitrite is then added, and enough of acetic acid to produce a strongly acid reaction. Stromeyer says that the mixture must be very concentrated, and be allowed to stand for some forty-eight hours to deposit all its cobalt. According to our own experience, a prompt precipitation of the cobalt can be effected from even moderately dilute solutions, by pouring the mixture (of metallic solution, nitrite, and acetic acid) into a fresh quantity of previously acidified reagent. After a few hours' standing, the cobalt is all precipitated as " Fischer's salt " ^ T^ Q 3 >5N 2 3 + x H 2 0), which must be washed, first with a solu- 108 EXERCISES IN ANALYTICAL METHODS. tion of acetate of potash, and lastly with alcohol, for the removal of the acetate of potash.* The alcoholic washings must not be allowed to mix with the nickel filtrate. From the filtrate the nickel may be recovered by precipitation with caustic potash ; but we have found it better to first destroy the nitrogen-acids by evaporation with hydrochloric acid, and then to precipitate the nickel by means of colourless sulphide of ammonium in the heat. The precipitate (NiS) is easily filtered off and washed. It is dissolved as sulphate, and the metal determined electro- lytically (see Ex. 32a). The Fischer's salt is dissolved in hot hydrochloric acid, the chlorine, &c., expelled, the metal precipi- tated as sulphide (as explained), the sulphide re-dissolved as sulphate, and the latter electrolyzed, we should say, were we quite sure that the method gives exact results. In a series of test-experiments on the electrolytic determination of cobalt by the old method (electrolysis of the sulphate in alkaline solution of sulphate of ammonia), we found it extremely difficult to obtain exact results with unmixed cobalt. The method worked admir- ably with nickel, even if moderately cobaltiferous ; with pure cobalt solution we always obtained dull-looking deposits, which weighed more than the cobalt contained in them. Besides, a small portion of the cobalt often escaped precipitation. Classen's oxalate process may give better results ;t but we do not know whether it does. Meanwhile, we prefer to determine nickel and cobalt conjointly by electrolysis, and, in the separation of the two metals by means of nitrite, to determine only the nickel directly, and the cobalt by difference. If a direct determina- tion is required, precipitate the cobalt as hydrate by caustic potash in the heat (which must not be too strong, or else cobalt remains dissolved), wash the precipitate thoroughly, reduce it in hydrogen in a Rose's crucible, wash away the alkali by water, and heat again in hydrogen. The ultimate product is metallic cobalt, which, however, had better be tested by dissolving * It is perhaps not superfluous to point out that in this case soda may not be substituted for potash. f From a series of test-analyses made in our laboratory by Mr. A. B. Robertson, after the above had gone to the press, it indeed appears that exact results can be obtained with Classen's process, though not as easily as with nickel. SEPARATION OF NICKEL AND COBALT. 109 it in nitric acid, to see if there is not a residue of silicon or ni'bon, or both. If so, the residue must be collected on a weighed filter, dried at 110, and weighed, to be allowed for. Perhaps a better method would be to make the cobalt into a strong, acid-free solution of chloride, precipitate by pure oxalic acid, and ignite the washed and dried oxalate. The residue is semi-oxidized metal, which is easily rendered purely metallic by ignition in hydrogen. A small quantity of cobalt may escape precipitation, but this is easily recovered as sulphide, and brought into a weighable form. Possibly the phosphate process, to be given below (under 5), would work well as a means for bringing the cobalt into a weighable form. (4) The Nitroso-ft-Naphthol Method, (M. ILINSKI and G. v. KNORRE.*) Reagent. A solution of nitroso-/3-naphtholt in acetic acid containing 50 per cent, of C 2 H 4 O 2 . The metals are brought into solution as sulphates or chlorides. The solution is mixed with a little hydrochloric acid, heated gently, and mixed with a sufficiency of a hot solution of the reagent. After cooling make sure that a further addi- tion of reagent gives no additional precipitate ; allow to stand for some hours, then wash, once with cold and then repeatedly with hot hydrochloric acid of about 12 per cent, until the nickel is removed, and lastly with hot water. The precipi- tate is very voluminous, but is easily washed. It is a cobaltic salt of the formula C 10 H 6 0(NO) 3 Co. /// When heated by itself it explodes ; but, if mixed with a few knife-points of (pure) oxalic acid, and then heated gradually in a covered porcelain crucible, its decomposition proceeds quietly and without loss. This operation should be carried out in a Rose's crucible, so that one can finally heat in an atmosphere of hydrogen, and thus reduce the ash to metallic cobalt, which goes to the balance. (See Ex. 19, Determination of Copper.) From the filtrate from * Fres. Zeitschrift for 1885, vol. xxiv., p. 595. Original in Ber. d. d. Chem. Gesellschaft, vol. xviii. , p. 699. f To l)e had from C. A. F. Kahlbaum, Berlin, Schlesische Strasse, 16-19, at ten marks the hectogramme. 110 EXERCISES IN ANALYTICAL METHODS. the cobalt, the nickel, it appears, can be precipitated quantita- tively by means of caustic potash. But the inventors prefer to determine nickel and cobalt first conjointly in some way and then to determine the cobalt by nitroso-/3-naphthol, to find the nickel by difference. In a few test-analyses made in this laboratory the method gave very fair results for the cobalt (the nickel was left undetermined). (5) The Phosphate Process. (DIRVELL'S as modified by Dr. JOHN CLARK.*) The solution containing the metals as chlorides, nitrates, or sulphates, which should be moderately concentrated, is mixed with phosphate of ammonia (about five parts of the salt, P0 4 (NH 4 ) 2 H for one part of Ni + Co), and with hydrochloric acid (about five parts of acid of 20 per cent, for every one of phosphate), and the mixture boiled for several minutes. It is then removed from the flame, and, while it is still hot, ammonia is added in small instalments till the precipitate first produced is re-dissolved. The solution is then stirred vigorously ; the cobalt separates out as a beautiful, purple, crystalline precipitate of the composition P0 4 Co(NH 4 ) + H 2 0. A few drops of ammonia are then added, and the beaker is kept on a water-bath for a few minutes to give the precipitate time to settle. The preci- pitate is collected on a filter, washed with cold water, dried, and ignited, to be reduced to and weighed as pyrophosphate, P 2 O 5 2CoO, containing 40'4 per cent, of cobalt. The filtrate con- tains the nickel, associated, however, with a little remnant of cobalt. It is heated to and kept at 100C. until a little of the nickel has come down. The precipitate (ammonio-phosphates of nickel and cobalt), is filtered off, dissolved in hydrochloric acid, and its cobalt estimated as above explained. The filtrate, of course, is united with the principal filtrate. From the filtrate the nickel is best precipitated by sulphuretted hydrogen, i.e., extemporised colourless (NH 4 ) 2 S, as NiS. (See above.) The test- analyses published by Dr. Clark gave very satisfactory results. From these analyses it appears that Dr. Clark prefers to use the metals as chlorides, although the sulphate and nitrate forms, as far as tried, worked well. * " Chem. News," for 1883, part 2, vol. xlviii., p. 262. PYRITES. Ill Dirvell-Clark's certainly is the most elegant of all known processes for the separation of nickel and cobalt. How it com- pares with the nitroso-/3-naphthol and nitrite methods in point of exactitude we are unable to say. In a few trials with Dirvell's original process we always found that a little of the cobalt escaped precipitation, while a little of the nickel passed into the cobalt precipitate. Ex. 35. Analysis of Iron or Copper Pyrites. SUBSTANCE. The student at first may practise on pure iron pyrites. We have a supply of "Rio Janeiro Pyrites" in the laboratory, which (apart from a fraction of a per cent, of chemically combined water, and a little otherwise combined oxygen) is chemically pure. The powdered substance is dried at 100C., and analysed in this condition. A. ASSAY FOB SULPHUR. In 1879, Mr. John M' Arthur, while a student in this labora- tory, made a thorough critical research into methods of pyrites analysis. Of all the methods for the determination of sulphur which he tested, the following gave the best results : Caustic Potash Method. Weigh out 5 grms. of pure caustic potash, free from sulphate, and 3'5 grms. of nitre, and melt these quantities in 1 cc. of water in a large silver basin.* Place 0'5 grm. of the pyrites (weighed in a preparation tube) on the surface of the mixture, stir up with a silver rod, and heat gently until the water is gone, ultimately to fusion, stirring up constantly to prevent spirting. Allow to cool, dissolve up in water, and filter off the oxide of iron ; wash with hot water until the last washings are free from alkali, and acidify the solution with hydrochloric acid. Boil for a few minutes, and add excess of barium chloride. Wash the BaSO 4 thoroughly ; ignite, and weigh. By theory, the weight of the precipitate, multiplied by (S -=- BaS0 4 = )0-13744, gives the weight of the sulphur. But a sulphate of baryta precipitate is always liable to be impure, * Or better, one made of Dittmar's alloy ; 91 of Ag, 7 of Au, and 2 of Ni. 112 EXERCISES IN ANALYTICAL METHODS. and we will take this opportunity for explaining generally what means we have for its purification. If the barium is the thing to be determined, the problem is easy. We need only boil the precipitate, after preliminary washing, in dilute sulphuric acid; or in more difficult cases, dissolve it in hot concen- trated sulphuric acid, and, after cooling, re-precipitate it with water to obtain it in the state of purity. But if the precipi- tate presents itself as a form of sulphur to be determined, the case is more difficult. The impurities likely to be present (we speak in reference to the general case), are nitrate of baryta, chloride of barium, foreign metallic oxides, Fe 2 3 , MgO, &c. The customary mode by which one seeks to remove these is this : The precipitate is pounded up with a rounded glass rod in the crucible, under 10 per cent, hydrochloric acid, and then heated with the acid on a water-bath for half an hour. The liquid is then diluted, allowed to settle in the crucible, the clear liquor is withdrawn with a plain (not a bulbed) pipette, and filtered through a very small filter. The precipitate is then washed twice by decantation with pure water in a similar manner. The liquid, in addition to the extracted impurities, contains some sulphate of baryta. To recover it, the liquid is evaporated in a Berlin basin to a few drops, to expel the free hydrochloric acid, and diluted with water, when the sulphate of baryta comes out as a precipi- tate. It is collected on the small filter, which, with its contents, is placed in the crucible over the main portion, to be dried, and ignited with it. The ignition of a sulphate of baryta precipitate must not be over-done. According to experiments made in this laboratory, long-continued strong ignition in a platinum crucible produces an appreciable proportion of sulphide of barium. A quarter of an hour's exposure to a dull-red heat is sufficient to drive off all the moisture. After determining the sulphur in pure pyrites, you may take up the case of a complex pyrites, and execute the determination in the same way. According to Mr. M'Arthur's experiments, copper, and such small proportions of lead as are likely to be present in an iron pyrites, do not interfere with the method at all. PYRITES. 113 Combustion Method. Before the introduction of the above method, we used to burn the pyrites in oxygen, in a boat stand- ing in a combustion tube, collect the sulphurous and sulphuric acids formed in bromine and water (contained in a bulbed U-tube, followed by another containing iodide of potassium, to catch the bromine vapours and any stray sulphurous acid), and determine the sulphuric acid produced, working up the contents of the two U -tubes separately. The cinders, even in the absence of copper and lead, are liable to contain sulphuric acid; they must therefore be dissolved in hydrochloric acid, and the sul- phuric acid in the solution determined and added on. The method is almost as exact as the caustic potash one, but more troublesome. It comes in handy when the pyrites to be analysed is contaminated largely with coally matter. But if, in such a case, the coally matter predominates, if, for instance, we have to deal not with pyrites proper, but with a pyritic shale or coal, Eschka's method* is preferable. One part of pure carbonate of soda is mixed intimately with two parts of pure calcined mag- nesia. The calcined magnesia of commerce, as a rule, is conta- minated with sulphuric acid ; but many kinds of magnesite are quite free from sulphate, and consequently furnish a suitable magnesia on ignition. Regarding the preparation of chemically pure magnesia, see the Author's " Memoir " in the " Challenger " Reports, p. 16.f 1 grm. of finely powdered substance is mixed intimately in a platinum crucible with To grms. of the mixture, and a layer of the latter is placed on the top of the whole. The open crucible is then placed in a slanting position and heated cautiously, so that the combustion goes on at a low temperature. The mixture is then stirred up with a platinum wire, and the roasting resumed at a higher temperature until the colour shows that the oxidation is substantially at an end. To make sure of all the sulphur being in the sulphate form, the roasted mass is mixed with 0'5 to I'O grm. of nitrate of ammonia, and heated in the closed crucible for ten minutes. It is then lixiviated with water, and in the filtrate * Fres. Zeitschr. for 1874, p. 344. f Perhaps the sulphuric acid of the commercial article could be extracted by boiling with solution of carbonate of potash. I 114 EXERCISES IN ANALYTICAL METHODS. the sulphuric acid determined with chloride of barium. Should it be impossible to procure perfectly pure reagents, the sulphuric acid in the weight of mixed reagent used must be determined by a blank, and allowed for. A similar remark applies to the sul- phates that may be present in the substance. Their sulphuric acid is extracted by continued boiling with carbonate of potash solution, and determined in the filtered solution. We have made a series of attempts to adapt the Eschka method to the analysis of pyrites proper by heating the mix- ture in a combustion tube in oxygen with granulated mixture in front but failed to obtain exact results. The results were from 12 per cent, too low. B. DETERMINATION OF THE METALS. (Only Iron, Copper, and Lead taken into account.) A convenient method of disintegration is to heat a known weight of the pyrites in hydrogen gas to redness, which removes about half the sulphur and part of the arsenic,* and next to apply dilute sulphuric acid to the residue. All the iron dissolves as ferrous sulphate, along generally with a small quantity of copper, which is recovered by precipitation with sulphuretted hydrogen. In the filtrate from the sulphide of copper the iron may be determined by titration with permanganate, unless lime, magnesia, &c., are present and are meant to be determined, in which case the iron must, after peroxidation, be separated from the protoxides and determined, as explained in Ex. 21. The residue left undissolved by the dilute sulphuric acid is dried in the filter, and, along with the filter ash, dissolved in sulphuric and nitric acids, f The nitric acid is removed by evaporation in a Berlin basin, and the residue taken up with a little water. The lead separates out as sulphate, the copper passes into solution, and from it is precipitated by sulphuretted hydrogen. The two sulphides of copper are ignited together in hydrogen, to be converted into Cu 2 S, and weighed in this form (see Ex. 19). * Not the whole, according to Mr. M' Arthur's experiments, f Or else the dry filter and contents together are treated with 1 -5 nitric acid, and the solution is evaporated with sulphuric acid, to obtain the oxides that were present in the quartz analysed. SILICATE ANALYM- 123 the filter, in a platinum spiral, and ignite precipitate and ash in a platinum crucible until constant. Observe that dry silica is a very light, mobile powder, which is easily blown away by the slightest draught of air, and that the ignited substance is highly hygroscopic. A trace of silica passes into the nitrate, from which it can be recovered and determined by adding a known weight of iron in the form of ferric chloride, precipitating with ammonia, filtering off, igniting and weighing the precipitate. What it weighs more than the calculated Fe. 2 O 3 is Si0 2 . The ferric chloride is easily prepared from a known weight (about 50 mgs.) of pianoforte wire, by solution in hydrochloric, and boiling with a few drops of nitric acid. The weight of the ignited silica should be equal to that in the quartz used to within less than half a per cent. (6) Analysis of Felspar, or any other silicate insoluble in acids. Powder a few grammes very finely in an agate mortar, and keep the powder in a preparation tube. Ignite about 1 grm. in a platinum crucible to determine the water (which, in the case of felspar, will amount to very little). The residue can be utilized for the determination of the SILICA AND BASES NOT ALKALIES. Mix with 6 grins, of mixed alkaline carbonates NaKC0 3 (for 1 grm. of substance), and proceed as shown above for " Quartz." Keep the silica in a preparation tube to examine it according to Note 1. Filtrate and Washings. Add ammonia, drop by drop, until alkaline, then boil off the excess of ammonia in a slanting flask, filter off the precipitate (impure oxides of iron and aluminium), wash it thoroughly with hot water, separate the alumina from the rest by caustic potash and determine it, as explained in Ex. 17. If the oxide of iron precipitate is large, it must be re- dissolved in hydrochloric acid, and re-precipitated by caustic potash. The (alumina-free) oxide of iron (as it still contains some alkali and small quantities of lime and magnesia) is again dissolved in hydrochloric acid, precipitated with ammonia (the 124 EXERCISES TN ANALYTICAL METHODS. excess of the latter being boiled off), washed, and weighed as usual. Keep the precipitate to examine it according to Note 2. The ammoniacal filtrate from the iron, containing part of the lime and magnesia, must be incorporated with the filtrate from the first sesquioxides precipitate. In the mixed liquor, after due concentration, the two protoxides are determined as usual. For the determination of THE ALKALIES, apply successively Methods I. or II. and Method III. I. Lawrence Smith's. Mix 0'5 grm. of the very finely pow- dered silicate in a mortar, with 0*5 grm. of sal-ammoniac and 4 grms. of pure granular (precipitated) carbonate of lime ; put the mixture into a platinum crucible, rinse the mortar with a little carbonate of lime, and put the rinsings on the top of the mixture in the crucible. Heat, lid on, first for half an hour over an Iserlohn burner, then for another half-hour as strongly as possible over a gas blow-pipe. The mass only sinters, yet the mineral, if it was finely enough powdered, is completely decom- posed, the alkalies assuming the chloride form. Soak the crucible lid (the lid fre- quently has a film of sublimed alkali chloride on it) and the crucible and con- tents in water until the alkalies can be assumed to have gone into solution ; heat, and filter. From the filtrate precipitate the lime by means of normal carbonate of ammonia at a boiling heat, filter off the carbonate of lime, evaporate to dryness, ignite the residue gently to remove the sal-amrnoniac, and weigh what remains. This residue should consist of only fixed alkali chlorides, but gene- rally it is contaminated with G - 31. lime and magnesia ; hence it is well to re-dissolve it in the crucible in a little water, to add a pinch of pure, lime, and again operate as before with the slaked SILICATE ANALYSIS. 125 mass. The purified alkali muriates are weighed, and the potash in them is determined by chloroplatinic acid. To obtain really exact results with this method, it is necessary to use a peculiar tube-shaped crucible of Lawrence Smith's invention, in which the disintegration can be effected without loss of alkali chloride by volatilization. Fig. 31 shows the whole of Lawrence Smith's apparatus, including the heating arrangement. The crucible a when being heated is suspended slantingly by means of a per- forated iron plate 6, fixed vertically to the slide of the stand. The top d of the Bunsen is shaped so as to yield a thin but broad sheet of flame, which, during the final heating, is kept steady and rendered more effective by means of a roof -shaped chimney c, which is suspended at the plate b over the bottom part of the crucible. The part projecting from the other side of the plate remains relatively cold, so that any volatilized chloride condenses in it and is saved. The Bunsen and chimney com- bined gives a sufficiently intense heat; there is no need of a blow-pipe. According to Lawrence Smith, however, a little of the alkali may remain in the precipitate (as undecomposed mineral). To recover it the precipitate is dried, mixed with a little sal-ammoniac, re-ignited in the crucible, &c. II. Dittmars Modification of the Baryta Process. Mix, say, 0'5 grm. of silicate with 3 grms. of pure hydrate of baryta (Ba(OH) 2 not crystals), and O'lo grm. of anhydrous chloride of barium, and heat the mixture in a platinum crucible so con- structed that an atmosphere of hydrogen can be maintained within it. An Iserlohn burner gives sufficient heat. The fuse may be worked up as in Lawrence Smith's method, but, of course, is available for a re-determination of the silica and of all the bases (except, perhaps, the lime), because the barium is easily eliminated by means of sulphuric acid. The method works very well as one of disintegration; also in this sense, that the platinum crucible remains unattacked. Whether there is really no loss of alkalies by volatilization still remains to be proved. Our experiments in this respect are not quite conclusive. III. Fluoride of Ammonium Method. Mix, say, 0*5 grm. of finely powdered silicate with 3'5 grms. of fluoride of ammonium in a platinum basin (best one with a concave lid, 126 EXERCISES IN ANALYTICAL METHODS. provided with a shielded perforation), and add a few drops of water, so that, on gentle heating, the whole assumes the form of a thin paste. Cautiously evaporate to dryness over an air- bath, stirring up with a platinum spatula, and keep the residue at a dull-red heat until vapours cease to come off. Then allow to cool, moisten the residue with a few drops of oil of vitriol, and heat again until the free sulphuric acid is expelled. All these operations must be carried out in an efficient draught place. After the mass has cooled down, dissolve it in hydrochloric acid. Take care not to completely decompose the sulphate of alumina, or else the alumina formed will not be soluble in dilute acid. There should be no residue ; if there is one, it must be filtered off, and worked up with fluoride of ammonium. In the hydro- chloric solution of the sulphates, all the bases can, of course, be determined by proper methods. If you care only for the alkalies, precipitate the sulphuric acid by chloride of barium ; then, without filtering off the sulphate of baryta, the insoluble oxides by means of pure (alkali-free) baryta water (hot), and filter. From the filtrate eliminate the barium and calcium by carbonate of ammonia, and manipulate the filtrate as just ex- plained for the Lawrence Smith method. Notes. 1. The silica must be quite soluble in strong boiling solution of carbonate of soda. When treated with fluoride of ammonium, like a silicate, it must leave no residue of sulphate. If these tests do not come out right, the analysis had better be commenced anew. 2. The oxide of iron ultimately obtained always contains a little silica, which remains when the ignited precipitate is dis- solved in strong hydrochloric acid. It must be collected on a filter, washed, ignited, and weighed, and allowed for. In the hydrochloric acid solution the iron may be determined titri- metrically for a check. If the weight of Fe 2 O 3 as thus deter- mined, together with that of the silica, is less than the total precipitate, the deficit must be put down as so much alumina. Calculation of the Results. Report your results in per cents, of the substance analysed, and then, in order to find the formula, divide each of the CHROME IRON ORE. 127 percentages by its formula value, and translate your analysis into a formula "Si0 2 (i.e., a times 60 parts of silica) + &M"' 2 3 + cM"O + tZR 2 O ; where M'" 2 3 means A1 2 3 or Fe 2 O 3 , M"O means CaO or MgO ; R 2 means K 2 O or Na^O. Next divide all the co-efficients by a, so as to bring the formula into the form !Si0 2 + b'W" 2 O s + c'W'O + d'R 2 O, and multiply successively by 2, 3, ... until the factors become sufficient approximations to integers. According to Tschermak, felspar is in general a crystal com- pound of three species, namely (1) Anorthite, an ortho-silicate of the formula (Al. 2 O 3 .CaO).2Si0 2 ; (2) albite, a sesqui-silicate (Al 2 3 .R,O).6SiO 2 , where R 2 is chiefly Na 2 ; and (3) microline, which has the same formula as albite, only R 2 is chiefly K 2 . [All triclinic ; orthoclase is an oblique prismatic isomer of microline.] Hence the correct way of calculating a felspar analysis is, after having found the co-efficients of the formula aSiO 2 + &M 2 3 , &c., to next combine the c (CaO or MgO) with as much silica and alumina as they need to become anorthite, and to deduct these quantities aSiO 2 and /3M 2 3 from the original values. The remainder (a a)Si0 2 + (b /3)M 2 3 + cZR 2 should fall in with the formula of albite. Silicates decomposible by Hydrochloric Acid. These we leave in the hands of the student. Most of them, however, are mixed with undecomposible silicates or quartz, so that what comes out as "silica" includes this admixture. To analyse such a mixture, boil it with concentrated solution of carbonate of soda, filter through a filter kept hot by a steam jacket, wash the residue, first with solution of carbonate of soda and, ultimately, with carbonate of ammonia. (With plain water a turbid filtrate is obtained). The residue is ignited, and weighed as silicate or quartz, as the case may be. In the filtrate the dissolved silica, i.e., the silica of the decomposible silicate, is determined as usual. 128 EXERCISES IN ANALYTICAL METHODS. Ex. 38. Analysis of Chrome Iron Ore. A. ASSAY FOR CHROMIC OXIDE. Preliminaries. (1.) Procure a real average sample, which should not be too small ; reduce it to a powder of about the fineness of sea sand without rejecting any residue, and bottle up the powder as "substance." In about 5 grms. of this "substance" determine the moisture by drying at 100C. It usually amounts to very little, but it must be reported. The analysis proper is referred to substance dried at 100C. (2.) Flux. Fuse together equal weights of borax -glass and the mixture NaKC0 3 in a platinum crucible, pour out the fuse into a platinum basin, and preserve it in a well-stoppered bottle as <: chrome ore flux." Analysis. Reduce somewhat more than 1 grin, to an impalp- able powder, dry this powder completely at 100C., and put it into a small narrow preparation tube ; cork up, and determine the exact weight of the whole, taking care to weigh also the emptied tube, so as to ascertain the exact weight of ore used. Fuse 8 grms. of flux (or 5 grms. each of the two components) in a platinum crucible of 50 cc.'s capacity over a powerful Iserlohn lamp, allow to solidify, and put on the top of the fuse the contents of the weighed preparation-tube. Heat the crucible, lid on, until the flux is quite fluid and the ore has sunk to the bottom. Now take off the lid, place the crucible slantingly on a platinum triangle, and the lid in front of it horizontally, and heat, stirring up frequently with a thick platinum wire, until all the ore is dissolved. This, unless the lamp is in first-rate condition and the gas pressure high, requires the use of the blow-pipe. Con- tinue heating (in the presence of air, and stirring up occasionally) for about half an hour, when the chromium can be assumed to be all converted into alkaline chromate. Now allow to solidify, add 5 grms. of double carbonate KNaC0 3 , fuse up again, and pour the fuse into a large platinum dish. Place the crucible in a basin with water, and heat until all the adhering mass is softened up ; wash what sticks to the crucible into the basin, and remove the crucible. Then add the bulk of the fused mass, and allow to CHROME IRON ORE. 129 stand on a water-bath until the whole is completely disintegrated ; then filter, and wash with hot water as long as the runnings are yellow. The brown precipitate consists of basic borates and hydrates of the oxides Fe 2 O 3 , A1 2 3 , MgO, and often includes silica. It always contains a few mgs. of oxide of platinum, although the crucible is not visibly attacked in the fusion process. It is dis- solved in hot hydrochloric or dilute sulphuric acid ; the platinic oxide dissolves along with the bases ; any undecomposed ore becomes visible ; it is washed, ' ignited, and weighed. In a carefully conducted fusion it will never amount to more than 2 nigs. Such small quantities may be taken into account as containing 50-60 per cent, of chromic oxide, without serious error. Larger quantities must be disintegrated and the products added to the main quantities. If a complete analysis be contemplated, the acid solution of the brown precipitate may be utilized for a determination of the iron. For this purpose it is treated with sulphuretted hydrogen hot, the mixed precipitate of sulphur and sulphide of platinum filtered off, and the iron in the filtrate titrated with perman- ganate or "bichrome." (See Ex. 18). The yellow filtrate contains the chromium as chromate, along with chiefly borate of alkali. For the determination of the chrortiium concentrate it by evaporation to about 200 cc. Bring it to an exactly known volume or weigh it, to be able to fractionate it quantitatively. A fraction corresponding to about O'l grm. of ore serves for a preliminary determination of the chromium. For this purpose the liquid is acidified strongly with dilute sulphuric acid, and then mixed with a known weight of standardized ferrous sulphate, sure to be more than sufficient to reduce the chromic acid to chromic oxide. 12 grms. of salt per 1 grm. of chromic oxide Cr 2 3 are more than sufficient. The surplus ferrosum is determined by titration with standard bichrome solution (see Ex. 18). The determination is then repeated on a larger scale, the preliminary trial enabling one to calculate the ferrous salt, and thus to avoid immoderate excess. What is left after the preliminary test suffices for two exact titrations. The chromium K 130 EXERCISES IN ANALYTICAL METHODS. is calculated as Cr 2 3 from the ferrosum oxidized by the chromic acid. 6Fe X 0'4530 = Cr 2 3 and Cr 2 s X 2'208 = 6Fe. As a matter of principle a synthetically standardized solution of bichrome ought to be taken as the fundamental standard, but bichromate of potash is not easy to obtain in a state of undoubted purity and correctness of composition.* B. COMPLETE ANALYSIS. For a complete analysis the ore might be disintegrated as explained for the assay, but it is better to effect a special dis- integration in which the addition of the extra 5 grms. of double carbonate is omitted to avoid excessive alkalinity, which brings unnecessarily large quantities of alumina and silica into the yellow solution produced from the fuse. Before going further, let us observe that even the best borax-glass of commerce usually contains silica and alumina. These do not interfere with a mere assay, but, in a complete analysis, must be determined by a "blank," and allowed for. Trommsdorff's "kali boricum" is generally free from these impurities, and 6'5 grms. of it, fused up with 3*5 grms. of NaKC0 3 , give a flux equivalent to what is produced from 10 grms. of the components of the ordinary flux. But the kali boricum cannot be expected to be of constant quality. I therefore propose for the future to make the flux for the complete analysis out of pure boric acid, B 2 3 3H 2 O and double carbonate NaKC0 3 . 124 parts of the acid and 163 of the carbonate should yield 174 parts of a flux equivalent to 100 of borax glass + 100 of double carbonate, i.e., 200 parts of ordinary assay-flux ingredients. Analysis of the Brown Precipitate. Dissolve in hydrochloric acid, evaporate to complete dryness, moisten the residue with strong hydrochloric acid, and give the alumina and iron sufficient time to become chlorides ; then dilute, filter off the silica and weigh it. From the filtrate eliminate the platinum by sulphuretted * The most certain method is to standardize the iron salt by means of a known weight of pure chromic oxide, fused up with flux, &c., like a chrome ore. Pure chromic oxide can be obtained by the ignition of pure mercurous chromate pro- duced best from pure bichromate of ammonia. But this method is very trouble- some. CHROME IRON ORE. 131 hydrogen, and, after filtration and expulsion of the dissolved sulphuretted hydrogen, re-peroxidize the iron by means of a granule of chlorate of potash. Then separate the sesquioxides ( ALO 3 and Fe. 2 O 3 ) from the protoxides by two successive applica- tions of the sal-ammoniac and ammonia process (see Ex. 21, p. 53). Separate the iron and alumina by caustic potash twice, according to Ex. 17, and separate and weigh the Fe 2 O 3 and the Al.,0 3 as there explained. The sal-ammoniac filtrate obtained from the iron is added to those produced before. The united .sal-ammoniac liquors are evaporated to complete dryness, the sal-ammoniac is chased away by calcination, and in the residue the lime and magnesia are determined by the usual methods. The solution of the calcined residue (in water and hydrochloric acid) may contain a little alumina; it is separated out by ammonia and weighed. The lime usually amounts to very little ; the magnesia to a good deal. The ignited oxide of iron must be dissolved in strong hydro- chloric acid. If a residue remains, it is collected, weighed, and in the calculations allowed for as so much silica. Analysis of the Yellow Filtrate. If a whole gramme was taken for analysis, one-half of this liquor suffices for the deter- minations, so that the other half can be reserved for checks. Add to it excess of sal-ammoniac (see Ex. 17, second paragraph), filter off the precipitate formed, ignite, and weigh it. If the flux pre- scribed for complete analysis was used, the precipitate amounts to very little, and may be put down as so much alumina. If the more strongly alkaline assay-flux was employed, the precipitate amounts to more, and it may be necessary to treat a finely-powdered aliquot part of the ignited precipitate with fluoride of ammonium (as in a silicate analysis, see p. 125), and determine the alumina in the desilicated residue by ammonia. The silica part is determined by difference. The filtrate from the precipitate produced by sal-ammoniac usually contains part of the magnesia. It is determined by adding enough of strongest ammonia, and then applying phosphate of ammonia as usual. The chromium we suppose to have been determined in a separate portion by the assay- method. 132 EXERCISES IN ANALYTICAL METHODS. The student, for his information, may use the reserve-half of the yellow filtrate for a determination of the chromium by means of mercurous nitrate. If the dissolved alumina was found (in the analysis of the first half) to amount to a mere trace, the yellow liquor is simply acidified with pure (N 2 O 3 -free) nitric acid ; the precipitant added and allowed to act at water- bath heat, until the precipitate, which has at first an impure colour, has assumed the form of red crystalline Hg 2 O Cr0 3 . If tangible quantities of alumina, &c., are present in the liquor, these must first be eliminated by means of excess of nitrate of ammonia (which in this case must be substituted for the chloride recommended above), and the mercurous nitrate applied to the filtrate. The mercurous chromate is collected on a small filter and washed with a hot very dilute solution of the precipitant ; it is then transferred, moist, to a platinum crucible (along with the filter), and ignited in a very efficient draught place, which is sure to carry away the poisonous mercury vapours. The residue is weighed as Cr 2 3 . A small quantity of the chromic acid, how- ever, escapes precipitation. To recover it, eliminate, first the mercurosum as calomel, then from the filtrate the mercuricum by sulphuretted hydrogen ; then boil down the filtrate to a sufficiently small volume, precipitate the chromium (now present as chromic salt) with ammonia, taking care to boil off the excess of ammonia before filtering, collect, and weigh the precipitate as Cr 2 3 . An impure chrome ore is liable to contain Chemically combined water. To determine it, heat some 3-4 grms. of the "substance" (see at the beginning of the article) in a platinum boat in a combustion tube, collect the water in a tared chloride of calcium tube, and weigh. A soda-lime (J-tube may be joined on to the chloride of calcium one to determine any carbonic acid that may come off. Calculation of the Results. We have good reasons* for assum- ing that all chrome iron ores contain their chromium as Cr 2 O 3 , while the iron in general is present, part (as a rule the greater part) as FeO, part as Fe 2 O 3 . Hence, in calculating results, we * See Rammelsberg's Mineral Chemie. CHROME IRON ORE. 183 naturally translate the data of the analysis into percentages of (SiO,), Cr,O 3 , A1A, MgO, (CaO), and FeO, and next contrast the sum of these percentages with 100, allowing for experimental errors, the aggregate effect of which on the sum may amount to 1 per cent, even in a good analysis. Occasionally it may amount even to a little more, but for greater definiteness in our explana- tions we will fix our toleration at 1 per cent., and say that, if the percentages (with all the iron as FeO) add up to more than 101, some component or components must have been over-estimated, and the analysis consequently must be repeated. If the results, even with all the iron calculated as Fe 2 3 , fall short of 99, this, of course, again condemns the analysis. Hence, only the inter- mediate cases need be considered in the following, but we prefer to pass to an example, and assume that the results of the analysis had come out, as shown in column I. of the following table: I. II. Chromic oxide, 55-14 - 0'3622 x Cr 2 O 3 Alumina, ... ... 575 = 0*0563 x A1 2 3 Magnesia, "... ... 9*39 0*2327 X MgO Iron (met.), 22*44 - 0*4007 xFe Oxygen, by ditf., ... 7*28 = 0*4550x0 100-00 A glance at column II. shows that the iron apparently is present as a mixture of xFeO and 2/Fe 2 O 3 , and the x and y are easily found. We have 0*4007 x FeO' + 0*0543 X O= j 0*0543 X O 4- 0*1086FeO = 0*0543 x Fe 2 3 \pius 0*2921 x FeO ; or 8*69 of ferric oxide plus 21*03 of ferrous, together = 2972 = 22*44 + 7*28, which, of course, brings up the results to exactly 100. But as our analysis reports no components foreign to the species, we had better test it further by seeing whether it is compatible with the general formula M 2 O 3 + RO for the pure mineral, which, referring to 100 parts of our ore, assumes the form ( 0*4185 x (Cr, or A1 9 )CU (0*2327 x MgO ) 1 xxFe 2 3 $ } 2/xFeO | Obviously x and y must be so chosen that 0-4185 + x = 0-2327 + 27 (% L ) and 2z + ?/ = 0-4007 (Eq. II.) 134 EXERCISES IN ANALYTICAL METHODS. both subject to the third condition that the iron-oxygen de- manded by the formula, which is (3x + y)x(O = 16), lies between (say) T'28-1 and 7'2S + 1. From Eqs. I. and II. we have 0-4185 + x-y = 0-2327, (I.) = 0-4007, (II.) by addition, 3x = 0*2149. Whence a? = 0*07163 and by substitution into II., ?/ = 0*2574. Hence (3x + y) = 0*4723; and (3x + y)I6 = 7'56 which falls within our limits. To check the computation we make the additions 0-4185 + 0-0716 = 0-4901 0-2327 + 0-2574 = 0-4901; (the result is as it should be) and compute 0-07163 x(Fe 2 3 = 160) = 11-46 per cent, of ferric oxide, 0-2574 x(FeO = 72) = 18'53 ferrous together, ... ... 29'99 instead of 2972, which brings up the results to 100*28 instead of 100. If the analysis of an apparently pure chrome ore does not submit to this mode of manipulation, and we have made quite sure of owr analysis, all we can do is either to calculate the FeO and Fe 2 3 from the analysis, and just report the results as they stand, or, if we find it worth our while, to try and interpret the results on the basis of some reasonable hypothesis on the proximate composition of the ore. With a palpably impure ore the latter course is the only one that is open to us, if we wish to discuss the analysis mineralogically. Unfortunately, there is no method for separating the gangue from the pure mineral, so that one could, for instance, see directly whether the silica is there as so much quartz or in some other form. From our large toleration for the sum of the percentages, the student must not conclude that any great error can be tole- rated in the chromic oxide ; its percentage in a good series of analyses should not be uncertain by more than 0*2 unit. For an exercise in chemical arithmetic, I append the results SEPARATION OF LEAD AND ANTIMONY. 135 of the analysis of a chrome ore which Mr. Cullen carried out lately under my direction : He found in 100 parts of ore, dried at 100C., Chromic oxide, ... ... ... ... 52*4, tower charged with pumice soaked in vitriol ; .S CO fl P Q p, C5 H pC 03 g i I- * . ^ ;| ^ s-> **? ' 1 ~ 5 c8 w PH 0) O ^ O g J ^ ^ S ! S ,3 f ill 2 f3 c6 t,_, 1 S fe I S S i ? S ^ " I u r*'i-? | -3 fe -f e FERROUS OXALATE. 