LIBRARY OF THE UNIVERSITY OF CALIFORNIA. 94 J 98 gale ^Bicentennial publication? RESEARCH PAPERS FROM THE KENT CHEMICAL LABORATORY OF YALE UNIVERSITY gale ^Bicentennial publications With the approval of the President and Fellows of TTale University, a series of volumes has been prepared by a number of the Professors and In- structors, to be issued in connection with the Bicentennial Anniversary, as a partial indica- tion of the character of the studies in which the University teachers are engaged. This series of volumes is respectfully dedicated to of tt) RESEARCH PAPERS FROM THE KENT CHEMICAL LABORATORY OF YALE UNIVERSITY EDITED BY FRANK AUSTIN GOOCH Professor of Chemistry in Yale University VOLUME II. NEW YORK: CHARLES SCRIBNER'S SONS LONDON: EDWARD ARNOLD 1901 Copyright, 1901, BY YALE UNIVERSITY Published, June^ /go/ UNIVERSITY PRESS JOHN WILSON AND SON CAMBRIDGE, U. S. A. CONTENTS VOLUME II. PAGE I. The Determination of Tellurium by Precipitation as the Iodide. By F. A. GOOCH and W. C. MORGAN .... 1 II. On the Application of Certain Organic Acids to the Estimation of Vanadium. By PHILIP E. BROWNING and RICHARD J. GOODMAN 4 III. The Determination of Oxygen in Air and in Aqueous Solu- tion. By D. ALBERT KREIDER 11 IV. A Method for the Separation of Aluminum from Iron. By F. A. GOOCH and F. S. HAVENS 20 V. The Estimation of Molybdenum lodometrically. By F. A. GOOCH 27 VI. The Application of lodic Acid to the Analysis of Iodides. By F. A. GOOCH and C. F. WALKER 33 VII. The Action of Urea and Primary Amines on Maleic Anhy- dride. By FREDERICK L. DUNLAP and I. K. PHELPS . 42 VIII. The Separation of Aluminum and Beryllium by the Action of Hydrochloric Acid. By FRANKS S. HAVENS .... 47 IX. The Titration of Sodium Thiosulphate by lodic Acid. By CLAUDE F. WALKER 52 X. The Combustion of Organic Substances in the Wet Way. By I. K. PHELPS 62 XI. The Estimation of Manganese as the Sulphate and as the Oxides. By F. A. GOOCH and MARTHA AUSTIN ... 77 XII. On the Condition of Oxidation of Manganese precipitated by the Chlorate Process. By F. A. GOOCH and MARTHA AUSTIN 85 XIH. On the Estimation of Manganese separated as the Carbonate. By MARTHA AUSTIN 96 XIV. The Action of Carbon Dioxide on Soluble Borates. By Louis CLEVELAND JONES . 100 94198 viii CONTENTS PAGE XV. Further Separations of Aluminum by Hydrochloric Acid. By FRANKE STUART HAVENS 106 XVI. The lodometric Determination of Molybdenum. By F. A. GOOCH and JOHN T. NORTON, Jr Ill XVII. On the Determination of Manganese as the Pyrophos- phate. By F A. GOOCH and MARTHA AUSTIN . . 121 XVIII. On the Detection of Sulphides, Sulphates, Sulphites, and Thiosulphates in the presence of each other. By PHILIP E. BROWNING and ERNEST HOWE . . . . 134 XIX. On the Separation of Nickel and Cobalt by Hydrochloric Acid. By FRANKE STUART HAVENS 141 XX. The Ethers of Toluquinoneoxime and their bearing on the Space Isomerism of Nitrogen. By JOHN L. BRIDGE and WILLIAM CONGER MORGAN 145 XXI. The Application of Iodine in the Analysis of Alkalies and Acids. By CLAUDE F. WALKER and DAVID H. M. GILLESPIE 162 XXII. The Estimation of Boric Acid. By F. A. GOOCH and Louis CLEVELAND JONES 172 XXIH. A Volumetric Method for the Estimation of Boric Acid. By Louis CLEVELAND JONES 182 XXIV. The Constitution of the Ammonium Magnesium Phosphate of Analysis. By F. A. GOOCH and MARTHA AUSTIN 190 XXV. The Influence of Hydrochloric Acid in Titrations by So- dium Thiosulphate, with special reference to the Esti- mation of Selenious Acid. By JOHN T. NORTON, Jr. 206 XXVI. -The Volatilization of the Iron Chlorides in Analysis, and the Separation of the Oxides of Iron and Aluminum. By F. A. GOOCH and FRANKE STUART HAVENS . 215 XXVII. The Titration of Oxalic Acid by Potassium Permanganate in presence of Hydrochloric Acid. By F. A. GOOCH and C. A. PETERS 222 XXVIII. The Estimation of Iron in the Ferric State by Reduction with Sodium Thiosulphate and Titration with Iodine. By JOHN T. NORTON, Jr 230 XXIX. The Determination of Tellurous Acid in presence of Haloid Salts. By F. A. GOOCH and C. A. PETERS . 238 XXX. An lodometric Method for the Estimation of Boric Acid. By Louis CLEVELAND JONES 244 CONTENTS XXXI. The Double Ammonium Phosphates of Beryllium, Zinc, and Cadmium in Analysis. By MARTHA AUSTIN IX PAGE 252 XXXII. Separation of Iron from Chromium, Zirconium, and Beryllium, by the Action of Gaseous Hydrochloric Acid on the Oxides. By FRANKE STUART HAVENS and ARTHUR FITCH WAY 266 XXXin. The lodometric Determination of Gold. By F. A. GOOCH and FREDERICK H. MORLEY 269 XXXIV. The Action of Acetylene on the Oxides of Copper. By F. A. GOOCH and DEFOREST BALDWIN . . . . 276 XXXV. Notes on the Space Isomerism of the Toluquinoneoxime Ethers. By WILLIAM CONGER MORGAN .... 283 XXXVI. On the Volumetric Estimation of Cerium. By PHILIP E. BROWNING 289 XXXVII. On the Estimation of Thallium as the Chromate. By PHILIP E. BROWNING and GEORGE P. HUTCHINS 300 XXXVIII. The Ethics of Isonitrosoguaiacol in their relation to the Space Isomerism of Nitrogen. By JOHN L. BRIDGE and WILLIAM CONGER MORGAN 804 XXXIX. The Constitution of the Ammonium Magnesium Arseni- ate of Analysis. By MARTHA AUSTIN .... 309 XL. On the Estimation of Thallium as the Acid and Neutral Sulphates. By PHILIP E. BROWNING 317 XLI. The Separation and Determination of Mercury as Mer- curous Oxalate. By C. A. PETERS 320 XLII. The Titration of Mercury by Sodium Thiosulphate. By JOHN T. NORTON, Jr 328 XLIII. The lodometric Estimation of Arsenic Acid. By F. A. GOOCH and JULIA C. MORRIS 336 XLIV. On the Qualitative Separation of Nickel from Cobalt by the Action of Ammonium Hydroxide on the Ferri- cyanides. By PHILIP E. BROWNING and JOHN B. HARTWELL 344 XLV. The Volumetric Estimation of Copper as the Oxalate, with Separation from Cadmium, Arsenic, Tin, Iron, and Zinc. By CHARLES A. PETERS 347 XL VI. The Sulphocyanides of Copper and Silver in Gravi- metric Analysis. By R. G. VAN NAME . . . . 359 x CONTENTS PAGE XL VII. On the Estimation of Caesium and Rubidium as the Acid Sulphates, and of Potassium and Sodium as the Pyro- sulphates. By PHILIP E. BROWNING 368 XL VIII. The Estimation of Calcium, Strontium, and Barium as the Oxalates. By CHARLES A. PETERS 373 XLIX. The Action of Sodium Thiosulphate on Solutions of Me- tallic Salts at High Temperatures and Pressures. By JOHN T. NORTON, Jr. . 384 INDEX 395 RESEARCH PAPERS PROM THE KENT CHEMICAL LABORATORY OF YALE UNIVERSITY OF" THE UNIVERSITY OF RESEARCH PAPERS THE DETERMINATION OF TELLURIUM BY PRECIPITATION AS THE IODIDE. BY F. A. GOOCH AND W. C. MORGAN.* IT was known to Berzelius that hydriodic acid and tellurous acid interact with the formation of tellurium tetraiodide, which is converted by water into an oxyiodide and by excess of an alkaline iodide into a soluble double salt. Wheeler f has shown that the double salt which is formed when tellurous iodide is boiled in a strong solution of potassium iodide in dilute hydriodic acid is definite and has the constitution represented by the formula 2KI . TeI 4 . 2H 2 O. We have observed, however, that when potassium iodide is added to a cold solution of tellurous acid containing at least one-fourth of its volume of strong sulphuric acid, no tendency toward the formation of a double salt becomes apparent until the potassium iodide amounts to more than enough to convert all the tellurous acid present into the tetraiodide according to the equation, H 2 Te0 8 + 4H 2 S0 4 + 4KI = TeI 4 + 4KHS0 4 + 3H 2 0. The tellurium tetraiodide which is thus formed is extremely insoluble in sulphuric acid of the strength mentioned, though soluble in excess of potassium iodide, and acted upon by water with the formation of tellurium oxyiodide and hydriodic acid. It is produced at first in the condition of a finely divided dark brown precipitate which upon agitation of the liquid containing it gathers in curdy masses and settles, leaving the * From Am. Jour. Sci., ii, 271. t Am. Jour. Sci., xlv, 267. VOL. n. 1 2 DETERMINATION OF TELLURIUM BY supernatant liquid clear. By taking advantage of this tendency to curd it is possible to determine without great difficulty the exact point during the gradual addition of potassium iodide when the precipitation of the tellurium iodide is complete, and we have been able to found upon this property a very simple titrimetric method for the direct determination of small amounts of tellurium. In our test experiments we used tellurium dioxide prepared by oxidizing presumably pure tellurium with nitric acid and igniting the residue at a low red heat. Weighed amounts of the oxide thus prepared were dissolved in Erlenmeyer beakers in a very little of a strong solution of potassium hydroxide, and dilute sulphuric acid was added carefully until the tellu- rous acid which was precipitated upon the neutralization of the alkaline hydroxide was just redissolved. To this solution sulphuric acid of half-strength was added in such amount that the solution finally obtained, after adding the aqueous solution of potassium iodide subsequently, should still contain at least one-fourth of its volume of strong sulphuric acid. The Erlenmeyer beaker was placed upon a pane of window glass supported upon strips of wood about 1 cm. above the level of the work table, which was covered with white paper. A solution of approximately decinormal potassium iodide free from iodate and carefully standardized in terms of iodine by a method described in a former paper from this laboratory * was introduced gradually from a burette into the middle of the Erlenmeyer beaker. As the drops of the potassium iodide touched the liquid the precipitation formed at the centre and travelled in rings toward the outer walls of the beaker. When the liquid became so opaque that the effect of the potassium iodide was distinguished with difficulty, the beaker was rotated and the curded precipitate permitted to settle, and then the process of titration was continued as before. We experimented with amounts of tellurium dioxide varying from approximately 0.025 grm. to 0.1 grm., the latter quantity being as large as can be handled with accuracy without intermediate removal of the * Am. Jour. Sci., rxxix, 188. Volume I, p. 1. PRECIPITATION AS THE IODIDE. 