145 tube to a length of about 35 ctms., measuring from the shoulder of the exit end. As shown in Fig. 33, charge the tube with oxide of copper which has been previously heated strongly in pure air until free from carbonic acid, attach an india-rubber tube to the exit end, and, by means of it, a small supplementary chloride of calcium LJ-tube. The india-rubber tube is wired to the combustion tube, but not to the supplementary U-tube. Then heat the whole gradually to redness, while a slow current of chloride-of-calcium-dried air passes through the tube. After a while the flames up to a point a little to the right of e are turned off, but the oxide of copper is kept red hot until the outgoing air ceases to precipitate baryta water. In order now to make sure that the joint g (which should be maintained at a heat little short of 100C.) is dry, tare the supplementary U-tube, attach it again for ten minutes, and weigh it. It should gain nothing, or at most half-a-mg. The drying of the joint g is effected within a small chamber of asbestos pasteboard, as described in Ex. 28. The chamber stands on a thin metallic plate and is heated by a special lamp, the temperature being controlled by a ther- mometer. During the above preliminaries the empty part of the com- bustion tube should be protected against radiant heat by means of a screen, and the time be utilized by weighing the absorption apparatus and the substance to be analysed. The latter is weighed in the tared boat, and then kept dry within a close tube made of two lipless test tubes fitting close one over the other. Begin by attaching the absorption apparatus, taking care to wire the U-tube end of g this time, and making sure of the tightness of joints generally. After having established, if necessary, a sufficiently low tem- perature in the respective parts of the tube, insert the boat with the oxalate, and at once let a slow current of air go through the apparatus. Then bring up the temperature of all the oxide of copper to redness, and next heat the empty part of the com- bustion tube by a single Bunsen lamp. The oxalate gets gradually heated and decomposed by radiation to some extent ; L 146 ELEMENTARY ANALYSIS. when the gas evolution becomes slack, heat the boat directly, but always cautiously, so that the gas enters the Liebig's bulbs at the rate of, at most, two bubbles per second. Finally, heat up the boat to full redness so as to insure a complete combustion. During the combustion the current of air should be slow, but the pressure of the air, from the gas holder to a point as close to the entrance end of the combustion tube as practicable, should be considerably higher than what is needed to over- come the resistance of the potash. This is best attained by means of a stopcock, but a screw-clip on a short india-rubber connection will do. When the combustion is completed, turn on the air more fully, and keep it going until, for every one volume of empty space from c to the potash bulb, about three volumes of air have been utilized for sweeping out the products. Then take off the absorption apparatus, close their ends with india-rubber caps, and take them to the balance. The supple- mentary chloride of calcium tube is at once substituted for the working one, so as to keep the moisture out of the combustion tube, and only then the gas turned off. The tube now is ready for another analysis. Weigh the absorption apparatus, and from their gains calculate the carbonic acid and water as such in terms of 100 parts of substance. W'eigh the boat containing the iron as Fe 2 3 , and, by multiplying with 07, find the corre- sponding weight of metallic iron. The three results conjointly should come up to 100 0'5, and agree with the numbers deduced from the standardization. Ex. 42. Analysis of Cane Sugar or Mannite. THE operation is the same as in the case of the preceding- Exercise, with this difference that the moisture in the substance is determined separately by drying at 100C. ; or that the substance is made dry at 100C., and analysed in this condition. It is as well also to use a somewhat longer column of oxide of copper, so as to be able to let the combustion proceed at a con- venient rate ; 20-25 ctms. of oxide of copper are sufficient. The BEN20IC ACID, ETC. 147 percentage of carbon should agree with that demanded by the formula to 0'15 per cent. ; the hydrogen to within 0'20 per cent. Ex. 43. Analysis of Benzole Acid or Naphthalene. Ix either case take the commercially pure substance, and purify it by resublimation. The modus operandi is the same as in the preceding Exercise ; only the column of oxide of copper must be 40-50 ctms. long, the combustion be conducted very slowly, and, at the end, the carbon concealed in the oxide of copper be burned off by means of pure oxygen. The latter, after having done its work (which is seen by the outgoing gas at the end kindling a glowing splinter of wood), is swept out by air. This is absolutely necessary, because oxygen is more soluble in caustic potash solution than air is, and the oxygen which would remain in the empty parts of the absorption tubes weighs more than its own volume of air. Ex. 44. The Old Method of Organic Analysis. THE student should not leave off this part of his laboratory studies without having made himself familiar with this method, because, in his practical career, it may be the only one which the apparatus at his disposal enable him to execute. Requirements. (1.) A combustion tube drawn out, as shown in Fig. 34. (2.) Powdered oxide of copper. (3.) Granulated oxide of copper. Each must be prepared for use by heating- it to redness in a crucible, and then bottling it up while still hot in small lipless phials pro- . , T .,, FIG. 34. vided with good dry corks, or what is better, by heating it in a combustion tube like tube a b of Fig. 35, in a current of dry air until all the carbonic acid is proved to be away ; the water then is sure to be gone. 148 ELEMENTARY ANALYSIS. These tubes should be pretty wide, and no longer than necessary. At the end of this preliminary operation, a is closed with an india-rubber cap, b with a dry cork and a protection tube passing FIG. 35. through it, which is charged with soda-lime at the outside, and with calcium chloride at the inside end. (4.) A small narrow tube for the substance. (5.) A brass or copper mixing wire. (Compare p. 80). Modus Operandi. After having shoved a small stopper of ignited asbestos into the narrower end of the combustion tube to keep the bayonet point free of oxide of copper, put into the tube about 10 ctms. of powdered oxide of copper, then pour in about one-half of the substance to be analysed; then add another 5 ctms. of oxide of copper, and mix the oxide of copper and substance with the wire. How this is done is better shown than explained in words ; it requires practice to do it well. Then put in more oxide of copper and the other half of your substance, and mix as before. Then insert an (anhydrous) asbestos stopper, and fill up the rest of the tube with granulated oxide of copper, leaving space for a terminal stopper of asbestos. This being done, the tube is closed by means of a dried cork and put aside while the emptied substance tube and the absorp- tion tubes are being weighed. A very carefully bored cork has been prepared to fit the outlet end of the combustion tube and accommodate the chloride of calcium tube, and all the time been in an air-bath at 105C., the hole being kept from collapsing by an inserted bit of rounded glass rod. To carry out the com- bustion, begin by tapping the tube horizontally on a table so as to produce a canal over the powdered oxide of copper ; then take the prepared cork out of the bath, and, after turning in the exit end of the chloride of calcium tube, insert it into the end of the combustion tube. Be sure of the absolute perfection of this joint, or the analysis will be a failure. The potash bulbs and soda-lime U-tube are now attached, the tube is placed in the combustion furnace, and the combustion COMBUSTION OF LIQUIDS. 149 started. First heat the granulated oxide of copper, proceeding from the open end of the tube towards the other. When this plain oxide of copper is red hot, next heat the relatively pure oxide of copper at the tail, and then, gradually, the mixture, proceeding from the end nearest the opening towards the closed end. After a time the carbonic acid will cease to bubble through the potash. At this stage attach the air gasometer to the end of the bayonet point by an india-rubber tube, secure the joints, and make sure that the air is at high pressure. Now, while keeping the closed air stopcock or clip in one hand, break off the tail of the bayonet end in the india-rubber, and then admit air at the proper rate. The rest requires no explanation. With solids, like sugar or mannite, very exact results should be obtained without the use of oxygen. Ex. 45. Analysis of Liquids. ABSOLUTELY or relatively non-volatile liquids, such as fatty oils, high-boiling paraffins, glycerine, &c., can be treated like solids, and burned in a boat, as explained in Exs. 41-43. Very volatile liquids, like aldehyde, &c., demand special methods, which we do not propose to consider. Liquids of the order of alcohol, benzole, &c., are manipulated in small bulbs with long capillary necks, as represented in Fig. 36. To make such a bulb, take a tube of 7-8 mms. outer diameter, and draw it out on both sides of a selected point, so as to produce two long cylindrical appendages. One of these is fused off, and then the cylindrical converted into a spherical or ovoid bulb before the blow-pipe. To fill a bulb, heat it over a flame, so as to expel part of its air, and dip it with its open end into a supply of the liquid while cooling. More or less of the liquid will, of course, be sucked in, but a single drop within the bulb proper is sufficient to enable one to drive out all the air with the vapour of the liquid ; and by re-inverting the bulb over the supply, to cause it to fill com- pletely. But it had better be only three-fourths full. The bulb is sealed up before it goes to the balance. The combustion 150 ELEMENTARY ANALYSIS. is conducted in a combustion tube, closed at one end, by means of granulated oxide of copper prepared for use, as shown in the last exercise. About 10 ctms. of this are introduced into the tube ; the bulb is opened by cutting off about two-thirds of the tail, and both it and the cut-off part are allowed to slide down the tube and rest on the copper oxide. Should there be a stopper of liquid in the neck of the bulb, it must be driven into the bulb proper by judicious heating of the neck, the whole allowed to cool, and only then opened. After the introduction of the bulb, the tube is filled completely with granulated oxide of copper, and, supposing the absorp- tion apparatus to have been attached, the front part of the oxide of copper brought up to a red heat as quickly as possible. A point quite near the closed end of the tube is then heated by means of a single flame to hinder unburned vapour from getting past this point. The rest requires no ex- planation ; any thoughtful worker will easily find out what we might say, by himself. The great difficulty, of course, is to keep a proper control over the rate of volatiliza- tion of the liquid. A couple of screens should be kept ready to be applied when and where necessary. Sometimes it is ex- pedient to distribute the liquid in two bulbs separated by a column of oxide of copper. Erdmann and Marchand long ago intro- duced a refinement upon this method, which consists in this, that the bulb is filled almost completely, so that only a part of the tail, which should be capillary, is left empty, and introduced in the sealed-up condition. After the front oxide and some of the close end have been brought up to redness, FIG. 36. NITROGENOUS SUBSTANCES. 151 the bulb is made to burst by temporary and local applica- tion of heat. The respective part of the combustion tube should not lie in a gutter, but be freely suspended to admit of prompt alteration of temperature. The bulb, of course, must be almost absolutely full of liquid, so that it is burst by the steady expansion of the liquid, and not exploded by vapour-pressure. The modification certainly is an improvement, but requires con- siderable experimental ability for its successful application. The combustion tube at its closed end is provided with a bayonet point, so that when all the vapour is burned, air can be driven through the tube to clear it of carbonic acid and water ; or oxygen, to burn away any deposited carbon, if the liquid is richly carboniferous. Ex. 46. Analysis of Nitrogenous Substances. IN the combustion of a nitrogenous substance with oxygen or oxide of copper, part of the nitrogen is converted into NO, which in its turn (in oxygen) passes into N 2 O 4 , which latter goes, part into the water of the chloride of calcium tube, part into the potash-bulb, and makes the weight of both too high. To avoid this error, place a closely-packed spiral of copper gauze in the exit end of the combustion tube, and keep it red hot during the combustion ; all the nitrogen oxides then are reduced to nitrogen. The combustion tube should be some 12 ctms. longer than usual to accommodate the copper spiral. The copper spirals are made as follows: Heat the copper gauze in a large Bunsen flame, one part after another, so as to burn away the trace of oil which is always on it. Then cut it into strips of the breadth of 1012 ctms., and roll up each into a closely-packed spiral, the thickness of which is such that it just goes into the combustion tube to be used. Place some half-dozen of such spirals in a combustion tube, and heat them in pure dry hydrogen gas until they are quite bright. Then turn off the hydrogen, and pass (air-free) carbonic acid gas over the spirals, through a T-piece inserted between the hydrogen apparatus and the tube, to expel the hydrogen. Then send a slow current of dry air over the spirals 152 ELEMENTARY ANALYSIS. until the one nearest to the air-entrance is half-spoiled by oxidation. The rest are then sure to be free from occluded hydrogen. While keeping the half-spoiled spiral hot, allow the rest to cool in a slow current of nitrogen. After cooling, pre- serve the spirals in a well-corked glass tube. If a spiral has been (even thus) preserved for a long time, it must be kept in a drying chamber at 110C. for about half an hour immediately before being used. The combustion is conducted as usual, only take care that the copper spiral is hot during the whole of the process, and if you use oxygen, see that it does not get oxidized prematurely. Ex. 47. Determination of Nitrogen in Organic Substances by Dumas' Method. Method. The substance is burned with oxide of copper in an atmosphere of pure carbonic acid, the nitrogen oxides are destroyed by red hot copper, and the out-going gas carbonic acid, steam, and nitrogen is collected over strong caustic potash solution, and measured. From the volume (V cc.), the pressure (P mrns.), and the temperature (tC.) of the gas, its weight in milligrammes is calculated. If 50 per cent, caustic potash ley was used to collect the gas, the tension of the vapour of water can be neglected ; hence, if the gas inside is brought to the atmosphere's pressure, the height of the barometer, reduced to 0C., can be substituted for P. If the barometer at t stands at B t , the equivalent height of mercury of is B , and B " = 1 + B 00018 - B ' (1 - - 00018 '>' ( P tic ^-) Requirements. (1.) A combustion tube, 80 ctms. long, with a bayonet point, as shown in Fig. 34. (2.) Copper spirals as above. (3.) Mixing wire. (4.) A carbonic acid apparatus, according to Kipp. To obtain pure gas, boil the pieces of marble in water DUMAS' METHOD. 153 immediately before use, and apply the air-pump after sufficient cooling so that the hidden air is sucked out through the water ; on the other hand, boil a supply of hydrochloric acid (it should contain 5-7 per cent, of HC1), and allow to cool in a stoppered flask. Wash the gas with water, and then filter it through cotton wool. Keep the gas going for a considerable time before using it, so as to be sure of its containing no air. (5.) A Schiff apparatus for collecting the gas, attached to the end of the combustion tube, as shown in Fig. 37. (6.) 50 per cent, caustic potash ley (not soda) to charge the Schiff. Modus operaind/i. Charge the tube with oxide of copper and substance, as explained in Ex. 44 for the old-fashioned method of organic analysis, but put a spiral of copper into the outlet end of the tube. To the end of the bayonet point attach the exit tube of the carbonic acid apparatus ; the other end communi- cates by a cork and quill tube inserted with the mercury joint m of the Schiff. An im- portant point to attend to is that the entrance of the carbonic acid into the tube is checked by a glass stopcock or screw-clip at a point as close to the entrance end as possible. The gas- pressure up to this clip or stopcock should be consider- ably higher than that prevailing in the tube. Begin by send- r - T FIG. 37. ing carbonic acid through until all the air is expelled, i.e., until a considerable quantity of gas evolved is completely absorbed by caustic potash. This point being reached, shut off the carbonic acid almost but not quite completely, and next heat the copper spiral to redness ; then heat the plain oxide of copper in front, then the plain oxide of copper in the tail end, and, lastly, the mixture, beginning at the outward end, and progressing towards the inward one. When the evolution of gases ceases, pass a 154 ELEMENTARY ANALYSIS. stronger current of carbonic acid through the tube until all the nitrogen can be assumed to be in the measurer. Now detach the latter from the combustion tube, shake the gas with the alkali to ensure complete absorption of the carbonic acid, and keep the apparatus in a place of constant temperature for about one hour, with the levels approximately adjusted. Lastly, establish exactly one atmosphere's pressure inside the nitrogen tube ; take t, P, and V, and calculate. To learn the method, begin with pure sugar, until you manage to obtain no more than, say, 0'5 cc. of nitrogen at most from 0*5 grm. of this substance ; then pass to hippuric acid, or some other suitable example. GAS ANALYSIS. GENERALITIES AND THEORY. GAS ANALYSIS, in the customary acceptance of the term, com- prises only the gas volumetric methods for the analysis of gaseous bodies. The mere volume of a gas, of course, conveys no information regarding its mass. In gas analysis, indeed, to " measure " a gas means to determine its volume V and its tension (pressure) P at a fixed temperature t. In reference to any given species of gas oxygen, air, the contents of a gas-holder, &c. the two quan- tities, V and P conjointly with the temperature mark t, do define the mass of a given quantum as definitely as the weight does, because, assuming in each case two of the variables to have fixed values, a greater weight of that kind of gas would demand a greater volume or a greater pressure, or a lower temperature. Gas analysis, however, as a rule, has only to do with compara- tive measurements ; and to define the relative masses of two or more gas quanta I., of this kind ; II., of a second kind ; III., of a third, &c., it suffices to state the volumes V 1} Y 2 , V 3 . . which they would occupy if they were measured all at the same tem- perature , and the same pressure P . This at once suggests two important questions : 1. Supposing the several bodies of gas were mixed together, would the volume V of the mixture (at t and P ) be equal to the sum V x + V 2 + V 3 . . ? 2. What is the relation between the ratios V x : V 2 : V 3 . . on the one hand, and the values P and f on the other ? In answering these questions, we may and will confine our- selves to the relatively limited area of combinations of tempera- ture and pressure which occur in practical gasometry. This 156 GAS ANALYSIS. comes to the same as saying that pressures above, say, two atmospheres for the vast majority of purposes the line might be drawn at one atmosphere and temperatures below 0C. are for us out of court. Let it also be understood that whenever we refer to a combination of t and P in a gas, this combina- tion is tacitly assumed to be such as to keep the gas on the safe side of condensation into liquid. On the basis of these restrictions, we will now proceed, in the first instance, to answer the second question in reference to the so-called permanent gases, i.e., to the gases H 2 , O 2 , N 2 , CO, and any mixture of two or more of them. In a body of permanent gas, whose temperature is kept con- stant at t m the volume or the pressure may assume any value ; but whenever the volume assumes a certain value V, a certain definite value P, for the pressure follows, and vice versa, and we always have VP = const. ... ... ... ... (Boyle's Law.) The constant obviously has two denominations, inasmuch as it represents numerically both the volume at unit pressure and the pressure at unit volume (assuming for calculating purposes the law were of unrestricted applicability). It depends on the magnitude of the units chosen whether the denomination is physically correct ; but in any case the V P for, say, a given quantum of air (there is unfortunately no name for the compound quantity), together with the given temperature t M defines its mass. The relation, at a given pressure P , between temperature and volume is not so easily explained. We could not reasonably expect a simple relation between the quantity V, and the mere numerical index t, as read from the mercury thermometer. For our purpose, however, it is sufficient to know that, as a matter of exact experimental research the relation between / and V is the same in all permanent gases, in this sense that supposing V x and V 2 to be the volumes at ^ and t, 2 degrees respectively, the ratio V x : V 2 depends only on t L and t%, and not on the nature of the gas. Hence, supposing we agree upon measuring temperatures by the volumes at these of a given quantum of any permanent gas (kept GENERALITIES AND THEORY. 15? at a constant pressure), and call these temperature values T, we have generally T 1 : T 2 = V t : V 2 (Gay-Lussac's Law). To h'nd a convenient unit for these T's, let us remember that according to the concordant experiments of Rudberg, Magnus, and Regnault, all permanent gases when heated from 0C. to 100C. (we will call the gas thermometer readings T and T 100 respectively) expand in the ratio of very nearly 273 : 373. Hence, taking T = 273, we have T 100 = 373, or, for the two cardinal points, T = 273 + t, where t stands for the reading of the centigrade mercury thermometer. By a happy accident the relation holds practically for all temperatures that we need care for. (It is good practice for the student to develop the corre- sponding formula for the Fahrenheit scale.) The values T are called "absolute temperatures," because we have theoretical reasons for assuming that they really measure the respective temperatures in the sense of mathematics. By combining the two laws we have VP =Q ..................................................... (!) where Q stands for a constant number whose value depends only on the units chosen, and which obviously bears three denominations arithmetically, being the value which V, or P, or 1 T assumes, if in each case the other two variables are at unit value. Let us see now how far eq. (1) applies to 7io?i-permanent gases. Their general behaviour may be stated thus: As long as we keep clear of condensation into liquid the formula applies to all gases approximately, and the deviations are always in the same sense, namely, the same as if the respective gases, while consisting sub- stantially of matter obeying equation (1), were contaminated with a small quantity (more or less) of a mist of the respective liquid. If we reduce the pressure from, say, P to J P, or increase the temperature from T to 2 T, the volume increases to a little more than twice its original value, because some of the " mist " gets converted into gas. Within the moderate range of pressures which practical gasometry discounts, the mist does not go far towards invalidat- ing Boyle's law. It tells more strongly on Gay-Lussac's ; but in 158 GAS ANALYSIS. reference to it, its influence can in all practical cases be reduced to practically nothing by giving the temperature a sufficiently high value. For an example, imagine a litre of saturated steam of 100C. (T = 373) as it rises from water boiling under 760 mms. pressure. If heated (at 760 mms.) from 100 to say 150, it expands into more than - 'TO -- litres; the volume at 150 423 is, say, = ~=g + A litres. And if we take away the surplus o7o volume Ao> an( i now pass from 150 to 200, the volume increases 473 473 to not T^, but , + Ai litres ; but Ai is considerably less " than Ao> an d so n - From a certain high temperature onwards, the A becomes practically equal to nil ; the steam in its expansion keeps pace with, for instance, its equal volume of hydrogen ; and from that limit temperature upwards it follows also Boyle's law, from at least the respective pressure downwards. The student will understand now what we mean by saying that, for any gas, there is an area of combinations of T and P, character- ized by a certain T as minimum temperature, and a certain P as maximum pressure, and extending on the one hand to P = nil, and on the other to T = oo, or rather the highest value which the gas stands without suffering dissociation, within which the gas, in a practical sense at least, obeys eq. (1) in all strictness? or to use customary phraseology, is perfectly gaseous. Obviously, to define a gas quantum numerically without reference to temperature or pressure, the right way is to bring it within its area of perfect gaseousness, to determine a set of simultaneous values for V, P, and T, and calculate the " Q " from these by equation (1). But how can we ascertain the boundary lines of that area in a given case ? Practical gasometry gives a very simple answer. If the gas is a gas at and 760 mms., simply measure it at some convenient temperature and pressure (taking care to keep the latter down as far as expedient if you have to deal with such things as carbonic acid or sulphurous acid), and calculate the Q from the V, P, and T thus obtained from eq. (1). The same rule is applied even if the gas is contaminated with such small GENERALITIES AND THEORY. 159 quantities of steam or benzole vapour, &c., as it might lick up at ordinary temperatures. Of course, in this case care must be taken to steer clear of condensation into liquid, provided the vapour belongs to the " gas to be measured." (The meaning of this clause will be made clear presently). By proceeding in this somewhat off-hand manner, we, in general, incur the risk of quite an appreciable error; yet this error, in the generality of cases, is less, or at least not of a higher order, than that involved in the measurement of V, P, and T, by means of the ordinary laboratory instruments. According to Amagat, carbonic acid from upwards expands at a greater rate than air, up to about 200, whence onwards it behaves like a perfect gas in reference to expansion caused by changes of temperature or pressure. At 760 mms., its expansion, from to 200C., is in the ratio of 1 : 1*74065. Hence, supposing we find for a quantity of carbonic acid the volume, at T = 273 and P = 760 mms., equal to Y , we have for the constant Q : (1.) By the ordinary routine mode of calculating, ^_ V x760 ^ = ~^73~~ (2.) For the true Q, V x 1*74065 x 760 ^ = 473 Whence Q = 1*0046 Q'. Carbonic acid is probably the most imperfect of ordinarily occurring gases, and and 760 mms. constitute about the lowest temperature and highest pressure customarily employed ; hence, 0'0046 might be put down as the maximum relative error intro- duced by neglecting the imperfection of gases in our calculations were it customary to measure gases in the state of dryness ; but such is not the case. A gas as it comes to be measured in the course of an analysis is, as a rule, contaminated with vapour of water, and it is the quantity of dry gas that is wanted. To obtain it, one way (and in the case of gases which like SO 2 or NH 3 are largely absorbed by water, the only way) is to remove the water by means of a suitable dehydrating agent, and measure the thus dried gas. But this is troublesome. With ordinary gases 160 GAS ANALYSIS. we always prefer another method, which is to, if necessary, add a drop of water, spread it out on the sides of the measuring tube so that the gas is saturated with water at the prevailing tem- perature, to measure the V, T, and P, and to correct the observed P by deducting the maximum tension of steam TT for the respec- tive temperature. The difference, P TT, is the tension which the gas would exhibit at that temperature T, if kept at the observed volume V ; very nearly, not exactly, and the error involved in the assumption often raises the error in the calcu- lated Q beyond O0046 of its value. A special case presents itself if the water produced in the combustion of a gas (say, of a hydrocarbon), with oxygen over mercury, has to be measured gasometrically. We then surround the measuring tube by a glass jacket, and blow steam through the jacket, so as to raise the temperature of the gas to some value near 100C. (which must be determined directly and exactly). If at this temperature the partial pressure of the steam is at less than, say, one-half of 760 mms. of mercury (the less it is the better), the hot gas, as a whole (steam plus excess of oxygen plus carbonic acid, &c.), may be assumed to be perfectly gaseous, the more readily as in such cases only a moderate degree of precision in the measurement of the pressure can be reached. A physical significance for Q is easily found. A glance at Q T eq. (1) shows V = , i.e., the volume of a gaseous body is T proportional to the ratio , the disgregation of the gas, as it has been called. Hence, Q may be defined as that value which V T assumes whenever - = 1, or T = P numerically. Supposing, for instance, we measure P in mms. of mercury, and T in centigrade degrees (let us at once adopt these units ; they are as good as any others), V = Q if tbr*= 100 200C. or T = 273 373 473 absolute temp. &c. P = 273 373 473 mms. T But P and V in the equation are interchangeable, i.e., P = - GENERALITIES AND THEORY. 161 Hence we may define Q as that pressure which the gas assumes whenever T = V, if (for instance) T = 273 ... 373 ... 473 absolute temp. &c., and V = 273 ... 373 ... 473 units of volume. A third definition will at once suggest itself; but we leave it on one side, because it is strained and unnatural. We prefer to explain that Q has a most important chemical significance. If we define the specific gravity S of a gas as being the number which tells us what number of times the gas is heavier than its own volume of some standard gas (say H 2 ) of the same disgregation, the specific gravity, by eq. (1), is independent of the value of the disgregation chosen, and d fortiori of the special P and T which happen to constitute the disgregation. ft consequently can depend only on the chemical constitution of the gas. It is the immortal merit of Avogadro to have divined what the dependence is. As found by him, and since confirmed by thousands of experiments, S is proportional to the molecular weight M. We always have M 1 :M 2 :M 3 . . . = S x : S 2 : S 3 . . ,- or generally, for given units, M = const. S . . .. (2). If the Q's of a series of gaseous bodies I. II. III. are Qj Q 2 Q 3 , then obviously the weights are (const, into) C^Mj Q 2 M 2 Q 3 M 3 , i.e., the Q's, in a relative sense, count the molecules present in the several gases I., II., III., &c. In what we have said so far in connection with Avogadro's law, we have been assuming that the several gases are chemical species, such as O 2 , CO, CO 2 , H 2 0, &c. But it is difficult to think that the law should not hold for perfect gas mixtures as well. So we should say, were it not proved by experience, that in the absence of chemical action, and at constant pressure and temperature, permanent gases exactly, M 162 GAS ANALYSIS. and other gases very nearly, mix without c.ontraction or expansion* The general relation between gas volume and gas weight, which is embodied in Avogadro's law on the one hand, and Regnault's determinations of absolute gas densities on the other, enables us to find a formula for the reduction of the volume of a gas in litres (or cc.'s) to its weight in grammes (or milligrammes), and vice versa. Of the several gas densities determined by Regnault, that for oxygen is probably the most exact. He finds that 1 litre of oxygen measured at 0C., and a pressure equal to that exerted by a column of 760 mms. of mercury of 0C. at sea-level and the latitude of 45, weighs 1'429 32 grammes. As this number simply denotes the specific gravity of oxygen at 0C., &c., referred to water of + 4 as equal to 1000, "gramme" may be taken as meaning any unit of mass, and litre as " the volume at 4C. of 1000 units of mass of water." From these data we easily calculate that generally 1 litre of gas weighs D grins. = /0-032089 ^^ - -fe . . (eq. 3) 27o + tj 2 where M stands for the molecular weight, as it is if = 16. For the volume of M grms. (32 grms. of oxygen, 18 grins, of steam, &c.), we have 273 4- 1 M grms. occupy 62*326 x --p litres, . . . (eq. 4). In Glasgow the constants of eqs. 3 and 4 assume the values 0-032120 .............................. ; log. = 2'506 776 62-267 ................................. ; log. - 1794258 To be able to utilize these formulae more directly, we must provide our gas-measuring tubes, manometers, and barometers with true millimetre scales, and never forget to reduce the height of a column of mercury measured manometrically to 0C. This reduction is easy. Imagine a column of mercury h mms. high at t C. to be contained in a cylindrical tube inexpansible by heat, By cooling down to in this tube, it shrinks into k = ~ -- r- 1 ~\~ K t * It is worth while to point out that this proposition has never been tested experimentally in the sense in which Boyle's law has been by Regnault, GENERALITIES AND THEORY. 163 or practically h (1 k ) mms. (where k is the cubic co-efficient of expansion of mercury), but continues to exert the same pres- sure as before. h 0) therefore, must be substituted for the ob- served h. For computing without, or with four-place, logarithms, write the equation h = h-hk t, and compute the second term to deduct it from h. If a six-place logarithm table is at hand, it is more expeditive to work the original equation with 1 + k t in the denominator, k = O'OOO 181 43. [A table of the values of log. (1 + k t) for t = 0, 1, 2, . . 30C. is given in the Author's " Tables to Facilitate, &c.," 2nd ed., p. 26. The same book, on p. 39, gives a table (Rosetti's) for the reduction of water-weights (in grms.) to volumes (in cc.)]. In practical gas analysis we generally find it expedient to let each gas- measurer have its own unit of volume, which is most con- veniently registered by stating the number of grammes of mercury of a stated temperature which corresponds to it. One gramme of mercury (uncorrected weight; brass standards) occupies at t the volume of IS'oOG" 1 x (1 + k t) cc. ; 1 : 13*596 = 0-073 5510 ; log. = 2'866 589. [A table giving the volumes at 0, 5, and 30 ready calculated is found on p. 40 of the book referred to]. In practical gas analysis we rarely have occasion to reduce gas- volumes to weights. As a rule, all we care for are the rela- tive quantities of a given series of gaseous bodies, and we may, of course, choose our own units. The natural unit for the pressures then becomes the pressure of a column of mercury 1 mm. high at the mean temperature that prevails during the series of measurements under consideration. That the temperature may have been, say, 15C. in one case, and 16, 17, or 14 in another, is of no consequence practically. In the rare case when one of the measurements was made at an exceptionally high tempera- ture, say, at 100C., and all the rest at the temperature of the gas room, say at 15C., the measurements involved in the one experiment at 100C. must be corrected for the expansion of the glass and of the mercury from 15 to 100. The linear expansion of glass is 8'3 units per million units per 1C. Hence, if a eudio- meter scale is correct at, for instance, 15C., the real length of 164 GAS ANALYSIS. 1000 millimetre-divisions at 100 is greater than 1000 mms. by (100 15) x 0*0083 = 071 mm. If a gas-measurer down to a certain mark holds V units at t, its capacity at a higher tem- perature t' is V. V = V jl + 8 (f - j where 8 = 0*000025. The reduction of the mercury column for temperature has already been explained. For the purposes of comparative measurement, we need only a set of relatively correct values for the " Q's " of the several gas- quanta concerned. Hence, of the three quantities, V, P, and T, in each set of cases, only one needs actually be measured. We may, for instance (1.) Keep T and P at constant (though unknown) values, T and P , and measure only the values V 1; V 2 , V 3 , &c.; i.e., for the Q's, substitute their values J V V V T O \ VI ' V2 ' Va ' P and strike off' the constant factor =-. (The old method of gas IG measurement.) (2.) Keep T and V at constant (though unknown) values, T and V , and measure the P's; i.e., substitute for the Q's their values P ......................... fee y and delete the constant factor ~. (Regnault's method.) IG (3.) We may allow P and T to vary, but take care to keep the disgregation D = T : P at a constant value D , and measure the volumes; i.e., substitute for the Q's, their values ,, V 2 , V 3) and delete the constant factor ^ . (Doyere's method.) -L'o In reference to a set of gas quanta I., II., III., &c., take q lt q 2 , q 3 , as representing the volumes at say T = 400, and P = 400 mms. (or to use general terms, for any combination of a temperature GENERALITIES AND THEORY. 165 and a pressure which are numerically equal to each other) ; take Q as having a similar meaning in reference to the substance produced by mixing I., II., III., &c., together, and we have by direct experience qi + q* + 93 + ......... =Q ......... (5), ; whence we might deduce, were it necessary, that at any tempera- ture and pressure, supposing both to be at the same values all round, the volume V of the mixture equals the sum of the volumes of the components v, + v, + v 3 ......... =V.., ...... (6). But our q's and Q bear the denomination of pressure at V = T (for arithmetical purposes, say V = 1 and T = 1), hence we have for the pressures which I., II., and their mixture respectively ? would exert at some temperature, T , if they were successively shut up in a vessel of the invariable volume V , _fcT _?,T _g 3 T Pi - -^r- 2?a - ~^T- Ps -TT- V O V O V O OT For the mixture P = -^, and obviously ^ O as q 1 + q. 2 + q s ............ = Q, pi + P* + p* ............ = P ............ (7), i.e., the pressure of the mixture is equal to the sum of the partial pressures of the components. Now, supposing all our equations from (5) to (7) correspond to the same set of gases, we have, for instance, for component I. and the mixture and i\ : V = g x : Q ; and (2) Pl = 2^s, and P = Q ^* V O V O whence p l : P = q l : Q, or quite generally for any component * We add the dashes in the second case to show that the T' is not necessarily the same as T in case (1). 166 GAS ANALYSIS. Hence, we see that the customary mode of stating the volu- metric composition of, let us say, atmospheric air is susceptible of three readings ; instead of saying (a) 100 volumes of air con- tain 21 volumes of oxygen and 79 of nitrogen, we may say (fc) 21 per cent, of the pressure is oxygen pressure, and 79 is nitrogen pressure ; and (c) 100 molecules of air contain 21 molecules O 2 , and 79 molecules N 2 . The statement (6) of the first two is obviously the more straightforward in all those cases, when one of the components at the ordinary temperature is a gas only through its association with gases properly so-called. The student may apply what we say to the case of air satu- rated with vapour of water at a given temperature, and he will see what we are driving at. From gasometry we now pass to gas analysis. PKOXIMATE ANALYSIS OF GAS MIXTURES. For the solution of this problem we have only one direct method. Its nature is best explained by an example : To analyse a mixture of 2 , N 2 , and C0 2 , we begin by measuring off a sample of the given gas (let the volume be V at T and P of dry pressure). We then remove the carbonic acid by caustic potash, and measure the residue ; i.e., determine its volume V x at T! and P x . This being done, we apply pyrogallate of potash solution (to absorb the 2 ) and determine the set of values V 2 , T 2 , P 2 , for the nitrogen. Obviously we have for the several Q's (reduced gas volumes) Original gas, ........................... =^^ = Q o *0 TheN 2 and0 2> ....................... . = ^ = Q: *1 TheN 2) . . .. T, Hence for the percentage of TheC0 2 , .............................. a; = Qo The0 2 , TheN PROXIMATE ANALYSIS. 167 To show what the method can do, we will enumerate the most important of the customary reagents, and for each name the gases for which it is available as an absorbent. 1. Water as such, or as Na2SO4.10H.30, absorbs HC1, HI, HBr very promptly. 2. Dilute Sulphuric Acid (besides acting as water) absorbs NH 3 , CH 3 NH. 2 , &c., very promptly. 3. Caustic Potash Ley, or strongly hydrated solid potash (a soft potash ball) absorbs all acid gases readily : CO 2 , SO 2 , H 2 S, HC1, HBr, HI. 4. Anhydrous KHO acts on CO 2 only slowly. The Author has no experience about its exact action on other acid gases. It is used chiefly for the absorption of vapour of water, especially in the suspected presence of small remnants of acid gases. No doubt it absorbs alcohol as C 2 H 5 KO. 5. Fused Chloride of Calcium may be named here as a charac- teristic absorbent for water. It absorbs also alcohol. 6. Oil of Vitriol (H 2 SO 4 + 1/12 H 2 0, Marignac's acid, a slight excess of water does not matter) absorbs water, alcohol, oxide of methyl, ether, propylene C 3 H<5, and its higher homologues. Ethylene is absorbed only on long-continued shaking (Berthelot). Hydrogen and marsh gas are not absorbed. 7. Fuming Oil of Vitriol (SO 3 dissolved in H 2 S0 4 ) does not absorb H 2 or CH 4: but absorbs all olejines C n H 2n very readily. The residual gas always contains vapours of S0 3 and S0 2 , which must be removed by potash before measuring. 8. Bromine (Br 2 ), in the presence of water and diffuse day- light, acts pretty much like fuming vitriol. It cannot be used over mercury, because it combines with the metal readily. The gas to be analysed is transferred to a glass-stoppered bottle over water. A sealed up bulb full of bromine is introduced, broken by shaking, and allowed to act. The liquid part of the residual bromine is then allowed to drop out into the trough, and the remaining bromine vapour absorbed by potash. 9. Pyrogallate of Potash (pyrogallic acid dissolved in caustic potash ley extemp.) absorbs oxygen largely and readily (Liebig), besides, of course, acting as KHO solution. 10. Cuprous Chloride dissolved in hydrochloric acid absorbs 168 CAS ANALYSIS. readily ; 2 , CO, C 2 H 2 , C 3 H 4 (Berthelot) ; spoils the mercury badly. 11. Cuprous Chloride dissolved in ammonia acts like the acid form of the reagent, but absorbs besides certain hydrocarbons, including (all the ?) olefines (Berthelot) ; does not spoil the mer- cury. 12. Chromous Sulphate (CrO salt) solution, mixed with sal- ammoniac and excess of ammonia, absorbs 2 , NO, C 2 H 2 , C 3 H 4 ; it does not act on CO, C 2 H 4 , C 3 H 6 (Berthelot). 13. Binoxide of Manganese (used as a compressed ball) absorbs H 2 S and S0 2 (Bunsen). (Permanganate or bichromate of potash solution, mixed with sulphuric acid, no doubt acts similarly). 14. Ferrous Sulphate, as concentrated aqueous solution, absorbs nitric oxide largely, but the compound has a measurable tension. We may mention in passing that NO can be absorbed also, as KN0 2 and KN0 3 , by the joint action of oxygen and caustic potash ley. There is no need of our drawing up a scheme for the syste- matic application of these absorbents, nor is it necessary to point out that all one can attain with them is not co-extensive with the problem. There are plenty of gas mixtures which, in oppo- sition to any chemical absorbent, behave as if they were unitary compounds. For the analysis of such gases only two methods are known ; we must either effect an ultimate analysis by the "method of combustion" (vide infra), if possible, and from the elementary try to guess out the proximate composition, or see what we can do by means of physical absorbents. But in the case of these we must do what Bunsen has taught us, namely, apply the solvents, and interpret their effect, in the light of the laws of gas absorption. THE LAW OF GAS ABSORPTION AND ITS APPLICATIONS. Imagine v volumes of a mixture of the gas species I., II.. III., to be shut up in a close vessel over h volumes of water (or alcohol or other liquid absorbent), an impervious diaphragm separating the two. As soon as the diaphragm is removed, the gas and liquid exchange molecules, and this goes on for ever ; but if a constant temperature t be maintained, a point is reached THE LAW OF GAS ABSORPTION. 