3 precipitate by filtration. With an Erlenmeyer beaker 10 cm. in diameter across the bottom and a final volume of liquid amounting to not more than 100 cm 3 , we were able to follow the precipitation most easily. The results of a series of determinations made according to the method described and recorded in the following table are closely accordant, and, in close agreement with the theory of the process if the atomic weight of the tellurium which we used is taken as 127. We feel justified in taking this number as the atomic weight of our tellurium, because the mean result of twelve oxidations by standard potassium permanganate of tellurium dioxide, prepared similarly to that which we used and from the same lot of material, and the mean result of twelve reductions by hydrobromic acid of the telluric acid thus produced,* point to this figure. Final volume. Strongest H 2 S0 4 Iodine value of KI used. TeOj taken. TeO, found. Error. present. cm 8 cm* grm. gTm. grm. grm. 50 17 0.0706 0.0223 0.0221 0.0002- 50 17 0.0764 0.0244 0.0239 0.0005- 50 17 0.1591 0.0496 0.0499 0.0003+ 60 17 0.1655 0.0517 0.0519 0.0002+ 60 17 0.1578 0.0498 0.0494 0.0004- 80 30 0.1591 0.0498 0.0499 0.0001+ 100 30 0.3179 0.1001 0.0997 0.0004- 100 30 0.3186 0.1008 0.0999 0.0009- 100 30 0.3208 0.1011 0.1005 0.0006- 100 30 0.3208 0.1010 0.1005 0.0005- From these results it is obvious that the method, which is very rapid, is accurate. * Am. Jour. Sci., xlviii, 377, 378. Volume I, pp. 279, 281. n ON THE APPLICATION OF CERTAIN ORGANIC ACIDS TO THE ESTIMATION OF VANADIUM. BY PHILIP E. BROWNING AND RICHARD J. GOODMAN * IN a former paperf by one of us a method for the determi- nation of vanadium was described in which tartaric acid was used to reduce vanadic acid to the condition of the tetroxide. The method may be briefly outlined as follows : Measured and weighed portions of a solution of ammonium vanadate, the standard of which had been determined by the evaporation and ignition of definite portions, were treated with tartaric acid in excess and boiled, when the appearance of the deep blue color indicated the reduction to the condition of the tetroxide. After cooling, the solution was neutralized with potassium bicarbonate and a moderate excess of that reagent added. To the alkaline solution an excess of a standard solution of iodine was added and the whole allowed to stand about one hour, when no further bleaching of the iodine was noticed. The excess of iodine was then destroyed with a standard solution of arsenious oxide, starch was added, and the blue color obtained with a few drops of the iodine solution. The total amount of iodine used, less the amount equivalent to the arsenious oxide solution used, is the amount necessary to oxidize the vanadium from the condition of the tetroxide to that of the pentoxide, from which, according to the following equation, can be calculated the amount of vanadium present : V 2 4 + I-I + H 2 = V 2 6 + 2HI. * From Am. Jour. ScL, ii, 355. t Zeitschr. anorg. Chem., vii, 158. t These determinations are best made in small Erlenmeyer beakers, closed with paraffin-coated corks while standing with iodine. ESTIMATION OF VANADIUM. The work to be described in this paper is in part an application of the work described in the paper above mentioned to a series of determinations of vanadium in the presence of molybdenum and tungsten. The solution of vanadium used was one of ammonium vanadate, and the standard was determined by evaporating and igniting, in the presence of a drop of nitric acid, measured and weighed portions, the mean of closely agreeing results being taken as the standard. Our first series of determinations was by the method previously described, that being the natural starting-point for the work contemplated. The results follow in the table : Bxp. V,0 6 taken. V 2 O 8 found. Error. Tartaric acid. grm. grm. grm. grm. (1) 0.1621 0.1618 0.0003- 2 2 0.1620 0.1624 0.0004+ 2 (3) 0.1614 0.1622 0.0008+ 2 (4) 0.1619 0.1606 0.0013- 1 (5) 0.1604 0.1597 0.0007- 2 (6) 0.1618 0.0615 00003- 3 (7) 0.1298 0.1305 0.0007+ 1 (8) 0.1294 0.1297 0.0003+ 1 (9) 0.1618 0.1618 O.OOOOi 2 (10) 0.2588 0.2575 0.0013 3 (11) 0.2722 0.2726 0.0004+ 2 (12) 0.3273 0.3269 0.0004- 2 We next tried the effect of treating a solution of sodium tungstate in the same manner. We found that after the boiling with tartaric acid, neutralizing, adding iodine and allowing to stand as before, the amount of free iodine present, as shown by the amount of arsenious oxide solution necessary to destroy it was the same as that originally added, showing that no reduction had taken place. Accordingly a series of determinations of vanadium in the presence of tungsten was made which is recorded in the next table. The results show that vanadium may be easily determined by this method in the presence of tungsten without any evi- dent interfering action on the part of the latter element. When the same method of treatment was applied in the presence of molybdenum in the form of ammonium molybdate, APPLICATION OF CERTAIN ORGANIC ACIDS Exp. V 2 O 6 taken. V 2 O 5 found. Error. Sodium tuiigstate. Tartaric acid. grm. grm. grm. grm. grm. (1) 0.1618 0.1615 0.0003- 1 3 (2) 0.1615 0.1606 0.0009- 1 3 (3) 0.1618 0.1624 0.0006+ 1 3 w 0.1619 0.1624 0.0005+ 1 3 w 0.1627 0.1623 0.0004- 1 3 (6) 0.1621 0.1624 0.0003+ 1 4 (7) 0.2587 0.2574 0.0013- 1 4 (8 0.2587 0.2689 0.0002+ 1 4 the majority of the determinations gave large plus errors, and a few experiments made with the molybdate alone seemed to show a noticeable reduction of the molybdic acid. In the following table the results are tabulated. In experiments (3), (4) and (5) the mixtures were not boiled with tartaric acid, but warmed on a steam bath, with, however, no very apparent prevention of the reducing action. E cp. V 2 O 6 taken. V,O 6 found. Error. Ammonium molybdate. Tartaric acid. grm. grm. grm. grm. grm. 1) 0.1620 0.1790 0.0170+ 2 2) 0.1624 0.1619 0.0005- 2 3 0.1294 0.1416 0.0122+ 2 4) 0.1296 0.1361 0.0065+ 2 5) 0.1291 0.1312 0.0021+ 2 6) 0.1293 0.1324 0.0031+ 2 7 0.1636 0.1760 0.0124+ 2 8) 0.1640 0.1724 0.0084+ 2 [9) 0.1622 0.1624 0.0002+ 3 (] 0) 0.1622 0.1632 0.0010+ 3 (1 1) 0.1619 0.1879 0.0260+ 3 (1 2) 0.1292 0.1360 0.0068+ 1 3 (1 3) 0.1860 0.1917 0.0057+ 1 3 1 4) 0.3274 0.3733 0.0459+ 1 4 (1 6) 0.2324 0.2383 0.0059+ 1 4 We next tried the action of tartaric acid upon the vanadium solution in the cold and found that the reduction could be carried on to completion under these conditions if the tartaric acid was in large excess, the time sufficient, and the volume of the solution small. The following series was made to deter- mine these points : TO THE ESTIMATION OF VANADIUM. Exp. V 2 O 5 taken. V 2 O S found. Error. Time. Tartaric acid. Total vol. grm. grm. gnu. days. grm. cms (1) 0.1646 0.1649 0.0003+ 1 4 25 (2) 0.1640 0.1606 0.0034- 1 4 65 3) 0.1293 0.1264 0.0029 2 3 55 4) 0.1633 0.1628 0.0005- 2 4 65 5) 0.1293 0.1288 0.0005- 3 2.5 60 6) 0.1298 0.1299 0.0001+ 3 2.5 50 7) 0.1295 0.1279 0.0016- 3 3 55 8) 0.1617 0.1597 0.0020- 4 2 70 9) 0.1623 0.1622 0.0001- 4 3 80 Solutions containing sodium tungstate and ammonium molybdate were allowed to stand from one to four days with varying amounts of tartaric acid without giving any evidence of reduction. In the series which follows may be seen the results of a Exp. V 2 5 taken. V 2 5 found. Error. Ammonium molybdate. Sodium tung- state. Time in days. Total volume. Tartaric acid. grm. grin. grin. grm. grm. cms grm. (1) 0.1552 0.1558 0.0006+ 1 25 5 (2) 0.1289 0.1301 0.0012-f < 1 25 5 (3) 0.2583 0.2587 0.0004+ m 1 50 5 (4) 0.1293 0.1299 0.0006+ 1 1 25 6 (5) 0.2582 0.2591 0.0009+ 1 t 1 50 6 (6) 0.2582 0.2588 0.0006+ 1 1 50 5 (7) 0.1297 0.1308 0.0011+ 1 1 25 5 (8) 0.1291 0.1289 0.0002- 1 1 25 6 (9) 0.2582 0.2568 0.0014- e 1 1 50 5 (10) 0.1293 0.1299 0.0006+ i' 1 1 25 8 (11) 0.2582 0.2579 0.0003- i 1 1 50 5 (12) 0.1550 0.1538 0.0012- . t 2 25 5 (13) 0.1556 0.1545 0.0011- e f 2 25 5 (14) 0.1289 0.1296 0.0007+ t f 2 25 5 (15) 0.1549 0.1527 0.0022- 0.5 2 25 5 (16) 0.1553 0.1548 0.0005- 1 m 2 25 5 (17) 0.1556 0.1554 0.0002- 1 2 25 5 (18) 0.1293 0.1310 0.0017+ 1 > 2 25 6 (19) 0.1295 0.1299 0.0004+ . y 2 25 6 (20) 0.1293 0.1289 0.0004- r i 2 25 7 (21) 0.1293 0.1301 0.0008+ ( 3 25 5 (22) 0.1289 0.1299 0.0010+ 0.5 t 3 25 5 (23) 0.1293 0.1292 0.0001- 1 f 3 25 7 (24) 0.1556 0.1567 0.0011+ 1 t 3 30 5 (25) 0.1291 0.1289 0.0002- 1 i 3 25 7 (26) 0.1550 0.1557 0.0007+ . ^ 4 25 5 (27) 0.1554 0.1557 0.0003+ 1 t 4 25 5 (28) 0.1556 0.1557 0.0001+ 0.5 4 25 6 8 APPLICATION OF CERTAIN ORGANIC ACIDS number of determinations of vanadium in the presence of molybdenum and tungsten made in the cold and allowed to stand from one* to four days. It will be noticed that the results on standing one day with five grams of tartaric acid are for the most part satisfactory, and an increase in the length of tune does not cause any apparent reduction of the molybdenum. Friedheim f has shown that vanadium is reduced from the condition of the pentoxide to that of the tetroxide by boiling with oxalic acid. The reduction is so complete that he has developed a method for the estimation of vanadium upon this reaction and shows that it may be applied in the presence of molybdenum and tungsten, the acids of these elements not being reduced by the oxalic acid. When the vanadic acid is reduced the oxalic acid is oxidized and a definite amount of carbon dioxide evolved according to the reaction. V 2 6 + H 2 C 2 4 = V 2 4 + H 2 + 2C0 2 . This carbon dioxide Friedheim conducts by an appropriate form of apparatus into potassium hydroxide and weighs. From this weight the amount of vanadic acid originally present may be readily calculated. We have applied the method of oxidation with standard iodine described in the tartaric acid process to the residue after boiling with oxalic acid. The method of treatment was identical with that outlined at the beginning of this paper. The results which follow in the table are for the most part satisfactory and the method is certainly more easily applied than Friedheim's process, the potassium hydroxide absorption apparatus being unnecessary. Having applied successfully both tartaric acid and oxalic acid in the manner described, the action of citric acid applied in the same manner suggested itself as a fitting conclusion to the study of the action of this class of organic acids. In this * Some of the determinations designated in the table as having stood one day in reality stood only about fifteen hours, from 6 p. M. to 9 A. M. t Zeitschr. anorg. Chem., i, 312. TO THE ESTIMATION OF VANADIUM. Eip. V,O 6 taken. V,O 6 found. Error. Oxalic acid. Ammonium molybdate. Sodium tungstate. gnu. gnu. gnu. gnu. gnu. gnu. (1) 0.1806 0.1803 0.0003- 1 2 0.1950 0.1955 0.0005+ 1 3) 0.1959 0.1955 0.0004- 1 , (4) 0.1950 0.1959 0.0009+ 1 t (5) 0.1954 0.1977 0.0023+ 1 t (6) 0.1956 0.1960 0.0004+ 1 m (7) 0.1956 0.1964 0.0008+ 1 (8) 0.1956 0.1957 0.0001+ 1 (9) 0.3900 0.3899 0.0001- 2 (10) 0.3897 0.3917 0.0020+ 2 t (11) 0.3903 0.3905 0.0002+ 2 (12) 0.1954 0.1959 0.0005+ 2 Y (13) 0.1957 0.1960 0.0003+ 2 i (14) 0.1954 0.1961 0.0007+ 2 i (15) 0.1806 0.1818 0.0012+ 3 e (16) 0.1807 0.1827 0.0020+ 3 (17) 0.1809 0.1803 0.0006- 3 Y (18) 0.1956 0.1961 0.0005+ 3 t 1 (19) 0.3611 0.3617 0.0006+ 5 , (20) 0.3616 0.3626 0.0010+ 5 i case, as in the others, the reduction of the vanadic acid is easily and quickly effected, but the oxidation with the iodine is slower than in the presence of alkaline oxalates and tar- trates. In the case of oxalic acid on standing about fifteen minutes with the excess of iodine, when tartaric acid has been used, the bleaching of the iodine continues from thirty to forty-five minutes, but in the presence of the alkaline citrate the time required is fully an hour. A large excess of tartaric or oxalic acids does not seem to materially affect the results, but hi the use of citric acid it is advisable to avoid a large excess, which tends to give high results. Accordingly in the following series of experiments it will be noticed that the amounts of citric acid do not exceed two grams except where ammonium molybdate or sodium tungstate are present, when the ammonium or sodium base combines with part of the free acid. The results follow, on p. 10. The mode of proceeding in the estimation of vanadium by the use of either tartaric, oxalic, or citric acid may be briefly summarized as follows : To a solution of a vanadate with or without a tungstate or molybdate, add approximately one 10 ESTIMATION OF VANADIUM. E *I V,O 4 taken. V 2 O 6 found. Error. Citric acid. Ammonium molybdate. Sodium tungstate. gnn. grin. gmi. gnn. grm. gnn. (1 i 0.1956 0.1956 0.0000 1 2 0.3905 0.3921 0.0016+ 2 3 0.1960 0.1960 0.0000 1 4 0.1953 0.1960 0.0007+ 1 5 0.2088 0.2082 0.0006- 2 6 0.2100 0.2098 0.0002- 2 0.2092 0.2107 0.0015+ 1 8 i 0.2092 0.2107 0.0015+ 2 9 i 0.2096 0.2082 0.0014- 2 0.5 (1 i 0.2099 0.2116 0.0017+ 3 0.5 1 0.2095 0.2101 0.0006+ 2 6.5 (1 2 0.2099 0.2095 0.0004- 3 .0.6 gram of the acid for every tenth of a gram of substance to be determined. Heat the solution to boiling, except in case tar- taric acid be present with molybdic acid, when digestion for from fifteen to twenty-four hours in the cold should be substi- tuted. To the cold liquid add about five grams of potassium bicarbonate for every gram of acid used. Add iodine in slight excess and set aside until no further bleaching is noticeable. Destroy the excess of iodine with arsenious oxide solution, add starch, and titrate back with standard iodine. The total amount of iodine used less the equivalent of the arsenious oxide is the measure of the oxidation. We have found it of advantage, when starting with a new solution of the vanadate, to make one determination roughly and to get from this rough determination the proportions of acid and iodine to be used in the determinations to follow. Large amounts of the acid and a large excess of the iodine have been employed in many determinations without any apparent unfavorable effect upon the results. The tendency, however, under these circumstances is toward plus errors, which may be avoided by following the above directions. Ill THE DETERMINATION OF OXYGEN IN AIR AND IN AQUEOUS SOLUTION. BY D. ALBERT KREIDER.* WHILE there is little to be hoped for by way of improve- ment in the accuracy of present known methods for the deter- mination of oxygen in the air, some choice as to manipulation may nevertheless be desirable, and a process which is not lim- ited wholly to the methods and apparatus of ordinary gas analysis will doubtless often be found serviceable. The very satisfactory results which I have obtained in the determination of perchlorates by the action of the liberated oxygen upon hydriodic acid through the medium of nitric oxidef has led me to test this action upon the oxygen of the air, where only the smaller amount of oxygen and its greater dilution with nitrogen might be expected to be unfavorable. However, with the apparatus and manipulation herein de- scribed it will be seen that the method affords a means for the determination of the oxygen of the air or of dissolved oxygen with ease and rapidity and with sufficient accuracy for all practical purposes. The method in brief is simply the conducting of a known volume of air through a strong solution of hydriodic acid in the presence of nitric oxide ; subsequently neutralizing the acid with potassium bicarbonate and titrating the liberated iodine with standard decinormal arsenic solution from which the equivalent volume of oxygen is readily calculated. By several simple devices, to be described, all calculations may be done away with and the percentage of oxygen seen imme- * From Am. Jour. Sci., ii, 361. t Am. Jour. Sci., 1, 287. Volume I, p. 316. 12 THE DETERMINATION OF OXYGEN diately by the volume of arsenic solution required for the titration. The volume of oxygen found by means of the arsenic solution is, of course, under the standard conditions of tem- perature and pressure (0 and 760 mm.), and it is therefore essential either to calculate this volume into that which it would occupy under the conditions of the experiment or to reduce to the standard conditions of temperature and pressure the volume of air taken. The latter plan is the more satisfactory since by Lunge's ingenious device* the reduction can be readily effected without any calculation and independently of changing temperature and pressure. For my purpose the following arrangement of two burettes answered admirably. One burette graduated to 120 cm 3 contained over mercury the same volume of moist air which 100 cm 3 of air at and 760 mm. would occupy under the given conditions, this stand- ard being very carefully determined. By means of a T-tube this standard burette was placed between and in connection with the burette in which the volume of air to be analyzed was measured, and a movable reservoir of mercury. Both burettes were firmly fastened to a movable iron rod and the zero marks accurately adjusted to the same level. By draw- ing into the measuring burette a volume of air greater than that required for which purpose a small bulb was attached to the lower end of the burette, and then by raising the reser- voir of mercury, compressing the air in the standard tube to the 100 cm 3 mark, at the same time allowing the excess of air to escape from the measuring burette, exactly 100 cm 3 of air under the standard conditions of temperature and pressure was obtained. To facilitate the adjustment, two strips of wood were fastened to the rubber connection by means of screw pinch-cocks in such a way that by closing one pinch- cock the flow of mercury from the reservoir could be shut off, and then by gradually tightening the other pinch-cock the mercury would be forced out of the rubber, and thus an easy and accurate adjustment to the 100 cm 3 mark be secured. * Zeitschr. angew. Chem., 1890, 139. IN AIR AND IN AQUEOUS SOLUTION. 13 The apparatus in which the action of the oxygen upon hydriodic acid was effected consisted of a 300 cm 3 bulb pipette, both ends of which were cut off short and sealed to glass stop-cocks. The tube from one of the stop-cocks was cut off short after being tapered and constricted so as to hold a rubber connector tightly, while the tube from the other stop-cock was left sufficiently long to reach to the bottom of a 500 cm 8 Erlenmeyer beaker. These tubes are preferably of about 3 mm. bore, since for the several connections all air may be expelled from tubes of this size by displacement with water. In order to expel all air from the flask, instead of passing a current of carbon dioxide as was done in the determination of perchlorates, tune was saved by first filling the flask with water, which was then displaced by pure carbon dioxide (prepared as described below) and the flask subsequently exhausted, which was accomplished instantaneously by the device described in the article on perchlorates. The required amounts of potassium iodide solution, hydrochloric acid and nitric oxide were drawn in in the order named, after which the measured volume of air was gradually admitted Awhile the bulb was constantly agitated so as to keep the hydriodic acid continually renewed along the surface of the bulb. The shaking was continued for a minute or two until the action was completed, when a dilute solution of potassium bicarbonate was admitted. The carbon dioxide liberated forces the liquid from the bulb into a beaker which contains bicarbonate in amount sufficient, as previously determined, to neutralize all the acid taken. When the exit is too slow more bicarbonate may be admitted through the other stop-cock, and after neutralization has been completed the bulb may be washed out without any danger from the admission of air. All the water employed, both for the solution of potassium iodide and for the various connections, was free of oxygen. It was prepared by filling a three-liter flask with distilled water and boiling until the volume of the liquid was reduced about one-third, when the flask was closed by a doubly perforated rubber stopper and fitted as a wash bottle. By 14 THE DETERMINATION OF OXYGEN means of the tube which reached below the surface of the water, pure carbon dioxide