169 sooner or later, when on both sides the changes of composition, caused by emission and reception, exactly compensate each other, so that it is the same as if the exchange had come to a stop. This point of dynamic equilibrium is reached almost instantaneously on violent shaking. The final result is that the gas space v is saturated with the vapour of the liquid, while of each of the components of the gas a quantity q is held in solution by the h volumes of liquid. This quantity q, at a given temperature, is in (more or less exact) accordance with the equation q = hftir I. where TT means the partial tension of the respective component in the residue, and /3 is a constant, which may be defined as being the value which q assumes when h = 1 and TT = 1 mm. The quantities q and /3 are, FlG - 38 - of course, of the same denomination ; if q means milligrammes, 8 means milligrammes likewise. But we will assume q to be measured by volume at 0C. (or T=273C.) and P = l mm., and on the basis of this assumption (with Bunsen) call the " co-efficient of absorption." Our equation I. has been tested experimentally only with water and, in a more limited sense, with alcohol, as a solvent ; and, in reference to either, may be assumed to hold, at pressures up to about 1 atm., and temperatures from to about 30C., for all gases which, under the circumstances, do not act chemically on, or dissolve very abundantly in, the respective liquid. With a given gas species, the constant j3, in general, increases when the temperature falls, or when alcohol is substituted for water. It has, in general, different values for different species of gas. Hence we at once see our way towards distinguishing a unitary gas from the isomeric mixture. Take, for instance, the case of marsh gas CH 4 as against a mixture of equal volumes of H 2 and C 2 H = CH 4 per 1 volume. With alcohol as an absorbent, the ft of C 2 H 6 is far greater than that for H 2 . Hence, if the mixture be dissolved partially by alcohol, the residue will contain less carbon per unit volume than J C 2 ; and similarly in similar cases. Before going further, let us draw an obvious but important conclusion from our equation. Given a gas-mixture containing 170 GAS ANALYSIS. m volumes of species I., in" volumes of species II., ...... in unit volume ; and supposing h volumes of (say) water to be shaken at a constant temperature and constant total dry pressure p, with successive (but always fresh) instalments of the mixture until absorptiometric equilibrium is established (or, mathematically speaking, supposing h volumes of water to be shaken with oo volumes of the mixed gas), then we have for the dissolved quantities of the several components q' = hpmfi'; q" = hpm"{', &c., &c.; and for the total quantity of absorbed gas Q = hp(m ft + m" ft' + ............ , &c. ). It is not quite so easy to calculate the composition and quan- tity of the gas absorbed if the h volumes of (let us say) water are shaken in a vessel of the constant capacity (h + v) with v volumes of the given mixture. One thing, however, is clear beforehand : unit volume of the absorbed part of the gas will, in general, contain not m', m", &c., volume of the components, but other quantities ri, n", &c. If r', r", r'", &c., stand for the undis- solved residues of the several components, P for the original (dry) pressure of the mixture, and p for the less (dry) pressure of the unabsorbed total residue, we have, since quite generally, pu = TT for any of the components, which we abbreviate into r = (np)v ; hence q + r=pn(v + file); but q + r may be expressed in function of P and m, thus : hence Pmv = pn (v + ftli) , or " = v ..II. m p which enables one to calculate n, n", n", from m, m", m", &c. THE LAW OF GAS ABSORPTION. 171 By summing up the specicd equations represented in Ila., we have P / m , , n+n+n Whence -= 1 - 2-^ ..III. P p jj and by substituting this for the in eq. Ila., we have m , / v m - *^*fimi iv - V~ \ V, -'o \ ^o The total quantity of dissolved gas is Q = Pv pv w whence <"*( l -*m) v - ] \ v / Q-^-kp is what, in the case of a single gas, would be the co- efficient of absorption. In the case of a mixture the quotient (which we will call the " solubility " of the gas mixture, and designate by " X") is variable, namely hp Eq. Va. affords a method for distinguishing a unitary gas from a mixture. We just determine the " solubility " X by measuring off, at t, V volumes at, say, P mms. dry pressure. We then shake the gas with h volumes of, say water, and determine the residual volume v and its dry pressure p. Then we have for the pressure P which the original gas would exhibit in v volumes, the equation VP = vP t whence 172 GAS ANALYSIS. and for the solubility the equation We repeat the experiment at a series of values for h : v or h : v, i.e., with varying quantities of water at the same v. If the gas is unitary we always find X = /3 = a constant number. If it is a mixture, X will vary when h : v varies. In general not necessarily. But in practice, it is not easy to distinguish between real and only approximate constancy in X. Hence, supposing X to come out practically constant, it is as well to change the solvent say use alcohol instead of water and see what now comes out. If the new solvent also leads to a constant X, the unitary nature of the gas is almost proved. Eq. [III.] shows the way to determine the composition of a mixture of two unitary gases, I. and II., if their /3's are known. We shake a known quantity of the gas with a known volume of water, and determine the pressures P and p (reduced to the volume of the unabsorbed gas-rest) of the original gas and unabsorbed gas-rest respectively, and then have (by eq. [III.]) m m' + ra" = l (2). p As is known, in' and in" can be calculated from the two equa- P tions. The easiest mode of solving them is to divide eq. 1 by m" and solve it in respect to result, in general terms, is m" and solve it in respect to n as the unknown quantity. The where = 1 + /^ and " = 1 + /3"^ v n v.. ULTIMATE ANALYSIS. 173 ULTIMATE ANALYSIS OF GASES GENERALLY. Among the several methods which fall under this heading, only one is of sufficiently wide applicability to be worth con- sidering in general terms. We allude to the method of com- bustion, which presumes that the gas to be analysed is in, or by addition of hydrogen or of oxygen or of either plus knallgas* can be brought into, such a condition that the mixture, when an electric spark is sent through it, is resolved completely into one of, in general, carbonic acid, nitrogen, water, and excess of either hydrogen or oxygen, as the case may be. Before going further, let us explain the customary mode in gas analysis of formulating the ultimate composition of a gas. Methylamine gas may serve as an example. Instead of trans- lating the formula CH 5 N into a gravimetric statement, we say Every 1 volume of this gas contains (potentially) J times 5 volumes of hydrogen, J X 1 volume of nitrogen, and half-a-volume of " carbon gas; " meaning, what would be half-a-volume of carbon, if carbon were a gas of the formula C 2 ; or, to put it more realistically, " the quantity of carbon contained in 1 volume of CO 2 ." Avogadro's law, of course, enables us to quite directly read the volumetric composition of any chemical species in its formula, but the system is clearly extensible to gas mixtures. If, in the sequel, we speak of a gas or gas mixture as being " (#C 2 , ?/H 2 , 00 2 ) + wN 2 = 1 volume," this means that 1 volume contains w volumes of nitrogen gas, mixed with a (unitary or mixed gas) containing (potentially) x volume of " carbon gas," y of hydrogen gas, and z of oxygen gas in (1 w) volume. For the theoretical development of the method, we will assume we had to deal with a gas mixture corresponding to this formula, and wanted to determine the co-efficients x, y, z, w, which consti- tute the volumetric elementary composition of the whole. For the determination of the carbon and the hydrogen, the general method is to measure off Vt units of the gas, add a sufficiency * Fulminating gas as produced by the electrolysis of water. We prefer the German word, on account of its shortness. t Meaning what represents V volumes at the disgregation 1, or some other adopted standard disgregation. We might say " N x V molecules," where N is an arbitrary very great constant ; but this would run contrary to established habits, 174 GAS ANALYSIS. (S units) of oxygen to produce V + S = V 1 of mixture, and fire it with an electric spark. We next measure the product cold, to obtain its quantity (V 2 units). In order now to determine the hydrogen, we expose the whole to a high enough temperature to convert all the water produced into a practically perfect gas, measure the hot mixture, and then find its quantity as V 3 units. Obviously the steam amounts to V 3 V 2 units, and consequently the hydrogen to V 3 V 2 units likewise ; whence 2/ = T (Vs ~ V2) ......... (1) " Of course the result can be correct only if all the materials used (including the mercury) were perfectly dry ; if knallgas was added, two-thirds of its reduced volume must be subtracted from V 3 V 2 as a correction. This method of hydrogen determination is quite generally applicable, but it is rather troublesome ; hence, whenever it is possible, we prefer to calculate the hydrogen from the " contrac- tion," meaning the contraction involved in the combustion, or the difference Vj V 2 = " C." So we may, if the gas I. Consisted entirely of hydrocarbons (including H 2 itself). The contraction being independent of the excess of oxygen added, we have (for the gas #C 2 , 2/H 2 = 1 volume) the equation xG 2) yH 2 + (2x + -) O 2 = 2x CO, + 2/H 2 O, and for the volumes, 1 (2x + -0 2x nil ; hence for the contraction per unit of gas, and for the actual contraction, C - Vc = V + I whence y = 2 - ^ -, (2) The carbon, in all cases, is determined by treating the product of combustion, V 2 units, with caustic potash, and measuring the ULTIMATE ANALYSTS. 175 residue V" units. ' We then have for the carbonic acid produced K = V 2 - V" (3) V V" whence x = - *-~ (4) II. We will now assume the gas is a mixture of hydrocarbons and free nitrogen. In this case, formula (2) may serve to calcu- late the hydrogen ; only we must substitute for V the volume V of the hydrocarbon part of the gas, to get which we must determine the nitrogen in the product of combustion, and deduct its volume $ from V ; V = V g. The determination of the nitrogen is effected after absorption of the carbonic acid by analysing the residual mixture of O 2 and N 2 , which may be done either by absorbing the 2 by pyrogallate, or firing with excess of hydrogen, and measuring the contraction ; one-third of it is the volume of the oxygen. That this method of nitrogen determination applies quite generally needs hardly be pointed out. It is not quite so obvious, and yet the case, that if III. The gas contains (besides hydrocarbons and nitrogen) even free oxygen in unknown quantity, the hydrocarbon cannot be calculated from the contraction. The only method in such a case is to remove the oxygen from the original gas by absorption with pyrogallate, and bum the residue unless it should be known that the oxygen and nitrogen are both present only as so much atmospheric air. In this case, of course, the problem becomes easy of solution. [See the Author's " Tables to Facilitate," &c., 2nd edition, pp. 30 and 32]. We refer to the same book, p. 32, for the consideration of the most general case, represented in the general formula, 1 volume = aC 2 ,/3H 2 ,y0 2 ,(5N 2 , for the gas to be analysed. We will only point out here that in the case of any carbon gas, which may contain components CH/30y, the only mode of determining the hydrogen is to measure the steam pro- duced in the combustion of the gas. To illustrate this by an example, assume the gas in one case were ethylene, C 2 H 4 , and in another, oxide of methyl, C 2 H 6 O. Both give the same proportion of CO.,, and the same contraction, and yet the second contains in every 1 volume what 1 volume of the first contains plus 1 volume 176 GAS ANALYSIS. of potential steam. If we have some knowledge of the proximate composition of the gas given for analysis, the problem, of course, may assume a different aspect. Supposing, for instance, we have to deal with a gas of the potential composition, #(aC 2 ,/3H 2/ )+2/CO = 1 volume, where (aC 2 ,/3H 2 ) stands for 1 volume of a unitary or complex hydrocarbon of this composition. In this case (reducing at once to unit volume of gas to simplify matters) we have, if k stands for the carbonic acid produced, and c for the contraction, k = 2xa + y I. c = x (l + i/3) + Jy II 1 = x + y III. i.e., we have only three equations for four unknown quantities. To be able to find these we need a fourth equation, which it may be possible to deduce from a known relation between a and ft. If, for instance, the hydrocarbon is known to be a unitary or mixed olefine, we have ft = 2 ; if the hydrocarbon part is a paraffin, we have /3 = %a + 1 for a fourth equation. If ft and a are known, if we know, for instance, that the hydrocarbon is pure marsh gas or pure ethylene, &c., two equations will do, say II. and III. We are here touching upon a large question, namely, the question of the extent to which the method of combustion is available as an indirect method of proximate analysis. To solve the problem in a given case, one way is to ascertain the formula for the elementary composition of the mixture, to establish equations between the a, ft, y, S, of the formula on the one hand, and the corresponding co-efficients for the several potential components on the other, and combine these with the obvious relation x 4- x" + x" 1, where the sc's stand for the perunitages of the several components. But it is evidently simpler to establish equations between the carbonic acid k obtained per unit of gas, and the sums of the portions of carbonic acid contributed, by hypothesis, by the ULTIMATE ANALYSIS. 177 several components, and to do the same in regard to the con- traction c, the vapour of water iv, the nitrogen n found, per unit of gas analysed, thus : k k' x' + k" x" + I c - c ' x ' 4. ( /' x " + II. w w' x' -f- w" #" + III. 72, 7i, r aj' -f- n" x" -f- ..IV. 1 = #' 4- a;" -1- ., . V. If the number of possible components does not exceed 5, we may be able to thus determine their perunitages x', x", &c. Of course there may be no nitrogen, and eq. IV. as a consequence collapses. We may in addition thereto know that there is no oxygen, when eqs. II. and III. become identical, because we can calcu- late the hydrogen, and consequently also the vapour of water from the contraction c, and vice versa. In this case, only three equations, I., II., and V. (or I., III., and V.) are left, and of these again one may be lost through the existence of a known relation between k and c. We may, for instance, know that all the components are olefines, when c can be calculated from the carbonic acid k. A similar result occurs if all the components contain the same number of hydrogen atoms, or the same number of carbon atoms per molecule. Supposing, for instance, all the components are di-carbon gases (C 2 H 2 , C 2 H 4 , C. 2 H 6 ), all the k's become equal to 2, and we have as eq. I. = which is evidently a repetition of eq. V. If all the components are of the formula C X H^ (ex. C 2 H 6 , CgHg, C 6 H 6 ), we know without experiment that all the c's are = 2'5, and eq. II. becomes a mere repetition of V. For illustrations see the Author's " Tables," &c., pp. 32-34. The following table gives the values, c, k, s w * w, and n, for a number of * * denotes the oxygen necessary and sufficient for the combustion of 1 volume of gas. X 178 GAS ANALYSIS. I. Gases Combustible by Oxygen. c. k. o- w. n. Hydrogen, H 2 , .... 1-5 0%5 1 Carbonic oxide, CO, . . 0%5 1 0-5 Methyl-aldehyde, CH 2 O, 1 1 1 1 Ammonia, NH 3 ,. T25 075 1-5 0-5 Methylamine, CH 5 N, . . 175 1 2-25 2-5 0-5 Cyanogen, N 2 C 2 , . . . 2 2 1 Hydrocyanic acid, NCH, 075 1 1-25 0-5 0-5 Marsh gas, CH 4 , 2 1 2 2 Acetylene, C 2 H 2 , . . . 1-5 2 2-5 1 Ethylene, C 2 H 4 ,. . . . 2 2 3 2 Ethane, C 2 H 6 , .... 2-5 2 3-5 3 Propylene, C 3 H G , . . . 2-5 3 4-5 3 Propane, C 3 H 8 , .... 3 3 5 4 Oxide of methyl, C 2 H G O,. 2 2 3 3 Benzol, C G H 6 , 2-5 6 7'5 3 Gas, C H8=l vol., . . 1+ 4 a + 4 T Gas, C H/3 Oy = l vol., 1+ ?-i a -H 2 II. Gases Combustible by Hydrogen* c. h. 'W. n. Nitrous oxide, N 2 O, . . 1 1 1 1 Nitric oxide, NO, f . . . 1-5 1 1 0-5 * A = hydrogen necessary for combustion. f Nitric oxide does not explode with hydrogen ; it does if mixed with it and a sufficiency of NjjO, but then the result is irregular (Buntn). NO + H^ + sufficiency of knallgas miyht work, but we have no personal experience. WORKING METHODS. 179 APPARATUS AND WORKING METHODS. We propose to leave Bunsen's physical absorption method on one side, and in regard to the chemical methods, to confine ourselves to those apparatus which are constructed for the use of mercury as a trapping fluid. Amongst the variety of such apparatus which have been invented, Bunsen's offer the great advantage over any of the rest, that they are cheap, easily procured, and not in any high degree liable to break. Bunsens Apparatus and Methods. Apart from auxiliaries, such as a barometer, thermometer, &c., Bunsen's apparatus consists only of a mercurial trough, the longer sides of which are made of plate glass, and two kinds of graduated gas tubes, namely, absorption tubes for the absorptions, and eudiometers for the ex- plosions. Before describing these and other apparatus, let us state that, to be able to do justice to Bunsen's method, one must work in a special room, the atmosphere of which is not liable to any sudden change of tempera- ture. The "gas-room" should not be heated arti- ficially, nor be contiguous to other rooms thus heated, and its windows should face the north to avoid direct sunlight. Close to the window stands a sub- 23 25 58 59 60 FIG. 40. Natural size. FIG. 39. Natural size. .stantial table, so constructed that any spilled mercury can be easily collected and recovered. On this table the analytical 180 GAS ANALYSIS. operations are conducted. Both kinds of gas tubes should be about 20 mms. wide (inside), because in narrower tubes the influence of capillarity becomes measureable. A strength of body of about 2 mms. suffices, even for the eudiometers. An absorption tube should not be more than 250 mms. long, and it should be provided with a spout (see Fig. 39) to facilitate the transference of its gas contents to another tube, full of mercury and standing in the same trough. The eudiometer, for ordinary purposes, should be 400 to 600 inms. long ; it must be provided with a couple of platinum wires, soldered in at opposite points near the vault. The wires without terminate in rings to append the pole ends of an induction coil when an explosion is to be effected ; their inside parts run along the vault, and their ends are at about 2 mms'. distance from each other (see Fig. 40). To be able to use the tubes as manometers and gas- volumeters at the same time, they are provided with millimetre scales running from one end to the other parallel to the axis ; and the gas volumes corresponding to the several marks are determined by calibration. To calibrate a eudiometer,* we make it rigorously clean and dry ; we then fix it on the table in an exactly vertical position (the open end upwards), pour in a succession of equal volumes of mercury, and after each addition ascertain the level of the top point of the meniscus by viewing it through a horizontal tele- scope, standing at a distance of some two metres, and seeing which point of the scale apparently coincides with the apex of the meniscus. It is easy, after some practice, to mentally divide each millimetre into ten equal parts, so that each reading in itself need not be uncertain by more than O'l mm. After each addition of mercury, any air-bells that may have got im- prisoned between the sides of the tube and the metal must be carefully removed by means of a long, smooth stick of whale- bone or vulcanite. An iron wire must not be used, because it is liable to produce scratches, which may subsequently expand into cracks. The most convenient mercury measurer is a short test tube, made of rather stout glass and provided with a well * The word eudiometer occasionally serves as a general term for either kind of gas tube. WORKING METHODS. 181 ground on flat lid (see Fig. 41). The measure is filled from out of a long-necked pipette provided with a stopcock. If, in filling the measurer, we take care that the whole of the metal, at any stage of the filling, forms one unbroken mass, the formation of air-bells is easily avoided. The measure is filled to overflowing, the lid pressed on to remove the outstand- ing hillock of metal, and the contents are then poured into the eudiometer. The capacity of the thimble should be so adjusted that each fill of mercury raises the meniscus about 20 mms. Within this short distance any decently cylindrical tube may safely be taken as being absolutely cylindri- cal. The unit being arbi- trary, we choose it so that every millimetre of differ- ence of level corresponds as exactly as possible to " 1 unit " of volume. Supposing the reading, after addition of 1, 2, 3, n measures of mercury, to have been R x , R,, R 3 , R,,, and v to be the number fixed upon as denoting the capacity of the standard measure, the eudiometer, from its close end up to R 1? R,, R 3 , ;v;-. . . R,,, of course holds v, 2v, 3v, nv units of volume of mercury ; but obviously the gas volume required to fill it from the top down to R,, R 2 , Rn is greater by the annular space between the two menisci corresponding to the calibration and gas measurement respectively. To determine its value we fill the eudiometer with mercury up to, say, " R mms. " of the scale, and then pour on a few drops of corrosive sublimate solution, which causes the miniscus to flatten out into a plane. 182 GAS ANALYSIS. We then take the reading of this plane let it be r (compare Fig. 42, in which the curvature of the meniscus is greatly exaggerated to make things plain). The total space up to plane R is obviously constant ; so is the volume of the total mercury. Hence the cylinder enclosed between R and r is equal in volume to the half of the annular space be- tween the two menisci. And suppos- ing R r is equal to S mms., then the annular space between the two menisci is = 2<5 units. As the tube is almost equally wide everywhere, S may be taken as constant, and the gas volumes corresponding to the above mercury volumes as being equal to (?;; 2?;; 3r ; nv) + 2& This correction having been effected, we next calculate for each interval the mean volume- value of 1 mm., and from the volume (nv-{- 2<5) = "V" easily deduce the exact value Ve, corre- sponding to the nearest millimetre mark. The values of the intermediate marks are found by interpolation on the assumption that the tube within each interval is exactly cylindrical. Supposing the exact capacity down to the point " 20'0 mms. " is 2278, and the average volume-value between 20 and 30 is 1-021, we compute 22780 + 1-021 to find the value for 21-0, we then again add on 1-021 to find the value for 22'0, and so on. We express ourselves thus because successive addi- tion (with the constant T021 written at the lower edge of a slip of paper) takes less time than the corresponding series of multiplications would. During the progress of a calibration the temperature of the mercury should be kept rigorously constant, or else the values for the lower marks may become uncertain by more than one can tolerate. What we mean to point out is, that it will not do on a dull day to have a gas name behind the eudiometer to enable one to read. In such a WORKING METHODS. 183 case the only correct thing to do is to keep the eudiometer immersed in a large cylinder full of water, and, during the whole of the process, to keep the temperature of this bath constant. Each gas-measuring tube should be calibrated twice (with well-agreeing results) before the calculation of the table is pro- ceeded with. In using the tubes we must never forget that each has its own unit of volume. It is expedient to supplement a calibration by ascertaining the exact weight of a till of mercury of the prevailing temperature, and from the weight and the well-known constants to calculate the value of " 1 volume " in, say, cubic centimetres or grammes of mercury of 0C. to be able to reduce one unit to another, and gas-volume to gas-weight generally, should such reductions become necessary, as they do occasionally. Bunsen's trough consists of a rectangular block of wood, scooped out semi-cylindrically above, and forming the base of a cistern, the sides of which are made of plate gla-ss, for an obvious reason. Our figure (43) represents an improved form, in which the framework is made of iron. The filling of a eudiometer is effected by means of a very long funnel tube, provided with a stopcock at the end of the funnel ; the funnel is used pre- cisely in the same way as the stopcock pipette is in the charging of the standard measure. The eudiometer is charged to over- flowing with mercury, the end is then firmly closed with the thumb, the eudiometer inverted in the trough, and laid down on the slanting support (see Fig. 43), to await the reception of the gas to be measured. If the gas to be introduced is meant to be measured moist, a drop of water is fixed to the inside of the vault of the eudiometer before pouring in the metal ; the latter flattens out the drop of water, and spreads it over a considerable portion of the inner surface. 184 GAS ANALYSIS. The usual mode of preserving a gas sample collected for analysis is to seal it up, at a pressure decidedly less than one atmosphere, in a short, cigar-shaped glass tube, provided with a short, narrow appendage at each end. To transfer the contents of such a tube to the eudiometer, one of the fused-up ends is pressed against the bottom of the trough, so as to break it off. The thus opened-up end is then brought below the end of the eudiometer in the trough, and the collecting tube subjected to a succession of jerking motions, when mercury gradually runs in, and drives the gas up into the eudiometer. Should gas-bells stick to the sides on their way up, they must be made to rise and unite with the bulk of the gas, which is done by alternately raising the eudiometer from its slanting support, and lowering it again in such a manner that the mercury inside alternately falls relatively slowly, and rises up again more quickly than it fell. Supposing the gas to be all in its proper place, the eudiometer is made exactly vertical, and fixed in this position by means of a clamp. In order now to be prepared for an exact reading of the mercury-level in the trough, a little paper screen, perforated as shown by the figure (44), is stuck in between the side of the trough and the metal, opposite the eudiometer, the thermo- meter suspended close to the gas space, and the barometer, if necessary, made FlG - 44 - vertical. The whole is then left to itself for half an hour or longer, to enable the gas to assume the temperature of the room. On returning to the gas-room take the necessary readings, by means of the telescope, in the following order : (1) That of the meniscus in the tube ; note it down as R (2) That of the surface of mercury in the trough, as R . The paper screen enables one to effect this other- wise difficult reading with a satisfactory degree of exactitude. (3) That of the thermometer, to obtain the temperature t. (4) That of the barometer ; this reading comes last, because the barometer requires to be " tapped " before being read, and this compels one to come close to the eudiometer. R R gives the height of the mercury column supported in the eudiometer, WORKING METHODS. 185 Hence, if the barometer stands at B mms., and the tension of vapour of water at t be TT mms., we have for the dry pressure of the gas in mms., P = (B + R) - (R + TT). For the observed upper reading R, the calibration table gives the volume of the gas V directly, or by an easy interpolation, so that we have for the " quantity " or the " reduced volume " of gas, VP Bunsen prefers to reduce to 0C, or T = 273, and 1000 mms., and accordingly calculates thus V- -- 1000 (1 + a t) where a = 1/273 = 0'003665. A table of the logarithms of (1 + a t) , proceeding by tenths of a degree from 1 to + 31C., is appended to his " Gasometrische Methoden." For the tension of vapour of water, see the Author's "Tables," p. 27. So much about Bunsen's method of gas-measurement ; let us now proceed to explain how he effects the absorptions and the combustions. Absorptions. To be able to effect a succession of absorptions, and all the measurements involved, in the same absorption tube, Bunsen, as a rule, employs the absorbents in the form of solid or semi-solid balls, fixed to long platinum wires, by means of which they can be introduced into, or withdrawn from, the gas to be analysed. Balls of caustic potash, chloride of calcium, and other fusible solids, are cast in a pistol-bullet mould, whose neck has been cut off. The coiled-up end of .the platinum wire is held into the mould, and the molten reagent poured around it. For the treatment of a gas with an intrinsically liquid reagent (e.g., oil of vitriol, alkaline pyrogallate solution, &c.), Bunsen fixes to one end of the wire a ball of coke or papier-macht, and soaks this ball with the respective reagent. To make a coke- ball, a finely-powdered mixture of 1 part of sulphur-free coal, and 2 parts of coke is pressed around the coiled-up end of the wire within a bullet-mould which has not been deprived of its neck, and exposed within the mould to a gradually increasing heat, terminating at redness. Should the bullet thus produced 186 GAS ANALYSIS. not be sufficiently solid, it is heated to about 100, plunged into sugar syrup or coal tar, and re-ignited ; this time in the open flame of the gas blow-pipe. Before using such a ball it had better be purified by treatment with, first, hot nitric, then hot hydrochloric acid, and lastly, water. The washed ball is, of course, dried and re-ignited before being soaked in, for instance, oil of vitriol. Papier-mache balls are easily made from paper pulp prepared extemp. by shaking bits of filter paper with water in a bottle. They serve chiefly for pyrogallate of potash. In introducing an absorbent ball into a gas, we must take care not to allow any air to slip in. How long should one allow a ball to act to be sure that the respective component is completely absorbed ? Unfortunately it is impossible to return a straightforward answer. The balls in general act so slowly, that it will not do to merely wait until the mercury ceases to rise visibly. The only certain method is to wait until one may presume the absorption to be completed, to measure the residue and then apply a fresh ball, and see whether there is any further contraction ; and so on. Fuming sulphuric acid (as an absorbent for olefines C n H 2n ) acts pretty promptly ; soft caustic potash (in absorbing C0 2 ) acts slowly, so does hard (dry) KHO as an absorbent for water (and C0 2 ). Analysis by Combustion. In the execution of such an analysis, a good many points have to be attended to. The following example is intended to dispose of some of these, and at the same time give a general idea of the order of operations. To deter- mine the percentage of oxygen in, let us say, the air of the gas- room, we measure off a convenient number of units of air, then add for every 100 units of air about 75 units or so of hydrogen, and measure the mixture. A high degree of precision in the pre- liminary measuring off of the two gases is not necessary ; yet it obviously will not do to charge the eudiometer with air down to the mark corresponding to n X 100 volumes, and then to add H 2 until the mercury is down at n x 175 ; because in the latter case the column of mercury suspended in the eudiometer is consider- ably shorter (and consequently the pressure of the gas higher) than it was in the first measurement. A correct method is to WORKING METHODS. 187 measure off the several gases roughly in a graduated test tube, at 1 atmosphere, and then let them up into the eudiometer.* Each of the gases, after having been introduced, must of course be "measured" in the sense of gasometry. This being done, the two gases must be mixed intimately by causing the mercury to oscillate up and down in the tube, or by closing the tube with the thumb very firmly, and letting the mercury run forwards and backwards. Before proceeding to the firing, the eudiometer must be closed firmly, by pressing it down on a properly shaped india-rubber " cork " (or actual cork coated over with a layer of vulcanized rubber), and holding it fast by means of a clamp, or the arm of a substantial stand, which arm is made to rest firmly on the top of the eudiometer, a layer of cork interposed between the two preventing breakage of the glass. The gas is now ready to be fired, for which purpose the two platinum wires are connected with the poles of a small induction coil, and the spark is made to pass through.! The effect is an almost noiseless instantaneous combustion, involving the conversion of the whole of the oxygen, and double its volume of hydrogen, into water. The very considerable shock involved is sure to send any air that may be sticking to the cork up into the eudiometer. Besides, the partial vacuum produced in the tube as a result of the explosion tends to suck in what there may be of air within, or on the top of, the cork. To avoid this error, the cork, imme- diately before being used, must be rubbed over with mercury and a little corrosive sublimate solution. The mercury of the trough then adheres firmly to the cork, so that the layer of air can easily be rubbed off under the mercury with the end of the finger. In lifting the eudiometer from its cushion after the explosion, we must be careful to see that the mercury does not rush in, but flow in slowly, or else it may draw in a good deal of air. We will now proceed to the general treatment of the subject, and next speak of the preparation of the gaseous reagents. 'Pure oxygen is easily made by heating a small quantity of pure chlorate of potash in a small bulb blown to the end of a * We shall come back to this point a little further on, f See Note (5) at end of volume. 188 GAS ANALYSIS. hard glass tube. After the introduction of the salt the glass tube is drawn out and bent in the shape of a long narrow delivery tube. The rest calls for no explanation. Pure hydrogen is obtained by decomposing dilute hydrochloric acid (of not more than 10 per cent.) or dilute 10 per cent, sul- phuric acid w T ith pure (especially arsenic-free) zinc, in the presence of platinum within a small phial, which, for obvious reasons, should just be large enough but no larger than neces- sary to prevent boiling over of the contents. The gas is passed through a narrow tube filled with fragments of caustic potash, to absorb water and sulphur- etted hydrogen, and to retain droplets of liquid, and then through a narrow delivery tube into the eudiometer. Of course, all the air must be sure to be expelled before the gas is fit to be used. For the preparation of knallgas Bunsen recommends to electrolize pure oil of vitriol, (diluted to the tenfold with water) in the apparatus repre- sented in Fig. 45. Two platinum electrodes a a are soldered into a cylindrical bottle of some 30 cc.'s capa- city, which terminates above in a funnel into whose neck the end e of the delivery tube and washing bulbs is ground in. A few drops of acid poured into the funnel seals the joint absolutely. The extent to which the bottle is charged with acid is seen in the figure. The bottle is suspended in a bath of water (or alcohol, to prevent freezing in winter time). The bath has two functions : one is to prevent over-heating of the wires, the other to maintain a constant temperature, so that absorptiometric equilibrium, when once established, remains undisturbed. Oxygen being more soluble FIG. 45. WORKING METHODS. 189 in the dilute acid than hydrogen is, the first instalments of gas contain an excess of hydrogen. Three or four Grove cells are sufficient to cause a sufficiently lively evolution of gas, and, after the gas has been going off for about 5-10 minutes, the gas which now follows can be accepted as normal. The bulbs b are charged with a little concentrated sulphuric acid to absorb the moisture and otherwise purify the gas. Knallgas thus prepared, even when exploded in large quantities, leaves no measurable residue. If it did, it would be of no use for gas-analytical purposes. In order now to avoid diffuseness in the discussion of an im- portant point, we will confine ourselves, at the outset at least, to the analysis of gases combustible by oxygen (H 2 and hydro- carbons), and introduce the following terms : " Fuel " means the gas to be burned (e.g., the hydrocarbon to be analysed): while " knallgas " in each case stands for the mixture of the fuel with the exact quantity of oxygen which it needs for its complete combustion. Thus: the mixture H 2 + J0 2 is hydrogen-knallgas ; the mixture CH 4 + 2O 2 of 1 volume of marsh gas with 2 volumes of oxygen is CH 4 -knallgas, &c. In practice, of course, we always take care to have an excess of oxygen present ; there is, however, no need in any case of a very large excess. The term " knallgas " includes only the calculated quantity of oxygen. Any pure knallgas, when fired with the spark at, say, J to 1 atmosphere's pressure, is sure not to miss fire; but in the case of hydrocarbons at least, the violence of the explosion may be more than the eudiometer can stand. Pure acetylene-knallgas, for instance, would shatter any eudiometer into minute fragments. To avoid such disasters we must take care to dilute our knallgas to a sufficient extent by addition of surplus oxygen (or air, which, of course, is the handiest diluent) before firing it. In practice we must go even a little beyond the safety line, because, if the temperature at the explosion is allowed to rise beyond a certain (uncertain !) point, part of the nitrogen, which in practice is always liable to be present, is converted into nitric acid, and the contraction becomes too high. On the other hand, we must not attenuate too much, or else the gas may miss fire or suffer only a partial combustion. 190 (IAS ANALYSIS. The question as to the extent to which a given kind of knallgas should be diluted before being fired, or, to speak more precisely, the question as to the degree of dilution which a given kind of knallgas must possess if its firing by a spark is to lead to a normal and safe combustion, cannot be answered in general terms. All one can say is that the required degree of dilution depends chiefly (not entirely) on the nature of the fuel. Supposing a certain degree to work well with ordinary knallgas (H 2 + \ O 2 ) ; marsh gas demands a higher degree ; ethylene a higher degree than marsh gas ; propylene a higher degree than ethylene. Let us at once add that the required attenuation cannot by any means be calculated from the molecular formula; acetylene (-knallgas), for instance, ex- plodes more violently than ethylene, although it contains H 2 less per molecule than, and as much carbon as, the latter gas does. In these circumstances we had better, at the outset, concentrate our attention upon one kind of fuel. We will select the case of hydrogen, and in reference to it formulate the problem thus : Given a mixture of air and ordinary knallgas (H 2 + J O 2 ), what proportions of knallgas may it contain if its combustion in a eudiometer is to lead to a normal result ? To solve this question, Bunsen and Kolbe many years ago made an elaborate series of experiments. Starting in each case with an exactly measured volume of air, they added a graduated variety of proportions of knallgas, measured again, fired (if possible), and measured the product. The combustion, as we see from the following table, took its normal course in experiments 1 to 5, when the percentage of knallgas lay between, say, 21 and 40, or the quantity of knall- gas per 100 of air between 26 and 64 ; and it is reasonable to presume, with Bunsen, that the limits would be substantially the same if, for instance, oxygen instead of air were used as a diluent. It is to be observed, however, that if the hydrogen plays the part of reagent so that the mixture is #(H 2 + J O 2 ) + ?/H 2 + 0N 2 , the percentage of knallgas may be raised considerably beyond 40 before formation of nitric acid sets in. WORKING METHODS. Results given by Bunsen and Kolbe*: 191 Number of Experi- ment. IN THE MIXTURE AS FIRED. IN THE PRODUCT. Number of units of Knallgas present per Pressure, mm. of Mercury. Tempera- ture, C. Numljer of units of gas per 100 of air used. 100 units of air. 100 units of Mixture. 13-45 11-86 500-6 177 No explosion. 1 26-26 20-80 521-0 177 100-02 2 34-66 25-74 534-4 17-2 100-15 3 43-52 30-33 552-1 16-3 100-07 4 51-12 33-83 5711 16-9 99-98 5 64-31 39-14 591-2 17'3 99-90 6 78-76 44-06 622-5 16-7 99-43 7 97-84 49-46 648-8 16-8 96-92 8 228-82 69-59 648-3 6-7t 89-32 Bunsen's directions regarding the application of the reaction H.> + J O.j = H 2 O generally are all based on the assumption of the general validity of the above limit values for the proportion of knallgas in the material to be burned. But the success of an explosion, it appears to us, is determined, not so much by the percentage of knallgas present as by its density in the mixture to be fired ; we mean the ratio of its absolute quantity to the total volume actually filled by the mixture. This ratio is conveniently defined by stating how many times the volume filled by the mixture is greater than that which its knallgas would occupy at and 760 mms. ; or, let us rather say (as we have a liberal margin to go upon), at the * " Gasometriche Methoden," 2nd edition, pp. 72 and 73. f This 6 '7 is not, as one might suspect, a misprint, obviously made at a different time from the rest. Experiment 8 was 192 GA8 ANALYSIS. prevailing temperature and 760 mms. pressure. This latter number we will call the attenuation of the knallgas, and the corresponding number for the hydrogen in the knallgas the attenuation of the hydrogen. We have calculated these attenuations for Bunsen and Kolbe's experiments, 6, and give them in the following table : Attenuation per unit volume at 760 nuns. Experi- ment. Tempera- ture t. and C. of and t C. of Knall gas. H* Knall gas. 12-81 Hi 177 13-64 20-46 19-22 No explosion. ij .l 16-3 to 177 7-47 to 3-49 11-20 to 5-24 7-01 to 3-28 10-52 to 4-93 | Normal I results. 6 167 2-94 4-41 | 277 4-16 HN0 3 produced. In the combustion of carburetted hydrogens a considerable attenuation of the knallgases becomes necessary apart from any other consideration to prevent destruction of the eudio- meter. The following table gives, for Marsh gas and for Ethylene, the percentages of the respective knallgases and fuels which Bunsen directs us to establish in the mixture to be fired, and, in the last two columns, the corresponding attenuations. It shows with sufficient directness what we have to do if the gas to be burned substantially consists of, or contains approximately known pro- portions of, H 2 , CH 4 , C 2 H 4 , and as we have considerable latitude in fixing upon the degree of attenuation to be established, a roughly approximate knowledge of the composition of the gas to be burned is sufficient to enable one to lay down a scheme of synthesis for the mixture to be fired which keeps us in both senses on the safe side. WORKING METHODS. 193 Refer- ences to Notes. Percentage of Pressure Tempera- in mms. ture. Attenuation referred to 1 vol. measured as 760 mms. and t of Knallgas. Fuel. Knallgas. Fuel. Marsh Gas. (i) (2) 27 to 20 35 9 to 67 117 5037 ordin'y 19'2 4-6 to 6-3 4-3 13-9 to 19 12-9 Ethylene. (3) 19-5 4-9 546'2 12'5 71 28-6 Hydrogen. . (4) 39 to 21 26 to 14 500 to 600 mm. 3-3 to 7-0 5 to 10-5 (1) General directions given by Bunsen in his " Gasometrische Method en," 2nd ed., p. 127. The attenuations are calculated by the writer on the assumption that the pressure of the mixture is 600 mms. (2) Analysis quoted by Bunsen hi his "Gasom. Meth. ," 2nd ed., pp. 128, 129. (3) Test analysis by Carius ; p. 132 of the same book. (4) Recapitulated from preceding tables for the convenience of the reader. To be able to effect the synthesis directly in the eudio- meter, Bunsen recommends to virtually provide the eudiometer with a reduced volume scale, in addition to the millimetre scale which it bears actually, by inverting the eudiometer full of mercury over the trough, letting a succession of equal small volumes of air all measured off at the prevailing atmospheric pressure rise up into the eudiometer, and noting the resulting positions of the meniscus, taking care to keep the trough-level reading approximately constant. But it is just as easy to find the several points of the supple- mentary scale by calculation. Let us assume the eudiometer scale runs from an imaginary zero (a little below the top of the vault) to 500 nuns, as the adopted constant trough-level reading (close to the open end). Suppose now the eudiometer is charged with a gas down to R mms. ; this reading, according to the cali- bration table, corresponds to v volumes of gas, which v in Bunsen's o 194 GAS ANALYSIS. system is not far removed from R itself. The corresponding volume fr, at 760 mms. pressure is fr= v (760 + R 500) -=- 760, and is easily calculated. It suf- fices to effect the calculation for, say,?; = 20, 50, 100, 150, 200, 500, to be able to embody the relation between v and fr in a curve laid down in a system of rectangular co-ordinates, i.e., on a sheet of curve - paper.