was passed through while the water was still boiling, which together with the escaping steam was sure to expel all oxygen. Then the heat was removed and the current of carbon dioxide continued until the boiling ceased, when the escape tube was closed by a piece of rubber tubing and screw pinch-cock. As the water cooled it was well shaken while still in connection with the carbon dioxide generator, and thus became saturated with the gas, which was then pumped in under considerable pressure by the little hand pump described in a previous paper from this laboratory. By this means the water could be drawn as needed without the introduction of any air. The escape tube was provided with a rubber tube and screw pinch-cock, and a long, slender nozzle which could be in- serted into the tubes of the absorption apparatus. A bottle thus charged sufficed for all the determinations and required only an occasional supply of carbon dioxide when large draughts of water were required for making the potassium iodide solution. The potassium iodide solution was made up to contain one gram of the salt in thirty cubic centimeters of water, and was contained in convenient form in an ordinary wide- mouthed bottle fitted as a wash bottle, and graduated ap- proximately for each thirty cubic centimeters' volume the amount usually taken. The potassium iodide was weighed into the bottle, which was then closed and all air expelled by a current of carbon dioxide, when the desired amount of water, free of oxygen, was drawn in, and attachment again made with the carbon dioxide generator. After allowing the gas to pass for several minutes the exit was closed and the gas pumped in by the little hand pump. Inasmuch as this solution, when it was used, was drawn into an exhausted bulb, the bottle could be emptied without ever exposing its contents to the air. Nitric oxide was generated very satisfactorily according to Professor Gooch's method by the action of nitric acid IN AIR AND IN AQUEOUS SOLUTION. 15 upon globules of copper in a Kipp generator. When the nitric acid is diluted with an equal volume of water the evolution of the gas is sufficiently rapid without the applica- tion of heat, but contamination by the higher oxide is more likely. However, since it is necessary, in order to be certain of purity, to pass the gas through an acidified solution of potassium iodide before applying it to the determination of oxygen, whatever higher oxide may be present will be reduced. By passing the gas, as it issued from the generator, through a set of Geisler bulbs containing an acidified solution of potassium iodide and washing with potassium iodide solu- tion, the perfectly purified gas was obtained. Theoretically, only a small amount of the nitric oxide is required for the transference of the oxygen to the hydriodic acid, but when too little is taken the action is very slow. On the other hand, too large an amount relieves the vacuum to such an extent as to interfere with the introduction of the air. A little device to measure the volume of gas taken was there- fore attached to the generator. It consisted of a tube filled with water and roughly graduated for every five cubic centi- meters, so attached to the generator that the gas would enter by displacement of the water, which would descend to a lower bulb, and as the gas was withdrawn the water would again take its place. Fifteen cubic centimeters of the gas was found a convenient and satisfactory amount for the analysis. Carbon dioxide was generated in a Kipp generator, the acid and marble of which had been previously boiled and contained a little cuprous chloride. To remove a trace of reducing matter which the gas was found to contain, it was first passed through a solution of iodine and washed with potassium iodide. For the titration a decinormal solution of arsenious oxide (4.95 grms. to the liter) was employed : 1 cubic centimeter being equal to 0.559846 cm 3 of oxygen at and 760 mm. when the weight of a liter of oxygen at and 760 mm. is taken as 1.42895 grin. When the volume of air taken 16 THE DETERMINATION OF OXYGEN is 100 cm 3 under standard conditions of temperature and pressure, as obtained by Lunge's device, the following table, calculated for the volume of oxygen equivalent to the volume of arsenic solution, shows directly the percentage of oxygen corresponding to the reading of the burette. The correction necessary for the fraction of a tenth of a cubic centimeter of the arsenic solution is obtained with sufficient accuracy by simply multiplying by 0.005. RELATION OF ARSENIC TO OXYGEN. Correction for SKA. Oxygen equivalent at and 760 mm. 0.01 cms AsjjO,, cms 37.0 cm 3 20.714 0.005 37.1 20.770 37.2 20.826 37.3 20.882 37.4 20.938 37.5 20.994 37.6 21.050 37.7 21.106 37.8 21.162 37.9 21.218 38.0 21.274 Table I shows the results obtained in a series of determina- tions. Experiments (1) to (11) inclusive were made upon por- tions of air collected over water on March 28, measured in an ordinary gas burette and reduced to the standard conditions of temperature and pressure. The remainder of the deter- minations were made upon air collected on April 8, each portion having been measured in the apparatus described, for the reduction to standard conditions. No correction was found necessary for the blank determina- tions, since when boiled water was used the solution was only faintly colored with iodine, which requires only a drop of arsenic solution to bleach it. As is evident from the table, the determinations according to this method are not reliable beyond 0.05 per cent, but for practical purposes this is sufficiently accurate. For the sake of comparison two determinations by IN AIR AND IN AQUEOUS SOLUTION. TABLE I. 17 Kxp. Volume of Air reduced to and 760 mm. l>" required. Volume of Oxygen found at and 760 mm. Per cent of Oxygen in Air. cm 3 cms cnjS (1) 91.18 34.06 19.07 20.91 (2) 91.73 34.47 19.30 21.04 (3) 90.84 34.25 19.17 21.11 (4) 90.60 34.20 19.16 21.13 (5) 86.06 32.55 18.22 21.17 (6) 85.96 32.40 18.14 21.10 (7) 86.49 32.53 18.21 21.06 (8) 87.85 33.00 18.47 21.03 (9) 44.17 16.60 9.29 21.04 (10) 44.11 16.70 9.35 21.19 (11) 44.54 16.80 9.41 21.12 (12) 100.00 37.44 20.96 20.96 (13) 100.00 37.54 21.01 21.01 (14) 100.00 37.50 20.99 20.99 (15) 100.00 37.57 21.03 21.03 (16) 100.00 37.47 20.97 20.97 (17) 100.00 37.50 20.99 20.99 the pyrogallic acid method were made upon a portion of the same air used in the last experiments, the results being 20.93 and 20.88 per cent respectively. While the pyrogallic acid method is capable of much greater accuracy when applied in Hempel's improved apparatus, in ordinary burettes it will probably not yield more closely agreeing results than the above method. Determination of dissolved Oxygen. A deter- mination of oxygen dissolved in water can be completed by the above method in about ten minutes by means of the apparatus illustrated by the accompanying figure. The apparatus consisted of a flask of about 300 cm 3 capacity, into the bottom of which was sealed a stop-cock with a long exit tube. Upon the neck was cut the fiducial circle c and immediately above this stop-cock e was sealed as shown. The neck of the flask was drawn out and sealed to stop-cock d and the bulb, a, of about 30 cm 3 capacity blown in it. The capacity of the apparatus be- VOL. II. 2 FIG. 17. 18 THE DETERMINATION OF OXYGEN tween stop-cock, &, and the fiducial mark, c, was carefully determined. The manipulation for the determination of dissolved oxygen was as follows : The flask was held in the position shown by a clamp fastened to a movable support. Stop-cock b being closed, the water was admitted through e and the air allowed to escape through d until the level of water was that indicated by the line /. (When the water to be examined is not saturated with air, the flask must first be filled with carbon dioxide and the water entered by replacement of that gas.) With d closed, sufficient water was allowed to escape through b to bring the surface to e, which was then closed. The nitric oxide generator was then attached to d, and by opening b the gas was allowed to replace the water until the meniscus coincided with c, when d was closed and the generator disconnected. Two cubic centimeters of strong hydrochloric were introduced through e by expelling nitric oxide through d, in which a drop of water formed an effective trap to prevent the entrance of air. Then the potassium iodide was admitted in the same way. The solution of iodide for this purpose was free of oxygen and contained one gram in three cubic centimeters. It was kept under pressure of carbon dioxide in the bottle previously described, and by means of a long nozzle could be conducted to the bottom of eh and thus be admitted with but momentary and slight contact with the air. The tube eh contained approximately three cubic centimeters. With all the stop-cocks closed, the flask was inverted several times and thoroughly shaken, at the same tune washing out the ends of the stop-cocks with distilled water. After again placing the apparatus in its position, enough potassium bicar- bonate solution was admitted through e to expel all the nitric oxide through d\ the bulb, a, holding sufficient of the bicarbonate to neutralize all the acid taken. The bicarbonate being heavier quickly diffuses through the contents of the flask and neutralizes the acid; d and e are kept closed for a minute with b open so as to allow sufficient of the liquid to escape into a beaker containing some bicarbonate to provide IN AIR AND IN AQUEOUS SOLUTION. 19 space for the carbon dioxide evolved. Then the flask is washed out and its contents titrated with arsenic. The bleaching, by the aid of starch for the final reaction, can be accurately read to a single drop Usually the reading was verified by adding a drop of $ iodine solution, which produced the characteristic color. Table II gives the results of a series of determinations. TABLE IL Volume of Water taken. Temperature. As 2 3 required. Volume of Oxygen dissolved in lOOOcmS of water at 760 mm. cm* C. cm 8 cm 3 314.63 20 3.42 6.04 314.63 20 3.45 6.09 314.63 20 3.40 6.00 314.63 20 3.41 6.02 314.63 20 3.43 6.05 314.63 20 3.40 6.00 314.63 20 3.36 5.93 314.63 20 3.40 6.00 314.63 20 3.40 6.00 314.63 20 3.50 6.18 314.63 20 3.38 5.96 314.63 20 3.40 6.00 The mean of these determinations gives 6.022 cm 3 of oxygen as the amount dissolved in distilled water at 20 C. and 760 mm., and while some of the determinations vary considerably from this mean, as a whole they are fairly accordant. This method, moreover, is applicable to carbonated water. IV A METHOD FOR THE SEPARATION OF ALUMINUM FROM IRON. BY F. A. GOOCH AND F. S. HAVENS.* OF the well-known methods for the separation of aluminum from iron by the action, for example, of an alkaline hydroxide in aqueous solution or by fusion of the mixed oxides in potassium or sodium hydroxide ; by reduction of the iron oxide to the metal by heating in hydrogen, with the subsequent solution of the metallic iron in hydrochloric acid; by boiling the nearly neutral solution of the salts of aluminum and iron with sodium thiosulphate either with or without sodium phosphate ; by acting with hydrogen sulphide or ammonium sulphide upon solutions of the salts containing also an ammoniacal citrate or tartrate no single process can be said to be ideal as regards directness, rapidity and accuracy of working. We have deemed it not superfluous, therefore, to attempt the utilization of a reaction which should apparently be capable of effecting directly and quickly the separation of aluminum from iron under conditions easily attainable. It is known t that the hydrous aluminum chloride A1C1 3 .6H 2 O is very slightly soluble in strong hydrochloric acid, while ferric chloride, on the other hand, is extremely soluble in that medium. It is this difference of relation of which we wished to take advantage. It appeared at the outset that crude aluminum chloride could be freed from every trace of a ferric salt by dissolving it in the least possible amount of water, saturating the cooled solution with gaseous hydrochloric acid, filtering upon asbestos * From Am. Jour. Sci., ii, 416. t Gladysz, Ber. Dtsch. chem. Ges., xvi, 447. SEPARATION OF ALUMINUM FROM IRON. 21 in a filtering crucible or cone, and washing the crystalline precipitate with the strongest hydrochloric acid. Aluminum chloride prepared in this way gave no trace of color when dissolved in water and tested with potassium sulphocyanide. The correlative question as to how much aluminum chloride goes into solution under the conditions was settled by taking a portion of the pure aluminum chloride, dissolving it in a very little water, diluting the solution with strong hydrochloric acid, saturating the cooled liquid with the gaseous acid, filtering on asbestos, precipitating by ammonia the aluminum salt in the nitrate and weighing the ignited oxide. From 10 cm 3 of such a filtrate we obtained in two deter- minations 0.0022 grm. and 0.0024 grm. of the oxide, the mean of which corresponds to 23 parts of the oxide or 109 parts of the hydrous chloride in 100,000 parts of the strong hydrochloric acid. This degree of solubility, though inconsiderable when the objective point is the preparation of the pure salt of aluminum, is obviously incompatible with the attainment of quantitative accuracy in the retention of the aluminum. We have found, however, that various mixtures of anhydrous ether and the strongest hydrochloric acid can be used satisfactorily as solvents for the iron chloride, while the aluminum chloride is insoluble to a very high degree in a mixture of hydrochloric acid and ether taken in equal parts and thoroughly saturated with gaseous hydrochloric acid at the atmospheric temperature. We found that 50 cm 3 of the solution of aluminum chloride, obtained by mixing about 0.1 grm. of the hydrous chloride (dissolved in 2 cm 3 of water) with the mixture of pure, specially prepared aqueous hydrochloric acid and ether in equal parts and again saturating the liquid at 15 C. with gaseous hydrochloric acid, left upon evaporation and ignition 0.0004 grm. in each of two experiments results which indicate a maximum solubility corresponding to 1 part of the oxide or approximately 5 parts of the chloride in 125,000 parts of the equal mixture of ether and aqueous hydrochloric acid of full strength. Pure aqueous hydrochloric acid of full strength mixes 22 A METHOD FOR THE SEPARATION perfectly with its own volume of anhydrous ether, but it is a curious fact that the addition to this mixture of any very considerable amounts of a solution of ferric chloride in strong hydrochloric acid determines the separation of a greenish oily ethereal solution of the ferric salt upon the surface of the acid. The addition of more aqueous acid does not change the conditions essentially, but more ether renders the acid and the oily solution completely miscible. The ferric chloride seems to abstract ether from the ether-acid mixture and, then dissolved in the ether, remains to some extent immiscible with the aqueous acid thus left until the addition of more ether restores to the mixture that which was taken from it by the ferric chloride. Our experiments show that, while for the separation of insoluble aluminum chloride from certain small amounts of soluble ferric chloride the mixture of the strongest aqueous hydrochloric acid and ether in equal parts serves a most excellent purpose, when larger amounts of ferric chloride are to be dissolved ether must be added proportionately in order to prevent the separation of the ethereal solution of ferric chloride from the rest of the liquid. Great care was taken to insure the purity of the aluminum chloride used in the test experiments. The so-called pure chloride of commerce was dissolved in the least possible amount of water and this solution was treated with a large volume of strong hydrochloric acid. The chloride thus obtained, free from iron, but possibly contaminated (as we found by experience) with some alkaline chloride, was dis- solved in water and converted by ammonia to the form of the hydroxide, which was thoroughly washed and dissolved in hot hydrochloric acid of half-strength. From this solution, after cooling, gaseous hydrochloric acid precipitated the hydrous chloride in pure condition. The chloride thus prepared was dissolved in water and the strength of the solution was determined by precipitating the hydroxide from definite portions, and weighing the ignited oxide in the usual manner. In the experiments recorded in Table I, measured portions of the standardized solution were submitted to the treatment OF ALUMINUM FROM IRON. TABLE I. 23 Bxp. A1 2 3 taken in solution as the chloride. found. Final volume. Error. grin. gnn. cm grm. 1) 0.0761 0.0746 50 0.0015- 2 0.0761 0.0745 50 0.0016- 3) 0.0761 0.0741 50 0.0020- 4) 0.0761 0.0734 50 0.0027- 5) 0.0761 0.0756 50 0.0005- 6) 0.0157 0.0149 45 0.0008- 7 0.0157 0.0147 40 0.0010- 8) 0.0157 0.0144 45 0.0013- (9) 0.0480 0.0481 30 0.00014- (10) 0.0960 0.0957 30 0.0003- with hydrochloric acid and ether. The essential thing in the process is to have at the end a mixture of the strongest aqueous hydrochloric acid with an equal volume of anhydrous ether saturated at a temperature of about 15 C. The most con- venient way to secure these conditions seems to be to mix the aqueous solution of the aluminum salt with a suitable volume of the strongest aqueous hydrochloric acid enough to make the entire volume something between 15 and 25 cm 3 to saturate this mixture with gaseous hydrochloric acid while the liquid is kept cool by immersing the receptacle containing it in a current of running water, to intermix a volume of ether equal to the volume of the liquid, and finally, to treat the ethereal mixture once more with the gaseous acid to insure saturation. The precipitated crystalline chloride was collected upon asbestos in a perforated crucible, washed with a previously prepared mixture of hydrochloric acid and ether carefully saturated with the gaseous acid at 15 C., and either ignited after careful drying at 150 or redissolved in water, converted to the hydroxide by ammonia in the usual way and weighed as the oxide after filtration, washing, and ignition. In experi- ments (1) to (4) the precipitated chloride was ignited directly ; in experiment (5) the ignition was made with great care in an atmosphere of superheated steam ; and in experiments (6) to (10) the chloride was dissolved, precipitated as the hydroxide, and weighed as the oxide. 24 A METHOD FOR THE SEPARATION The experiments in which the chloride was converted to the hydroxide before ignition show upon the average an absolute loss of about 0.0006 grm. : the single experiment in which the ignition took place in steam shows about the same loss 0.0005 grm. ; while in those experiments in which the chloride was dried and then ignited directly, the average loss amounts to about 0.0020 grm. The error of the process which involves the precipitation of the aluminum as the hydroxide, falls within reasonable limits, but it is plain that the direct ignition of the chloride is liable to error, which may possibly be explicable as a mechanical loss occasioned by the too rapid evolution of the hydrochloric acid and water of crystallization, or, possibly, as the result of a very slight volatilization of the aluminum still holding chlorine in spite of the decomposing action of the water upon the chloride. In either case, it would seem to be reasonable to suppose that a layer of some easily volatilizable oxidizer placed upon the aluminum chloride might serve to obviate the difficulty in the one case, by serving as a screen to diminish mechanical transportation of the non-volatile material ; and in the other, by acting as an agent to promote the exchange of chlorine for oxygen on the part of the aluminum chloride. We have tried, therefore, the expedient of covering the aluminum chloride before ignition with a layer of mercuric oxide, which of itself left no appreciable residue when it volatilized. The hydrous chloride was collected as usual upon the asbestos in a perforated crucible, dried for a half-hour at 150 C., covered with about 1 grm. of the pure mercuric oxide, gently heated with great