* From this curve we read off the values v, for ft = 10, 20, 30, 500, and enter them on a scale of v's in the way shown by Fig. 46. t To show the way in which the double scale is meant to be used, we will assume that the gas to be burned is known to be approx- imately half nitrogen and half marsh gas, and fills the eudio- meter down to R = v = 100 mms. J This, according to the scale, corresponds to b = 48, and con- sequently we have for the fr's : of marsh gas present, 24 ; oxygen needed, 48 ; mixture of substance and oxygen 48 + 48 = 96. But fe = 96 by the scale corresponds to R = 170, and down to there we must let in oxygen. But the gas, to afford a normal explosion, requires to be diluted down to, let us say, 24 x 15 = 360 volumes, V to V QAA to 10 oOU 310 230 20 320 240 30 10 330 250 260 40 340 - 270 50 20 350 280 60 360 290 70 30 370 300 310 80 380 320 90 40 390 330 340 100 400 50 350 110 410 - 360 1 OA l\f\ 370 380 130 430 390 70 140 440 - 400 150 80 450 410 - 420 160 90 460 i - 430 440 170 470 450 180 100 480 460 470 190 - 110 490 480 490 200 120 AA 500 OUU FIG. 46. * To be had from Messrs. A. & K. Johnston, Geographists, Edinburgh. fin the actual diagram, as used, the scale of "V's" should be a real milli- metre scale, the scale of "to's" subdivided, and both be more correctly drawn than our figure can pretend to be. J We assume, for simplicity's sake, that R = v. Anybody will easily see his way towards constructing a scale which gives R, v, and to in function of one another. WORKING METHODS. 195 which we will assume correspond to R = 360 exactly. Down to this point, 360, accordingly, we must let in air. This is our own modus. Bunsen would say : Add a reduced volume of air, so that the b of the mixture becomes equal to so-and-so many times, say 13 times, the original fr of the marsh gas, which would make the b of the mixture 13 x 24 = 312, corresponding, according to our scale, to v = R = 373. Of course, there is considerable latitude, or else practical gas analysis by combustion would be an impossibility. The addition of large volumes of air to the gas to be fired detracts from the precision of the work generally, and especially from that of the determination of the nitrogen in the ultimate residue (we mean the burned gas minus carbonic acid). Thomas * was the first to conceive the happy idea of effecting the necessary attenuation by mere expansion through reduction of pressure, which the kind of apparatus he used f readily lent itself for. Lothar Meyer and Seubert J subsequently took up the same idea, and by the invention of a supplementary con- trivance rendered the Bunsen eudiometer available for putting it into practice. Their apparatus is represented in Fig. 47, which requires only a few words for its explanation. As shown by the figure, the trough is pierced by two perforations in its bottom, which widen out above into cylindrical sockets for accommodating the wider ends of two india-rubber corks. The outstanding ends of the corks are fitted, one into the eudiometer, the other into the wider limb of a long Gay-Lussac-burette-like combination of tubes, which bears a millimetre scale and serves as a barometer. The tube g (which should go down so far that the highest india-rubber joint below the small bulb is under the pressure of about 760 mms. of mercury, so that there is no fear of air leaking in there) serves to make the eudiometer and barometer * "Chem. Soc. Journal." Transactions 1879, vol. xxxv., p. 213. f It was a " Frankland." Described and figured in Thomas' " Memoir ;" also (more briefly and without a figure) in the Author's article " Analysis " in the new edition of Watt's " Dictionary of Chemistry," vol. L, p. 243. J " Chem. Soc. Journal." Transactions 1884, vol. xlv., p. 581. 196 GAS ANALYSIS. communicate with one another, and, by means of a long stout india-rubber tube, with the mercury reservoir Q. To prepare the apparatus for use, the trough is emptied so far that the bottom end of the narrow limb s is free, the eudio- meter being away meanwhile. A small quantity of water is then sucked up through s, and the eudiometer (moist and full of mercury) put in its place. By alternately raising and lowering the reservoir it is easy to fill the barometer completely with mer- cury, expel the air from it, and moisten its walls with water, so that the readings are, as they come, corrected for the tension of the vapour of water. How the apparatus is utilized for expanding a gas contained in the eudiometer is too obvious to require to be explained. In using the apparatus as a measurer, Lothar Meyer and Seubert (rightly) recommend to so place the reservoir that the surface of the mercury in B is about at a level with the zero of the eudiometer. The gas volume then becomes nearly equal, numerically, to the dif- erence of level between the meniscus in the eudiometer and that of the barometer, i.e., equal to the dry pressure p of the gas ; and, as a consequence, the relative error in pv (and con- sequently in the "reduced volume") is at its minimum. By means of their apparatus, Lothar Meyer and Seubert have FIG. 47. WORKING METHODS. 197 ascertained, for a series of different knallgases (1) the minimum partial pressure at which the respective knallgas, when a spark is sent through it, catches fire ; and (2) a range of values of this partial pressure within which the explosion is both safe and effective. The following table gives their results in the latter respect, and besides, in so many additional columns, the corre- sponding partial pressures of the fuels and the attenuations of these and the knallgases : Fonnulie of fuels and Knallgases. Partial Pressure in mms. Attenuation referred to unit vol. at 760 mms. Knallgas. Fuel. Of Knall- gas. Of Fuel. H 2 + |(X 176-127 117-85 4-3-6 6-5-9 CO + 10 2 CH 4 + 2O 2 243-219 140 80-70 162-146 47 20-17-5 3-1-3-5 5-4 9-5-10-8 47-5-2 16 38-43 C 2 H 2 + 2-5O 2 50-40 14-3 _ 11-4 15-2-19 53-66 C 3 H 6 + 4-50 2 89-80 16-14-5 8-5-9-5 47-52 C 3 H 8 + 50 2 80 13-3 9-5 57 For the sake of more convenient comparison, we contrast the attenuations (implicity) recommended by Bunsen, with the cor- responding ones of Lothar Meyer and Seubert : Attenuation (per unit vol. at 760) of Gas named in Column I. Name of Gas. Formula. Bunsen. L. M. and S. Hydrogen, . . . . H 2 4-9-10-5 6-5-9 Orel. Knallgas, . H 2 + JO 2 3-3-7-0 4-3-6 Marsh gas, .... CH 4 13-19 16 Ethylene, .... C,H 4 29 38-43 , 198 GAS ANALYSIS. With Lothar Meyer and Seubert's, or any other apparatus admitting of the method of expansion, the order of operations with a gas of unknown composition is as follows : After having added a sufficiency of oxygen, we expand so far as to be sure to be on the safe side of the safety line, and then apply the spark. If no explosion occurs, we increase the pressure somewhat and try again, and so on until we succeed. If no explosion occurs even at (let us say) one atmosphere's pressure, we must add a sufficiency of ordinary knallgas (H 2 + JO 2 ) to render the gas explosive, and fire then. To calculate the requisite quantity of knallgas, we assume (1.) In the first instance, that the gas as given is substantially free of fuel of any kind. In this case an attenuation of 3'3 is sufficient. Hence, at 760 mms. pressure, we may have 1 volume of knallgas and 2*3 of diluent, or for 1 volume of diluent (given gas) l/2'3 = 0*435 volume of knallgas. Hence, supposing the ft of the given gas to be = $", the addition of 0'435 X of knallgas should produce 1'435 of a mixture fit for explosion at 760 mms. But then (2.) The given gas may contain some kind of knallgas which only missed fire on account of its being too largely diluted with other things. In the absence of any knowledge of the quality and quantity of this hidden knallgas, let us assume that the (reduced to 760) volumes of given gas contain (virtually) 1/10 x of (H 2 + i0 2 ). If so, the total knallgas amounts to (0'435 + 01) = 0'535 units, and consequently must be expanded into at least 3*3 x 0'535 $ = 1765 units of actual volume by reduction of pressure before the spark is tried for the first time. Certain gas analysis apparatus (e.g., Reg- nault's and the Author's) admit of the pressure being raised to more than one atmosphere, and thus afford an additional resource. The Author's apparatus, besides, enables one to con- veniently fractionate the given gas, so that a projected scheme of operations can be rehearsed with a small sample before being applied to the bulk of the gas. Bunsen's methods leave nothing to be desired on the score of elegance and potential precision, but they are very wasteful of time, and demand a special room for their execution, WORKING METHODS. 199 They, moreover, almost prohibit the use of gas flames as a means of artificial illumination, a serious drawback in a city like, for instance, Glasgow, where the sun does not shine evei*y day.* The desire to avoid these inconveniences has led to the con- struction of quite a series qf other apparatus than Bunsen's, including (as the latest arrival) one of the Author's invention,! which we will now proceed to describe, not, however, without having first stated that its leading feature is borrowed from Doyere, J in this sense, that the author, like Doyere, utilizes the Ettling gas-pipette as a means for the treatment of gases with liquid reagents, and as a means for the transference of gases generally. The gas-pipette, however, is such a useful instrument in itself that we will first of all describe it in the form which it assumed in Doyere's hands. A glance at Fig. 48 gives a sufficient idea of the construction of the instrument ; there is no need of our explaining how it is used for the mere trans- ference of gases from one tube to another, or of our defining the conditions and limits of its avail- ability for this purpose. To ex- plain how it is utilized for analyses by absorption, let us assume that we had measured a gas in a (sufficiently short) eudio- meter, and now wished to deter- mine, let us say, its carbonic acid by means of solution of caustic potash. For this purpose we begin by charging the pipette with mercury to the extent shown in Fig. 48. To be able to go any further we need a pneumatic trough provided with a deep well, * A Swan-Edison electric lamp would help us over this difficulty ; but it could only be thought of where an electric current of sufficient strength is always ready. f Dittmar's "Report on the Composition of Ocean Water" (Challenger Memoirs), Part III. As these reports do not appear to be widely diffused, I will again discharge the agreeable duty of thanking my then assistant, Mr. Robert Lennox, for the great assistance he gave me in the realization of my ideas. % Ann. Ch. Phys. [3], 28, p. 1. FIG. 48. 200 GAS ANALYSIS. which latter must be shaped so as to admit of the following operations : To introduce the absorbent we immerse the (J of the pipette in the well of the trough, and next blow out the air (without sending any more mercury after it than we cannot help). We then suck in a small quantity of the reagent from out of a test-tube inverted over the trough, taking care not to let any more mercury follow than necessary to seal the contents by a mercury thread, i I, in the (J. The pipette, if not wanted immediately, may, of course, be put aside for an indefinite time, because the reagent is fully protected against the action of the atmosphere. But assuming the gas to be ready for treatment, we transfer it in its tube to the pipette trough (by means of the contrivance represented in Fig. 55, p. 205), slide the tube over the outer limb of the (J, press it down so that the exit of the latter touches the top of the tube, and suck at a to dislodge the mercury thread, and then again until drops of mercury are seen falling into the bulb, but no longer. We then lift the pipette out of the trough, and, by shaking it judiciously, soon cause the carbonic acid to be com- pletely taken up by the reagent. All that now remains to be done is to re-transfer what remains of gas to the measuring tube, without allowing any of the reagent to follow. This, however, is a somewhat delicate operation, which it takes an apprenticeship to learn. The first step is easy. We place the tube intended to receive the gas over the outer limb of the U, and blow in at a to dislodge the mercury thread. If the quan- tity of mercury in the pipette is still (substantially) what it was at first, only part of the gas follows. To get out the rest we lift the pipette, so that its exit becomes visible within the gas space of the eudiometer, and while maintaining this con- dition for the pipette, cautiously raise the eudiometer. The more the eudiometer is raised, the more energetically it " sucks " at the pipette, and (after a deal of practice) one learns to so govern the (up or down) motion of the eudiometer that, when the thread of liquid reagent begins to show itself in the outer limb of the U, he is able to make it move up or down, or stand still, as he wills. We arrest it, of course, at say 2 mms.' distance from the outflow, then re-place the pipette on the table, WORKING METHODS. 201 which at once seals the outflow by a superincumbent layer of mercury, suck at the pipette until mercury is seen to flow in, and then lift it out and place it on the table. The gas is now transferred, and is ready to be measured. Whoever tries the operations detailed will find that they are not by any means as easy as they look on paper. A little device of Mr. Buchanan's facilitates them greatly ; it consists in this, that the bend at d is made capillary, so that any error 'of judgment committed does not find such immediate punishment as it would otherwise. The writer, some years ago, finding it difficult to work the Doyere pipette, satisfactorily modified its construction in a manner shown by Fig. 49, which, however, represents an improved form of the Author's contrivance which was devised subsequently by Mr. Lennox. In preparing the pipette for use the first step, of course, is to fill it completely with mercury, and the second to charge it with a small quantity of the respective reagent. This is done pretty much in the same way as in the case of the original Doyere, but the moveable mercury reservoir greatly facilitates the work. Both remarks apply to the mode of introducing the gas. To FlG - 49 - blow back the gas residue into its test tube, the reservoir is raised, the stopcock of the pipette neck opened, and the gas allowed to flow on until the liquid reagent has come to about a millimetre on the safe side of the point where the side tube joins it. The principal stopcock is then closed, and that of the side tube opened to drive out the remaining thread of gas by a current of mercury. 202 GAS ANALYSIS. The modified gas-pipette, combined with a measurer and exploder, and the two necessary troughs, complete the author's apparatus. The measurer (Fig. 50) is a combination of a wide with a narrow glass tube, after the manner of Gay-Lussac's burette. As shown by the figure, it is fixed within a square water-bath, the front and back of which are made of plate glass. The back pane had better be made of ground glass, which affords a more satisfactory view of the meniscus when gas-light is used. The wide tube by its lower contracted end, and a long capillary india-rubber tube attached to it, communi- cates with a mercury reser- voir, the support of which slides up and down at D. At their upper ends both tubes are provided with Geisler stopcocks. To the exit end of that of the wider tube is soldered the capillary U, characteristic of the Ettling pipette. The wide tube bears a millimetre scale ; the gas volumes corresponding to the several marks are deter- mined by gravimetric cali- bration. The bit of capillary tube between the stopcock and the top of the measurer joins on to the latter quite abruptly; the volumes are counted from the point of junction, because after the introduction of a gas the narrow canal retains its thread of mercury. FIG. 50. WORKING METHODS. 203 In the calibration of the measurer it is expedient to use a special (light) water-jacket, which is easily improvised out of that of a glass Liebig's con- denser, but should be free of flaws and irregularities of form. More convenient than such an extemporised water -bath is a stout, four-legged stool of iron, so constructed that it supports the ordinary measurer's bath with safety, and raises the latter's outlet to a sufficient altitude. In a laboratory where there is a continuous supply of w^ater direct from the street pipes, the maintenance of a con- stant temperature is easy. In the absence of such a water supply the temperature must be kept constant to within 1C. about, somehow. The jacket being fixed over the measurer, the lower end of the latter is connected with a capillary stop- cock, by means of a bit of very stout india-rubber, in such a manner that only a mere (circular) line of india-rubber is under the influence of the pres- sure of the mercury. The reser- voir tube is slipped over the exit end of the stopcock, and raised until the metal has just got above the stopcock of the narrow side tube, which cock is now closed. More mercury is FIG. 51. (Side View). FIG. 52. (Top View). 204 GAS ANALYSIS. then let in until the metal has just passed the stopcock of the wide tube. After having made sure that there are no air-bells imprisoned anywhere, we close the cock below, remove the reservoir, and next, while keeping cock b closed, let out mercury very cautiously until the metal stands exactly at, or perhaps a millimetre above, the zero point referred to. The business of calibration then commences. We tare a convenient vessel (say a porcelain basin) on a fine balance of sufficient carrying power, run in mercury down to about 10-20-30, &c., rnms., and in each case, after having taken the exact reading with the telescope, weigh the whole of the mercury which has been discharged so far. In a second series we check the points 15 25 35, &c., and from the two series combinedly deduce a cali- bration table, the volume of 1 grm. of mercury at the tempera- ture prevailing in the bath during the calibration serving as a convenient unit ; unless we prefer to so adjust the unit of volume that the numbers for the volumes agree as nearly as possible with the readings, which in the subsequent use facilitates interpolation. The calibration- table in any case, to be con- venient, must proceed from millimetre to millimetre. The measurer (in its bath) holds a fixed position on one side of a trough (Figs. 51 and 52) provided with two deep wells, one for the accommodation of the U of the measurer, and the other for that of the exploder, which stands on the other side. FIG. 53. SVORKING METHODS. 205 The construction of the exploder is seen in Fig. 53. It must be relatively strong in the body, and so wide as to enable one to expand a gas transferred to it from the measurer to a suffi- cient extent before firing it. It is convenient to graduate the exploder roughly, to enable one to establish a predetermined ratio of expansion. The graduation, however, had better not be laid down on the tube, but on a slip of wood or Bristol board, which is held against the measurer when needed. The absorber has a trough of its own. A gas to be analysed, to be within reach of the measurer, must be contained in a short enough stout test-lube inverted over a mercury trough. Fig. 54 represents a convenient suction tube, invented by Doyere, for taking the air out of such a short gas tube while it stands over the well of the trough inverted, and thus filling it with mercury without producing any air-bells at the sides. A small portable iron trough (see Fig. 55) serves to convey the gas to the trough of the measurer, which we assume to have already been filled com- pletely up to the end of the U with mercury. The cock of the side tube is kept shut until the gas has been sucked over into the measurer, which is then closed by turning its cock. The gas then is brought to very nearly the pressure of the atmosphere by placing the reservoir in the proper position. This being done, the side tube is made to communicate with the atmosphere, and the reservoir readjusted until the two mercury menisci, in the graduated and side tube respectively, are in the same horizontal plane. A horizontal "wire" in the focus of the telescope facilitates this adjustment greatly, but, as the two menisci almost touch each other, the wire is not indispensable. The gas now is at the pressure B + b mms., where B stands for the height of the barometer, and b for the small excess of capillary depression in the narrow, as compared with the wide, tube. Only for absolute measure- ments (reduction of gas- volume to weight) needs b be known ; FIG. 54. 206 GAS ANALYSIS. it is easily determined by letting both branches of the measurer communicate with the atmosphere and noting the difference of level. With an apparatus which served in our laboratory for a long time, b was equal to 0*6 mm. We ought to have stated before that the inside of the measurer is kept moist, so that the dry pressure of the gas PisP = B + 6 ?r; where TT signifies the maximum tension of vapour of water at the temperature of the gas, i.e., of the bath as given by an immersed thermometer. As the pressure and temperature remain nearly constant throughout an analysis, it is not expedient to reduce to unit disgregation, but better to reduce to a set of convenient integer values P and T lying close to the observed values for P and T by means of the formula Where A? and AT are the excesses of the observed P and T over the adopted standard values P and T , and both increments are reckoned positive. If the reciprocals of the practically needed values of P and T are at hand, the calculation becomes so easy that it is not worth while to look out for any mechanical arrangement which would effect the joint correction automatically. In calculating the -=^- the capillary depression term b may be neglected. In many short analyses P arid T are constant, so that the volumes as measured are " reduced volumes " in themselves. All the rest of the modus operandi may be left in the hands of the student. For his convenience we append a table of recip- rocals of practically occurring values for P and T . EXAMPLE. Found. V = 456-3 at P = 762'3 and t = I4*'o. Wanted. V for 760 mms. and 15. Both corrections are positive. That for AT is 0'5 x "00347 = '00174 A? is 2-3 x -00132 = -00303 The two conjointly, . . . = '00477 Hence A V = '00477 X 456 = 218, or V = 458'48. WORKING METHODS. RECIPROCALS OF (273 + t.) 207 t. 1: (273 + Q. t. 1: (273 + t. 1:(273 + *. 003663 10 003534 20 003413 1 003650 11 -003521 21 003401 2 003636 12 003509 22 003390 3 003623 13 003497 23 003378 4 003610 14 003484 24 003367 5 003597 15 003472 25 003356 6 003584 16 003460 26 003344 7 003571 17 003448 27 003333 8 003559 18 003436 28 003322 9 003546 19 003425 29 003311 RECIPROCALS OF P. P. 1:P. Diff. 710 001408 -20 720 001389 -19 730 001370 -19 740 001351 -19 750 001333 -18 760 001316 -17 770 001299 -17 780 001282 -17 790 001266 -16 208 GAS ANALYSIS. EXERCISES. WHATEVER kind of gas apparatus the laboratory where the student goes through his course of instruction may afford, he should begin by making himself familiar, as far as the resources of the laboratory may permit, with Bunsen's apparatus, because Bunsen's system is the only one which he could think of extem- porising in his future practice as a professional chemist. Hence, the Author, although his own laboratory hardly affords all the facilities for doing full justice to Bunsen's method, recommends to his students the following series of exercises as an intro- duction to the subject: Ex. 1. Making of a Bunsen Eudiometer. Procure a good piece of glass tubing about 600 mms. long, 20 mms. wide inside, and of 2 mms. strength of body, which is as nearly as possible straight and cylindrical, and free of irregularities generally. By means of a pair of calibers it is easy to test a series of tubes quickly and select the best. Close the tube at one end and grind the other flat, first on a flat wet sandstone, and then on a lead plate with emery and turpentine. Then fuse in the platinum wires. To make the necessary perforations heat the respective part of the tube in the blow-pipe flame, stick on a piece of stiff platinum wire, and by means of it pull away a small portion of the glass laterally, so as to produce a very narrow capillary branch, which then cut off close to the tube. The rest is more easily explained by demonstration than in words. If you are not enough of a glass blower to solder in the wires properly, procure a ready-made eudiometer, but do not forget to test the wire joints thus : Fill the eudiometer with mercury completely, insert it in the trough, and repeatedly tap the bottom of the trough with it as sharply as compatible with the fragility of the instrument. If air-bells are seen to rise from the wires, the joints are leaking, and the eudiometer is useless. [Professional glass-blowers now-a-days often use a thread of a special kind of glass for, so to say, cementing in the wires, but we do not know what sort of glass it is]. Assuming now BUNSEN EUDIOMETER. 209 the eudiometer still requires to be graduated, empty out its mercury into a graduated cylinder (or into a tared vessel) to determine its total capacity directly (or indirectly from the weight of the metal). From the volume found, and the total length of the eudiometer in millimetres, calculate the average volume value of 1 mm. of eudiometer space. In order now to tind a convenient point for the zero of the scale, pour mercury equal to 20 mms. into the eudiometer, and mark the level of the top of the meniscus by means of red ink thickened with gum. The eudiometer is supposed to have been cleaned and dried before this is done. Close the tube with a cork, or better, by means of a round stick of wood fixed in its open end as a handle, heat it in its entire extent to beyond the fusing point of wax, and then paint it all over with wax kept molten in a basin, and allow the wax to freeze, taking care to turn the tube round frequently to prevent one-sided accumulation of the wax. After having thus produced a thin uniform coating of wax on the tube, scratch in a millimetre scale, so that the glass is laid bare at the intended marks, taking care that the 20-stroke coincides with the red ink mark made (which is quite clistincty visible through the wax) ; number the scale from the close end downwards, and etch it in with hydrofluoric acid. Gaseous H F gives the more satisfactory scales, the liquid acid is the more convenient agent.* The scale needs not be etched in very deeply, because it is always read with a telescope which magnifies considerably ; but this being so, the figures must be very small and stand close to the general scale, each on its projecting centimetre stroke. For the construction of the scale a screw-engine, so contrived that every revolution of the wheel advances the cutting tool by exactly one or one-half of a millimetre, is by far the most satis- factory apparatus. But such engines are very expensive, and for the mere making of millimetre scales Bunsen's copying engine (if provided with a really correct standard scale) is almost as good. A drawing of it is in Bunsen's book, p. 28 of the second edition; but its construction can be made clear enough by a verbal description. Imagine a table about 2 metres * Both are dangerous substances, which require to be handled cautiously. Ask the Demonstrator to instruct you. P 210 GAS ANALYSIS. long with a flat semi-circular, or better, roof-shaped gutter run- ning from one end to the other. In this gutter the eudiometer to be graduated is fixed at the right, the standard scale (which is etched pretty deeply into a hard glass tube of about the diameter of a eudiometer, and should have a range of about 800 mms.) at the left end. Both tubes are held fast by means of strips of strong sheet brass screwed to the table, so that their outer edge, which must be exactly rectilinear, is in the middle line of the respective tube. The ruler for the standard scale is a plain straight edge ; that for the eudiometer is provided with notches constituting a half-centimetre scale, every second notch being a little longer than the preceding one. The tubes being fixed, an exact copy of the standard scale is produced on the eudiometer by means of a long beam compass. While one foot is made to slide along the straight edge of the standard from mark to mark, the other foot (governed by the right hand) serves to scratch the marks into the wax coating. The standard-foot of the compass should have the shape of a strong but sharply pointed needle, the other that of a hatchet with an oblique edge which cuts in the marks through the wax instead of scraping them off" as a needle would. To convert his copying into a real dividing engine, good enough for making burette scales or the scales of gas tubes divided into cc., &c., Bunsen has added to it a glass plate bearing a system of etched-in lines so laid down that they all (virtually) pass through the apex of an isosceles triangle, whose base they divide into (say 22) equal parts of convenient length. The writer has modified the Bunsen dividing engine in the following points : The system of lines is on an exactly rectangular plate of plate glass, and so arranged that they constitute a rectangular triangle and divide the lesser cathete of this triangle into 50 or 100 equal parts. The glass plate, when used, lies between two brass rails fixed on the engine table, a brass spring attached to one side of the glass plate causing the other to lie flat against its guiding rail. At a convenient place a flat thin ruler with a bevelled edge goes across the plate from rail to rail ; it is held fast by two thumbscrews, in such a way, however, as to admit of CALIBRATION OF EUDIOMETER. 211 slight alterations in the adjustment of the edge in reference to the system of lines. The undivided cathete of the triangle is parallel to the brass guiding rail against which the glass plate is being pressed by the spring. Along it runs a scale of equal parts, so adjusted and figured that it gives the exact length of one degree of the scale produced by the intersection of the bevelled edge with the system of lines. To divide a given length into, say, 10 equal parts, measure it out in millimetres, and divide by 10 to find the exact place where the bevelled edge must go across the system to produce degrees of the intended length. The beam compass is constructed so that, while one foot is fixed, the other can be made to slide forwards and backwards ; the final adjustment being effected by means of a micrometer screw governing the respective part. Such beam compasses are to be had in commerce. The apparatus, to be quite convenient, must be provided with a number .of. glass plates, whose systems of lines overlap and supplement one another. One of the glass plates may bear a plain millimetre scale for the making of millimetre scales, but for the graduation of eudiometers or barometers the original Bunsen contrivance is better. Let us now return to our subject. The millimetre scale being completed, we proceed to the calibration of our instrument. Ex. 2. Calibration of the Eudiometer. In regard to this operation we have little to add to what we said on pp. 180-183. The thimble for measuring off the equal instalments of mercury is easily made. A sufficient length of stout tubing is closed at one end, a quantity of mercury equal to about 20 mms. of eudio- meter space poured in, and the tube cut off at a level with the top of the meniscus. It is then ground down so as to hold very nearly the predetermined quantity of mercury, and fit the ground glass lid without rocking. Of course, there is no need of a sepa- rate thimble for each eudiometer, but it is useful, in addition to the ordinary (" 20 volume ") thimble, to have a second which holds exactly five times the weight of mercury. The larger 212 GAS ANALYSIS. thimble enables one to find the points corresponding to 100, 200, 500 volumes promptly and more exactly than would be possible with the smaller one. This being done, a second cali- bration is made with the small thimble, and the results are discounted for the subdivision of the 100- volume intervals. The details of the calculation will easily be found out by the student on a little reflection. Ex. 3. Preparation of a Reduced Volume Scale for the Eudiometer (see pp. 193 and 194). Our dividing engine enables one to graduate the scale with ease and exactitude. Ex. 4. Making and Calibration of a Bunsen Absorption Tube (see p. 179). The student, after having made the two gas-tubes, may utilize them for carrying out the first of the following exercises in gas analysis repeatedly until he has learned to work them properly. This being done, our own students pass to the use of our apparatus. Ex. 5. Calibration of the Measurer in the Authors Apparatus. We keep a stock of ready-graduated measurers, so that each student can have one of his own, but he must do the calibration himself. (For the modus operandi, see pp. 202 and 203). Gas Analysis. We presume our own apparatus to be used. Ex. 6. Analysis of a Mixture of Air and Carbonic Acid. Fill the measurer about one-third full of pure air, and measure this air exactly ; then add some pure carbonic acid, and measure again. In the gas mixture thus produced determine the carbonic acid by absorption with caustic potash solution, in the pipette, and, in the residual air, the oxygen by explosion with hydrogen, MARSH GAS, ETHYLENE, &c. 213 as explained above. Compare the percentages found with those demanded by the synthesis. Atmospheric air contains 21*0 per cent, by volume of oxygen. Ex. 7. Analysis of a Mixture of Carbonic Oxide and Carbonic Acid. Prepare some of the mixture CO + CO 2 = 2 volumes by decom- posing oxalic acid with oil of vitriol in a small flask, and analyse the gas thus : (1) Absorption of the C0 2 by potash, and measur- ing of the residue ; (2) Combustion of the residue with oxygen ; (3) Determination of the C0 2 produced in (2), after addition of a measured volume of air to enable one to measure any residue. The theoretical composition is seen in the formula. Ex. 8. Analysis of Marsh Gas. To obtain pure marsh gas heat a mixture of 1 part of anhy- drous acetate of soda and 4 parts of soda-lime in a combustion tube,* and purify the gas by passing it first through a small U-tube or wash-bottle (either had better have glass joints only) charged with bromine, and thence through a U-tube charged with pumice soaked in caustic soda solution. After having expelled the air collect what follows in a mercury gas-holder, and preserve it therein for analysis. Analyse the gas by combustion, taking care that the mixture to be fired is at about 15 times the volume which the CH 4 in it would occupy at 760 mms., or else you may shatter your exploder. Calculate the volumes of " carbon vapour " and hydrogen gas per unit volume, and see if the results agree with " CH 4 ." Ex. 9. Analysis of Ethylene. The best method of preparing pure ethylene is Mitscherlich's. It consists in passing vapour of pure 85 per cent, alcohol (not of * C. A. Brinley recommends to dissolve 750 grms. of caustic soda and 750 grms. of acetate of soda in 800 cc. of water, and add 1250 grms. of coarsely powdered quick-lime. The mixture is cautiously evaporated into dryness in an iron basin, then powdered coarsely and heated in an iron flask or in a combustion tube. Yield, 125 litres. 214 GAS ANALYSIS. methylated spirit) through boiling vitriol so diluted with water that it boils at 165C.* To free the gas from ether and alcohol vapour, pass it through (1) an empty bottle kept cold in ice, and (2) oil of vitriol. Collect the gas, in the first instance, over boiled- out water alkalinized strongly with KHO in a small Pisani gas- holder. From it transfer it without too much loss of time to a mercury gas-holder or distribute it in tubes trapped by mercury. To make quite sure of the absence of ether and other foreign vapours, treat a quantity in a pipette with ordinary vitriol. Of the gas thus finally purified pass a quantity into the measurer, measure it exactly, and dilute it with air to an exactly known (say twice its original) volume. Analyse this mixture thus : (1.) Determine in a portion the ethylene by absorption in fuming oil of vitriol. The kind of crystalline, partly hydrated S0 3 , which was referred to in the exercise on Kjeldahl's method, p. 83, is well adapted for the purpose. A measured volume of the gas is transferred to a plain tube of suitable dimensions, and treated therein with a coke-ball soaked in the reagent. The residual gas (air) is sucked into a pipette charged with caustic potash, to be freed from SO 3 and S0 2 vapours. It is thence transferred to a clean tube, and from it to the measurer. (2.) After having thus determined the percentage of C 2 H 4 in the mixture, and thus indirectly that of the real C 2 H 4 in the ethylene used/r measure off a small quantity of the same mixture and subject it to the process of combustion with oxygen. Take great care that the mixture to be fired is at the proper attenuation. According to our experience, an expansion into 30 volumes (volume of real C 2 H 4 at 760 = 1) is sufficient. As long as you are comparatively inexperienced in this opera- tion, better arrange matters so that if the exploder itself should explode you are personally safe. Determine the contraction and the carbonic acid produced, and in the residue, after absorption of the latter, the oxygen left unburned to obtain the nitrogen. Compare all your results with the corresponding values * Bulb of the thermometer in the liquid. f With Mitscherlich's method it is not easy to obtain a gas which is quite free of air, EXTRACTION OF ABSORBED GASES. 215 demanded by the synthesis as supplemented by the proximate analysis. Ex. 10. Analysis of a Mixture of H 2 , CO, CH 4 , and N 2 . Pure carbonic oxide is easily prepared by decomposing a for- mate with (not fully) concentrated sulphuric acid, or else from the CO + CO 2 produced from oxalic acid, by removal of the C0 2 in a gas-pipette ; nitrogen, by passing purified (C0 2 and H 2 O free) air over red-hot metallic copper. From supplies of the pure gases kept shut up over mercury, prepare a mixture of them by exact quantitative synthesis, and then analyse the mix- ture by combustion. The " attenuation " for the mixture to be exploded is calculated from the synthesis ; if you were quite ignorant of the quantitative composition of your gas, you would naturally try the spark first at the attenuation 15, and then, if necessary, proceed to less volumes. From the values (C, K, |t) found, and the relation x + x" + x" + x"" = \, calculate these per-unitages, and compare them with those demanded by the synthesis. Ex. 11. Extraction of the Absorbed Gases from, a Water, and their Determination* Substance. To obtain a suitable water to operate upon, pre- pare a supply of perfectly pure water by " torturing " ordinary distilled water with strongly alkaline permanganate, boiling off the ammonia and re-distilling. Place one or two litres of such water in a glass-stoppered bottle of about twice the capacity, and shake it violently with some 20 successive fills of the bottle of (always fresh) normal atmospheric air,f taking care to maintain, as nearly as possible, a constant temperature, which is best done by means of a large water bath kept at the temperature of the laboratory, into which bath the bottle is immersed for a sufficient time whenever the contents, when tested with a thermometer * We refer to only such waters as derive their gas contents from the normal atmosphere ; natural strong absorptions of CO-2, for instance, such as Apollinaris water, &c. , demand special methods. f We mean not the air of the laboratory. 216 GAS ANALYSIS. immersed in them, show a tendency to get too warm. The air is sucked from the open atmosphere through a glass tube passing through a perforation in the window frame into a pair of suction bellows, and thence blown into the air-space of the bottle. At the end of the absorption, note the temperature of the water operated upon and the height of the barometer. For the Extraction of the gases, Bunsen's method, as modified by Jacobsen, is the best. Jacobsen's apparatus is represented in Fig. 56. A is a globular flask of 700-800 cc.'s capacity, provided with a well-fitting soft india- rubber cork whose one per- foration accommodates the stem of a pear-shaped bulb a. The cork is pressed down to a fixed point on the neck. The exact capacity of the flask up to this point must be determined, best by weighing the quantity of water which it holds when filled so far. The stem of bulb a is closed below, but has a small lateral per- foration at e. Tube b serves for the reception of the gases. One litre of water absorbs less than 30 cc. of air gases, even at 0C. ; hence a capacity of 60 cc. per one litre of water operated upon should suffice for the gas-tube b ; but it is better to adopt the same size (60 cc. about) for a flask of 700-800 cc.'s capacity, as supposed to be represented in the figure. The order of operations is as follows : The flask is filled with the water to be analysed through a very wide funnel tube, which goes down to its bottom. The water is poured in cautiously so as to FIG. 56. EXTRACTION OF ABSORBED GASES. 