care under a suitable ventilating flue, and finally ignited over the blast. The results are given below (see Table II). It is obvious, therefore, that the precipitation of the crystal- line hydrous aluminum chloride from solutions of the pure salt is perfectly feasible and very complete, when effected by aqueous hydrochloric acid and ether thoroughly saturated with the gaseous acid and kept cool ; and that the conversion of the chloride into the weighable form of the oxide is best effected OF ALUMINUM FROM IRON. 25 TABLE II. Exp. Al,0,, taken in solution as the chloride. Al a O 3 found by ignition with HgO. Final volume. Error. grm. grm. cm 3 grm. (1) 0.0761 0.0758 25 0.0003- 0.0761 0.0754 25 0.0007- (3) 0.0761 0.0761 25 0.0010- by ignition under a layer of mercuric oxide, or by dissolving it in water and precipitating it as the hydroxide to be afterward washed, dried, and ignited. Of the two methods the former is by far the more convenient. The precipitation of the aluminum chloride in pure condition from solutions containing ferric chloride ought not, it would seem, to present any difficulty, providing only that the precaution is taken to have present a sufficient excess of ether. The question was put to the test of experiment with the results recorded in Table III. TABLE III. Exp. A1,O 3 taken in solution as the chloride. A1 5 O 3 found by ignition with HgO. Fe 2 3 present as chloride. Final volume. Error. grin. grm. grm. cm 3 grm. (1) 0.0761 0.0757 0.15 25-30 0.0004- 2 0.0761 0.0756 0.15 25-30 0.0005- 3 0.0761 0.0755 0.15 25-30 0.0006- 4) 0.0761 0.0755 0.15 25-30 0.0006- Measured portions of the standardized solution of aluminum chloride were evaporated nearly to dryness in a platinum dish, an amount of pure ferric chloride equivalent to about 0.15 grm. of the oxide was added in a very little water, 15 cm 3 of the mixture of strong hydrochloric acid and ether in equal parts were introduced, the liquid was saturated at 15 C. with gaseous hydrochloric acid (the dish being held in a convenient device for cooling it by running water), 5 cm 3 more of ether were added to secure complete miscibility of the solutions, and 26 SEPARATION OF ALUMINUM FROM IRON. more gas passed to perfect saturation. The aluminum chloride was collected upon asbestos in a perforated crucible, washed with a mixture of ether and aqueous hydrochloric acid thoroughly saturated with the gaseous acid, dried at 150 C. for a half -hour, covered with 1 grm. of pure mercuric oxide, and ignited at first gently and finally over the blast. The results show plainly a very satisfactory limit of error. THE ESTIMATION OF MOLYBDENUM IODOMETRICALLY. BY F. A. GOOCH * IN a former paper from this laboratory f several modes of applying hydriodic acid to the reduction of molybdic acid were studied. It was found, first, that the digestion process of Mauro and Danesi f is of very limited applicability, owing to the fact that the reaction of reduction is reversible. Secondly, it appeared that the use of the same reaction by Friedheim and Euler in a distillation process, so arranged that the iodine set free in the reduction might be caught in the distillate and titrated to serve as the measure of the reducing action, was not sufficiently regular because of in- attention to minor details. It was shown that by taking care to adjust the conditions constant results might be obtained. Thirdly, the fact was developed that by simply boiling the solution under well defined conditions in an ordinary Erlenmeyer flask, partly closed by a simple trap, the reduction of the molybdic acid proceeded regularly, and that the addition of standard iodine to the solution made alkaline with sodium bicarbonate served to restore the original condition of oxidation of the molybdic acid. The results of this treatment were shown to be accurate. In a recent paper || Friedheim has seen fit to make our modifications of the distillation process the subject of attack. Friedheim's comments upon the third method discussed (as * From Am. Jour. Sci., iii, 237. t Gooch and Fairbanks, Am. Jour. Sci., ii, 157. Volume I, p. 375. J Zeitschr. anal. Chem.. xx, 507. Ber. Dtsch. chem. Ges., xxviii, 2066. || Ber. Dtsch. chem. Ges., xxix, 2981. 28 THE ESTIMATION OF MOLYBDENUM. well as upon a subsequent application of the process) * are evidently prompted wholly by personal opinion and demand no further attention. With reference to Friedheim's denial of the necessity of modification in the Friedheim and Euler treatment the case is different. The process of Friedheim and Euler consists, it will be remembered, in treating the soluble molybdate, or the solu- tion of molybdic acid in sodium hydroxide, with potassium iodide and hydrochloric acid in a Bunsen apparatus, boiling until the solution is of a clear green color, collecting the iodine distilled in potassium iodide, and titrating it with sodium thiosulphate. We found that the development of the green color was not a sufficient criterion of the exact reduction of the molybdic acid to the condition of the pentoxide and of the removal of the iodine which should be theoretically set free. To accomplish that end we found it safer and more convenient to start the distillation with a definite volume (40 cm 3 ) of liquid and boil until a definite volume (25 cm 3 ) was reached, care being taken with regard to the strength of acid and the excess of potassium iodide employed. Experience showed unmistakably that in order to avoid the decomposing action of the air upon the hot vaporous hydriodic acid in the retort, it was necessary to go beyond the measures advised by Friedheim and Euler (namely, to warm the retort and its contents slowly, heating to boiling only when the connecting tube was well filled with iodine vapor and the tendency toward back-suction of the liquid in the receiver began to appear) and to conduct the operation in a simple little apparatus (the retort holding about 100 cm 3 ) put together entirely with sealed and ground joints, as shown in the figure of the former paper, so ar- ranged that a current of purified carbon dioxide could be passed through retort and receiver during the distillation. With this apparatus we were able to determine with accuracy the point of concentration at which the free iodine left the liquid, the molybdic acid having been converted to the con- * Am. Jour. jSci. ii, 181. Volume I, p. 391. THE ESTIMATION OF MOLYBDENUM. 29 dition of the pentoxide. It was found that if dependence is placed upon the occurrence of the so-called clear green color of the liquid to determine the end of the distillation, it may frequently happen that free iodine remains in the residue. This takes place, it will be observed, in the atmosphere of carbon dioxide, so that the presence of the free iodine can by no possibility be attributed to the action of atmospheric air upon the hydriodic acid remaining after the distillation is complete. On the other hand, it appeared that, if the distillation is pushed too far, the molybdenum pentoxide may be still further reduced with consequent evolution of more than the expected amount of iodine. The attainment of an exact degree of reduction with the expulsion of the corre- sponding amount of iodine becomes, therefore, a matter of chance unless further precautions are taken. We found in our experiments that, if amounts less than 0.3 grm. of the molybdic acid are introduced in soluble form into the 100 cm 3 retort with a not too great excess of potassium iodide, and the 40 cm 3 of liquid so constituted that 20 cm 3 of it shall be water and 20 cm 3 the strongest hydrochloric acid, the reduc- tion proceeds with a fair degree of regularity in the manner expected. We found it important to restrict the excess of potassium iodide so that it shall never exceed the theoretical requirement by more than 0.5 grm. Our determinations with the pure molybdenum trioxide showed errors varying from 0.0010 grm. -|- to 0.0007 grm. ; the variations from theory in the experiments with ammonium molybdate ranged from 0.0011 grm. -f- to 0.0011 grm. . If these results are compared with those given by Friedheim and Euler, the advantage is a little in favor of the latter; but a scrutiny of the figures given by Friedheim and Euler develops the fact that the apparent accuracy of their work is founded upon miscalculations. This fact was known to us at the tune of our former writing, but we did not consider it essential then to make the matter public. The recent attack of Friedheim makes that course now necessary. Herewith is reproduced a table of results obtained by Fried- 30 THE ESTIMATION OF MOLYBDENUM. heim and Euler in the test of their method upon ammonium molybdate, shown by analysis to contain 81.49 per cent of molybdenum trioxide. The figures which are incorrect are enclosed in brackets: OEIGINAL FIGUKES OF FRIEDHEIM AND EULEB. Per cent of Molybdate taken. Na 2 S 2 3 used. Mo0 8 found. MoO s referred to molybdate taken. grm. cm 8 grm. 0.2674 30.8 ) 1 cm 2 = 0.2184 [81.71] 0.4418 50.8 > 0.00709 0.3601 81.51 0.4075 [40.71*1 Mo0 8 . 0.3317 81.40 0.3281 0.4340 0.4098 0.4305 37.33 1 T o ACk AQ I 1 Cm = I?'!? [ 0.007086 46.63 I ii/r^r* 49.08 J Mo - 0.2644 0.3502 0.3304 0.3478 t 81.85-1 81.69 81.67 81.78J Appended is a recalculation of the percentage of the trioxide found, with columns showing the percentage error and the error stated in fractions of a gram. Changes from the figures of Friedheim and Euler are in heavy-faced type. RECALCULATION OF THE RESULTS OF FRIEDHEIM AND EULEB. Corrected per cent of MoO, found, referred to the molybdate. Error in per cent of Mo0 3 found compared with Mo0 3 taken. Error of MoO a . grm. 81.68 0.23+ 0.0005+ 81.51 0.03+ 0.0001+ 81.40 0.12- 0.0004- 80.58 1.12- 0.0030- 80.69 0.99- 0.0035- 80.62 1.05- 0.0035- 80.79 0.86- 0.0030- These figures of their own (properly calculated) are suffi- cient to show the inadequacy of the method of Friedheim and Euler. We ourselves were occasionally able to get results from the method of Friedheim and Euler quite as good as Probably 46.7. THE ESTIMATION OF MOLYBDENUM. 