217 avoid the formation of gas-bells, and, when the flask is full, an additional 200 cc. or so are poured in by the funnel and allowed to run over the edge of the flask. The flask having been thus charged to overflowing, the cork is screwed in down to that fixed point, and the stem of the pear-shaped bulb (which has previously been charged with a little pure water) pressed into the perforation of the cork, in such a way, however, that the lateral perforation is within the cork. The gas-tube b is now joined on by means of a good india-rubber tube secured with wire, and then the air is expelled from the bulbs by heating the water in a to boiling and keeping up the ebullition until the air can be assumed to be all away. The bit of india-rubber tube at the end of b is now closed with a clip, the lamp with- drawn, and the end of b sealed up. The stem of bulb a is now pushed down so that the perforation is within the water, and the flask then placed in a heated water-bath, to expel the gases. As there is a strong vacuum in the flask at first, the boiling commences far below 100C., and as the whole of the gas of the water at 760 mms. pressure would fill less than half the space at its disposal, as long as the gas tube remains sufficiently cool* there is no fear of the pressure within rising beyond that of the atmosphere even at the end. The whole of the operation, indeed, can be conducted without ever raising the temperature of the bath to 100C. After 1-2 hours' boiling the gases can be assumed to be ex- pelled, and the next step is to drive the whole of them into bulb b and shut them up there. With some practice it is easy, at the end of the operation, to almost completely fill bulb a with hot water, so that little eras is there. The stem of a is then drawn ' o up so as to bring the lateral perforation within the cork, f and the top stratum of water in a kept boiling for a while so as to drive the gases into b, which is then closed by pinch- ing the indiarubber tube which connects it with a the o , t lamp, of course, being removed immediately after. Should a little water have gathered in 6, this is easily let down into a * This condition almost takes care of itself. t If a naked flame is used instead of a water -bath, this flame of course must 1)6 removed immediately afterwards, 218 GAS ANALYSIS. by cautiously opening the screw clip at the joint for a few moments, until the water is just sucked down. The gas tube is now detached, and its lower end sealed up to bring the gas into a condition in which it may be preserved unchanged for any time. For its Measurement it is transferred first to a plain tube in Bunsen's trough (see p. 183). It is then sucked into the measurer and measured. The results are reduced to cc., measured dry at and 760 mms. pressure, and referred finally to 1000 cc. of water analysed. For the reduction to Bunsen's table of the logarithms of (1 H-aO comes in handy ; if it is not at hand calculate thus : Log. 0-359211=0-555 349-1, or, to put it into an easily remembered form : reduce to 1000 mms. pressure and 359"2* absolute temperature. Analysis. The carbonic acid is absorbed by solution of caustic potash in the pipette, and determined by re-measuring the residue. The latter is divided into two parts for two determinations of the oxygen by explosion with hydrogen. In calculating the latter, remember that water-air contains some 33 per cent, of oxygen, not 21, as ordinary air does. For 100 volumes of gas to be analysed add some 80 volumes of hydrogen, and as the knallgas is certainly less than 3 x 40 volumes, but not much less, an ex- pansion into 3'3 x 120 = 400 volumes before firing will work well. Finally, compare your results, for the absolute composition of the air, in 1000 cc. of water analysed, with the values calculated from the co-efficients of absorption and the barometric pressure at the saturation of the water with air, and the law of gas absorption. The proportion of carbonic acid in ordinary air (as taken from, for instance, a Glasgow street) may rise to 0"05 per cent, by volume instead of the normal value of 0"03 ; but this does not disturb the percentages of oxygen and nitrogen in the carbonic acid free part of the absorbed air. Consult Dittmar's " Tables to facilitate," &c., 2nd edition, pp. 41-43. * 359 = number of degrees in the circle minus 1, for mnemonic purposes, PROMISCUOUS EXERCISES IN APPLIED ANALYSIS. 1. Analysis of a Sea Water.* Preliminary Note. The results of a sea water analysis should be given in two ways, namely : (1) in reference to 1000 grms. of the sea water; and (2) in reference to 100 grms. of chlorine, meaning total halogen, determined titrimetrically by silver, and calculated as chlorine. The several samples of sea water re- quired for the determinations should be measured off in pipettes, but weighed exactly immediately afterwards. The weight of the sample enters the final calculation. In the sequel, "Take n cc. " means, weigh out a quantity equal to n cc., by pipette measurement. CHLORINE. Memoir, p. 4. I. For an approximation, Mohr's method is very convenient. Requirements: (1) Neutral silver solution, prepared by dissolv- ing 17 grms. of pure fused nitrate of silver in water, and diluting to 1000 cc. (2) A solution of chromate of potash K 2 CrO 4 , containing 5 grms. of the salt per 100 cc. To analyse a sea water, take 5 cc., add a few drops of chromate solution, and then, from the burette, nitrate of silver solution until the red colour of chromate of silver, which appears transi- torily from the first, becomes permanent on stirring. Every ,1 cc. of solution used indicates 1 10 Cl = 3'545 mgs. of chlorine, very nearly. A far more exact result is obtained if the silver * Mostly abridged from Dittmar's Report on the Composition of Ocean Water, in "Report on the Scientific Results of the voyage of H.M.S. ChaJle.nyer" by Sir C. Wyville Thomson, Knt., F.R.S., &c., and John Murray, F.R.S.E. ; "Physics and Chemistry," vol. i. To be referred to as " Memoir," 220 EXERCISES IN APPLIED ANALYSIS. solution is standardized empirically with a sea water whose chlorine has been previously determined with high precision, and the rule be followed in both the standardization and analyses to maintain constant ratios between the total silver solution used on the one hand, and the total volume of sea water and reagents at the end, and also the volume of chromate added as an indi- cator, on the other. That a certain depth of redness must be adhered to, as indicating the "end" of the reaction, stands to reason. Taking S as a symbol for the volume of total silver solution required, and M as that of the final volume of the mix- ture, the relation M = 1/25 S can be maintained in all cases by addition of water ; of course, M needs not to be measured with high precision. II. Exact method (Volhard's modified). Requirements : (1) Perfectly pure chloride of potassium. Prepare pure perchlorate, and heat the salt in a platinum basin until the oxygen is apparently gone. Then fuse the residue in a platinum crucible, and keep it in fusion until every trace of oxygen is sure to be away. The salt then may contain a little surplus alkali. There- fore, dissolve the fuse in water, add a few drops of pure hydro- chloric acid, and evaporate in platinum (or Berlin porcelain, which is almost as good) to perfect dryness. Next, keep the dry salt in a crucible, close to but below its fusing-point, until the water may be presumed to be expelled. Now, powder the salt coarsely, and resume heating until the weight is constant. Preserve the salt in a good preparation-tube or bottle ; but before having kept it long, prepare from it a (2) Standard solution of Chloride of Potassium, by dissolv- ing 1/10 KC1 = 7'459 grms. in a tared litre-flask in water, and diluting to 1000 cc., thus : After having added about 0*9 of the water, mix by rotating, then fill up to the mark, determine the exact weight of the solution, and only then mix by shaking. Note down that 1/10 KC1 grms. = for instance, 1006'3 grms.* = (by intention to) 1000 cc. (3) A Solution of Nitrate of Silver, containing, as nearly as possible, 1/10 Ag = 10793 grms. of silver per 1000 cc., besides about 20 cc. of free nitric acid of T4 spec. grav. (See below.) * We mean the weight of the " 1 litre " of solution, as found, CHLORINE IN SEA WATER. 221 (4) A Weaker Solution of Nitrate of Silver, containing 1/100 Ag = r0793 grms. of silver (and 2 cc. of 1'4 nitric acid) per litre. If an exactly standardized supply of solution (3) is not at hand, this solution is best made by direct synthesis from weighed pure silver. (5) A Weak Solution of Sulphocyanate of Ammonium, con- taining 1/100 NCS grms. per litre in this sense, that, under the circumstances which prevail when it is used, it just saturates solution (4) volume for volume. This point must, of course, be made sure of by repeated direct trials. (6) Solution of Iron Alum. 50 grms. of iron-ammonia-alum, 20 cc. of 60 per cent, sulphuric acid, and enough of water to produce about 1 lit. of solution. Note. Make sure that neither silver solution contains nitrous acid, and that both the sulphocyanate and the iron alum are free from chloride. To test the sulphocyanate, precipitate a sample completely with nitrate of silver, wash the precipitate by decan- tation, and heat it (in a draught place) with concentrated sul- phuric acid and a little water until the visible reaction is at an end. Then allow to cool, and pour into much water. If chlorine is present, a milky opalescence will appear. The Standard Silver Solution. Its preparation is easy. We dissolve 1/10 AgNO 3 =17 grms. of pure fused nitrate in water, add 20 cc. of T4 nitric acid, and dilute to 1 litre. Or else we dissolve 1/10 Ag grms. of metallic silver in 29 cc. of 1*4 nitric acid, previously diluted to double its volume, and dilute to 1000 cc. As the solution must be standardized with the chloride of potassium solution in any case, absolutely pure nitrate or metal need not be used. To determine the exact titre both by weight and by volume, tare a 200 cc. phial, and weigh into it, first, 50 cc. of the standard chloride of potassium, and then 50 cc. + an additional 1 cc. all exactly measured of the silver solution ; shake violently, and allow the mixture to stand in the dark until the precipitate has settled so completely that (practically) the whole of the supernatant liquor can be decanted off clear into a beaker.* Do so, and determine the small remnant of * A little reflection will show whether or not it is worth while to allow for the dissolved silver remaining in the phial as part of a moist precipitate. 222 EXERCISES IN APPLIED ANALYSIS. dissolved silver by adding 5 cc. of iron alum, and then weak sulphocyanate out of a burette until the mixture becomes dis- tinctly red. With the help of the weak silver solution determine the end point several times by to-and-fro titration, and take the mean of the last 3-4 results, which ought to agree with their mean to within 1/20 cc. If everything was well done, far less excess of silver solution than 1 cc. will be sufficient to make sure of the existence of a remnant of dissolved silver at the end, so that, in a second and third trial with 50 cc. of KC1, 501 cc. of strong silver or 50 cc. of strong plus 1 cc. of the centesimal solution will do. Should it be the case that the first few drops of sulphocyanate strike a red tint, do not take it for granted that you have hit the point, but add a measured sufficiency of the weak silver solution, pour the whole back into the phial, shake up, allow it to settle, &c. The volumes of added reagents must, of course, be noted down and allowed for in the calculation. From the mean of a number of well-agreeing experiments, calculate the volume-titre and the weight-litre thus : Supposing 50 cc. of chloride solution to have required 50 cc. of silver and 3'2 cc. of weak NCS solution, the excess of silver obviously is = 0*32 cc. of strong solution, and 49*68 cc. of silver solution arc equivalent to 50 cc. of chloride solution = by intention to 50 x 1/10 Cl mgs. It is better, however, to base the calculation upon the dry KC1 as calculated from the weight of the 50 cc. and the weight of 1 litre of chloride solution as ascertained in its prepara- tion. But the silver solution itself is intended to be measured by weight in the analyses. To calculate the weight-litre we reduce the 0'32 cc. of excess of strong silver solution found in the titra- tion to weight by multiplying by T024 (which is a good enough approximation to the specific gravity ; in the present case this produces practically no change), deduct the result from the weight of silver solution used, and put down the rest (call it S grins.) as representing the equivalent to the C grins, of real chloride of potassium used as 50 cc. of solution. We then have C grms. of chloride of potassium = S grins, of silver solution. .'.7-459 =Sx 7-459 C ~ W ' CHLORINE IN SEA WATER. 223 Note down that W grms. of silver ( = to so-and-so many cc. as calculated before) precipitate 1/10 Cl = 3'545 grms. of chlorine. (W conies to about 1025.) For the analysis of a sea water, determine the chlorine first in 5 cc. by method I. (Mohr's) as an approximation, then twice in 10 cc. (weighed) by method II. in its gravimetric form. The burette serves only to measure out the proper quantum of silver solution as calculated from the preliminary analysis. Report in terms of 1000 parts by weight of sea water analysed. Appendix. If a large number of sea waters has to be analysed for chlorine, it is convenient to prepare the (strong) silver solu- tion on a large scale, not merely to save trouble, but chiefly to avoid the loss of precision necessarily involved in the successive use of different standard solutions in different analyses. The method of preparation we used in the course of the "Challenger" research was as follows: A supply of the purest nitrate of silver was reduced to powder, mixed, and standardized by keeping a known weight of about 10 grms. of salt in a tared crucible near its fusing point, then fusing, allowing to cool, and weighing the dry residue. From the result it was easy to calculate the quan- tity required for (let us say) 40 litres of solution. But such a large volume as 40 -litres is difficult to measured toto, and the volume of water (or rather very dilute nitric acid, 20 cc. of 1*4 acid per litre) required cannot be calculated. A near guess at the required volume, however, is easily made. Supposing the salt (weighing about 680 grms.) to be dissolved in 39 lit. of solvent, the solution produced will amount to 39 lit. + about 680-^4-35 = 156 cc., say 3916 lit., so 0'84 lit. of solvent should complete the 40 lit. To be on the safe side we add only (say) 0'4 lit., and after having mixed the whole by blowing (dust-free) air through it, proceed to analyse a measured small volume somehow. As "tenth- normal " sulphocyanate is always at hand, the simplest method is to apply Volhard's method (see Ex. 13) to, say, 50 cc. ; but, to obtain the highest precision, it is better to measure the sulphocyanate by weight, which is done by delivering it from a kind of ungraduated Mohr's burette, made out of a wide glass tube, so that the burette becomes short enough to be suspended from, and tared on, a fine balance. The solution is standardized 224 EXERCISES IN APPLIED ANALYSIS. expressly with, say, J grin, of directly weighed metallic silver, which gives its titre per 1 grm., and then applied to the silver solution in hand. Supposing we find that the silver solution contains 1*017 times the intended weight of silver per litre, our 40 litres (or what it may be now, say V lit.) have to be diluted to 1-017 X V by addition of 0'017 V of water, or better, say 0.015 V to keep on the safe side. The final correction is made on the basis of a series of titrations with 50 cc. (weighed) of standard chloride of potassium solution by our method II., as above described. LIME AND MAGNESIA. Memoir, p. 32. Weigh out 500 cc. of sea water, add 15 cc. of 20 per cent, hydrochloric acid, and boil for fifteen minutes to expel the carbonic acid. Then allow to cool, supersaturate the acid by addition of 100 cc. 10 per cent, of ammonia, then add 180 cc. of oxalate of ammonia (1 cc. = 11 '2 mgs. of CaO), and allow to stand cold over two nights. Then filter off the precipitate, wash it first with cold, then with hot, water, dry, and ignite. This precipitate, apart from any remnant of carbonic acid, contains about 9 per cent, of impurities, chiefly soda, also magnesia. To purify the crude lime, slake it in a beaker with water, add 5 cc. of hydrochloric acid, boil away the carbonic acid, allow to cool, add ammonia drop by drop until alkaline, and then boil off the excess of ammonia to bring down the sesquioxides, which collect, wash hot, re-dissolve in 2 cc. of hydrochloric acid, and re-precipitate by ammonia, boiling off' the excess of the latter as before. Filter off the precipitate (it amounts to 12 ings.), and to the united filtrates add 20 cc. of ammonia and 40 cc. of oxalate of ammonia, to re-precipitate the lime in the cold. Allow to stand over night, then heat the mixture on a water-bath, filter off' the oxalate, and ignite it in a small platinum crucible as usual, using the blow-pipe for finishing, until the weight is rigorously constant. To determine the "magnesia, mix all the lime filtrates and wash-waters, and weigh them on a large precision balance. Take about one-tenth of the whole, weigh it, add 20 cc. of 10 per cent, ammonia and 12 cc. of phosphate of ammonia POTASH IN SEA WATER, 225 (1 cc. = 20 mgs. of MgO), and allow to stand for at least twelve hours. Filter off the phosphate of ammonia and magnesia, wash it with 3 per cent, ammonia, dry, ignite, and weigh. The most exact mode of manipulating the filter is, after removal of the bulk of the precipitate, to re-place it in the funnel and dissolve off the adhering precipitate by means of the least sufficient quantity of dilute acetic, strengthened by a little nitric acid, to evaporate filtrate and wash- water in a tared crucible to dry- ness, and ignite the residue. The bulk of the precipitate is then added, ignited till constant, and weighed. THE POTASH. Memoir, pp. 12 and 234. Since the completion of the " Challenger " analyses, the Author, conjointly with Mr. John M' Arthur, has made an extensive critical research in the Finkener method of potash determination, and on the basis of this work we now recommend the following form of the Finkener process for the analysis of sea water (or similar) salts. Take 100 cc. of sea water, add a volume of standardized sulphuric acid equivalent by calculation to the chlorine, as determined, for instance, by Mohr's method (supposing 10 cc. of sea water had required 60 cc. of silver solution containing 60 x Ag/10 mgs., 100 cc. obviously demand 60 x 40 mgs. of sulphuric acid calculated as S0 3 ), and double the calculated volume of chloroplatinic acid solution (i.e., for normal sea water 200 mgs. of platinum Pt). Evaporate on a water-bath towards the end with constant stirring to a very small volume, and stir diligently while this cools down to prevent formation of large crystals. Then add 30 cc. of absolute alcohol, and, after half an hour's standing, 15 cc. of absolute ether.* Mix well, and allow to stand, best on a glass plate under a small bell-jar with ground edges, for two hours. (According to our experience, with potash salts generally, five minutes' standing with the alcohol and about half an hour's with it and the ether should suffice, but we have so far not had occasion to apply this modification to sea water at all extensively). Filter off, and wash with ether-alcohol as * Make sure that both the ether and alcohol are free from ammonia. Q 226 EXERCISES IN APPLIED ANALYSIS. usual (see Ex. 15, p. 27). The precipitate always includes some foreign chloroplatinates. To remove these, give the salts a final wash with plain ether, and allow the remnant of this to dry off in the air ; then re-dissolve in hot water, again evaporate to a very small volume, and re-precipitate with ether-alcohol as before. The thus purified precipitate is washed, freed from at least the ether by drying at a gentle heat, transferred with 150 cc. of hot water to an Erlenmeyer flask of 400-500 cc. capacity, and the platinum precipitated by hydrogen, as ex- plained in Ex. 15. According to our experiments, the weight of platinum obtained if multiplied by 0751 gives, in cases like the present, a close approximation to the weight of potassium sought, calculated as KC1 ; hence the factor 0*4744 gives the potash K 2 0. (The factor K 2 O : Pt calculated from the present atomic weights (Pt = 194'8 ; O = 16) is 0-48386). TOTAL (BASES AS) SULPHATES. Memoir, p. 17. Weigh out 20 cc. of sea water, add a quantity of standard 20 per cent, sulphuric acid barely equivalent to the chlorine (see " Potash "), and evaporate to dryness in a flat platinum crucible of about 25 cc. capacity, provided with a perforated lid and a shield under the perforation, as represented by Fig. 57. The evaporation is conducted on an air bath, with the lid off, until the residue is nearly dry. In order now to expel the remnant of volatiles, put on the perforated lid, and a large piece of platinum foil over the latter, and let a large Bunsen flame play down on the foil so that the mass is heated chiefly from above. Towards the end the heat from below may be increased. When visible vapours have ceased to come off, heat the still covered crucible over a naked flame to redness for a time, allow to cool, and weigh. Then moisten with a little of the acid and water so that a small excess of sulphuric acid is sure to be present, evaporate, ignite again, and weigh this residue. If vapours of SO 3 were seen to go off, the chlorine is sure to be all away, and all that is necessary is to finish by repeated ignition until the weight is constant, CARBONIC ACID IN SEA WATER. 227 i Take care not to use an excessive heat ; on the other hand, make sure that the aqueous solution of the residue is neutral to litmus. From the "total sulphates," subtract the lime, magnesia, and potash, all calculated as sulphates ; the rest gives the sulphate of soda, and thus the soda independently of the determination of the chlorine, sulphuric acid, and carbonic acid. After having thus determined the principal components, check your results by expressing the bases and the acids found per 1000 parts as multiples of C1 2 , SO 3 , Na, 2 O, K 2 0, CaO, MgO respectively, and balance the number of base-equivalents against that of the acid-equivalents. There should be a small excess of base, some O'OOl x R 2 O, about per 1000 parts of sea water. THE CARBONIC ACID. Memoir, p. 107. Method. From 1 lit. of sea water the carbonic acid is eliminated by boiling with excess of sulphuric acid in a flask under an inverted condenser, and collected in a vacuous vessel previously charged with a measured (excessive) volume of standard baryta-water. The excess of unprecipitated baryta left is determined by titration with standard hydrochloric acid, and from the results the carbonic acid is calculated. The apparatus is represented in Fig. 58. C is a Pisani gas- holder containing air free of carbonic acid (only one of the two bottles is shown in the figure). To prepare it, almost empty one bottle into the other, clip the connecting india-rubber tube, drop a stick of caustic potash into the small remnant of water left, put on your hand, or the india-rubber cork with glass rods instead of the tubes in the perforations, and shake vigorously for two or three minutes. All the carbonic acid is absorbed. The rest follows from the inscriptions on the figure. Reagents 1. Decinormal Hydrochloric Acid. Distil pure acid of 1*1 spec, grav., collecting only the middle portion, which is free from carbonic and sulphurous acids, &c., on the one hand, and ammonia on the other. Dilute it to a strength convenient for its acidimetric analysis (with standard alkali), and from the thus standardized acid prepare one which contains, say, I'Ol times 1/10 HC1 grms. per litre. Determine its exact strength 228 EXERCISES IN APPLIED ANALYSIS. by silver, exactly as shown under " Chlorine," method II. ; make it as exactly as possible volumetrically decinormal, and record both the volume-litre and the weight-litre. 2. Decinormal Baryta Water. 15 75 grms. of pure Ba(OH) 2 -h 8H 2 O should give 1 litre of decinormal solution. To allow for the carbonate, dissolve n X 17 grms. in hot (distilled) water, filter, and dilute with boiled-out water to about n litres in a large bolt- FIG. 58. Between d and B, mercury trap ; /, decomposition flask ; h, inverted condenser ; b, india-rubber tube ; r, india-rubber cork ; ra, vacuum flask ; q, baryta-water. head. To determine the strength of the solution, measure off from a burette, say, 30 cc. of the acid, add a few drops of turmeric tincture, and then run in (from another burette protected by a soda-lime tube) the baryta solution until the liquid becomes brown ; then again, drop by drop of the acid until the liquid just becomes clear light yellow. After the result of such tests, dilute CARBONIC ACID IN SEA WATER. 229 FIG. 59. the baryta so that it contains about I'Ol times as much baryta per litre as it is intended to. Then transfer the solution to Winchesters ; the one to be used more directly must be provided with a syphon with a Mohr's clip at the end, and with a protecting soda-lime tube. Allow the solu- tion to clear up completely in this Winchester, and then determine its titre finally by weighing out, say, 100 cc. of the acid, and thus indirectly the exact weight of real acid, HC1, present in the 100 cc. according to the weight-titre, and titrating as before. But the bulk of the baryta had better be measured in a pipette. The best kind of pipette for this purpose is one constructed on Stas' principle (see Fig. 59). The pipette delivers a known volume (50 to 60 cc.) when completely tilled from below to overflowing, and then allowed to run out spontaneously, the last drop being removed as far as possible by short contact of the lower end with the wet side of the receiving vessel. The last instalment of acid and baryta are added from burettes graduated so that they give the tenths directly, and that 0'02 cc. is very plainly visible. The Analysis. To obtain a really exact result we must operate upon a whole litre of sea water, which may be presumed, in general, to contain less than 110 mgs. = 5 x JCO 2 mgs. (per litre), corresponding to less than 50 cc. of the baryta water. But we must start with a decided excess, so that the Stas' pipette had better be adjusted to about 60 cc. To prepare the vacuum flask, wash it first with some ordinary baryta water, and then with water. Then exhaust it at the air-pump,* and fill it with (puri- fied) air from out of the Pisani. It is now ready to receive its baryta water, and be evacuated for the analysis. The sea water is measured into the flask with the least amount of shaking, then mixed with 5 cc. of 20 per cent, sulphuric acid (which is ample), and immediately connected with the condenser, and through the latter with the vacuum flask, the stopcock of which * When you exhaust a flask for the first time, it is as well to place it within a box, or wrap it up in a towel, and see that it does not collapse. Bolt-heads sometimes are very thin in the body. Bottles as a rule stand better. 230 EXERCISES IN APPLIED ANALYSIS. is turned on cautiously, so as to establish a slow current of air. The contents are now brought up to the boil, and kept boiling for about twenty minutes, when all the carbonic acid can be assumed to be eliminated. The liquid then is allowed to cool ; the current of air is made a little faster, and continued until it stops by itself. During the whole of the process the pressure in the Pisani should be a little short of that which produces a current of air when the pressure in the flask is at one atmo- sphere. When the vacuum is undone, close the vacuum flask, detach it from the apparatus, and by making it rotate for a time, collect all the stray carbonic acid in the baryta. Then open one of the two perforations, run in some turmeric, insert the (long) outlet tube of the hydrochloric acid burette, and let acid run in until the alkaline reaction is almost gone. If neces- sary, re-establish a decidedly alkaline reaction by adding baryta from the burette. Now, close the flask again, and again collect what there may be of liberated carbonic acid in the flask. At last empty the contents into a tared Erlenmeyer or phial, which should not be larger than necessary ; put on a good cork, and allow the mixture to stand until the precipitate has settled. Weigh the mixture to 01 grm., pour off as much as possible of the clear liquor, determine its weight, and finish your titration by the zig-zag mode. The calculation requires no explanation. Do not rely on the constancy of the titre of the baryta; it diminishes gradually, all protection tubes notwithstanding. Note. Before operating on the sea water, rehearse the method with 1 litre of boiled-out distilled water, plus a suitable weight of pure carbonate of soda. ALKALINITY. Memoir, p. 106. All sea water is alkaline, in this sense that the sulphuric and hydrochloric acids do not quite suffice to neutralize the bases. The surplus base is present as a mixture of carbonate R 2 C0 3 and bicarbonate RHC0 3 , and therefore, for calculating purposes, can be represented as consisting of xR. 2 C0 3 + yCO 2 ; i.e., so much normal carbonate plus so much " loose carbonic acid." " Alka- linity," in a quantitative sense, is customarily defined as meaning ALKALINITY OF SEA WATER. 231 the weight of carbonic acid present in the normal-carbonate part of the carbonate. It is referred customarily either to 1 litre of sea water or to 100 parts of total salts or to 100 parts of chlorine. For its determination Tornoe has given the following method: Requirements. (1.) A dilute hydrochloric acid so standardized that 1 cc. contains 1/22 HC1 mgs., and consequently corresponds to 1 mg. of carbonic acid (C0 2 ). 22 cc. of such acid contain 1 mg. equivalent of hydrochloric acid, or as much as 10 cc. of the decinormal acid. (See " Carbonic Acid.") Hence, ten volumes of the latter diluted with water to 22 volumes or 454'5 cc. (better the weight of decinormal acid containing by silver titration 45*45 x HC1 mgs.), diluted to 1000 cc., gives standard acid of the required strength. (2.) A Solution of Pure Carbonic Acid-free Potash, exactly or nearly equivalent to the acid, volume by volume, prepared from recently causticised pure alkali. To determine the strength of the alkali take, say, 40 cc. of the acid, add a few drops of alco- holic aurine solution, heat in a Berlin basin, and add alkali until the violet colour appears and remains. Use the to-and-fro mode of titration, and from the mean of the last 3-4 results (w r hich should agree very closely) calculate that 1 cc. of alkali = k cc. of the acid, i.e., corresponds to k mgs. of CO. 2 . To analyse a sea water, measure off 250 cc. in a Berlin (or silver or platinum) basin ; glass is not safe. Add an excess of the acid and aurine, boil for fifteen minutes to expel all the carbonic acid, and titrate back with alkali, using the to-and-fro method. Note. Ordinary good lime water is stronger than the above standard alkali, and could probably be substituted for it without loss of precision. The greater basilousness of the potash is no advantage, because the end-reaction in any case is brought about by the magnesia of the sea water. If there is enough of sea water, take 500 cc. of it for the analysis, and use decinorrtial hydrochloric acid and (standardized) lime water as reagents. TOTAL SALTS. Tornoe s Method. Memoir, p. 40. Evaporate 30-40 cc. of sea water to dryness in a large (tared) porcelain crucible, provided with a good lid, over a water-bath. 232 EXERCISES IN APPLIED ANALYSIS. Then put on the lid, and heat over a Bunsen for about five minutes, allow to cool in an exsiccator, and weigh. The weight, however, requires a correction, because, in the last stages of the evaporation, and especially in the ignition, part of the chloride of magnesium (by the action of water) becomes magnesia. To determine this magnesia, dissolve the residue in water (in a Berlin basin) along with a sufficiency of decinormal hydrochloric acid, add aurine, heat to boiling, and titrate back with stan- dardized lime water or alkali. Supposing the magnesia was found equivalent to 31 cc. of decinormal acid, deduct the small volume corresponding to the original alkalinity, which will bring down the number to, let us say, 30 cc.; then the magnesia produced in the dehydration amounts to m = 30 x 1/20 x MgO mgs., and the original chloride to c = 30 x 1/20 x MgCl 2 mgs. ; hence we have to add r5x(MgCl 2 -MgO) = l'5x(Cl 2 -0) mgs., or To x 54'9 = 82'4 mgs. to obtain the total salts, exclusive of all carbonic acid. The result affords a valuable check for the sum total of S0 3 , (C1 2 O), and bases as determined separately. Note. It still remains to be ascertained how much of the hydrochloric acid which goes off in the dehydration is hydro- bromic acid. THE BROMINE. Memoir, pp. 98 and 239. Weigh out 1 kilogrm. of sea water, and add four hundredths of the quantity of decinormal (acid) silver solution, which, according to the chlorine determination, would bring down all the halogen.* Shake vigorously in a Florence flask or clear- coloured bottle, allow to settle in the dark, syphon off the liquor as clear as possible, and to it add the same volume of silver solu- tion as you used for the first precipitation. The two precipitates are wrought separately, but in the same manner. Wash by decantation with water, and collect the washings in a wide- necked bolthead. They are allowed to stand until the small turbidity has settled completely. The supernatant liquor is then * There is, of course, no need of wasting precisely standardized silver for this. 17 grms. of fused nitrate + 20 cc. of 1'4 nitric acid dissolved to 1 litre will do. BROMINE IN SEA WATER. 233 syphoned off clear, the precipitate from the washings collected on a filter, and treated like an ordinary chloride of silver pre- cipitate ; to be weighed as AgCl. The principal precipitate is transferred to a Berlin basin, and dried over a water-bath in the dark. The dried precipitate is then transferred to a large porcelain boat which has been pre- viously tared within, and along with a piece of combustion tubing about two inches longer than the boat itself. Within the tube the dehydration of the haloid is effected by embedding it in magnesia in the gutter of a short combustion furnace, and cautiously heating it to fusion in a current of dry air. After cooling, the outside of the tube is cleaned of adhering magnesia and weighed. The tube is then re-placed in the magnesia bed over the furnace, and the haloid fused and kept at fusion heat in a current of dry chlorine until the bromine appears to be away. The gas is then turned off, the chlorine displaced by dry air, and the (cold) tube weighed again. The chlorination is repeated until the weight of the resulting chloride is constant. There must be no cork at the combustion tube ; the chlorine (and air) are led in by a bit of drawn out combustion tubing, the wide end of which almost fits into the combustion tube like a piston. As there is no surplus pressure to overcome, the joint needs not be air tight. From the loss which the haloid suffered in the chlori- nation, the bromine is calculated. As AgBr yields AgCl, every (Br Cl) parts of loss correspond to Br parts of bromine. (Br - Cl) X 179675 = Br and (Br - Cl) X 4'2225 = AgBr. As in a given haloid precipitate the loss involved in the chlori- nation (I) is proportional to the ultimately obtained chloride of silver (c), it is easy to see how we allow for the small weight of stray haloid filtered off and weighed as chloride. Obviously (its weight) X is the correction to be added on to the loss suffered by the bulk of the precipitate. After having determined the bromine, we are in a position to calculate the true chlorine from the chlorine determined, as shown in the first section. As we determined the chlorine by. titration, the number of gramme equivalents of chlorine found, 234 EXERCISES IN APPLIED ANALYSIS. minus the number of gnu. equivalents of bromine found, per kilo., gives the true chlorine. AVERAGE COMPOSITION OF OCEAN-WATER SALTS. Memoir, pp. 137 and 138 ; also pp. 203 and 204. Per 100 parts of Total Salts. Per 100 of Halogen calculated as Chlorine. Dittmar. Dittmar. Forchhammer. Chlorine, 55-292 j 0.1884 ( 6-410 0-152 1-676 6-209 1-332 41-234 (-12-493) 99-848 0-3402 11-576 0-2742 3-026 11-212 2-405 74-462 Not determined. Not determined. 11-88 Not determined. 2-93 11-03 1-93 Not determined. Bromine Sulphuric Acid, SO 3 , . . Carbonic Acid, CO.,, . . Lime, CaO, Magnesia, MgO, . Potash. K 2 0, . . . . Soda, Na.,0, (Basic Oxygen equivalent to the Halogens), TOTAL SALTS, . . . 100-000 180-584 181-1 * Equal conjointly to 55'376 parts of chlorine, which accordingly is the per- centage of " halogen reckoned as chlorine " in the reed total solids. Combining acids and bases, we have (Dittmar) Chloride of Sodium, 77758 Chloride of Magnesium, 10'878 Sulphate of Magnesium, ... ... 4737 Sulphate of Lime, 3*600 Sulphate of Potash, 2'465 Bromide of Magnesium, ... ... 0"217 Carbonate of Lime, ... ... ... 0*345 Total Salts, 100-000 STASSFUR1H POTASH SALTS. 235 In the open ocean the quantities of bromine, sulphuric acid, magnesia, potash, and soda, per 100 of chlorine, are almost con- stant ; that of the lime is subject to slight, that of the carbonic acid to very tangible, variation. The loose carbonic acid (not reported in this table) is eminently variable. The weight of salts per kilo, varies from about 33"7 to 37'6 grins. The following Table gives the specific gravity of sea water at 15'56C.= 60F. (water of 4C. = 1000) ; and the weight of " Chlorine," in grammes per litre : in function of the weight of "Chlorine," in grammes per kilogramme : "Chlorine >! grnis. per "HiflP Specific Gravity. Diff. Jjm. Kilo. Litre. 17-0 17-386 1022-68 17-5 17-909 523 23-38 70 18-0 18-433 524 24-07 69 18-5 18-958 525 2476 69 19-0 19-484 526 25-46 70 19-5 20-010 526 26-15 69 20-0 20-536 526 26-85 70 20-5 21-064 528 27-54 69 21-0 21-593 529 28-24 70 Abridged from tables in Memoir, p. 81. 2. Stassfupth Potash Salts. (Commercial chloride of potassium, kainite, carnallite, erates ammonia from the nitrogenous organic components. 254 EXERCISES IN APPLIED ANALYSIS. quite ready and greased beforehand, is put over the whole, and the apparatus left to itself for 48 hours. The ammonia is then all in the acid. The total Nitrogen, in the absence of nitrates, is best determined by Kjeldahl's method ; in their presence, by Dumas' method (see Exs. 30 and 47.) The Nitric Acid by Schloesing's method. (See Ex. 31). Uric and Oxalic Acids. Boil a known weight with carbonate of potash solution, filter, and wash. Both acids go into solution as potash salts. To precipitate the uric acid, acidify with hydro- chloric acid, and allow to stand for 24 hours ; collect the pre- cipitate on a weighed filter dried at 100C., wash with small instalments of cold water, dry at 100C., and weigh. (Method used in Urine analysis.) The oxalic acid is in the filtrate. Supersaturate with ammonia, add acetate of calcium and allow to stand until the precipitate has settled completely. Decant through a filter and digest the precipitate with dilute acetic acid. The precipitate now is impure oxalate of calcium. Perhaps weighing it as such would afford a sufficient approximation ; but it is probably better to place it in a flask along with some pure binoxide of manganese, conveniently made from sulphate by precipitation with permanganate, to decompose by added sul- phuric acid, and weigh the carbonic acid evolved (directly). (Not tried yet.) Some chemists recommend to apply this method direct to the guano. In this case we must, of course, allow for the carbonic acid of the carbonates. For the separate determination of the soluble parts treat, say, 10 grms. of substance with about 200 cc. of cold water, collect what remains on a weighed filter dried at 100C., and wash with small instalments of cold water. The residue is dried at 100C. and weighed. With regard to the analysis of the two parts we have nothing to add to what the student will readily find out by himself, except perhaps the remark that all the alkalies at least, practically pass into the solution. Note regarding the incinerations. Mixtures of phosphates and charcoal are apt to attack platinum vessels. Hydrate of baryta is sure to attack them very perceptibly. Porcelain (in the case of baryta or alkali being used) may yield up silica. MILK. 255 Vessels of Dittmar's silver alloy (see foot-note, p. Ill) will probably work best. 6. Milk. The methods to be given are invented for fresh milk, which is still perfectly sweet. Before measuring off a sample for a determi- nation, agitate the milk to make sure that the fat is uniformly diffused. It is advisable in hot weather it is necessary to start all the determinations at the same time. Total Solids. A requirement for this and the following deter- mination is purified sand. A supply of fine white sea-sand is digested with strong hydrochloric acid, then washed with abund- ance of water to remove the dissolved parts and the light dust, and at last ignited in a platinum basin. A platinum crucible of 30 cc. is charged with such sand up to about one- third of its height, the sand moistened with water, and then dried again at 100C. until constant in weight. 2 cc. of the milk are then dropped on the sand, so as to wet it as uniformly as possible, and the whole is re-weighed (lid on) to determine the weight of the added milk. The crucible is then kept (open) in a drying chamber at 100C. until the weight is constant. This gives the percentages of total solids and water. The Fat. 10 cc. of the milk (weighed) are poured on 20 grms. of sand contained in a flat Berlin basin, and over it evaporated to dryness with continual agitation with a glass or porcelain spatula to prevent formation of lumps. The residue is dried fully by keeping it for a time in a drying chamber at 100C., and then transferred to a dry flask of about 200 cc.'s capacity. What sticks to the basin is removed by softening it up with a little water, and evaporating the liquid to dryness over an additional small quantity of sand. In order now to extract the fat, the united sandy residues are boiled with 50-60 cc. of anhydrous ether under an inverted Liebig's condenser, the con- tents allowed to cool and filtered into a tared, short-necked flask of about 100 cc.'s capacity, and from out of it subjected to distillation to recover the ether, which is used for a repetition of the extraction. The ether recovered this time (with more ether, 256 EXERCISES IN APPLIED ANALYSIS. if necessary) is used for washing the sand (cold) until the last runnings are free from fat, which is ascertained by allowing a few drops to evaporate on a watch-glass, and rubbing the residue against the watch-glass with the little finger. The least trace of fat then becomes visible as a coating on the glass. The fat solution is deprived of as much of its ether as possible by dis- tillation, and the bulk of the rest is next removed by heating the flask on a water-bath (in the open laboratory) and blowing air through it from time to time.* At last the flask is kept at 100 in a chamber (its vapour being sucked out from time to time) until its weight is constant. Should the (molten) fat not be quite clear, it must be dissolved in absolute ether, filtered into another tared dry flask, and in it brought to dryness. The above method may be modified as follows to save time without much loss of exactitude : Place the dry sand-residue in a phial of, say, 150 cc.'s capacity, tare the whole on a balance, add 100 cc. of ether, stop up well, and allow to stand for some hours, with occasional agitation. Then allow to settle completely. Weigh the whole, decant part of the ethereal solution into a tared fat flask, and, by weighing what remains, ascertain the weight of ethereal liquid poured out. Determine the fat con- tained in the fraction, which incidently gives the weight of the solvent, which is associated with the fat. Supposing the fat amounts to n mgs., the ether associated therewith to p grins., the whole of the ether present in the original sand mixture just before fractionating to P grms., then the total fat is _ n X P P The total ash might be determined in the residue from the total solids, but it is better to evaporate a fresh quantity of milk (say 5 cc.) to dryness in a porcelain crucible without sand, and incinerate at the lowest sufficient temperature. ALBUMENOIDS AND SUGAR. These are best determined by Ritthausen's method. Requirements. (1) A standard solution of sulphate of copper, * If the flask were placed in the drying chamber as it comes off the condenser, an explosive atmosphere might be formed within the chamber, MILK. 257 containing 63*5 grms. of CuSO 4 5H 2 O per litre : (2) A solution of caustic potash equivalent to (1), volume for volume, in the sense of the equation, CuSO 4 + 2KHO = Cu(OH) 2 + K 2 S0 4 . There is no need of high precision in either standardization. 10 cc. of the milk are weighed out, diluted to 200 cc. with water, and mixed first with 5 cc. of the copper solution, and then with a volume of the caustic potash so adjusted that a small trace of the copper remains dissolved in a slightly acid liquid; 0'7 X 5 cc. will, as a rule, be sufficient. The whole of the albumenoids and the fat are now in the precipitate in combination (or association) with cupric hydrate. The precipitate is collected on a weighed filter, and washed successively with (1) water, until the sugar is out ; (2) alcohol of 85 per cent, by weight ; (3) absolute alcohol ; (4) ether, until all the fat is removed; (5) again absolute alcohol, to remove the ether.