31 these; it must be said, however, that most of our results obtained by their unmodified method have been even worse than their own. In another series of six determinations, in which molybde- num trioxide was the starting-point, Friedheim and Euler were more successful, the errors varying from 0.0006 grm. + to 0.0006 grm.. Thus Friedheim and Euler establish by their own results the fact that the hitting of the right point at which to stop their process of boiling is a matter of chance. In spite of the probability that some of the iodine which they found in the receiver was liberated by atmospheric action, the fact remains that their results are in many cases very low. That is, they did not boil long enough. The difficulty appears again in the modification of their method which Friedheim and Euler apply to the determination of molybdenum trioxide associated with vanadium pentoxide,* namely, the distillation with phosphoric acid and potassium iodide of the residue left after reducing the vanadium pentox- ide by hydrochloric acid and potassium bromide, according to the method of Holverscheit. We reproduce the part of their table which refers to the determination of the molybdenum, adding, however, columns containing the errors and corrected percentages. MoO a taken. MoO ? Per cent Mo0 3 Error. Per cent Mo0 3 . P. and E. Recalculated. grin. gnu. grm. 0.15037 0.15005 99.79 0.00032- 99.79 0.16895 0.16879 99.90 0.00016- 99.90 0.17758 0.17729 99.84 0.00029 99.84 0.24975 0.24962 99.95 0.00013- 99.95 0.33151 0.33607 [99.87] 0.00456+ 101.38 Four of the five determinations are accurate, but the fact that all figures are carried out to the fifth decimal place does not keep three good-sized figures out of the error column for the fifth determination. * Ber. Dtsch. chem. Ges., xxviii. 2072. 32 THE ESTIMATION OF MOLYBDENUM. It is hardly necessary, in the light of a comparison of the results of Friedheim and Euler with ours, to discuss further the unreliability of the unmodified process. The necessity of a proper control of the volume, strength of acid, and excess of potassium iodide, as well as proper protection from atmos- pheric oxidation, is real. VI THE APPLICATION OF IODIC ACID TO THE ANALYSIS OF IODIDES. BY F. A. GOOCH AND C. F. WALKER.* IT has long been understood that iodic acid is easily and completely reduced by an excess of hydriodic acid with the liberation of iodine according to the equation: HI0 8 + 5HI = 61 + 3H 2 O. To apply this reaction to the quantitative estimation of iodic acid, it is only necessary to add to the free iodic acid or solu- ble iodate an excess of a soluble iodide, to acidify best with dilute sulphuric acid and to titrate the iodine thus set free with sodium thiosulphate, one-sixth of the iodine found being credited to the iodic acid. It has been shown recently by Rieglerf that this reaction may be also applied to the quantitative estimation of iodides, the iodine set free upon the addition of a known excess of iodic acid to the iodide solution being removed by petroleum ether, and the residual iodic acid titrated directly with sodium thiosulphate. The present investigation was undertaken to define more particularly the limit of applicability of the reaction and to establish, if possible, a direct method for the quantitative esti- mation of iodides, dependent upon the action of iodic acid or an iodate in the presence of free sulphuric acid, neutralization of the solution by means of an acid carbonate, and titration of the free iodine by arsenious acid five-sixths of the iodine thus found being credited to the iodide to be estimated. It * From Am. Jour. Sci. iii, 293. t Zeitschr. anal. Chem., xxxv, 305. VOL. II. 3 34 THE APPLICATION OF 10DIC ACID has been found that by fulfilling certain necessary conditions, the proposed method is entirely successful, so far as concerns the estimation of iodine in iodide solutions free from large amounts of chlorides or bromides. In a system containing a considerable quantity of free iodine with variable amounts of the other reagents mentioned, as well as possible impurities, it is conceivable that secondary reactions may occur, depending largely on conditions of mass, tune, and temperature, and of a sort likely to alter the amount of recoverable iodine, or to exert an excessive oxidizing influ- ence on the arsenious acid finally titrated. It has been estab- lished by Schonbein, Lunge and Schloch, and others, that iodine forms compounds with the alkalies of the type R-O-I, and Phelps* has recently found that the formation of some such compound, accompanying the iodate naturally expected, is distinctly recognizable when iodine and barium hydroxide interact at ordinary temperatures. It has been shown, also, in a former paper from this laboratory! that free iodine or an iodide interacts very easily with iodic acid in the presence of dilute hydrochloric acid with the formation of iodine mono- chloride, according to the equations: HI0 3 + 21, + 5HC1 = 3H 2 + 6IC1. HI0 3 + 2KI + 5HC1 = 3H 2 + 2KC1 + 3IC1. Moreover, organic compounds containing the groups 1 = and I ~ Q, in which iodine seems to be analogous to nitrogen, result in great variety from the oxidation of halogen sub- stitution products. It would seem, therefore, that the formation of inorganic reduction products of iodic acid under the conditions likely to obtain in this analytical process might be by no means beyond the bounds of possibility. A few simple qualitative tests to determine the possibility of interaction between small quantities of iodine and iodic acid alone met with negative results. Thus, a single drop of a decinormal solution of iodine, made as usual in potassium * Am. Jour. Sci., ii, 70. Volume I, p. 370. t Roberts, Am. Jour. Sci., xlviii, 157. Volume I, p. 257. TO THE ANALYSIS OF IODIDES. 35 iodide, gave when added to 10 cm 3 of decinormal iodic acid a distinctive color to chloroform. Similar results were obtained when the iodine was employed in aqueous solution in which there was no alkaline iodide. A few drops of an aqueous solution of iodine treated (in either order) with 10 cm 3 of a saturated solution of potassium bicarbonate and 10 cm 3 of decinormal iodic acid gave the same distinctive color to chloroform as came from the same amount of iodine in the absence of the iodic acid. So it appears that if in the system under consideration reactions do occur between iodic acid and iodine to alter the amount of iodine recoverable, such action is not appreciable between small amounts of these materials. This, however, does not preclude the possibility of perceptible changes under the mass-action of a large amount of iodine. The reactions of hydrochloric acid, and probably of hydro- bromic acid, hi the presence of varying amounts of iodic acid, iodine, and iodide, as well as the reaction of the alkaline carbonate upon such mixtures are doubtless complex, more or less reversible, and dependent upon proportions and dilution. The tendency of the former reactions is toward the reduction of the molecule of iodic acid, and the formation of the chloride or bromide of iodine. Thus, Miss Roberts * demonstrated that a solution of hydrochloric acid, so dilute that by itself it is without effect on iodic acid, acts upon a mixture of iodic acid with either free iodine or an iodide to form iodine monochloride. The action of the acid carbonate upon the iodine chloride or bromide may produce a salt of the oxy-acids and free iodine. The practical effects, under the conditions of analysis, of the reaction between iodine, iodic acid and the halogen acids in presence of sulphuric acid, and of reactions which may occur upon neutralization by an acid carbonate, were studied in detail in a number of experiments. The preliminary experiments of Table I were made to bring out the effect of neutralizing with the acid carbonate and subsequently titrating with an alkaline arsenite a solution * Loc. cit. 36 THE APPLICATION OF IODIC ACID [5 cm3 TABLE I. EFFECT OF THE CARBONATE. (1:3). Total volume of liquid, 250 cm 8 .] Ezp. I (in KI) taken. KHC0 3 in excess. I found. Error. grin. cm 8 grin. gnu. (1) 0.0713 Very small. 0.0707 0.0006- (2) 0.0715 Very small. 0.0710 0.0005- (3) 0.0713 10 0.0710 0.0003- (4) 0.0710 10 0.0706 0.0004- (5) 0.0723 10 0.0717 0.0006- (6) 0.0713 20 0.0709 0.0004- (7) 0.0713 20 0.0709 0.0004- (8) 0.3565 Very small. 0.3560 0.0005- (9) 0.3568 Very small. 0.3561 0.0007- (10) 0.3667 10 0.3563 0.0004- (11) 0.3596 10 0.3588 0.0008- (12) 0.3565 10 0.3565 0.0000 (13) 0.3572 20 0.3560 0.0012- (14) 0.3567 20 0.3569 0.0002-f- containing sulphuric acid and a considerable amount of free iodine. The danger of mechanical loss of iodine during the effervescence accompanying neutralization, as well as by spontaneous volatilization from the surface during the process of titration, was minimized by effecting the neutralization in the trapped Drexel washing-bottle to be described later, and making the titration in the same tall washing cylinder without transfer. To varying amounts of a recently standardized decinormal solution of iodine were added successively 5 cm 3 of dilute sulphuric acid and varying amounts of potassium bicarbonate in excess of that necessary to neutralize the free acid, decinormal arsenious acid in slight excess of the iodine, 5 cm 3 of starch emulsion, and decinormal iodine to coloration, the total volume of the liquid being not greater than 250 cm 8 . The results show plainly that while the loss, mechanical or otherwise, in the treatment of reasonably large amounts of fairly concentrated iodine is perceptible, it is still well within permissible limits (amounting to a little less than 0.0005 grm. in the mean), and obviously independent of the excess of the carbonate in the solution, and of the amount of free iodine present. TO THE ANALYSIS OF IODIDES. 