* [The filtrate and aqueous part of the washings are mixed and preserved for the determination of the sugar]. The precipitate is dried, first over oil of vitriol, in the cold, then at 125C. until its weight is constant. It is then ignited (the filter being incinerated separately), and lastly heated in contact with air until constant in weight. This weight, when deducted from that of the dried albumenate of copper, gives the weight of the albumen,oid8 probably with as high a degree of precision as can be attained by any direct gravimetric method for their determination. A higher degree of constancy in the results is obtained by determining the total nitrogen by Kjeldahl's method in, say, 10 cc. of milk ; 1 part of nitrogen corresponds to 6'452 parts of milk albumenoids (Ritt- hausen). Yet Ritthausen's process is invaluable as furnishing a very convenient material for the determination of the milk sugar. THE SUGAR. The aqueous filtrate from the albumenate of copper is diluted to a known volume, and next small aliquot parts are devoted to the approximate determination of the volume of Fehling's solu- tion (see Note 2), which a given volume of the liquor is capable * Ritthausen utilizes the alcoholic and ethereal washings for a determination of the fat. We prefer the sand method. S 258 EXERCISES IN APPLIED ANALYSIS. of reducing (with formation of a red precipitate of cuprous oxide). A quantity corresponding to 1-2 grms. of milk is now mixed with about 1^ times the volume of Fehling's solution which it is capable of reducing, the mixture heated to boiling, and kept hot on a water-bath until the cuprous oxide has settled. The pre- cipitate is then collected on a filter, washed with hot water, dissolved in an excess of an acid solution of iron alum, the solution cooled down promptly in the absence of air (best in carbonic acid), and titrated with standard permanganate diluted so as to correspond to about 5 '6 mgs. of iron per cc. From the results the weight of the milk-sugar is calculated. 2Fe = 1x0 = Cu 2 exactly, and one part of cuprous oxide, by experience, corresponds to about 0'6756 parts of milk-sugar (C 12 H 2 20 n H 2 0) ; .*. 1 mg. of iron = 0'860 mg. of milk-sugar. To obtain an exact result, repeat the analysis, and at the same time carry out a parallel experiment with a known weight of pure milk-sugar equal to the quantity to be determined according to the pre- liminary analysis, taking care to let both processes and titrations go on under precisely the same conditions. The experiment with pure sugar affords the factor for the reduction of the number of cc. of permanganate used to mgs. of sugar under operation. Note 1. Commercial milk is very frequently adulterated with added water, or by removal of part of the fat (as cream). As the composition of even genuine milk is subject to great variation, such adulteration cannot, in general, be proved with certainty. But it is easy on the other hand to ascertain what proportions of water and fat would have to be added, or taken away from, a standard milk of assumed fixed composition to produce a given kind of milk. All that is necessary is to determine the fat and the solids not fat, and to refer the results on both sides to one part of solids not fat; i.e., of that component which, in the presumed manipulation, would remain constant in an absolute sense. To explain this by an example, let us adopt what Mr. Wanklyn puts down as the average composition of genuine milk as our standard, and then contrast the numbers with the results in a given analysis of suspected milk : SOAP. 259 Waiiklyn's Suspected Milk Standard. (say) Fat, 3-2 318 Solids not fat, 9'3 8'43 Water,., 87'5 88-39 100-0 100-00 Reducing to 1 part of solids not fat, we have Solids not fat. Fat. Water. Milk. Standard, ............. 1 ... 0'344 ... 9'409 ... 10753 Suspected, ............ 1 ... 0'377 ... 10'483 ... 11-860 Additions to 10*753 . 1 of Standard reqd. + 1 ' Note 2. Fehling's solution is best made ex temp, from the following two solutions : (1) 34'65 grms. of crystallized sulphate of copper are dissolved in water, and made up to 1 litre. 1 cc. = 5 mgs. of glucose (not milk-sugar, or even inverted milk-sugar) ; (2) 173 grms. of Rochelle salts and 70 grms. of caustic soda (sticks) are dissolved to 1 litre. Equal volumes of the two mixed together constitute " Fehling's solution " (of half the customary strength). 7. Soap. SUBSTANCE. In the analysis of any impure substance the first point to be attended to is to prepare a good average sample ; in the case of a soap we have to contend with the additional diffi- culty that in a given sample it is difficult to keep the percentage of water constant for any time. The best course is (from a homogeneous sample) to at once weigh out all the portions of substance required for the several determinations on so many watch-glasses, and note down their weights. WATER. Dry, say, 3-5 grms. at 105-110C. until constant. IMPURITIES. While the drying process progresses, treat a weighed sample with boiling strong alcohol in a flask. A pure soap (meaning 260 EXERCISES IN APPLIED ANALYSIS. a soap containing only alkali salts of fatty acids and perhaps glycerine) dissolves completely ; any admixture of alkali car- bonate, sulphate, silicate, or gelatine (which occurs occasionally) remains behind. Collect the insolubles, if any, on a weighed filter, wash with alcohol, dry at 100C. till constant, and weigh. If the insolubles amount to anything considerable, devote one such precipitate to a qualitative analysis. We now proceed on the assumption that the soap was practically soluble without residue in hot alcohol. Where this assumption does not hold, the first step in each of the following determinations is to dissolve the respective sample in hot alcohol, filter, and evaporate away the alcohol if necessary. The solution obtained is treated as prescribed. FATTY ACID. Dissolve, say, 5 grms. of the soap in a tared Erlenmeyer fiask in 100 cc. of water, and add a sufficient volume of standard sul- phuric acid out of a burette, to produce a decidedly acid solution. Then heat on a water-bath until the fatty acid has gone to the top as a clear oil, leaving a clear aqueous liquid below. Then allow to cool, when the fatty acid in many cases will become solid or semi-solid, and filter the aqueous part through a ivet filter. Re-melt the fatty acid in hot water, with frequent agita- tion, to give the soluble parts a chance of going out, again allow to cool, and filter as before. Continue this with smaller and smaller instalments of water, until the last water is neutral to litmus and gives no precipitate with barium chloride. Take care that the bulk of the fatty acid remains in the Erlenmeyer, and only very little goes on the filter. This small quantity is recovered by drying the filter at 100C. (a test tube serves to support the funnel, so that any fatty acid that runs out collects in it), dissolved in hot alcohol, and added to the bulk of the fatty acid in the flask. The fatty acid is then dried at 100C. until constant, and weighed. The fatty acid is customarily calculated as hydrate, i.e., it is stated in the report that 100 parts of soap gave so-and-so many parts of fatty acid (C n H 2n 2 , or C n H 2n _ 2 O 2 ) as weighed ; the alkali combined with the acid into soap as Na 2 O or K 2 0. But 2C n H 2n 2 combine with Na 2 into SOAP. 261 2(C n H 2n _ 1 2 )Na-hH 2 O ; hence the sum of the several per- centages must be corrected by subtracting H 2 = 18 grms. for every Na>O or K 2 O grms. of alkali, present as neutral soap in 100 grms. of substance* (vide infra). The calculation cannot be based upon the weight of the fatty acid, because its molecular weight is unknown. The Total Alkali is determined in the aqueous filtrate and washings obtained in the determination of the fatty acid by titrating in the heat with standard alkali. The result, when corrected for the surplus alkali, gives the datum required for the combined water (JH 2 O per molecule) in the fatty acid. The Surplus Alkali used to be determined by dissolving the soap in water, then salting it out, and determining the alkali in the salt liquor produced. But the process has been proved by Wright and Thompson f to be erroneous, because even a neutral soap, when thus treated, yields free alkali by hydrolysis. Wright and Thompson recommend the following process : A known weight of soap is dried, and then boiled with a sufficiency of strong or, better, absolute alcohol. The filtrate is titrated with alcoholic acid or alkali, J using phenol-phthalein as an indicator. In the residue on the filter the alkali present as carbonate is titrated as usual. The two alkalinities added together (alge- braically in equivalents) give the alkalinity of the soap. THE ALKALI SALTS GENERALLY. The chlorine (present as chloride of potassium or sodium) can be determined in an aliquot part of the sulphuric liquor obtained in the preparation of the fatty acid. For the sulphuric acid, potash, and soda, decompose a special portion of the soap with hydrochloric acid, and use aliquot parts of the aqueous solution for (1) sulphuric acid, and (2) potash and soda. The alkalies are recovered in the first instance as (in general impure) chlorides ; these are converted into and weighed as sulphates, and in them the potash is determined by Finkener's * This comes to the same as saying that in adding up percentages the fatty acid must be calculated as anhydride. f "Soc. of Chem. Ind. Journ.," 1885, p. 634. J As the solution may contain free fatty acid in lieu of alkali. 262 EXERCISES IN APPLIED ANALYSIS. method. Observe that most " soda " soaps contain more or less of potash. A shorter but less exact method is this : Char a known weight of the soap in a platinum crucible, and extract the char with water. Then incinerate the charcoal at the lowest sufficient temperature, treat the ash with water, add the solution to the first, and analyse the mixture ; an insoluble residue usually con- sists of lime. The incineration method is always adopted when we have to deal with a soap contaminated with silicate. THE GLYCERINE. To determine it dissolve the soap in water, decompose with a slight excess of sulphuric acid, remove the fatty acid, which wash with the least sufficient quantity of water. Neutralize the aqueous liquor with carbonate of potash, and evaporate on a merely simmering water-bath to a syrup. Exhaust this syrup with absolute alcohol to obtain (a residue of sulphates and) a solution of impure glycerine. Evaporate away the alcohol, dis- solve the residual glycerine in a mixture of two volumes of absolute alcohol and one volume of ether, filter, .evaporate to dry ness, dry at 100C., without insisting upon constancy of weight, as glycerine is volatile, and weigh. A more exact method probably is to oxidize the glycerine in the solution with alkaline permanganate, and determine the oxalic acid produced.* In this case it is better to eliminate the fatty acid by hydrochloric acid. The aqueous filtrate from the fatty acid is neutralized with caustic potash and mixed with caustic potash, 5 grms. of solid alkali per O25 grm. of glycerine present. Powdered per- manganate of potash is now gradually added until the solution is permanently pink at a boiling heat, which is maintained for half an hour. The excess of reagent is then decomposed by cautious addition of sulphurous acid or sulphite of soda, and the precipi- tate of peroxide of manganese filtered off. The solution is acidified by acetic acid, and the oxalic acid precipitated by acetate of calcium in the heat. The precipitated oxalate is * Fox and Wanklyn, " Chem. News," 1886, I., p. 15. SOAP. 263 collected on a filter and washed with hot water. It might be dried at 100C., and weighed (as C 2 4 Ca + H 2 Fresenius), or ignited and weighed as CaO. Fox and Wanklyn, however, prefer to titrate the oxalic acid with permanganate. For this purpose dissolve in the minimum of pure (nitrous acid free) dilute nitric acid, add water and dilute sulphuric acid, and then standard permanganate from the burette, until the liquid is permanently pink. 2Fe = 1 x O = C 2 O 4 H 2 . According to Fox and Wanklyn, C 3 H A + O fi = C 2 H 2 4 + CO,"+ 3H 2 O. RESIN. Many cheap soaps are made from a mixture of resin and fat, instead of pure fat. Resin is substantially a mixture of acid anhydrides of high molecular weights ; ordinary colophony con- sists chiefly of the anhydride of abietinic acid, C 44 H 62 O 5 . H 2 . In the process of saponitication, the resin anhydrides are converted into their alkali salts, which latter, on decomposition with acids, yield a precipitate consisting of the (hydrated) acids. Hence, were our process of analysis applied to a resinous soap, the resin acid would go with the fatty acid, and be weighed as such. Even the mere detection of resin in a soap is not easy. Barfoed,* who investigated the matter, has given the following, amongst other, methods : (1.) Detection of Resin in a fatty acid, containing much Stearic and Palmitic, beside little or no Oleic. The acid-mix- ture is dissolved in warm alcohol of 70 per cent, (by volume), and the solution allowed to stand, cold, for 24 hours. The fatty acids separate out almost completely as a crystalline precipitate ; the resin acids remain dissolved. The resin is isolated by evaporation, or by addition of water and some hydrochloric acid, and shaking, with judicious application of heat. By repeatedly boiling the resin precipitate with water, it becomes more compact and easier of identification. It is contaminated in general with some oleic acid, but not to the extent of veiling its characteristic properties. (2.) If Oleic Acid is present in quantity, the following method is used : The acid mixture is saponified at 100C. with * Fres. Zeitschrift, for 1875, p. 20. 264 EXERCISES IN APPLIED ANALYSIS. very dilute caustic soda (10 per cent, solution diluted with about seven times its volume of water), an immoderate excess of alkali being avoided.* The soap solution is evaporated to complete dryness, the dried soap powdered finely, and dried completely at 100C. It is then treated for a few hours with absolute alcohol (5-10 cc. per 1 grm. of soap) in a close bottle at about 80C.f The mixture is now allowed to cool (when some fat-soap separates out), and, after having made sure that no noteworthy quantity of the alcohol has been lost, 5 cc. of absolute ether are added for every 1 cc. of alcohol. This produces a voluminous precipitate of previously dissolved fat-soap, which, after 24-48 hours' standing, has settled completely so as to leave a clear liquid above itself. An aliquot part of the clear liquor is then measured off and eva- porated to dryness, the residual resin soap dissolved in water, and the resin acid precipitated by hydrochloric acid. After 24 hours' standing the precipitate is collected on a weighed filter, washed first with cold and then with warm water until all the hydrochloric acid is removed, dried at 100C. for 5-6 hours, and weighed. The fatty acid might be found by difference. Barfoed, however, prefers to operate upon two portions of the dry soap weighed out at the same time, and to use one for the resin as explained, and the other for the determination of the total acid (fatty and resinous), so as to obtain the ratio of the two quan- tities independently of the weight of the original substance. In calculating the results, 1 grm. of resin (abietinic, &c., anhy- dride) may be taken as corresponding to 1'03 grms. of resin acid as it is weighed on the filter (Barfoed). If the resin is given as part of a soda soap, the second method is liable to an obvious simplification ; we must make sure, however, of the absence of potash and glycerine. Thomas S. Gladding j has given a method founded upon the * Alcoholic soda, used conjointly with phenolphthalein as an indicator, would work better. f Alcohol boils at 78C. No doubt, an open flask attached to an inverted condenser and heated in a water-bath would work as well without danger. J "Chem. News" for 1882, I. (vol. xlv.), p. 159. Taken from "American Chem. Journal," vol. iii., No. 6, BUTTER. 265 presumption (confirmed by him experimentally) that resinate of silver is soluble, while the silver salts of the ordinary fatty acids are insoluble, in ether. We refer to the memoir cited. 8. Butter. TOTAL FAT AND NON-FAT. 10-20 GRMS. of substance are weighed out in an Erlenmeyer flask, and dehydrated by heating on a water^bath, and occasional addition of absolute alcohol. The water soon goes so far that none of it remains in a visible form. The rest is expelled in a drying chamber at 100C. This gives the percentage of water. To determine the Curd and Salt, the product is passed through a dried and weighed filter, by means of a hot- water funnel (the filtered pure fat being put aside for further tests), and the fat adhering to the filter removed by placing the funnel in an ordinary stand, trap- ping its outlet end with mercury contained in a narrow beaker (or bit of a test tube standing in a beaker), and then filling the filter to the edge with anhydrous ether. A concave glass plate is placed on the funnel to prevent loss by evaporation. After some time, the fat is completely dissolved ; the solution then is allowed to run off, and the residue washed completely with ether. The (uncovered) funnel is allowed to stand in a warm place in the open laboratory until the bulk of the ether is gone (see foot- note, p. 256), and the filter and contents are then dried at 100C. until their weight is constant. This gives the curd and salt. The latter is extracted and determined by evaporating to dry- ness, and weighing of the dehydrated residue, or indirectly, by titration with silver by Mohr's method. FOREIGN FAT. Of the many methods invented for the determination of foreign fats in butter, jReichert's is the easiest and best. The butter is dehydrated and filtered (see above), and of the fused fat, 3 cc. are measured off in a graduated pipette, and weighed, to be saponified with 266 EXERCISES IN APPLIED ANALYSIS. 1 grm. of caustic potash, and 20 cc. of ordinary alcohol on a water-bath. The alcohol is then completely removed by evapora- tion, the residue dissolved in 50 cc. of water, transferred to a distillation flask, and mixed (cold) with " 20 cc. of sulphuric acid, prepared by diluting 1 volume of oil of vitriol with 10 volumes of water." No doubt far less acid will do. As 56 mgs. of pure KHO neutralize 1 cc. of normal sulphuric acid, 1 grm. neutralizes 17'8 cc. of normal acid ; hence 25 cc. of normal acid, i.e., 25 x 49 mgs. of H 2 S0 4 , are ample. The flask is attached to a Liebig's condenser, a slow current of air passed through the liquid, and the latter distilled, the distillate being filtered into a graduated tube. When 10-20 cc. are over, the distillate is poured back, the distillation resumed, and kept on steadily until exactly 50 cc. of distillate are produced. The (filtered) distillate is titrated with decinormal caustic soda (1 cc. = 1/10 NaOH = 4 mgs.), litmus being used as an indicator. According to Reichert, the number of cc. of decinormal alkali neutralized is For genuine butter, 14*0 0*45 lard (pig's), 0'3 ox fat, 0-25 coco-nut fat, 370 Excluding the last named fat, we have for a mixture containing x grm. of butter and 1 x grm. of ordinary fat in 1 gramme: Alkali required by calculation = # x 14 -+ (1 a;) 0*3 = V, which symbol shall stand for the cc. of alkali used in the analysis of the V 0*3 mixture. Hence, x = = or, to be on the safe side, in pronouncing judgment on a suspected butter, say, x = - according to which formula, fat requiring 13*3 cc. (to 14'45) is genuine. It is easily seen that a mixture of x of butter and 1 x of coco-nut oil requires these 13*3 cc. if x = 0'93. As the method is purely empirical, rehearse it (before actual use) with (1) pure butter, made by yourself out of cream (by shaking a supply in a wide-necked bottle, &c.) ; and (2) say suet, and adhere to the " letter of the law " in all the trials. [For further infor- mation on methods of butter analysis generally, see " Dammer, Illustrirtes Lexikon der Verfalschungen " (Leipzig); p. 152]. TANNING MATERIALS. 267 RANCIDITY. Rancid butter contains free acid which can be determined by dissolving, say, 10 grms. of the filtered dehydrated butter in 30 cc. or q. s. of ether, adding phenol-phthaleine, and titrating with decinormal alcoholic alkali in the cold, taking care to stop when the liquid, by addition of the last drop and stirring, becomes violet throughout, for however short a time. On standing, the violet colour disappears through saponification of the fat by the excess of alkali. The result is conveniently reported, as so much butyric acid, C 4 H 8 O 2 . 9. Assaying- of Tanning Materials. THESE materials owe their availability for tanning to the pre- sence in them of some kind of "tannin." All "tannins" agree in the following properties : (1) They are solids, soluble in water, and exhibiting feebly acid properties ; (2) the solution of the potash or soda salts absorbs oxygen from the air with great avidity, and formation of dark-coloured soluble compounds ; .(3) they precipitate gelatine from its solution ; (4) they are precipi- tated from their aqueous solutions by animal raw hide with formation of "leather;" (5) they strike a dark-blue or green colour with solution of ferroso-ferric salt, preferably the acetate; (6) when dissolved in water and mixed with a sufficiency of sul- phuric acid they are readily oxidized in the cold by addition of permanganate of potash. Upon the propositions (4) and (6) Lowenthal has founded a method for the determination of the " tannin " in sumach, hem- lock* extract, gall nuts, oak bark, &c., which we will now proceed to describe, assuming, for fixing ideas, that we had to deal with sumach. ASSAYING OF SUMACH. Sumach consists of the dried shoots and leaves of rhus corifiria. The commercial article usually has the form of a powder. 7 grms. of the air-dry powder are extracted three times with boiling- * An American tree, belonging to the order of Conifers, 268 EXERCISES IN APPLIED ANALYSIS. hot water on a water-bath ; the mixed extracts are filtered and diluted to 1 litre, so that 1 cc. corresponds to 7 mgs. of substance. Reagents Required. (1.) A solution of Permanganate of Potash, so adjusted that 1 cc. oxidizes about 1'8 to 2'0 mgs. of ferrosum. According to Neubauer's experiments with gallo-tannin, 1 grm. of ferrosum corresponds to 07423 grm. of tannin ; hence 1 grm. of tannin corresponds to 1-7-07423 = 1*3471 grms. of ferrosum. Hence, supposing the permanganate to be so adjusted that 1 lit. = 2'021 grms. of ferrosum ( = ri40 grms. crystals of KMn0 4 by calcu- lation), 1 cc. of such solution would correspond to exactly I'D mgs. of tannin. Neubauer's factor does not necessarily hold for other tannins,* hence the most rational mode of reporting would be to state that 100 grms. of sumach contain tannin equivalent to so and so many grammes of ferrosum (in refer- ence to oxidation by permanganate). This, however, is not customary; and, as the general practice is to report per cents, of " tannin," we will adopt Neubauer's factor as if it held for the tannin in question. (2.) A solution of Indigo Carmine (sulphindylate of soda) in water, acidulated with sulphuric acid, which contains in every litre 60 cc. of oil of vitriol and so much of the pigment that the solution requires very nearly its own volume of perman- ganate for oxidation. With a certain kind of very superior indigo-carmine, which we have in the laboratory, this means about 12-13 grms. of carmine. All carmines, of course, are not of the same strength. (3.) Hide Shavings. Raw ox hide as obtainable from any tanner is deprived of its hair by moistening the hairy side with dilute caustic soda, and then plucking oft* the hair. The unhaired hide is washed with abundance of water, and next steeped in ordinary alcohol for about two days to take out part of the water. The treatment with alcohol is repeated, absolute alcohol being used at the end to obtain a water-free hide, which is dried over sulphuric acid at the ordinary tera- * It indeed is highly questionable whether it is correct for gallo-tannin itself, TANNING MATERIALS. 269 perature. The thus dehydrated hide has all the properties of leather.* It shows no tendency to putrefy, even if kept in the air, and not even, it appears, if moistened, and then allowed to dry in the air. By means of a sharp plane it is easily reduced to shavings. It may be comminuted also by means of a rasp ; but the heat evolution in this case is so considerable that the raspings become appreciably singed. The shavings are preserved in a bottle. Modus operandi. Observe that the process to be described is a purely empirical one, and in its execution nothing must be taken for granted except the proposition that the volume of permanganate required to oxidize, say, unit weight of a certain kind of tannin under given conditions is constant as long cts the conditions remain the same, including even the rate at which the permanganate is dropped into the solution to be oxidized. We assume that the permanganate solution has been standardized exactly with ferrosum, and that the indigo solution has been approximately adjusted to the in- tended strength. (1.) Measure out 20 cc. of indigo, add 1 litre of water and 10 cc. of 60 per. cent, sulphuric acid ; mix in a round-bottomed, short- iiecked flask (" bolt-head "), and drop in permanganate with constant agitation until the mixture becomes golden-yellow, free from all admixture of green. Repeat this several times, and take the mean of the last 4-5 titrations, which ought to agree to within dbO'l cc. of permanganate. (2.) After a preliminary titration of the sumach extract, measure off exactly a volume corresponding to about 10 cc. of permanganate, add 20 cc. of indigo, 1 litre of water, 10 cc. of 60 per cent, sulphuric acid, and titrate again. Deduct the permanganate corresponding to the indigo according to (1), and put down the balance as corresponding to the permanganate reducers in the sumach. This titration also must be done several times ; the results are not quite so constant as in the case of unmixed indigo. * This process was invented by the author some years ago, and proposed for the preparation of costly skins, such as tiger skins, which stand the expense of alcohol. 270 EXERCISES IN APPLIED ANALYSIS. (3.) Measure off five times as much sumach solution as you used in titration (2), and for every 10 cc. of permanganate which this would require weigh out 1 grin, of hide shavings, which, if our directions were followed, means 5 grms. of the latter. Place the hide in a flask, add a volume of water equal to the sumach solution, wait until the hide is softened up; then add the sumach solution, and allow to stand overnight so as to give the hide time to imbibe the tannin. Then filter through a dry filter, measure out a volume equal to twice the volume of sumach solution used in titration (2) (i.e., corresponding to the weight of sumach titrated), and titrate with permanganate under the exact conditions which prevailed in titration (2). (4.) The nett permanganate used this time corresponds to the gallic and other permanganate reducers, not tannin. It is deducted from the nett permanganate used in titration (2) as a correction. The difference is calculated into " tannin " by Neu- bauer's factor. An example may serve to make the mode of calculation, &c., clear. The substance worked upon was genuine sumach of first quality. 7'003 grms. of sumach were exhausted with water, and the extract diluted to 1 litre. 1 cc. of permanganate by standardization against pure ferrous sulphate was found to correspond to T855 mgs. of iron. 20 cc. of indigo solution required 19'53 cc. of permanganate. 77 cc. of sumach extract + 20 cc. of indigo, in presence of 10 cc. of 60 per cent, sulphuric acid and 1 litre* of water, required, as the mean of several titrations, 29'63 cc. of permanganate. Deducting 19'53 cc. for the indigo, we have 1010 cc. Treatment with Hide. 5 grms. of hide were moistened with 381 cc. of water, 381 cc. of sumach extract were added, and the mixture allowed to stand. After standing over night 15 '4 cc. of the filtered solution (as the mean of two numbers) required 1 -22 cc. KMnO 4 (corrected for indigo). Hence permanganate, corrected for gallic acid, &c., = 1010 1-22 = 8'88 cc. Now 8'88 x 1/855 = 16'47 mgs. of iron = 07423 X 16'47 = 12'23 mgs. of tannin, according to the factor. 77 cc. of extract = 53'9 ings, of sumach ; therefore percentage of tannin = 22-69. TEA. 271 10. Partial Analysis of Tea. PROCURE a good average sample, mix it well, and keep it bottled up. The Water. Weigh out 23 grins, of the sample between watch-glasses, and heat in an air-bath to 105-1 10C. until the weight is constant. The Theine may be determined either in the original substance or in the aqueous extract from it. In the former case 2 '5 grms. of the finely-powdered leaves are mixed with 0'5 grm. of ignited magnesia. The mixture is trans- ferred to a shallow porcelain basin, about 30 cc. of water are added, and the whole is evaporated on a simmering water-bath, on constant stirring, until the mass has almost but not com- pletely dried up. If the evaporation be carried too far, there is a danger of some of the theine being lost by volatilization. The mass is then scraped out by means of a metal spatula, and trans- ferred to a dry, conical flask of about 150 cc.'s capacity. 50 cc. of chloroform are added, the flask is connected with a small, inverted condenser, and the extraction of the theine effected by heating the flask and contents in a water-bath. After about a quarter of an hour's heating the flask is cooled down, detached from the condenser, and the extract decanted through a filter made of loose paper into a small, tared, conical flask. The chlo- roform is then distilled off and used for a second extraction of the residue. After three treatments in this way the theine is extracted as far as possible. The chloroform is finally distilled off, the residue dried at 105C. and weighed once, merely for a check, as " crude theine." It is largely contaminated with resinous matter, which, is removed by treatment with boiling water and filtration. The solution of theine is evaporated to dryness in a tared platinum or porcelain crucible, dried at 105 C., and weighed. When the aqueous extract of the tea is used for the deter- mination of the theine, the total amount of residue is much smaller than in the case just mentioned, and the extraction with chloroform is much more quickly and easily effected. Another advantage is that the theine obtained from the first contains 272 EXERCISES IN APPLIED ANALYSIS. only a small, unweighable quantity of resinous matter (insoluble in hot water). For the execution of this modification 2 grms. of the tea (unpowdered) are extracted several times with boiling- water, each instalment of water being allowed a sufficient time to act, and the flask in which the extraction is carried out being kept on a water-bath. The extract is evaporated with magnesia as before, the theine extracted repeatedly from the residue by chloroform, dried, and weighed. Mr. John M'Arthur, in each of two tea analyses, I. and II., applied the two modifications of the method to the same sample, and obtained the following percentages of theine : I. II. From the leaves, 3'80 4'00 From the infusion, 373 4'20 The Tannin. 7 grms. of original tea are extracted several times with boiling water, the solution is filtered, and diluted to 1 litre. In the extract the tannin is determined exactly as described under " Assaying of Sumach," the treatment with hide not being forgotten. The end-point of the reaction is, as a rule, more difficult to determine exactly than in the case of sumach ; a number of titrations must be carried out in order to eliminate the irregular errors as far as possible. 11. Analysis of Wood Spirit. WOOD SPIRIT consists mainly of methyl-alcohol CH 3 OH, acetone CO (CH 3 ) 2 , and water ; besides these, there are a host of minor components, of which acetate of methyl, CH 3 . COOCH 3 and {O OH O TH 3 are ^ es ^ known. When we speak of the " analysis " of a wood spirit, we generally mean the determination of the methyl-alcohol present as such, or, potentially, as acetate of methyl, and of the acetone. In the examination of what pretends to be pure methyl- alcohol, e.g., the kind of alcohol sold for the making of aniline colours or for the use of chemists, it is expedient to begin with qualitative tests to prove, if possible, the absence of certain WOOD SPIRIT. 273 impurities. Pure methyl-alcohol has only a very faint and not unpleasant smell. It mixes with water in all proportions without formation of a milky mixture. 1 cc. when mixed with 2 cc. of concentrated oil of vitriol gives a colourless, or almost colourless, mixture. 1 cc. when mixed with 5 cc. of water, and then one drop of bromine water, gives a yellow liquid ; if the colour of the bromine disappears, this indicates the presence of allyl-alcohol C 3 H 5 OH. For the detection of acetone we may use the method given below in a quantitative form. In regard to ethyl-alcohol see the next exercise. Supposing qualitative tests for impurities to have given negative results, or only proved the presence of traces of the respective substances, a determination of the vapour-density by Gay-Lussac's method volatilization of a known weight over mercury in a graduated tube, and measurement of the volume, temperature, and pressure of the vapour affords an excellent test for impurities generally ; because, apart from water (which we will assume to be absent), of all components of wood spirit methyl-alcohol has the lowest vapour-density. Given a mixture of volatile substances, I., II., Ill , containing p l grms. of component I., > 2 of component II., &c., and we have obviously for each of these P = where M is the molecular weight, i.e., p grms. represent p/M. gramme-molecules, and consequently at any fixed upon dis- gregation, e.g., that corresponding to 100C. and 500 mms. pressure for the vapour-volume of the component under con- sideration -^ X C litres, M , where C is a constant as long as the disgregation remains the same. Hence, we have for the vapour-volume of 1 grm. of our mixture at standard disgregation 274 EXERCISES IN APPLIED ANALYSIS. But H 2 = 2 grms. of hydrogen occupy C litres likewise ; hence, the specific gravity of the mixed vapour on the hydrogen-scale is Example : The given liquid consists of 0*8 grm. of methyl- alcohol and 0'2 grm. of water. In this case we have 32 + 1 The divisor computed is 0'07222, and its reciprocal, i.e., S = 13-85. In this particular case, however i.e., if we know that, sub- stantially speaking, water is the only impurity present we need not go to the trouble of determining the vapour density ; the specific gravity of the liquid, taken at, say, 1556C. ( = 60F.), or at 0C., gives us all we want, with the help of the alcoholo- metric table appended to the end of this article. For the determination of the specific gravity of a liquid like methyl-alcohol the Westphal balance is a handy instrument ; but more exact results can be obtained by means of a cylindrical narrow-necked specific-gravity bottle provided with either one mark, or a calibrated millimetre scale, at the narrow neck, and with a well -fitting stopper for the funnel-shaped appendage in which the neck terminates ; because with such a bottle volatilization is easily prevented by inserting the stopper, and a pre-determined fixed temperature established by means of a water-bath. In the examination of a wood spirit more especially, the determination of the boiling-point curve comes in as a useful preliminary. To effect it, place 100 cc. of the spirit in a fractionating flask of about 17 times the capacity, and attach the flask to a Liebig's condenser, after having fixed a thermo- meter in its neck, so that the bulb is fairly below the lower edge of the side tube. Then distil the contents over a naked flame at the lowest rate compatible with the condition that the thermometer bulb is always immersed in fully saturated vapour, and at regular short intervals note down the reading of the WOOD SPIRIT. 275 thermometer and the corresponding volume, v, of total distillate obtained at that stage. Thus v= 5 cc. 10 cc 90 cc. 95 cc. t = (say)640 640 641 66 671 As methyl-alcohol has quite a characteristic tendency to bump on distillation, it is necessary to prevent this somehow. The best method is to send a very slow but continuous current of air or hydrogen through the boiling liquid, but this means a relatively complex apparatus. More convenient expedients are to put some fragments of an alloy of tin and (little) sodium into the liquid, or, in the absence of this alloy, a few pieces of the stem of a clay tobacco pipe. Supposing the distillation to be completed, lay down the v's as abscissae and the t's as ordinates in a system of rectangular co-ordinates (i.e., on a sheet of " curve-paper," as sold by Messrs. W. & J. K. Johnston, Geographers, Edinburgh), and draw a continuous curve, so that it takes in the t's as nearly as possible, to obtain the boiling-point curve. Before going further, we will give a brief statement of the physical characters of the principal components of wood spirit. Methyl-alcohol. The specific gravity 4 S t , meaning the sp. gr. at t referred to water of 4C. as standard and taken as = 1 is as follows *: t= 5 10 15 1556 20 4 S t = -81015 -80557 '80098 79640 79589 79181 Boiling point under 760 mms. pressure = 64'96, say 65; 29 mms. of difference of pressure corresponds to 1 of difference in the boiling point, or from about 760 30 to 760 + 30 mms., 1 mm. of increment in pressure corresponds to 00345 in the boiling point. Acetone. A colourless liquid possessing a strong fragrant smell, and mixing with water in all proportions. Extractable from its aqueous solution, if not too dilute, by addition of dry carbonate of potash as an upper layer. 4 S = '81858, T. E. Thorpe. * According to an investigation by Dittmar and Charles A. Fawsitt com- municated to the Roy. Soc., Eclin., in 1887. 276 EXERCISES IN APPLIED ANALYSIS. According to H. Kopp, S = '8144; S 13 . 9 = 7995. Boiling point, 5653 at 760 mms. (Thorpe). Acetate of Methyl. A colourless mobile liquid of a strong agreeable smell. According to H. Kopp, S = *9562; boiling point, 5653 at 760 mms. According to experiments made in our laboratory by Mr. James Robson (on a hot summer day), 100 volumes of water dissolve 35*8 volumes of the ester; or 1 volume of the latter requires 2*8 volumes of water for its solution. The Ethyl-ester, according to Mr. Robson's experiments, requires 11 '7 times its volume of water. Di-Methyl-Acetal. A colourless liquid possessing a peculiar ethereal smell. According to Kramer and Grodzki (Ber. d. Deutsch. Chem. Ges. for 1876, p. 1930), S 15 = '8554 ; boiling point, 63'2 to 64 <0 8 at 760 mms. Soluble in 15 parts of water. DETERMINATION OF THE METHYL- ALCOHOL. For this determination Krell, some years ago, invented a con- venient method, the principle of which is to convert the methyl of the alcohol into iodide and to measure, i.e., to indirectly weigh, the latter. The method was subsequently modified by Kramer and Grodzki. In the following we reproduce the directions given by the latter in his article on methyl-alcohol in Dammers Handworterbuch der Verfdlschungen : 15 grms. of biniodide of phosphorus are placed in a flask of 30 cc. capacity connected with an inverted condenser, and 5 cc. of the alcohol to be analysed are gradually dropped on the iodide through the condenser while the flask is being kept cold in a water-bath. This being done, 5 cc. of a solution of 1 part of iodine in 1 part of aqueous hydriodic acid of 1*7 specific gravity are added, and the whole is digested at a gentle heat for a quarter of an hour. The condenser is then turned downwards, and the water-bath heated up to cause the iodide of methyl formed to distil over. It is collected in a tube, graduated into tenths of a cc., under water. When the iodide is over, as far as possible practically, the aqueous (upper) layer in the tube is made up to 15 cc. by addition of more water, shaken with the (heavy) iodide and the latter allowed to settle, to be measured as it WOOD SPIRIT. 277 stands. According to Kramer .and Grodzki's standard trials 5 cc. of pure methyl-alcohol give 7'45 cc. of iodide instead of the 7*8 cc. required by theory and the known specific gravity of the iodide. The method, however, valuable as it is, is invested with numerous errors, which defy all calculation. Hence, to give reliable results, it must be wrought in a strictly empirical style, and this includes that, instead of relying on Kramer and Grodzki's results, we had better prepare perfectly pure methyl- alcohol ourselves, work 5 cc. of it in the apparatus meant to be used, and determine our own constant. According to Grodzki (I.e.), a methyl-alcohol, to be fit for aniline-colour making, should yield at least 7'3 cc. of (crude) iodide. If acetate of methyl is present, a quantity of iodide of methyl is produced from it which is almost exactly the same as if the FIG. 61. ester in the course of the process were decomposed into acetic acid and methyl-alcohol, and the latter acted on by the hydri- odic acid as if it had been there as such from the first. This, however, is no reproach to the method, because acetate of methyl, being so readily saponified by alkalies, is really worth its equivalent of methyl-alcohol. If a separate determination 278 EXERCISES IN APPLIED ANALYSIS. be desired, heat a known quantity of the spirit with a sufficient measured volume of standard caustic soda in a closed bottle to 100 for a sufficient time, i.e., until all the acetate is decomposed into alcohol and soda-salt; then allow to cool, and determine the excess of caustic soda left by titration with standard acid (and alkali). Every NaOH part of soda neutralized corresponds to CH 3 , COOCHg parts of methyl-acetate = CH 4 parts of methyl- alcohol. The presence of acetone or di-methyl-acetal produces appreciable (positive) errors ; but we cannot afford here to dis- cuss the matter any further, and must refer to Kramer and Grodzki's original memoir (cited under Di- Methyl- Acetal). See also Bardy and Bordet Comptes Rendus 88, 236. The distillation apparatus must be so constructed that the condenser can be turned either upwards or downwards without undoing any joint. Fig. 61 represents the apparatus which we are in the habit of using in this laboratory. Preparation of Iodide of Phosphorus. 3*1 grms. of red phosphorus and 2 5 '4 grms. of iodine (both powdered and dry) are put into a small round-bottomed flask, mixed by shaking, and heated cautiously until they have fused and united into a compound (P I 2 ). This is allowed to freeze, then taken out by breaking the flask and preserved in a dry -glass-stoppered bottle, whose stopper has been smeared over with vaseline before insertion to produce a hermetic joint. According to our experience, however, there is no need of this preparation ; 1*6 grms. of red phosphorus, conjointly with 13*4 grms. of iodine, act on methyl-alcohol exactly as if they were combined into 15 grms. of P I 2 . Hydriodic Acid of 1*7 specific gravity. This specific gravity corresponds to about 57 per cent, of HI, i.e., the percentage of that aqueous acid which boils without change of composition at 127C. For its preparation the most convenient method is the following : Dissolve 5 grms. of iodine and 10 grms. of iodide of potassium in 20 grms. of water in a mortar, and add other 80 grms. of water. Through the brown liquid thus produced pass sulphuretted hydrogen until the liquid is decolorized, i.e., until the reaction H 2 S + 1 2 = 2HI + S is accomplished. Then dissolve an additional quantity of iodine in the liquid, again decolorize WOOD SPIRIT. 279 by means of sulphuretted hydrogen, and so go on until a fresh instalment of iodine added refuses to be decolorized by sul- phuretted hydrogen, which will be the case before the liquid is quite up to the desired percentage of HI. When that point is reached heat gently to cause the sulphur to coagulate and the dissolved sulphuretted hydrogen to escape, and filter off the sulphur. Then add the necessary supplementary iodine and enough of red phosphorus to convert it into hydriodic acid, and distil the mixture, taking care to change the receiver as soon as the boiling point has become constant at (or very near) 127C. DETERMINATION OF THE ACETONE. An important determination, because the presence of more than 0*5 per cent, of acetone in a given methyl-alcohol unfits it for the purposes of the aniline-colour maker (Grodzki). The method described in the following paragraph was worked out by Kramer and Grodzki. Reagents required : 1. A doubly -normal solution of iodine in iodide of potassium. 25'4 grms. of iodine and 50 of iodide of potassium are dissolved in water and diluted to 100 cc. 2. A doubly-normal caustic soda containing 2NaOH = 80 grms. per litre. 3. Alcohol-free ether. Ordinary ether, according to our experience, can be made almost absolutely alcohol-free by long digestion with a large excess of powdered caustic soda in the cold. If required at short notice, the crude ether is boiled with the reagent at the " wrong end " of a condenser. Copper flasks,* as sold for the making of coffee, are well adapted for either purpose. After elimination of the alcohol as NaOC 2 H 5 , the mixture is distilled from out of a water-bath. To remove the last remnant of alcohol the distillate is tortured with sodium under an inverted condenser and then distilled. To analyse a wood spirit, mix 1 cc. of it with 10 cc. of the caustic soda in a graduated test tube ; then add gradually 5 cc. of the iodine solution. lodoform is produced, sometimes visibly, by the reaction CO(CH 3 ) 2 + 3I 2 + 3NaOH = (CO.CH 3 ) OH + 3NaI + 2H 2 O + CHI 3 Acetone. Acetic Acid. lodoform. * Glass flasks, being attacked by the alkali, are not safe. 280 EXERCISES IN APPLIED ANALYSIS. CO(CH 3 ) 2 = 58; CHI 3 = 394; hence 1 grm. of iodoform produced indicates 0'1472 grm. of acetone. To extract the iodoform shake the mixture with 10 cc. of ether, allow the ethereal solution to rise, measure off an aliquot part of it, let it evaporate spontaneously on a tared watch-glass, and weigh the iodoform after having kept it for a while over concentrated sulphuric acid, without, however, aiming at con- stancy of weight, which is unattainable on account of the appreciable volatility of the compound. The iodoform obtained is liable to be contaminated with non- volatile matter. To deter- mine it, volatilize the (weighed) product at a gentle heat, weigh the residue, and deduct it as a correction. In calculating the whole of the iodoform present from the part weighed multiply with the ratio of the volume of the ethereal layer (not that of the total ether taken) to the volume of ethereal solution measured off for evaporation. The following table is extracted from the Memoir, quoted in the foot-note on p. 275. It is based on duplicate determinations of the specific gravities at 0, about 10 and about 20 of pure methyl-alcohol, and of aqueous alcohols of 5, 10, 20, 30, 40, 50, 60, 70, 80, 90. 95 per cent, derived from it by gravimetric syn- thesis. The methyl-alcohol used was proved to be free of acetone ; it suffered no diminution in specific gravity on re-dis- tillation over anhydrous sulphate of copper ; its vapour-density was practically equal to 16. The values S given by the table are probably correct to 0*0001 ; they are sure not to be out by more than 0'0002. As stated in the heading, they are taken in reference to water of 4 as = 100000. To find the specific gravities in reference to water of any other temperature t as a standard, divide with the specific gravity ( 4 W t ) of water at t. Taking the weight of one volume of water of 4 as = 1, that of one volume of water at t is 4 W t = 1 e. For t = 4 15 15-56 e = nil -000 129 '000 840 '000 928 and generally, tS = 1 with sufficient exactitude. WOOD SPIRIT. 281 DlTTMAR AND FAWSITT'S TABLE OF THE SPECIFIC GRAVITIES OF AQUEOUS METHYL-ALCOHOLS AT AND AT 15-56C. ; WATER OF 4- 4 = 100000. I._From to 30 per cent, of CH 4 O. 4 S - 4S t = at + bt2 Per- centage. Sp. Gravity at 0C. Diff. a b Sp. Gravity at 15'56C. Diff. 999 87 - 6-0 + 705 999 07 i 998 06 -181 5 '4 694 997 29 -178 2 996 31 175 4-8 6*1 995 54 '75 3 994 62 169 3 '9 670 993 82 172 4 992 99 163 3*o 659 992 14 168 5 991 42 J57 - 2'2 648 990 48 1 66 6 989 90 IS 2 1'2 '634 988 93 155 7 988 43 i47 0'2 621 987 26 167 8 987 01 142 + o'9 609 985 69 157 9 985 63 138 2'I 596 984 14 155 10 984 29 !34 + 3'3 58i 982 62 J 5 2 ii 982 99 130 4'8 569 981 ii J5 1 12 981 71 128 6-2 552 979 62 149 13 980 48 123 7-8 536 978 14 148 14 979 26 122 9'5 '5*9 976 68 146 15 978 06 120 + II'O 500 975 23 i45 16 976 89 117 12-5 480 973 79 144 i? 975 73 116 I4-5 461 972 35 144 18 974 59 114 l6'2 440 970 93 142 i9 973 46 JI 3 18-3 '420 969 50 143 20 972 33 JI 3 + 20*0 398 968 08 142 21 971 20 H3 22'2 '373 966 66 142 22 970 07 H3 24-3 '35 965 24 142 23 968 94 H3 26'4 321 963 81 H3 24 967 80 114 29'0 291 962 38 143 25 966 65 H5 + 31*3 261 960 93 145 26 9 6 5 49 116 33'8 230 959 49 144 27 964 30 119 36*0 191 958 02 147 28 963 10 120 38-8 I5 1 95 6 55 147 2 9 961 87 123 41-1 106 955 6 149 30 960 57 I 3 44-0 063 953 55 151 282 EXERCISES IN APPLIED ANALYSIS. II. From 30 to 100 per cent. 480 4t = a t. Per- centage. Specific Gravity at 0C. Diff. a Specific Gravity at 15*56C. Diff. 30 960 57 -130 + 44'36 953 67 31 959 21 136 45-66 95 2 ii -156 32 957 83 138 46-93 95 53 158 33 95 6 43 140 48-17 948 94 159 34 955 143 49'39 947 3 2 162 35 953 54 146 50-58 945 67 165 36 952 04 150 5175 943 99 1 68 37 95 5 1 153 52-89 942 28 171 38 948 95 156 54"oi 940 55 i73 39 947 34 161 55-io 938 77 178 40 945 7i 163 56-16 936 97 1 80 4i 944 oo 171 57-20 935 1 187 42 942 39 161 58-22 933 35 i75 43 940 76 163 59-20 93i 55 1 80 44 939 ii 165 60*17 929 75 1 80 45 937 44 167 6i'io 927 93 182 46 935 75 169 62-01 926 10 183 47 934 03 172 62*90 924 24 186 48 932 29 174 63-76 922 37 187 49 93 5 2 177 64*60 920 47 190 5 928 73 179 65'4i 9i8 55 192 5 1 926 91 182 66*19 916 61 i94 5 2 925 7 184 66*95 9H 65 196 53 923 20 187 67*68 912 67 198 54 921 30 190 68-39 910 66 201 55 919 38 192 69-07 908 63 2O3 56 917 42 196 69-72 906 57 206 57 915 44 198 70-35 904 50 207 58 913 43 201 70-96 902 39 211 59 9n 39 204 7^54 900 26 2I 3 60 909 17 222 71-96 897 98 228 61 907 06 211 72-37 895 80 218 62 904 92 214 72-91 893 58 222 63 902 76 216 73-45 891 33 225 64 900 56 220 73-98 889 05 228 WOOD SPIRIT. II. From 30 to 100 per cent. Continued. 283 Per- centage. Specific Gravity at 0C. Diff. a Specific Gravity at 15'56C. Diff. 65 898 35 - 221 74*5 J 886 76 - 229 66 896 II 224 75^5 884 43 233 67 893 84 227 75-57 882 08 235 68 891 54 230 76-10 879 70 238 69 889 22 232 76-62 877 14 256 70 886 87 235 77-I4 874 87 227 71 884 70 230 77-66 872 62 225 72 882 37 233 78-18 870 21 241 73 880 03 234 78-69 867 79 242 74 877 67 236 79-20 865 35 244 I 75 875 3o 237 79-71 862 90 245 76 872 90 240 80-22 860 42 248 77 870 49 2 4 I 80-72 857 93 249 78 868 06 243 81-23 855 42 25 1 79 865 61 245 8r73 852 90 252 80 863 14 247 82-22 850 35 255 81 860 66 248 82-72 847 79 256 82 858 16 250 83-21 845 21 258 83 855 64 252 83-70 842 62 259 84 853 10 254 84-19 j 840 01 261 85 850 55 255 84-67 837 38 263 86 847 98 257 85-16 834 73 265 87 845 39 259 85-64 832 07 266 88 842 78 26l 86-12 829 38 269 89 840 15 263 86-59 826 68 270 90 837 5i 264 87-07 823 96 272 834 85 266 87'54 821 23 273 9 2 832 18 26 7 88-01 818 49 274 93 829 48 270 88-48 815 72 277 94 826 77 271 88-94 812 93 279 95 824 04 273 89*40 810 13 280 96 821 29 275 89-86 807 31 282 97 818 53 2 7 6 90-32 804 48 283 98 815 76 277 90*78 801 64 284 99 812 95 28l 91-23 798 76 288 IOO 810 15 280 91-68 795 89 287 284 EXERCISES IN APPLIED ANALYSIS. 12. Determination of Ethyl- Alcohol. FOR the determination of the percentage of real, in a substantially pure aqueous, alcohol, we need only determine its specific gravity at a definite convenient temperature, and by means of one of the tables, based on standard experiments, which have been drawn up, translate the result into the corresponding percentage. The following table has been calculated by the author from Men- delejeff's famous experiments,* not quite directly though, but in this way : Landolt and Bornsteins Physikalische Tabellen include a table (on their p. 151) which, proceeding from per cent, to per cent., gives the specific gravities, 15 S 15 , of aqueous alcohols to five decimal places. According to its heading, the table has been drawn up by the " Kaiserliche Normal-Aichungs-Commission" and is " basirt auf den von Mendelejeff berechneten Formeln" From this table I have calculated the percentages corresponding to the specific gravities "999, '998, '997, 794 to originally three decimals by simple interpolation. Hence, apart from the errors introduced by neglecting second differences and cancelling the third decimals of the percentages, the specific gravities given in column 1 of our table as functions of the stated percentages should be as exact as the original table is. To check my com- putations I calculated the specific gravities for all integer per- centages backward from my table, again by simple interpolation, and, comparing them with those in the original table in Landolt's book, found that The difference nil occurred 51 times, 1 44 2 5 Water = 100000, which showed, incidentally, that our table, in point of arith- metical precision, is practically at a level with the one in Landolt and Bornstein's book. The third column gives the differences of the consecutive _p's to facilitate interpolation ; column 4 gives the reciprocals of * Poggendorff's Amialen, vol. cxxxviii, pp. 103 and 230. ETHYL-ALCOHOL. 285 these differences to facilitate backwards interpolation. Each of such a couple of differences belongs to the interval between the p or S, which it stands on a line with, and the next pre- ceding entry. The numbers given in column 5 as AS for 1 of change of temperature belong to the respective values S or p individually. These values AS/A^ were calculated from two tables included in Mendelejeff's Memoir, which give the specific gravities 4 S referred to water of 4 at 0, 10, 20, and 30, one for all integer percentages from 5 to 5, as found virtually by direct experiment : the other for all integer percentages from 10 to 10, as calculated by means of interpolation formulae. From the respective entries it was easy to calculate the values (i 5 S 10 - 15 S2o)-rlO for 5, 10, 15, &c., per cent, alcohol, and these values I adopted as approxi- mations to any AS/A^ from = 10 to = 20, for the respective p. To form an idea of the degree of approximation afforded by these values, I used them for calculating the specific gravities 15 S 15 from the values 15 S 10 as given (virtually) in Mendelejeff's Memoir, and compared them with the entries in the table in Landolt's book, which had served me for the calculation of the entries p in my table. The differences ranged as follows : Forp = 100 to 50 45 15 35,30,25,20,10,5 -5to + l +10 -15 +6 to -9 units. Water =100000 units. This may not do full justice to Mendelejeff's work, but it surely suffices for all practical purposes. Relying chiefly on the data of Mendelejeff's calculated table, I found the values AS/A for all the specific gravities given in my table under S by graphic interpolation, and entered them in column 5. In the table the specific gravities are referred to water of 15 <0 . 1. To find the specific gravity 4 S referred to water of 4C. from 15 S, calculate thus 4 S = 16 S(1- O'OOO 840). 2. To find the specific gravity referred to water of 15 0< 56C.= 60 C F. calculate thus - 0-000 088). 286 EXERCISES IN APPLIED ANALYSIS. TABLE GIVING THE RELATION IN AQUEOUS ETHYL-ALCOHOL BETWEEN SPECIFIC GRAVITY AT 15C. ON THE ONE HAND AND THE PERCENTAGE BY WEIGHT OF REAL ALCOHOL, OR THE SPECIFIC GRAVITY AT ANY TEMPERATURE FROM 10 TO 20, ON THE OTHER. Calculated after Mendelejeff's Determinations, by W. Dittmar. Water of 15C. = 1000. Spec. Grav. of 1 t\' Percent- age. fp for q _ Diff. Gn in Spec, iv. for Spec. Grav. af 1 K Percent- age. A_p for / q _ Diff. Grs tn Spec. iv. for dill 1O 15Sl5. p. A O 1. &p = At- = l. at 10. 15815. p. ZA io 1. *C" A* = l. 1000 0*00 0-144 975 I7'25 84 19 "33 999 "53 '53 1*89 146 974 18-08 "S3 20 348 998 i '06 '53 1*89 147 973 18*91 "S3 '20 367 997 1*61 '55 82 150 972 1973 82 *22 387 996 2*16 '55 82 'JS 2 971 20*54 *8 1 23 405 995 273 '57 75 J 54 970 2i'35 *8 1 23 421 994 3'3o '57 '75 ^57 969 22*13 78 28 "434 993 3'9i *6 1 64 161 968 22*89 76 -32 "45 992 4'S 1 *6o 6 7 166 967 23-65 76 -32 466 99 1 5' 12 *6 1 64 170 966 24-39 "74 "35 480 990 577 65 '54 174 965 25" J 3 "74 "35 496 989 6'43 66 5 2 180 964 25-84 -71 '41 510 988 7'9 66 5 2 186 963 26-53 6 9 '45 522 987 7-78 69 '45 192 962 27*23 70 '43 "535 986 8-49 7i HI 199 961 27*90 67 "49 548 985 9*20 7i HI 206 960 28-55 65 "54 560 984 9-96 76 32 214 959 29*21 66 -S 2 570 983 10*72 76 32 223 958 29*84 63 '59 58i 982 11*50 78 28 231 957 30*46 62 6 1 "593 981 12-28 78 28 242 956 31*09 63 '59 "604 980 13*10 82 *22 254 955 31*69 *6o 67 '614 979 13*92 82 *22 267 954 32-3 *6 1 64 623 978 1474 82 '22 281 953 32*88 58 72 632 977 15-58 84 I 9 296 952 33-45 "57 '75 641 976 16*41 83 *2O 311 95 1 34-03 58 -72 650 ETHYL- ALCOHOL. 287 Spec. Grav. at 15 * 15815. Percent- age. P- AP for . Diff. in Spec. Grav. for Spec. Grav. at 15 * 15815. Percent- age. P- Ap for A Q Biff, in Spec. Grav. for 1. A r A* = l. 1. a r A = l. 95 34-59 56 1-79 657 915 51*68 -45 2*22 *800 949 35^5 56 1-79 666 914 52-I3 -45 2*22 *80 1 948 3570 55 1-82 673 913 52-58 "45 2*22 *802 947 36-25 55 1-82 680 912 53-02 "44 2-27 80 3 946 3678 53 1-89 687 911 53*47 "45 2*22 804 945 37-31 53 1-89 "694 910 53'9i "44 2*27 806 944 37'84 53 1-89 -700 909 54-35 "44 2-27 j '807 943 38-36 52 1-92 -707 908 54-8o "45 2*22 ; *8o8 942 38-88 52 i- 9 2 712 907 55^4 "44 2*27 '809 941 39*39 "5 1 i { 9 6 718 906 55-68 '44 2*27 *8lO 940 39 '90 *5 J 1-96 723 905 56-12 "44 2-27 *8n 939 40*40 *5 2'00 727 904 56-56 "44 2*27 "812 938 40-90 50 2-QO 733 903 57-00 "44 2-27 -813 937 41-40 *5 2'00 737 902 57-43 "43 2-33 -813 93 6 41-89 *49 2-04 741 901 57-87 '44 2-27 -813 935 42-38 *49 2*04 '746 900 58-31 '44 2*27 "814 934 42-86 48 2-08 75 899 58-74 "43 2-33: -815 933 43'35 '49 2*04 753 898 59*18 i -44 2*27 "816 932 43'83 48 2-08 -756 897 59-6i '43 2*33 -817 93i 44'3 '47 2-13 760 896 60*05 "44 2-27! -817 930 44-78 48 2-08 763 895 60*48 "43 2-33 8l 7 929 45'25 '47 2-13 767 894 60-91 "43 2-33 ; -818 928 j 4573 48 2-08 770 893 6i-35 '44 2*27 -819 927 46*20 '47 2-13 773 892 61-78 '43 2'33 819 926 46-66 46 2-17 774 891 62*21 "43 2'33 820 925 47'i3 "47 2-13 778 890 62*64 "43 2'33 820 924 47*59 46 2-17 78i 889 63-07 '43 2-33 -820 923 48-05 46 2-17 783 888 63*5 '43 2'33; "820 922 48-51 46 2-17 -786 887 63-93 '43 2*33 -821 921 48-96 '45 2'22 788 886 64-36 '43 2-33 822 920 49-42 46 2-17 790 885 64-79 "43 2'33 822 919 49-88 46 2-17 -792 884 65-21 42 2*38 -823 9i8 50-33 "45 2-22 -794 883 65-64 "43 2*33 ; -824 917 5078 -45 2'22 -796 882 66-07 '43 2*33 -825 916 5!"23 '45 2*22 798 88 1 66-49 42 2-38 826 288 EXERCISES IN APPLIED ANALYSIS. Spec. Grav. at 15 - 15815. Percent- age. p. A? for AS = 1. Diff. in Spec. Grav. for Spec. Grav. at 15 - 15815. Percent- age. p. AP for A ^ Diff. in Spec. Grav. for &p = A=r A io 1. &> = At = l' 880 66*92 *43 2'33 825 845 81-46 *40 2-50 854 879 67'34 42 2-38 826 844 8r86 *40 2*50 855 878 6777 *43 2*33 826 843 82-26 40 2-50 856 8 77 68-20 '43 2*33 827 842 82-66 40 2-50 857 876 68-62 42 2-38 828 841 83-06 40 2-50 858 875 69*04 42 2-38 829 840 83-46 40 2-50 859 874 69*46 42 2-38 830 839 83-86 *40 2-50 860 873 69-88 42 2-38 830 838 84-26 40 2-50 860 872 70-30 42 2-38 830 837 84*65 '39 2*56 860 871 7o'73 *43 2*33 831 836 85-05 40 2-50 861 870 7i'i5 *42 2-38 832 835 85*44 *39 2*56 862 869 7i*57 *42 2-38 833 834 85*83 *39 2*56 862 868 71-99 *42 2-38 834 833 86-22 *39 2-56 862 867 72-40 *4i 2-44 835 832 86-61 *39 2-56 863 866 72-82 42 2-38 836 831 87-00 *39 2-56 863 865 73*24 *42 2-38 836 830 87-38 38 2-63 864 864 73-66 *42 2-38 837 829 87-77 '39 2*56 865 863 74-08 42 2-38 838 828 88-15 38 2-63 865 862 74*49 *4i 2-44 "839 827 88-53 38 2*63 865 86 1 74-90 *4i 2*44 840 826 88.91 38 2*63 865 860 75*32 *42 2*38 840 825 89.29 38 2*63 865 859 75*73 *4i 2-44 840 824 89-67 38 2-6 3 866 858 76-15 *42 2-38 841 823 90-05 38 2-63 866 857 76-56 *4i 2-44 842 822 90-42 *37 2-70 866 856 76-97 *4i 2 '44 842 821 90-79 *37 2-70 866 855 77*38 *4i 2-44 8 44 820 91*16 *37 2-70 865 854 77*79 *4i 2-44 845 819 9 r 53 *37 2-70 864 853 78-21 42 2-38 846 818 91-89 36 278 864 852 78-62 *4i 2*44 -8 47 817 92*26 '37 2*70 863 851 79-02 40 2-50 848 816 92*62 36 278 862 850 79*43 *4i 2-44 848 8i5 92*98 36 2'78 862 849 79*84 *4i 2-44 8 49 814 93*34 '36 2*78 861 848 80*24 40 2*50 850 813 93*7o 36 2*78 86 1 847 80-65 *4i 2*44 852 812 94*05 *35 2*86 860 846 8 1 -06 *4i 2-44 "853 811 94*41 36 2*78 859 ETHYL- ALCOHOL. 289 Spec. Grav. Percent- L P for Diff. in Spec. Grav. for Spec. Grav. Percent- AP for Diff. in Spec. Grav. for n> 1 ^' age. A <3 a> 1 ^ t0 age. A Q ali !) 15Sl5. P- A o 1. A r A = l- at 10 15815. P- 1. Ap = A = l' 810 9476 "35 2-86 859 800 98-16 33 3'03 '850 809 95' 11 '35 2-86 859 799 98-49 33 3'03 8 49 808 95*45 '34 2-94 -858 798 98-81 -32 3-13 8 4 8 807 95-80 '35 2-86 i -857 797 99-14 -33 3-03 847 806 96-15 '35 2-86 856 796 99-46 32 3-I3 846 ! 805 96-48 '33 3'3 856 795 99-78 32 3" J 3 845 804 96-82 '34 2-94 855 794 lOO'IO 32 3-i3 844 803 97-16 '34 2-94 854 794-32 100 802 9 7 '49 '33 3'03 i '852 801 97*83 '34 2-94 8 5 I Referring to water of 4C. as = 100000, the specific gravity 4 S t of absolute alcohol, according to Mendelejeff, is 4 S t = 80 625 - 83-4* - 0'029 2 ; hence (by Mendelejeff's computation) we have for the Specific Gravities of Absolute Alcohol t 5 10 15 20 25 30 Sj 806 25 802 07 79788 793 6 7 78945 78522 780 96 The degree of relative exactitude which we reach in the deter- mination of alcohol by the specific gravity method depends, of course, on the degree of exactitude with which we determined the specific gravity, and on the strength of the alcohol solution operated upon. Supposing, for instance, in one case, we operate upon a 1 per cent., and in a second upon a 50 per cent, alcohol, and our experi- mental specific gravities are right to within dbO'OOOl. i.e., 01 by our table, then the uncertainty in the first case is 0'053 per cent, per 1 per cent., i.e., about 1/20 of the value to be determined. U 290 EXERCISES IN APPLIED ANALYSIS. In the second case it is only 0*046 per cent, absolutely, i.e., 0-046 -f 50 = '00092 of the value to be determined. Hence, if the given solution is weak, it is expedient to strengthen it by fractional distillation, to, for instance, drive all the alcohol contained in 200 grms. of given solution into, say, 20 grms., and then to determine the specific gravity of the strong liquor. Distillation as a preliminary to specific gravity determination is indispensable if the given liquor contains (non-volatile) dis- solved solids. In the analysis of undistilled alcoholic beverages (wine, beer, &c.), we must not forget that these in general contain free acetic and other volatile acids which, if allowed to pass into the distillate, would raise its specific gravity. This source of error is avoided by alkalinizing the liquor with caustic soda before dis- tilling it. The distillate then obtained, however, is liable to contain ammonia, which, if present, must be removed by re- distillation after addition of a slight excess of (preferably phos- phoric) acid. In the case of liquors rich in carbonic acid, it is best to begin by eliminating the greater part of this component by shaking the liquor repeatedly with, always renewed, air in a flask of sufficient capacity. Apparatus. For most purposes a fractionating flask attached to a small Liebig's condenser works well enough. There is in general no need of any refined means for preventing volatili- zation of alcohol from the distillate formed ; but it is necessary to have a suitable adapter at the outflow end of the condenser so that the distillate as it comes over has not to fall through more than, say, a height of 1-2 mms. of air before it unites with what is already over. The receiver must be tared before use so that the weight of distillate obtained can be determined without transvasation. How far must the distillation be pushed to be sure that all the alcohol is in the distillate ? In answer to this question let us say, first of all, that ccet. par. the efficiency of the process as a method of separation is the greater the more slowly the distillation is made to progress, or the greater the proportion of the vapour which suffers condensation before it reaches the condenser tube. With beer, cider, Rhine wine, or ETHYL-ALCOHOL. 291 claret if one distils at a moderate rate, there is no need of more than one-half to two-thirds of the liquid being drawn over. But it is better in any case to strengthen one's judgment by fixing a good thermometer in the neck of the flask (as explained under " Wood Spirit," in the section on the boiling-point curve) and to continue distilling for a while after the thermometer has reached the point which it would show with pure water boiling in the same apparatus* In the case of liquids very poor in alcohol we must work on a relatively large scale, and, if necessary, strengthen the first distillate by re-distillation. With such liquids it is as well to, from the first, use a distillation apparatus so constructed that a very considerable portion of the vapour is made to condense before the rest gets over the hill. Of the multitude of depldey- mators invented Hempel's is the simplest, and it is as efficient as any other not based upon the use of liquid baths kept at prescribed temperatures. It consists of a (sufficiently) long, narrow tube (provided near its upper end with a soldered-in side tube, like the neck of a fractionating flask, and drawn out into a narrower appendage below so that it can be fixed in the cork of the distillation flask), which is filled with pea-sized glass beads up to near the side tube. The upper open end serves for the insertion of a thermometer. For the Determination of minute quantities of alcohol in aqueous liquors (such as, for instance, pass into the urine after the imbibing of alcohol), chemical methods must be resorted to. For the mere detection of the alcohol a good method is to first concentrate the alcohol by distillation in, say, a Hempel appa- ratus, and then to apply the iodoform test, as described for acetone under " Wood Spirit " (p. 279). Whether thfe method as there given would afford reliable quantitative results has not yet been ascertained. Supposing it does, every CHI 3 parts of iodoform correspond to C-JIeO parts of alcohol, because one of the two carbon groups in CH 3 .CH 2 OH becomes formic acid. Perhaps the following adaptation of Wanklyn and Chapman's Method of limited oxidation would give reliable results. After a preliminary distillation, boil an aliquot part of the distillate * This point must be determined beforehand, and quite directly. 292 EXERCISES IN APPLIED ANALYSIS. with a known sufficient volume of a standard solution of bichro- mate of potash and the corresponding quantity of standardized sulphuric acid (4H 2 S0 4 for !K 2 Cr 2 O 7 ) in a flask under an inverted condenser until all the alcohol can be assumed to be oxydized into acetic acid : C 2 H 6 + 20 = H 2 + C 2 H 4 2 . Hence 1K 2 O 2 7 -3x0 = 3/2C 2 H 6 O. In the resulting green liquid the residual chromic acid is deter- mined by adding a properly adjusted known excessive weight of standardized ferrous sulphate (see p. 34), and then titrating back with standard bichrome solution, using ferricyanide of potassium as a drop-indicator for ferrosum. From the quantity of chromate reduced, the alcohol is calculated according to the above equation. According to J. C. Thresh,* small quantities of alcohol con- tained in an aqueous liquor can be detected by subjecting it to distillation (if necessary), and then distilling the distillate with a little chromic acid. The distillate now formed contains alde- hyde, and consequently, when boiled for a time with caustic potash, and allowed to stand for a few hours, assumes a yellow colour. In this way as little as 0*02 per cent, of alcohol can be detected. SEPAEATION OF ETHYL AND METHYL ALCOHOL. The word " separation " in our heading must be taken in the sense of side-by-side determination, because a method of real separation is not known. It certainly cannot be effected by fractional distillation. Some years ago the Author had occa- sion to inquire into this matter, and for this purpose worked out an approximate indirect method of quantitative separation, which is based upon the facts that the iodides of the two alcohols can be separated fairly well by fractional distillation, and that methyl-iodide has a higher specific gravity than the ethyl compound. 20 cc. each of pure ethyl-alcohol, pure methyl- alcohol, and of the alcohol mixture under analysis are converted separately into iodides thus : * Pharm. Journ. and Trans. [3], ix., p. 408. Not having this journal at hand, we quote from Jahresbericht for 1878, p. 1074. SEPARATION OF ETHYL AND METHYL ALCOHOL. 293 20 cc. of the alcohol are poured on 30 grms. of iodine, contained in a small flask kept cool by a water-bath, and 4 grms. of red phosphorus are then added in small instalments. The mixture is distilled by means of a water-bath, the distillate washed, first with water (to avoid formation of iodoform), then with aqueous caustic soda, and then again with water, to be dried by addition of fused chloride of calcium, and distilled. The specific gravities of the three iodides are determined side by side of one another at the same temperature. In an analysis thus made the follow- ing results were obtained (by Mr. James Robson) : Pure CH 3 1. PureC 2 H 5 I. Mixture. S = Spec. grav. at 180,. . . 2*2882 1'9426 1'9975 1^-S, 0-43703 0-51477 0-50063 Percentages of CH 3 I=_p= 100 18'2 The percentage was calculated on the assumption that the two iodides mix without contraction or expansion, i.e., thus : 1 00 0-43703? + 0-51477(100 -p) = Q , . S of mixture. That assumption, it is true, is not proved to be correct, but we are pretty safe in presuming that the specific gravity of any mixture of the two iodides lies between those of the two com- ponents. Hence the result will always be right in a qualitative, and probably afford at least a rough approximation to the truth in the quantitative, sense. Besides, if we have to deal with a doubtful case, it is open to us to prepare the respective iodide in a large scale and fractionate it by distillation under a bead tower. By this refinement even a small admixture of methyl- alcohol to ethyl-alcohol can be detected with certainty and vice versa. The degree of ease with which the two iodides part on dis- tillation (as compared with the alcohols) may be seen from the following experiments : 20 cc. of iodide of ethyl, boiling at 73'3 to 73'4, and 1 cc. of iodide of methyl, boiling (under the same circumstances) at 42'5 to 43 C> 5, were mixed by Mr. Cullen and distilled from out of a Hernpel (bead-tower) apparatus. The boiling point ranged as follows : 294 EXERCISES IN APPLIED ANALYSIS. Volume over, in cc., 1 2*5 3 Almost all. Boiling point, ... 67 70 71 71 73 The methyl compound obviously accumulated in the earlier fractions, so that, supposing the experiment had been made on a larger scale, the specific gravity of, say, the first tenth of the distillate, would have left no doubt about the presence of methyl-alcohol at any rate. By way of contrast let us state shortly what we observed in the fractionation of a mixture of the two alcohols, which was carried out by Mr. Robson. 360 cc. of a high-class spirit of wine of 92*5 per cent, by weight were mixed with 40 cc. of pure methyl-alcohol from Kahlbaum in Berlin. 380 cc. of the mixture were distilled under a bead tower, and thus divided into five fractions : Fraction, I. II. III. IV. V. Boiling point,* 72 -76 76-4-76'9 76-9 75-77 77'3 Volume, 20 260 20 48 22'5 cc. Portions of Fractions I. and V. and of the original mixture (it was it whose iodide figured in the above statement beside the pure ethers) were made into iodides, and the specific gravities of all determined side by side with those of the two pure ethers. The percentages of CH 3 I in the respective iodides were as follows : Orig. Mixture. Fr. I. Fr. V. Percentage, 18'2 28'2 4*0 It is not worth while to reduce these numbers to percentages of methyl-alcohol in the corresponding liquids analysed. The numbers as they stand show that although the two alcohols when distilled together tend to separate in accordance with their boiling points, the actual separation does not amount to much ; even the first twentieth of distillate (as iodide) contained only 28, and the last twentieth still contained 4 per cent, of the methyl compound. The above experiences were gathered incidentally in the course of an experimental critique of a method for the detection of * Unconnected, SEPARATION OF ETHYL AND METHYL ALCOHOL. 295 methyl-alcohol in ethyl-alcohol, which was invented in 1876 by Riche and Bardy. Eiche and Bardy' s method (Comptes Rendus for 1876*) is as follows: 10 cc. of the alcohol, previously rectified over car- bonate of potash if necessary, are placed in a small distillation flask with 15 grms. of iodine, and 2 grins, of red phosphorus are added gradually, while the flask is kept immersed in a cold water-bath to prevent loss of alkyl-iodide.t The mixture is distilled by means of a water-bath, and the iodide collected under 30 cc. of water. The iodide is separated from the aqueous part and transferred to a small flask containing 5 cc. of aniline, with which it unites readily into hydriodates of alkyl-anilines. Should the reaction become too violent, it must be moderated by immersion of the flask into cold water, to be subsequently sup- ported by gentle heating. After an hour's standing the (solid) product is dissolved in water, any excess of either co-reagent boiled off, and the solution, after cooling, decomposed by addition of an excess of caustic potash or soda. The alkyl-aniline sepa- rates as an oil, which, if it does not rise by itself, can be made to rise by adding some solid caustic alkali to increase the specified gravity of the mother-liquor. By adding some water or alkali ley it is driven up into the neck of the flask, and allowed to clear up there. 1 cc. of the oil is incorporated with 10 grms. of a mixture made of 100 parts of sand, 2 of common salt, and 3 of nitrate of copper, the mixture placed in a test tube, and in it kept at 90C. for 8-10 hours by means of a water-bath. The product is lixiviated with instalments of warm alcohol so as to produce 100 cc. of filtrate. This latter is always intensely coloured ; but while pure ethyl-alcohol yields a dark brownish- red colour, free of all admixture of violet, if methyl-alcohol was present the tincture exhibits a more or less pronounced violet colour. With 2*5 per cent, of methyl-alcohol the shade of violet is very distinct. To detect minor quantities, 5 cc. of the tincture are diluted with water to 100 cc., and 5 cc. of this solution * Not having this periodical at hand (I am writing in the country), I quote from Allan, "Commercial Organic Analysis," 2nd ed., vol. i., p. 51. f Alkyl, a generic term for methyl, ethyl, &c. 296 EXERCISES IN APPLIED ANALYSIS. (corresponding to O25 cc. of tincture) are diluted with 400 cc. of water in a porcelain basin, to be heated to near boiling and used for the dyeing of a bit of white merino- wool cloth (about 10 by 10 ctms.), which must be free of sulphurous acid. The cloth is kept in the liquor at water-bath heat for about half an hour. It is then taken out, washed with water, and dried. If the alcohol under analysis was pure, the cloth remains white, but if it was contaminated with methyl-alcohol, the cloth exhibits a more or less intense violet colour. When the present writer, some years ago, applied this method to a number of samples of spirit of wine, suspected to be adulte- rated with small proportions of methylated spirit, he observed no violet tinge in the alcoholic tinctures of the pigments pro- duced, but the merino test gave, in all cases, distinctly positive results. This, in a sense, was satisfactory ; but when we pro- ceeded to test-experiments with spirits of undoubted genuineness and methylated alcohols prepared from such spirits by addition of measured proportions of pure methyl-alcohol, we obtained puzzling results. Even with the pure spirits the merino cloth in most cases (not always) exhibited a distinct colour similar to that yielded by mixtures containing about half a per cent, of added methyl-alcohol. With mixtures containing 2 per cent, of methyl-alcohol or more there was no difficulty ; the colours obtained from these could not have been misinterpreted. With the assistance of Messrs. John M'Arthur, Arch. Kling, William Cullen, and James Robson, I tried hard to discover the cause of the irregularities referred to, and found, as a matter of high probability at least, that they were owing to the presence in the spirits of traces of higher alcohols. Special experiments made with really pure aniline showed that they were not caused by any impurity in the commercially " pure " aniline used in the majority of the trials. The following description of the form of the method which we ultimately adopted shows indirectly how the difficulty was overcome : To test a (strong) spirit of wine for methyl-alcohol, convert at least 40 or 60 cc. (the more the better) into iodide by means of the proportions of iodine and red phosphorus recommended by Riche and Bardy (vide supra), but wash the iodide, first with SEPARATION OF ETHYL AND METHYL ALCOHOL. 297 water,* then with dilute caustic soda, and lastly again with water; dry it with granulated fused chloride of calcium, and re-distil it. For a preliminary test, measure off 4 to 6 cc. of this iodide into a small tared flask, weigh it, add for every gramme of iodide 0'58 cc.t of pure aniline, and heat the mixture under an inverted condenser by means of a water-bath for a few minutes ; then allow to stand for an hour in the cold. Dissolve the solid product in water, boil off any excess of aniline or iodide, allow to cool, eliminate the alkyl-aniline, oxidize 1 cc. of it, from the product produce 100 cc. of alcoholic tincture, and dilute 5 cc. of the tincture to 100 cc. with water. Run 5 cc. of this dilute solution into 400 cc. of boiling hot water contained in a Berlin basin, which place on a full-going water-bath. Immerse a square decimetre of previously washed white merino cloth in the hot liquor for live minutes ; then take out the cloth, add other 5 cc. of the dilute colour solution, mix well, re-place the cloth in it, and continue heating for half an hour. Then take out the cloth, wash it with water, and dry it. Of the various cases that may present themselves, only one needs be considered here. We will assume that the colour taken on by the merino is that feeble shade, which may or may not be owing to the presence of methyl-alcohol in the original spirit. Assuming as a hypothesis that it is, proceed to concentrate your "methyl" by a series of fractional distil- lations of the bulk of our alkyl-iodide. Supposing you had operated upon 50 cc. of spirit. In this case the yield of iodide will amount to some 40 cc., of which we will assume that 32 cc. are still at your disposal. Distil these out of a small Hempel apparatus, and collect the first 16 cc.; then distil these 16 cc. and collect the first 8 cc., and from these 8 cc. again extract the most volatile 4 cc. Convert these into alkyl-aniline, &c., &c., as above explained. If the colour obtained in the preliminary test really owed its origin to methyl, the proportion of methyl in these 4 cc. is now about the eight-fold of what it was in the * To prevent formation of iodoform. fA little more than the quantity demanded by eq. C2H 5 I + C6H 5 NH 2 = HI.N(C 6 H 5 ).(C2H5).H. Whether only mono-ethyl-aniline salt is produced is another question which we do not propose to consider. 298 EXERCISES IN APPLIED ANALYSIS. original iodide, and far more intense violet colour will come out. If it was owing to some irregularity, say in the process of oxida- tion or to some impurity in the aniline, the doubtful shade in the preliminary test will reappear. If it was owing to the presence of propyl, amyl, &c., the cloth will take on no colour at all except perhaps a faint shade of grey, with no violet tinge in it. A large number of critical experiments with genuine spirits and mixtures of such with known proportions of methyl- alcohol (some of the trials were carried out with 140 cc. of spirit) enables us to affirm that the second case only rarely presents itself. As a rule, a non-methylated spirit, in its thrice distilled iodide, yields no colour ; while a spirit containing as little as a fourth of a per cent, of methyl- alcohol furnishes a decidedly violet cloth. Wanklyn and Chapman's method of limited oxidation includes a number of methods for testing a spirit for an admixture of methyl-alcohol. The one that suggests itself most forcibly is based upon the (assumed) fact that, in their general process of oxidation, ethyl-alcohol yields only acetic acid ; while methyl- alcohol yields carbonic acid (and water) as final products. To test this method, Mr. M 'Arthur made a number of trials in the following manner : An oxidizing solution was prepared by dissolving 98'2 = l/3K 2 Cr 2 7 grms. of bichromate of potash in water, adding 210 grms. of a 62*3 per cent, vitriol = 130*7 grms. of real H 2 S0 4 , and diluting to 1 litre, to obtain a solution of which 1 cc. = 16 mgs. of available oxygen. (To be referred to as "bichrome mixture"). To test a spirit, 1 grm. of it was mixed with 60 cc. of bichrome mixture in a small flask provided with an inverted condenser, the upper end of which communicated with a bulbed U-tube charged a known volume of standard baryta water, and the mixture boiled until the oxidation was apparently completed. A slow current of carbonic-acid-free air served to sweep the stagnant carbonic acid into the barytic reagent. The carbonic acid was determined by titrating the excess of baryta left as Ba(OH) 2 . (See p. 229.) The following analyses may be quoted: (1.) 1 grm. of a high-class spiritus vini of 85 per cent, (by weight) yielded 10'3 mgs. of carbonic acid, SEPARATION OF ETHYL AND METHYL ALCOHOL. 299 (2.) 1 grm. of a mixture of 95 volumes of the same spirit, with 5 cc. of pure methyl-alcohol, gave 47'9 mgs. (3.) 250 cc. of a suspected spirit us vini was distilled under a bead tower and the first 50 cc. collected; these were again dis- tilled in the same manner and 10 cc. drawn over. Of these 10 cc. 1 grm. was operated on as before. The carbonic acid obtained amounted to 12'2 mgs. Of these analyses No. 1 is the most important for us here ; it clearly shows that even genuine spiritus vini yields a larger pro- portion of carbonic acid than could be accounted for by the presence of traces of (non-normal) higher alcohols. Hence the method is not available for the detection of small quantities of methyl-alcohol. Another method which readily suggests itself is to aim at the production of formic acid from the methyl-alcohol, and test for it by means of nitrate of silver. To test this method (which was recommended by J. T. Miller), Mr. M' Arthur mixed 3'5 cc. of the above spiritus vini with 25 cc. of bichrome mixture and 2 cc. of water, allowed to stand a quarter of an hour, and distilled off 25 cc. The distillate was alkalinized with carbonate of soda and next boiled down to 10 cc. to eliminate the aldehyde (C 2 H 4 0) produced. The residue was acidified with acetic acid, filtered, mixed with a solution of O'l grm. of nitrate of silver in 3 cc. of water, and boiled. Only a faint dark turbidity was produced. 3*5 cc. of the 5 per cent, methylated alcohol used for the above experiment (2), when treated in the same manner, yielded a strong precipitate of metallic silver. This method, as we see, is more reliable than the carbonic acid one. But after all, Wanklyn and Chapman's general process, it appears to us, should lend itself better for the Detection of Ethyl-alcohol in Methyl-alcohol than for the solution of the inverse problem. To detect an admixture of ethyl to methyl-alcohol, all that is required is to boil a known weight of the suspected alcohol with a sufficient excess of bichrome mixture under an inverted condenser, or, what is better, to keep the mixture of alcohol and reagent in 300 EXERCISES IN APPLIED ANALYSIS. a close vessel* at 100 for a sufficient time and distil the result- ing liquid to dryness, best from out of a retort immersed in a sperm-oil bath and connected with a Liebig's condenser. To prevent "bumping" put a few fragments of clay tobacco-pipe into the retort, or, what is better, let a slow current of air bubble through the distilling mixture. The residue is re-dis- solved in a little water and again distilled to dryness to recover a remnant of acetic acid. The united distillates are titrated with baryta water to ascertain the weight of acetic acid produced. For a check add a few drops of surplus baryta to the neutra- lized liquid, evaporate to dryness over a water-bath, re-dissolve in a little water, filter off the carbonate of baryta produced, evaporate the filtrate to dryness, dry the acetate at 110 until constant, and weigh it. Then determine the percentage of baryta by igniting the acetate and weighing the carbonate. We have not yet had occasion to try this method in real earnest. In applying it to a suspected methyl-alcohol one must not forget that acetone, if present, yields acetic acid as well as ethyl-alcohol does ; C 2 H 4 2 parts per CO(CH 3 ) 2 parts of the ketone ; as one of the two methyl groups burns into carbonic acid and water. Berthelot recommends to heat the suspected methyl-alcohol with concentrated sulphuric acid, as if one intended to prepare oxide of methyl gas from it. This ether is readily absorbed by ordinary oil of vitriol, which does not act very promptly on the ethylene produced from the ethyl-alcohol. The residual ethylene is identified as bromide C 2 H 4 Br 2 or by gas analysis. In regard to this method also we have no personal experience. The same remark applies to the methods which now follow. Riche and Bardy use a method which is founded upon the formation of aldehyde from the ethyl-alcohol. Its most charac- teristic feature, however, is the peculiar manner in which they identify the aldehyde. Special Reagents: (1) A solution of thiosulphate of soda of 1*29 specific gravity. (2) A solution of * A sealed-up tube, if not too full and at all carefully prepared, is not likely to burst. A "pressure bottle," meaning a bottle whose stopper is held down in some way or other, is less safe ; it must be tied up in a strong linen bag while being heated in its water-bath to avoid accidents. SEPARATION OF ETHYL AND METHYL ALCOHOL. 301 permanganate of potash of 1*028 specific gravity.* (3) A solu- tion of 0'02 grm. of rose-aniline in a litre of water. 4 cc. of the liquid under examination are mixed with 6 cc. of concentrated sulphuric acid and 10 cc. of water (with the view, we presume, of fixing part of the methyl-alcohol as methyl- sulphuric acid), and the mixture is distilled, 8 to 10 cc. being drawn over and run into 10 cc. of water. This liquid is mixed with 5 cc. of sulphuric acid and 10 cc. of the permanganate solution, and the mixture is allowed to stand for ten minutes. 4 cc. of the thiosulphate solution and 4 cc. of the rose-aniline solution are then added. In the absence of ethyl-alcohol the mixture is yellowish-white ; in its presence it becomes more or less intensely violet. Acetone, formic acid, and isopropyl-alcohol give no similar reaction. Upon the observation of A. W. Hofmann's that iodide of methyl is far more readily attacked by ammonia (with forma- tion of chiefly iodide of tetra-methyl-ammonium) than ethyl- iodide is, Tiemann some years ago based a method for the separation of the two ethers, or rather alcohol-radicals, which worked well in his hands. Hence, a good method for detecting methyl in ethyl-alcohol probably would be to first convert the given alcohol into iodide, and next to concentrate the ethyl in a small fraction of the whole by a series of fractional distillations. The concentrated iodide (say the least volatile 1/8 or 1/16) is treated in Tiemann's way,t i.e., digested with aqueous ammonia at the ordinary temperature for 6-8 hours. The methyl-iodide now has assumed, at least substantially, the form of solution of iodide of tetramethyl-ammonium, and is easily removed by sucking off the aqueous layer. The residual iodide is washed, dried, distilled, and identified as (pure or impure) C 2 H 5 I by its specific gravity and boiling point. * Why define the strengths of these solutions by their specific gravities, instead of simply stating the weight of dry reagent per litre ? f Kindly communicated to the Author by Professor Tiemann in a private letter. NOTES. NOTE (1) on Exs. 1 to 3. On the Theory of the balance. The study of this subject may (and, indeed, had better be) postponed until the student has become quite familiar with the instrument by its practical use. The following notes are intended to supplement what he may find in his text-books on mechanics or physics : Imagine a fine Oertling balance, in which the knife-edges, bear- ings, and suspension arrangements are ideally perfect. Imagine the left side to be charged with P grms. and the right side with P grms. likewise (including the weights of the pans, &c.), and it is clear at once that the dis-arrested beam can remain at rest only in one position, because the conjoint effect of the weight W of the beam and the charges P and P is the same as if they were all concentrated in a fixed point C, which lies below the axis of rotation in a line passing through this axis and standing perpendicular on the plane of the two end edges. If disturbed in that position, the beam vibrates about it as a pendulum. Imagine now a small overweight A to be added on, say, the right side ; the effect is that the common centre of gravity of the system shifts from C to a point C' lying a little to the right of C ; the balance, to gain what is now its position of rest, must turn through an angle a, which ccet. par. is the greater the greater A . If the three axes lie in exactly the same plane, the angle of deviation a is independent of the magnitude of the charges P and P, because these conjointly act like one point equal to P + P in the axis of rotation. If this axis lies above the plane of the end-edges, angle a for a given A is the less the greater P ; and if the plane of the end-edges lies above the axis of rotation, the angle a for a given A is the greater the greater P. As easily shown with the help of a little algebra where I is the arm-length of the balance (which we have tacitly assumed to be exactly the same on both sides), s the distance of the centre of gravity of the empty beam from the axis of rotation, and h the distance of the axis of rotation from the plane of the terminal edges. The + in the denominator corresponds to the case when the axis of rotation lies above the plane of the end-edges j the - to the opposite case. Taking J as designating the length of the needle, measured from the axis of rotation to the zero of the scale, in degrees of the scale (which runs horizontal and is divided into degrees of equal length, say, into milli- NOTES. 303 metres), and n the number of degrees through which the needle moves in describing the angle a, we have _!L = E"= U A ~ W s 2PA The value which E assumes when the milligramme is taken as the unit of weight we will call the "sensibility" (German Empfindlichkeit) of the balance. By means of the gravity-bob (without which no balance is complete) we can give s any value however small that we could reasonably care for ; hence we can give to E, i.e., to the deviation per milligramme of overcharge, any value we like ; we can render E so great as to make the deviation corresponding to even O'OOl mgs. distinctly visible ! In other words, there is no limit to the potential precision of a balance if it is absolutely perfect. But no balance is ; the knife-edges are no abso- lutely straight rigid lines, the bearings no absolutely rigid planes, &c., and as a necessary consequence the position of rest, instead of being absolutely constant, is anywhere within a small angle /3. This indifference-angle is the greater the greater the sensibility, but for a given charge (P, P) its weight-value TT is constant, and it is obviously of no use to increase the sensibility beyond the value which renders that angle /3 just visible and no more. With a really good balance it would, as a rule, be a mistake even to go so far, for a number of reasons, of which we will content ourselves with naming one which is perhaps less obvious to a beginner than some of the rest. The time of vibration t of a balance is the same as that of a mathe- matical pendulum, the length R of which equals the momentum inertise of the (charged) beam divided by its momentum staticum ; hence, if R is the length of the second pendulum, we have But the momentum of inertia of the charged beam is the momentum staticum stands before us in the denominator of the formula for E. Hence the time of vibration is the greater the heavier and the longer the beam, and it increases rapidly with the charge P. Any student who is at all familiar with this chapter in mechanics will easily see that we are right in concluding that for a given balance, charged with a given P on both sides, the time of vibration is propor- tional to the square root of the sensibility, or, in symbols, that f 2 = const. E. By increasing the sensibility to the 4, 9, 16 fold, we increase the time of vibration to the*2, 3, 4 fold. * k is a numerical factor which depends on the shape of the beam ; with the customary beam form it is very near 1/3. 304 NOTES. To the student who has gone through Exs. 1 to 3 with a fine balance constructed for heavy charges, the practical bearings of this need not be explained. If we screw down the bob so far that the time of vibration assumes the small value that would just suffice for exact readings of the excur- sions of the needle, the decimilligramme (as an overweight) ceases to be visible, even with the longest needle one could reasonably use. But this difficulty can be overcome easily by optical means. The Author some years ago* devised an arrangement, which was executed for him by Mr. Oertling, and has since done him excellent service. It consists of a narrow ivory scale, divided into very small degrees, which is fixed slantingly (at about the angle at which you hold a book for convenient reading) to a point near the lower end of the needle, and a (feebly magnifying) microscope fixed to the pillar of the balance, which microscope is provided with one vertical wire in its focus, and pervades the central (fixed) part of the front pane of the balance-case to enable one to read while the case is shut. The ordinary scale which does duty as usual, if required, is divided conveniently into millimetres, the micro-scale into degrees, each of which has the angular value of 0*1 degree of the principal scale. The apparent motion of the " wire " in the microscope on the micro-scale is in the same sense as that of the real motion of the needle on the ordinary scale below, which pro- vides against blunders in the . The microscope, to be more than an illusion, must be very steady, and for this reason be fixed to the pillar of the stand of the balance. It adds about 3 to the cost of the instrument. A cheaper arrangement is the following : The micro-scale, divided conveniently into fifth-millimetres, is fixed slantingly to the pillar, and the needle at the corresponding part is shaped thus Q^ ; a hair, by means of capillary perforations at a and 6, is stretched out between a and b so that it is parallel to the surface of the micro-scale, and at only 0*2 to 0*5 mm.'s distance from it. A (short) terrestrial telescope, fixed in the central (fixed) part of the front pane of the balance-case, serves for reading. The telescope needs not be very steady, as the hair is so close to the scale. The object-glass is a single lens, which, under the circumstances, produces an image within the tube which is rather less in size than the scale itself; the eye-piece is a feebly magnifying compound microscope, which magnifies this image as far as necessary for a distinct reading of tenths of degrees. I have applied this arrange- ment to one of my balances and found it to work very well ; but it is not equal to the one described before. Whoever uses the method of vibration in its exact form (Ex. 2) prefers, of course, to so place the centre of gravity of the beam that 1 of deviation corresponds to exactly 1, 0*5, O2 mgs., (fee., to avoid com- putation. The Author many years ago contrived for the purpose an auxiliary gravity-bob) which is fixed to the upper end of the needle by * Zeitschrift fiir Instrumentenkunde. February number of 1882. NOTES 305 mere friction so that it can easily be moved up and down. The respec- tive part of the needle has the shape of a triangular prism, which turns one of its sides towards the observer. This side is divided into milli- metres. With a hectogramme-balance the auxiliary bob needs not weigh more than 2-3 grms. Supposing the upper edge of the bob to be y mms. away from its highest position, and we have for the weight- value of 1 of deviation, i.e., for - - the equation rj where (E" 1 ),, and a are constants, easily determined by ascertaining the value E" 1 , first at y = 0, and then when y is at its maximum value. For further information, see the Author's Memoirs Ueber die Waage des Chemikers, Zeitsckrift fur Instrumentenkunde, Oct., 1881 ; also his article " Balance," in the new edition of the Encyclopaedia Britannica. NOTE (2) to Ex. 10, p. 13. The table referred to on the last line of p. 13 was intended originally to be nothing more than an extract from Kolb's well-known table ; but it subsequently struck me that it would be better to base it on deter- minations of our own. I accordingly caused Mr. Archibald Kling to carry out the following experiments : A perfectly pure hydrochloric acid of about 1*11 specific gravity was made from pure materials with great care, and put aside well-stoppered as " acid A." The exact percentage in A was determined by diluting a known weight with water to a weight so adjusted that the solution lent itself for the application of our gravimetric modification of Volhard's method. (See p. 220.) By means of it it was ascertained that acid A contained 22 '525 per cent, of HC1. From acid A two more dilute acids were derived by gravimetric synthesis, which were intended to contain about 20 per cent, and 18 per cent, of HC1 respectively ; but through a blunder in the calculation of the proportions of water to be added, the percentages came out as follows : In A, as stated, 22*525 In B, by synthesis, 18-172 In C, by synthesis, 16-983 and, unfortunately, the calculations of these two latter numbers were made only after all the specific gravity work had been completed. Nor could Mr. Kling have checked the percentages of B and C by direct analyses without delaying the publication of the book. But from Kolb's table it is quite evident that the relation between percentage p and specific gravity S is governed by an almost linear function ; besides, I know Mr. Kling's work to be sound, and his results for S and p fall in so well with the formulae which I deduced from them, that I have little hesitation in giving the following table based on his work, feeling sure X 306 NOTES. that it is, at the worst, more than at a par in point of precision with any of the tables given in the handbooks of chemistry. The specific gravities were determined by means of two test-tube- shaped specific gravity bottles, provided with capillary necks bearing etched-in (calibrated) millimetre-scales. The temperatures were estab- lished, and kept constant, by means of a large water-bath. The bottles had a capacity of about 20 cc., and 1 mm. of the scale corresponded to a little less than O'Ol cc. ; one fill of the bottle enabled one to deter- mine the specific gravity at three temperatures, namely, at 8-10, then 15, and lastly 20 (about), by combining the scale readings and the one weighing. After weighing the bottle and contents, the last temperature was re-established, the liquid in the neck brought down to near mm., the bottle and contents weighed again, and thus a fourth determination effected. For each acid two such series of observations were made, one with bottle I., the other with bottle II. From the observational results I calculated the following table by means of equations of the form (1) S t = S -a; where a = a b p (2) p = c + (S 16 -l)rf, which were found to do sufficient justice to the observations. All the specific gravities must be understood to be referred to water of 15C. as = 1. Reduction to the vacuum omitted. TABLE FOR FINDING THE PERCENTAGE p OF HC1 FROM THE SPECIFIC GRAVITY nK S AT 10 TO 20. 15815. p- A p. For A** = l, AS- 15815. p. &p. For A = l, AS = (O'OOOI X ) (0*0001 x ) 1-085 16*96 3' I'lOO 19^5 19 3-5 I -086 17-16 '20 3-0 I'lOI 20*04 19 3-6 1-087 iy35 I 9 3' I'I02 20*24 *20 3-6 I -088 I7-54 I 9 3'i I-I03 20-43 19 3-6 1-089 1773 I 9 3'i I'I04 20*62 19 3'7 1*090 i7'93 "20 3'2 IT05 20*81 19 3'7 "091 18-12 19 3'2 i"io6 21*01 20 3-8 092 18-31 19 3'2 1*107 2I'2O I 9 3'8 093 18-50 19 3'3 1-108 21-39 I 9 3-8 094 18-70 '20 3'3 1*109 21-58 I 9 3'9 *95 18-89 I 9 3'3 I'lIO 21-78 *20 3'9 096 19*08 I 9 3'4 I'm 21-97 19 3'9 097 19-27 I 9 3 '4 I'II2 22"l6 19 4-0 098 19-47 "20 3'5 1*113 22-35 I 9 4-0 099 19-66 19 3'5 1*114 22'55 '20 4'i NOTES. 307 To form an estimate of the degree of precision attained by Mr. Kling in his work, I deduced from each of his 24 determinations of specific gravities 15 S t , and the <'s, the percentage p of the respective acid by means of the table, and compared it with the p found by analysis of A, or synthesis from A. The results were as follows : Acid A. (p-p ) varies from -005 to -04; mean of the 8 values = -028, corresponding to AS = '00014. Acid B. (p-p ) varies from '02 to -10; mean of the 8 values = '049, corresponding to A S = "00024. Acid C. The individual values 'A/> were: +'01; +*03; -O4; - -05 ; + -06 ; - -12 ; - '16 ; - '21 ; general mean = -085, corresponding to A S = '00042. Series C obviously includes at least three bad determinations. As soon as I found this, I based my constants chiefly on series A and B, and I believe that the table (from 22 '5 to 18 per cent., and for each kind of acid from about 8 to 20) may be relied on as being correct, as a table for finding the p for a given 15 S t , to within 0'04 per cent, of HC1, and as a table for finding the S for a given p to within '0002. The Author's differential method of specific gravity determination may here be referred to as coming in useful occasionally in the preparation of standard solutions. Its requirements are : (1) A test-tube-shaped specific gravity bottle, which should not hold less than 30 cc., and must be provided with a well ground in, narrowly perforated, glass stopper. (2) A cylinder large enough for the bottle (1) being freely suspended within a liquid contained in the cylinder. (3) A precision balance so arranged that a thing to be weighed can be suspended from one of the pans below the table. (Regnault's arrangement). The cylinder is charged with liquid I. after this has been brought to the temperature of the balance-room; the specific gravity bottle, charged with the same liquid, is suspended within the mass of liquid in the cylinder by means of a fine platinum wire, hooked on to the balance-pan, and its tare taken in grammes as soon as it has assumed a constant value. The bottle is now taken out, emptied, charged to almost overflowing with liquid II., and next kept plunged into liquid I. up to the neck to assume the temperature of I. The stopper is then put on, the overflowing liquid wiped off quickly with filter paper, and the bottle suspended within I. to be weighed again, care being taken to wait until the balance gives constant readings, and to see that the temperature remains constant throughout. Supposing the bottle holds V grms. of water at t, and the second tare is, say, greater than the first by A, then, designating the two specific gravities ( t S t ) by S' and S", we have * O/ A = V with a degree of precision which is not easily reached by any other known method. 308 NOTES. NOTE (3) to Exs. 14 and 15. As shown by the Author,* the two double salts MgS0 4 .K 2 S0 4 -1- 6H 2 and FeS0 4 .K 2 S0 4 + 6H 2 0, when dissolved in at least hot water, suffer partial decomposition into the two component sulphates. At his request Messrs. James Robson and Andrew Hodge have tried, by means of syste- matic synthetical experiments, to ascertain the conditions under which a solution of the two sulphates (FeS0 4 and K 2 S0 4 , or MgS0 4 and K 2 S0 4 ) deposits pure crystals of the respective double salt on cooling. The case of the magnesia salt was taken in hand by Mr. Robson. According to his experiments, a solution of n x K 2 S0 4 grms. of sulphate of potash and n, or I'l n, or 1'2 n times MgS0 4 in a quantity of hot water, in- sufficient for holding the double salt in solution after cooling, when allowed to cool, deposits in general, besides a crop of double-salt crystals, more or less of a sandy mass of small crystals of sulphate of potash, which stick chiefly to the lower faces of the large crystals of the double salt. The proportion of solid sulphate of potash produced is the less the greater the excess of magnesia salt used. The following, it appears, is an infallible recipe for obtaining unmixed crystals of the double" salt MgK 2 S 2 8 + 6H 2 0: 87*15 grms. of pure powdered sulphate of potash and 153 '45 grms. of sulphate of magnesia (7H 2 salt) are dissolved in 350 grms. of hot water ; the solution is filtered into a perfectly clean flask, made up to 640 grms., and allowed to stand under a stopper of cotton wool thrust into the neck until the temperature has fallen to about 60C. without deposition of crystals. The liquor is then transferred to a glass basin, covered up and allowed to stand overnight. The next morning one finds a crop of transparent crystals, which are quite free of all visible admixture of free sulphate of potash. Two analyses of crystals pro- duced in this manner gave the following results : I. II. Mean. Theory. Water, 26-90 2678 26-84 26*82 Sulphuric Acid, 39-60 39*83 39-71 3975 Magnesia, 10'26 10-03 10*14 10-02 Potash, not determined. 23-40 100-00 Mr. Hodge's results with the iron salt are quite analogous. It takes about 1*25 to 1'3 times FeS0 4 per 1 K 2 S0 4 in a hot solution to obtain cooling crystals of the double salt which are not contaminated with free sulphate of potash. To obtain pure crystals dissolve 29 grms. of sul- phate of potash (powdered) arid 56 grms. of pure ferrous sulphate (FeS0 4 + 7H 2 0) in 100 grms. of hot water, acidulated with 0-5 cc. of 20 per cent, sulphuric acid to prevent precipitation of ferric compounds, and proceed otherwise as directed above in regard to the magnesia salt. The crop of crystals obtained is mechanically pure, and gives the right * Proceedings of the Royal Society of Edinburgh for 1886-7. NOTES. 309 percentage of iron on titration with permanganate. In the case of either double-salt, however, should a crust of suspicious-looking small crystals come out (beside normal looking ones), sift these off after drying. The case of the double sulphate of copper and ammonia (the subject of Ex. 19) still requires to be looked into. From our present incidental experience it appears that the salt made from a solution of the calculated proportions of the two sulphates as prescribed in the exercise is normal and unmixed. We may state on this occasion what Mr. Hodge has found out in experiments on potash alum. 100 grms. of pure potash alum were re-crystallized from hot water, the mother liquor evaporated down, allowed to deposit its crystals, the second mother liquor treated simi- larly and so on until the ultimate mother liquor was so far reduced that it just sufficed for duplicate determinations of the potash (by Finkener's method) and of the alumina (by precipitation with sulphide of ammonium). The ratio of potash to alumina was the same in the ultimate mother liquor as it was in the original alum. But this does not prove that the alum is not decomposed by water. A cold saturated solution was placed in a battery cell, which was immersed up to near its edge in pure water, and the latter renewed from day to day. After about a week's standing the liquor of the cell was analysed for potash and alumina, and the latter found to predominate largely over the former. This confirms an old experiment of Graham's, who showed that if alum solution diffuses into pure water (quite directly) the potash salt wanders up faster than the alumina salt does. NOTE (4) to Ex. 24, p. 63. In the determination of the carbonic acid in a bleaching powder, a mere bulb-tube charged with stannous chloride does not suffice for the absorption of the chlorine ; it must be followed by a \J-tube charged with pumice soaked in the stannous solution. NOTE (5) to Ex. 26, p. 74. The method for the separation of nickel and cobalt from iron given in the foot-note was introduced by Schwarzenfo/y. NOTE (6) to Ex. 34. Since the above section of the book was written, Mr. Frank Lyall has, at my request, carried out a number of experiments for seeing to what extent the Dirvell-Clark method is available as a means of bringing pure cobalt (given, say, as double nitrite of cobalticum and potassium) 310 NOTES. into a weighable form. To obtain pure metallic cobalt a supply of Fischer's salt was made into oxalate, the latter into metal, the metal into chloride, the chloride into purpureo-chloride. This last was re-crystallized and reduced in hydrogen to obtain pure metallic cobalt, and from this a standard solution of chloride was made synthetically. For each test analysis a known weight of standard solution was boiled down to have it as strong as possible, and expel the dissolved air which otherwise (after addition of ammonia) gives rise to the formation of cobaltic compounds.* To the boiling-hot mixture ammonia is added in successive instalments until the precipitate, which is first blue and amorphous, has assumed the characteristic crystalline form and violet colour of the double salt P0 4 CoNH 4 . This is allowed to settle (which it does very quickly), filtered off, and washed with small instalments of cold water, but no longer than necessary, as the precipitate is not quite insoluble in water. Three determinations were made, with the follow- ing results : Weight p of ignited Weight of metal Volume of Metal taken. precipitate. found Error. nitrate P 2 O 5 2CoO. =px 0-404. and washings. (1) -2285 -5635 -2277 - -0008 135 cc. (2) -0914 -2248 -0908 - -0006 95 cc. (3) -0485 -1188 -0481 - -0004 70 cc. The filtrates from (1) and (2) were mixed with sulphide of ammo- nium, heated, allowed to stand, and the (slight) precipitate collected, roasted, and weighed as CoSxOy. Its weight in the case of (1) was 1 "3 mgs. ; in the case of (2) it was 1 mg. From these experiments it appears that, although the method involves a slight loss, it may be used as a supplement to the nitrite method, namely, for bringing the cobalt of the " Fischer's salt " into a weighable form. All that is necessary is to dissolve the yellow salt in hot hydro- chloric acid, evaporate to dryness (to render any silica insoluble) to dissolve the residual chlorides in a little acidulated water, filter, and apply the phosphate process as explained. Special experiments showed that the proportion of phosphate recommended by Clark (five parts of solid P0 4 (NH 4 ) 2 H for one of cobalt metal) works well. NOTE (7). On Platinum Solution and Platinum Residues. The best material for the making of platinum solution is pure spongy platinum, which is to be had from Messrs. Johnson, Matthey & Co., but, of course, costs more than the ordinary metal. The customary mode of converting the metal into platinum chloride, i.e., chloroplatinic acid * Dr. Clark recommends to boil the solution with HC1 " to convert any meta- or pyro- into ortho-phosphate." There surely is no occasion for this, as the P 2 O r, is added in the ortho-form. NOTES. 311 PtCl 6 H 2 * solution is to heat with aqua regia (3 volumes of hydrochloric acid of 1-1 and 1 volume of nitric of 1-2) until the metal is dissolved. The solution is evaporated as far as possible on a water-bath, and then re-evaporated with added hydrochloric acid to destroy the remnant of nitric acid. The residue contains more or less of the nitroso-compound PtCl 6 (NO) 2 . To remove it the residue is taken up with water, and the solution re-evaporated; the N of the compound goes off as N 2 3 [PtCl 6 (NO) 2 + 2HO.H = 2NO.OH + PtCl 6 H 2 ]. But the N 2 3 by the action of the water is converted partly into nitric acid, which remains, and it reproduces nitroso-compound ; to destroy it the evaporation with hydrochloric acid must be repeated and supplemented by one with water. It is questionable whether all the nitroso-compound can be thus destroyed. Mr. M' Arthur and the Author, in the course of a research on the com- bining constant of platinum in the chloroplatinates,f came to adopt the following method : The platinum is placed in a large glass-stoppered colourless bottle (for 20 grms. of metal or less a litre bottle is of con- venient size), a quantity of hydrochloric acid of I'l is poured on it, and the bottle then filled with well washed chlorine gas by displacement, and closed. In the course of 1 224 hours the chlorine is mostly, if not all, absorbed with formation of PtCl 6 H 2 . The bottle is then refilled with chlorine and allowed to stand until the chlorine is gone, the bottle again refilled with the gas, and so on until all the platinum is dissolved. The "chlorine Kipps" referred to on page 137 came in useful here. With one of these at hand the method is not so troublesome as it appears at first sight. Of course it is not a method for making platinum solution ex temp.; but in a properly regulated laboratory the reagent rarely needs be made on short notice. The solution produced is evaporated on a water-bath, the residue is re-evaporated with water, and the chloro- platinic acid thus obtained dissolved in water and diluted to a con- venient volume, say to 20 cc. per gramme of metal, to produce " 5 per cent." solution. Pure chloroplatinic acid solution is intensely, but clear, yellow ; it is free of all tinge of brown. If there is such a tinge, this indicates the presence of iridium or platinous chloride. To make pure platinum out of broken-down crucibles, &c., the best laboratory method is Schneider's. It is as follows : The crude metal is dissolved in aqua regia (if any black powdery residue remains, which obstinately refuses to dissolve, it may be put aside as iridium), the solution evaporated as far as possible on a water-bath, and the residue re-dissolved in water. The solution is neutralized with pure carbonate of soda,| the carbonic acid driven off, and the solution (of PtCl 6 Na 2 ) * It is remarkable that to this day a considerable minority of chemists are under the impression that what goes as platinum chloride is PtCl.|. Even the most neutral solution preparable contains PtCl-i. EUC^, and the H2C12 cannot be removed by evaporation on a ivater-bath at any rate. f Communicated to the Roy. Soc., Ed., in the summer of 1887. J Trammsdorffs ' ' purissimum " crystals of Na 2 CO 3 + 10H 2 O, are very pure. 312 NOTES. mixed with a quantity of pure caustic soda, which had better be made expressly from pure crystals of carbonate in a nickel basin. The solu- tion is kept at a boiling heat for an hour or longer, and a strong alkaline reaction kept up by occasional additions of caustic soda. The effect is that, while the platinum retains the chloroplatinate form, the iridium and other foreign platinum metals are reduced to lower chlorides (Ir 2 Cl 6 ; PdCl 2 , fcc.), with formation of hypochlorite of soda, which latter must be reduced by addition of a little alcohol in the heat. (My impression is that this last operation might well be omitted ; but I have so far always carried it out, as it gives little trouble.) After the reduction of the hypochlorite, the liquid is acidified strongly with hydrochloric acid, which often leads to the formation of a green insoluble iridium compound, any precipitate filtered off, and the plati- num of the filtrate precipitated by adding pure powdered sal-ammoniac until the mother liquor consists of a saturated solution of this salt. After some hours' standing all the platinum is down as chloroplatinate of ammonia. This is filtered off, washed first with concentrated sal- ammoniac solution, then once or twice with small quantities of water, and lastly with alcohol. It is then dried and ignited cautiously in a porcelain crucible, best in small instalments, because the heating of a large porcelain crucible is an inconvenient operation. The residual spongy metal is placed in a basin and washed very thoroughly with, first hot water to extract any trace of fixed alkali chloride, then boiling hydrochloric acid (which always dissolves a trace of platinum), and lastly again with water. The residual metal is re-ignited, and is now ready to be used as pure metal. WORKING UP OF PLATINUM RESIDUES. These may be classified as shown in the following : 1. Spongy metal, as obtained in the Finkener method, or from PtCl 6 (NH 4 ) 2 , in nitrogen determinations. This kind may be allowed to take care of itself. We would only remind the student that small proportions of impurities, such as may well be neglected in an analysis, cannot be tolerated in a material intended to be used for making pure solution. 2. Solid chloroplatinates of potassium, ammonium, &c., as obtained analytically. These are reduced with carefully purified hydrogen either in the dry way or in the wet way, explained in Ex. 15, p. 28. The powdery metal obtained is carefully and exhaustively washed, and ignited in porcelain. 3. Alcoholic and Ethereal Filtrates. These are subjected to distilla- tion by means of a water-bath until all the ether and the bulk of the alcohol is over. The residual aqueous liquid is united with what there may be of 4. Aqueous Filtrates. These are best reduced with hydrogen in the wet way, &c. (See 2.) 5. Small quantities of platinum diffused throughout a large volume of aqueous liquid can be precipitated with zinc as platinum black, NOTES. 313 which is washed with acid and then water. As the product is liable to be contaminated with lead (from the zinc), it must be dissolved in aqua regia, the surplus acid expelled by evaporation, the residue re-dissolved, and the solution mixed with sulphuric acid to precipitate the lead if present. The PbS0 4 is allowed to settle, filtered off, and the platinum of the filtrate precipitated by means of sal-ammoniac, or as metal by hydrogen in the wet way. The following method of Mr. Tatlock's may be quoted here as coming in useful in many cases : Given an impure solution of chloroplatinic acid (i.e., a solution free of iridium, &c., but contaminated with ordinary metals), precipitate the platinum as such by adding caustic soda and alcohol, and boiling. Whether the platinum thus obtained is pure or not depends, of course, on the nature of the impurities present. The method, no doubt, is valuable as a first step in recovering platinum from impure solutions of all kinds. NOTE (8). Determination of the Sulphuric Add in Sea Water. (Forgotten to be inserted on p. 224 after section on "Chlorine.") Weigh out 100 cc. of the sea water, add 25 cc. of a chloride of barium solution containing 0'25 x BaCl. 2 gnus, per litre and 10 cc. of 20 per cent, hydrochloric acid. Heat the mixture on a water-bath for a time, and allow to stand over night ; decant the clear liquor through a small Swedish filter, previously washed with hydrochloric acid, wash the pre- cipitate, first by decanting filtration with hot water acidulated with hydrochloric acid, and then with plain hot water on the filter. Dry the precipitate, detach it from the filter, incinerate the latter by itself in a platinum spiral, add the ash to the precipitate, ignite in a small platinum crucible not too strongly or too long (see p. 112) and weigh. BaS0 4 x 0-34323 = S0 3 . To make quite sure of the purity of the reagents, make a blank analysis with 100 cc. of pure water and the exact volumes of reagents prescribed. If a (small) precipitate is formed on standing over night, collect it on a filter of the same kind as used in the analysis, incinerate the filter and contents in a tared platinum crucible, and deduct the weight of the ash from that of the sulphate of baryta precipitate obtained in the analysis as a correction. NOTE (9). The adjoining figure represents a new kind of glass two-way cock, which was introduced lately by Messrs. Greiner and Friedrich, of Stultzerbach, Germany. Its application in combustions and other analytical operations is obvious. INDEX. 53 276 275 279 256 289 Acetate method for separation of sesquioxides and protoxides Acetate of methyl, properties Acetone, properties determination Albumenoids in milk - Alcohol, absolute ; specific gravities atO-30 - - - - aqueous ; specific gravities (table) - - - 286-289 detection of traces - - 291 determination in liquors 284-290 separation from methyl-alcohol 292 detection in methyl -alcohol 299-301 Alkali, free and combined, in soap 261 Alkalies in silicates ... Alkalimetry Alkalinity of sea water Alumina, separation from iron Amagat, experiments on carbonic acid Ammonia in its salts Determination 43 do. Kreusler's apparatus 84 Ammonia in guano - - 252, 253 Ammonium, chloroplatinate ; man- ipulation 44 Antimonious acid, determination by iodine - 59 Antimony, separation from lead 135-140 Arsenic, determination in pyrites - 115 Arsenious acid, determination with iodine 58 determination as sulphide - 116 B Balance, theory of the - Dittmar's improvements 304-305 Barium, determination - - - 18 chloride ; analysis - - -17-2C 124 46 230 33 159 sulphate ; purification Bichrome method of iron deter- mination . . - - 37 Biniodide of phosphorus, preparation 27 Bleaching powder, analysis - - 60-6"3 Assay for oxygen - Bleaching powder, Duflos' gravi- metric method 62 Titrimetric methods - - 60-62 Gas-volumetric method - - 64 Soiling-point curve - - - 274 Brass, analysis - - - 101-104 Bromine, determination in sea water - - - - - 232 Bunsen's gas apparatus - - 179-193 Butter, analysis - - - 265-267 Calcium, separation from phos- phoric acid 30 determination in sea water - 224 separation from iron and manganese - - - - - 52 Carbon, determination In cast-iron /- - 238-240 In organic substances - 143-152 In steel - -. * - 247 Carbonates, analysis - - -47-52 Carbonic acid in sea water - 227-230 Cast-iron, analysis - - 238-248 Chloracetic acid, a reagent - - 105 Chloride of platinum, preparation 310 potassium, preparation - - 220 Chlorides, determination of chlorine gravimetric method - - 17 titrimetric method (Mohr's) - 219 do. (Volhard's) 21, 220 combination of gravimetric and titrimetric method - 21 Chlorination, method of metal analysis - - - 136-140 Chlorine, active ; in bleaching powder - - - -60-67 as chlorate and total in bleach 63 Chlorine, gas ; generation of - Chrome iron ore ; analysis - 128-135 do. assay - - 128 do. calculation of results - - - 132-135 Cobalt, determination as phosphate 309 separation from nickel - 106-111 separation of small quantities from iron (foot-note) - - 74 316 INDEX. Copper, determination as sulphide do. as sulpho- cyanate - determination by electrolysis do. by titrimetric methods By cyanide of potassium - By iodine - ... separation from nickel - PAGE 42 104- 100 117 75 104 do. from zinc 101-102,104 Copper-voltameter 97 Cupellation - - - 140-142 D Di-methyl-acetal, properties - - 276 Dittmar's apparatus for gas analysis - - 201-207 silver alloy (note) - - - 111 Doyere's gas-pipette - - - 199 Dumas' method of nitrogen deter- mination - - - -152 E Electrolysis, quantitative 91-101 Electrolytic apparatus - - 92-93 Ethylene, analysis - - - 213 Eudiometer, Bunsen's - - 179, 208 calibration of - - 180,211 Fat in milk 255 foreign in butter - - - 265 Fatty acid, in soap - - - 260 Felspar, analysis - - - 122-127 do. calculation- - 126 Ferricum, direct determination - 39 reduction to ferrosum - - 36 Ferrosum, standard sulphate of - 34 Filter-ashes, determination - - 16 Finkener's method of potash deter- mination - - 26, 225, 236 Fluoride of ammonium in silicate analysis - - - 125 Foreign fat in butter - - 265 Fulminating gas, apparatus for - 188 G Galena, analysis - extraction of silver from 119-121 - 120 - 95 Galvanometer ... Gas absorption, physical ; theory 168-172 Gas analysis Exercises - - - 208-218 Theory - - - 155-178 Bunsen's apparatus and methods - - - - 179 Gas analysis Lothar Meyer and Seubert's apparatus and method - 195 Dittmar's method s - 20 1 -207 Doyere's method by absorbents 199 Disgregation - - - 160 Extraction from waters 215-218 Proximate analysis - 166-168 Relation of weight and volume 162 Ultimate analysis - 173-178 combustible - Table giving combustion con - staiits per unit - - 178 Their analysis, necessary atten- uation before firing 190-198 Gasometry, volumetric; theory of 155-166 Gas-volumetric analysis, example- 64 German silver, analysis - 104-111 Glycerine, determination in soap - 262 Gold and silver alloys, analysis by cupellation - - - 140-142 Greiner and Friedrich's new stop- cock 313 Guano, analysis - - - 252-255 H Haen, De ; titrimetric determina- tion of copper 75 Hydriodic acid, preparation - - 278 Hydrochloric acids, specific gravity of aqueous - - - 305-307 standard - - - 13, 21 Hydrogen, determination in organic substances - - - 143-152 293 278 56 I Iodides of ethyl and methyl, their specific gravities, etc. - Iodide of phosphorus, preparation Iodine, pure ; preparation standard solution, preparation and applications - - 56-60 Iron, cast; analysis - - 238-248 gravimetric determination - 23 titrimetric determination 34-42 separation from alumina - 33 separation from manganese and calcium - - 52-55 traces in nickel, determination 105 Jacobsen's apparatus for extracting the gases from a water - 216 Kjeldahl's method of nitrogen de- termination - - 83-85 INDEX. 317 Knallgas apparatus - - 188 Kreusler's apparatus for Schlcesing's method - - - 88 for ammonia distillation - 84 Lead in galena, determination - 120 separation from other metals as sulphate - - 101, 104, 120 separation from antimony 135-140 Liebig's potash bulbs, manipula- tion of ... 50 Lime, determination in sea water - 224 separation from phosphoric acid 30 Limited oxidation method of analy- sis - - - - 298-300 Liquids, their combustion with oxide of copper - - - 149 volumetric apparatus for 10-12 determination of specific gravity 12 Lothar, Meyer, and Seubert's gas analysis apparatus - - 195 Lunge's vitrometer - 90 M Magnesia, determination - - 24 determination in sea-water - 224 Manganese, native oxides ; analysis 67-75 native oxides, their assay for active oxygen Bunsen's method , - - 70 Fresenius and Will's - - 68 Permanganate methods 69 determination of traces of nickel and cobalt - - 74 separation from iron and cal- cium - 52-55 titrimetric detenu ination Pattinson's method - - 245 Volhard's method - 244 Manometer, to test gas evolution apparatus 44 Marsh gas, analysis - - - 213 Measuring flask, graduation - - 11 Measurement of liquids by volume 10-12 Methyl-alcohol, determination of - 276 detection in spiritus vini 295-299 properties 275 aqueous, specific gravities and percentages (table) - 281-283 Milk, analysis - - - 255-258 do. calculation - - 258 N Nickel, electrolytic determination 100 separation from cobalt - 106-111 separation from much iron (foot-note) - 74 Nitrates and nitrites, their analysis By Schkt'sing's method - 85 By Walter Crum's method - 89 Nitrogen in organic substances, determination By Dumas' method - - 152 By Kjeldahl's method 83-85 By Varrentrapp and Will's method - - 79-83 Nitrometer, Lunge's - - 90 Organic analysis, ultimate - 143-154 Oxalate of iron, preparation and analysis - - - - 143 Oxalic acid in guano - - - 254 Permanganate method of iron de- termination - - - 34 Phosphate of lime, analysis - 30-32 Phosphoric acid, separation from lime 31 determination by molybdate- 31 titration with uranyl-acetate 249 Phosphorus in cast-iron - - 242 Pipettes, graduation of - - - 12 Platinum, preparation of solution and recovery from residues 310 Potash, determination by chloro- platinic acid Finkener's method 26, 225, 236 Tatlock's method - 25, 236 Fresenius' (foot note) - - 237 Potash, separation from iron - 23 Precipitates, weighing of, on tared filters 45 Pyrites, analysis - - - 111-118 R Rancidity of butter, determination 267 Reciprocals of 273 + t and 7 10 to 790 207 Reduced volumes scale for eudio- meter - --- 193-194 Reporting of analysis 19 Resin in soap .... 263 Rheostats .... 98-99 Riche & Bardy's test for methyl- alcohol - - - 295-298 test for ethyl-alcohol - 300 Rose's mode of manipulating heavy metallic sulphides - - 43 S Salts, total; in sea water - - 231 Schlcesing's method - - 85 318 INDEX. PAGE Sea water, analysis - - 219-235 composition - ... 234 chlorine and specific gravity - 235 Silica, amorphous ; separation from quartz, etc. 127 Silicates, analysis - - - 122-127 Silicon in cast-iron - - - 246 Silver, determination ; gravimetric 20 determination; titrimetric - 20 extraction from galena - - 120 and gold alloys, analysis by cupellation - - - 140-142 Slag iii cast-iron .... 247 Soap, analysis - - - 259-265 Spathic iron ore, analysis - 55-56 Specific gravity of liquids Determination of - 12-13, 274 do. change of standard 280, 289 do. the author's differen- tial method - - - 307 Specific gravity tables for Aqueous ethyl-alcohol 286-289 do. hydrochloric acid - 306 do. methyl-alcohol 281-283 Standard solutions, adjustment - 13 do. preparation of large supplies - - 223 Stas' pipette 229 Stassfurth salts, analysis - 235-238 Steel, analysis .... 247 Sugar in milk .... 257 Sulphate of baryta, purification - 112 - of copper and ammonia, analysis - - - 42-47 of iron and potash, analysis 22-24 do. do. preparation 308 - of magnesia and potash, analysis - - - 24-32 - of magnesia and potash, preparation - - - 308 Sulphuric acid, determination - 22 do. do. in sea water - ... 313 Sulphurous acid, determination with iodine 59 Sumach, assay of - - - - 267 Superphosphate of lime, analysis 248-252 PAGE 267 Tanning materials, assaying of Tatlock's method of potash, deter- mination - - - 25, 236 Tea, determination of tannin and theine- - - - 271 Thiosulphuric acid, determination by iodine - - - 59 Total bases in sea water, as sulphates 226 U Uranyl acetate, for phosphoric acid determinations - - - 249 Uric acid in guano - - - 254 V Voltameter (knallgas) - - - 188 (copper) ... 97 Volumetric apparatus for liquids 10-12 W Walter Crum's method for deter- mining nitrous and nitric acid 89 Water, determination by direct method - - - 76-79 Water gases, extraction - - 215 Weighing, routine method - - 1 method of vibration - - 3 absolute - - - 5 reduction to vacuum - - 8 of pre-determined quantities of liquids - - - - 10 Wood spirit, analysis - - 272-283 do. components - 275-276 do. determination of acetone - - - 279 do. examinatioiiforethyl- alcohol - - 299-301 Zinc, determination by electrolysis 102 separation from copper - 102-104 do. from iron, nickel, cobalt, manganese, by chlora- cetic acid - - - - 105 William Hodge & Co., Printers, Glasgow. OTHER WORKS BY THE SAME AUTHOR. I. Report on the Composition of Ocean Water: Part of "The Physics and Chemistry of the Voyage of H.M.S. Challenger." Part I. Printed for Her Majesty's Stationery Office. Sold by Adam and Charles Black, and Douglas & Foulis^ Edinburgh ; and others. 1884. II. A Manual of Qualitative Chemical Analysis : Douglas & Foulis, Edinburgh, 1876. TABLES belonging thereto, same publishers, 1877. HI. Analytical Chemistry : A Series of Laboratory Exercises, constituting an Elementary Course of Qualitative Analysis. W. & R. Chambers, Edinburgh ; 2nd edition ; 1886. IV. -Tables to facilitate Chemical Calculations: Printed for the Author by James Maclehose & Sons. Sold by Williams & Norgate, London. UNIVERSITY OF CALIFORNIA LIBRARY BERKELEY Return to desk from which borrowed. This book is DUE on the last date stamped below. 18 1947 MAY '948 NovlO'49?! JAN 10 REC'C -u OCT 29 1960 LD 21-100m-9,'47(A5702sl6)476 YC 21865 UNIVERSITY OF CALIFORNIA LIBRARY