37 TABLE IL EFFECT OF DILUTION. Bxp. El taken. KI found. Error. Approximate volume upon addition of H 2 S0 4 . Volume H 2 S0 4 (1:3) used. (1) (2) (3) (4) (5) (6) (7) (8) grin. 0.0772 0.0772 0.1544 0.1544 0.3087 0.3087 0.3859 0.3859 grlu. 0.0768 0.0765 0.1546 0.1541 0.3090 0.3088 0.3864 0.3860 grm. 0.0004- 0.0007- 0.0002+ 0.0003- 0.0003+ 0.0001+ 0.0005+ 0.0001+ 1SSSSSSSS cm** 6 5 5 5 5 6 5 5 (9) (10) (11) (12) 0.0772 0.0772 0.1543 0.1544 0.0754 0.0757 0.1532 0.1524 0.0018- 0.0015- 0.0011- 0.0020- 300 300 300 300 5 5 5 5 (13) (14) (16) (16) (17) (18) 0.0772 0.0772 0.1544 0.1544 0.3859 0.3859 0.0744 0.0737 0.1521 0.1512 0.3827 0.3831 0.0028- 0.0035- 0.0023- 0.0032- 0.0032- 0.0028- 600 500 500 500 500 500 5 6 5 5 5 5 (19) (20) (21) (22) 0.0772 0.0772 0.3859 0.3859 0.0744 0.0757 0.3828 0.3827 0.0028- 0.0015- 0.0031- 0.0032- 500 500 500 500 10 10 10 10 In the experiments of Table II the proposed process of analysis was tested upon potassium iodide taken by itself in varying amounts of a J^ normal solution and carefully standardized by the method formerly elaborated in this laboratory.* The apparatus employed was a Drexel washing-bottle of 500 cm 3 or 1000 cm 3 capacity, according to require- ments, with stop-cock and thistle-tube fused to the inlet tube and a Will and Varrentrapp absorption trap sealed to the outlet, as shown in the accompanying figure. The iodide for the test was drawn from a burette into the bottle and carefully washed down, and potassium iodate in excess of the amount theoretically necessary (namely, FIG. 18. * Gooch and Browning, Am. Jour. Sci., xxxix, 188. Volume I, p. 1. 38 THE APPLICATION OF IODIC ACID 5 cm 3 of a 0.5 per cent solution for every portion of 20 cm 3 of the iodide solution), was added and the volume of the liquid was adjusted to the volume at which it was desired that the iodic and hydriodic acids should react. The stop- per with the thistle-tube and trap was now placed on the bottle and the trap was half filled by means of a pipette with a 5 per cent solution of potassium iodide. Five cen- timeters of dilute (1 : 3) sulphuric acid were added through the thistle-tube and washed down ; the stop-cock was closed, .and the solution gently agitated, if necessary, to insure a ; complete separation of iodine. Potassium bicarbonate in saturated solution to an amount about 10 cm 8 in excess of that required to neutralize 5 cm 8 of dilute (1 : 3) sulphuric acid, was poured into the thistle-tube, and allowed to flow into the bottle slowly enough to avoid a too violent evolution of gas. The stop-cock was closed and the solution agitated by giving to the bottle a rotary motion, at the same time keeping the bottom pressed down upon the work-table, to prevent a possible splashing of the iodide out of the trap into the yet acid solution. When the neutralization of the solution had been completed, the bottle was shaken until the last trace of violet vapor was absorbed in the liquid. The greater part of the solution in the trap was then run back into the bottle, the stopper removed, and the tube and trap carefully washed, the washings being added to the bulk of the solution. Deci- normal arsenious acid was introduced from a burette to the bleaching point, 5 cm 3 of starch emulsion were added, and the solution was titrated back with decinormal iodine (usually only a few drops) to coloration. Blank tests made upon a solution obtained by mixing the maximum amount of the iodate with 5 cm 3 of dilute sulphuric acid (1 : 3), neutralizing as usual with potassium bicarbonate, adding the iodide from the trap and 5 cm 3 of starch emulsion, showed that a single drop of iodine was invariably sufficient to bring out the starch blue. Occasionally it was found that the mixture, particularly when chlorides or bromides were present, of itself developed a trace of color, but by no means TO THE ANALYSIS OF IODIDES. 39 a reading tint. A correction of the one drop of iodine necessary to bring out the color reaction in the blanks, was applied uniformly in the analytical process. The number of centimeters of decinormal arsenious acid required to bleach the free iodine, multiplied by 0.01383 (log. 2.140822) gives the number of grams of potassium iodide taken for analysis, being equivalent to five-sixths of the iodine liberated in the solution. From these results it appears that the degree of dilution of the solution at the time when the mixed iodide and iodate are acidified has an important influence on the completeness of the reaction. Thus, the mean error of the determinations in which the volume at the time of the reaction did not exceed 150 cm 3 was practically nothing, while the errors at volumes of 300 cm 3 and 500 cm 3 amounted to 0.0016 grm. and 0.0028 grm. respectively. It is obvious that the doubling of the amount of sulphuric acid used in acidifying does not increase the amount of iodine liberated at the highest dilution. The plain inference is that the interaction between the iodide and iodate should be brought about in a volume of liquid not much exceeding 150 cm 3 . In the following series of experiments, recorded in Table III, the effect of the introduction of a chloride or bromide TABLE III. EFFECT OF CHLORIDE AND BROMIDE. Exp KI taken. KI found. Error. NaCl taken. KBr taken. grin. grm. grm. grm. grm. (1) 0.0772 0.0795 0.0023+ 0.2 , (2 0.0772 0.0784 0.0012+ 0.2 (3 (4 0.0771 0.0773 0.0823 0.0819 0.0052+ 0.0046+ 0.5 0.6 (6 0.1544 0.1588 0.0044+ 0.5 ( 6 0.1544 0.1590 0.0046+ 0.5 o 0.0772 0.0802 0.0030+ , , 6.2 (8 0.0773 0.0853 0.0080+ . , 0.2 (9) 0.0772 0.0873 0.0101+ t 0.5 (10) 0.0772 0.0861 0.0089+ 0.5 11) 0.1544 0.1646 0.0102+ t 0.5 (12) 0.1543 0.1626 0.0083+ 0.6 40 THE APPLICATION OF IODIC ACID into the iodide (before the iodate is added) was studied. The volume of the liquid at the time of acidifying was fixed at 150 cm 3 approximately, and 5 cm 3 of the dilute sulphuric acid (1 : 3) were used. The mode of procedure was otherwise similar to that of the foregoing series. The influence of sodium chloride and potassium bromide in increasing the amount of iodine liberated is plain. The increase comes without doubt from the iodate, and is doubt- less due to the formation of iodine chloride or bromide, during the acidifying, by the interaction of the free iodine, the iodic acid, and the hydrochloric or hydrobromic acid, according to the reactions previously discussed. It is plain, therefore, that the value of the process in the determination of iodine in an iodide is restricted of necessity to those cases in which it is known that chlorides or bromides are not present to TABLE IV. ANALYSIS OF PURE POTASSIUM IODIDE. Exp. KI taken. KI found. Error. grm. gnn. grm. (1) 0.0814 0.0816 0.0002+ (2) 0.0814 0.0813 0.0001- (3) 0.0814 0.0805 0.0009+ (4) 0.0815 0.0809 0.0006- (5) 0.0814 0.0808 0.0006- (6) 0.0814 0.0806 0.0008- (7) 0.0814 0.0812 0.0002- (8) 0.1628 0.1624 0.0004- (9) 0.1628 0.1617 0.0011- (10) 0.1628 0.1621 0.0007- (11) 0.1628 0.1619 0.0009- (12) 0.1628 0.1624 0.0004- (13) 0.1628 0.1621 0.0007- (14) 0.1628 0.1626 0.0002- (15) 0.2442 0.2451 0.0009+ (16) 0.2442 0.2442 0.0000 (17) 0.2442 0.2439 0.0003- (18) 0.3256 0.3258 0.0002+ (19) 0.3256 0.3256 0.0000 (20) 0.3256 0.3258 0.0002+ (21) 0.3256 0.3272 0.0016+ (22) 0.3256 0.3256 0.0000 (23) 0.4071 0.4076 0.0005+ (24) 0.4071 0.4080 0.0009+ (25) 0.4071 0.4073 0.0002+ TO THE ANALYSIS OF IODIDES. 41 any considerable extent. For determining the standard of a solution of nearly pure potassium iodide, employed in so many laboratory processes, it should find useful application. In Table IV are comprised a number of experiments made exactly like those which seemed to give the best results in the series of Table II. The iodide and an excess of iodate (5 cm 3 of the 0.5 per cent solution to every 20 cm 3 of ^ iodide) were made to interact in a volume of about 150 cm 3 , 5 cm 3 of sulphuric acid (1 : 3) were used to bring about the reaction, 10 cm 3 of potassium bicarbonate were added after the neutrali- zation of the sulphuric acid was complete, and the free iodine was estimated by titrating decinormal arsenious acid, the manipulation being like that previously described in detail. The average result of a series of several determinations in which a great excess (0.1 grm.) of potassium iodate was used, proved to be practically identical with that of a similar series in which only a small excess of the iodate was employed, so that it appears to be unnecessary in any practical work to restrict the amount of iodate below the amount necessary to decompose the maximum quantity of potassium iodide which we have handled, namely, 0.4 grm. It appears that for the estimation of iodine in a soluble iodide free from notable amounts of chlorides or bromides, this method, depending as it does upon a single standard solution, is simple, fairly accurate, and rapid. vn THE ACTION OF UREA AND PRIMARY AMINES ON MALEIC ANHYDRIDE.* BY FREDERICK L. DUNLAP AND ISAAC K. PHELPS. IN a former article, f a method was described by one of us for obtaining imides by the action of urea on the anhydrides of dibasic acids. It was shown that the formation of imides by the interaction of urea and anhydrides was to be explained by the addition of the urea and the anhydrides to form an acid, which when heated, decomposed with the formation of the imide, carbon dioxide and ammonia. These reactions can be shown by the following equations : COOH . ^O.^T In some cases this intermediate product, formed by the addition of the reacting substances, was readily isolated, and upon heating to certain temperatures, it was found that this addition-product decomposed with the formation of the imide. It was found, in particular, that maleic anhydride formed an addition-product with urea, and we hoped to obtain the unknown imide of maleic acid by the decomposition of this addition-product. Male-uric Acid. When equal molecules of maleic anhydride and urea are heated to 100 -105 C. in an oil-bath, the mixture melts, and if this temperature is maintained for a short time, * From Am. Chem. Jour., xix, 492. t Am. Chem. Jour., xviii, 332. Volume I, 355. ACTION OF UREA AND PRIMARY AMINES. 43 the liquid solidifies. This product was purified by crystalli- zation from alcohol, and when pure was perfectly white and melted at 167.5-168 with decomposition. Upon analysis the following results were obtained: 0.2148 gram substance gave 0.2968 gram C0 2 and 0.0751 gram H 2 0. Calculated for C 37.97 37.69 H 3.80 3.88 Maleiiric acid is soluble in hot water and alcohol, fairly insoluble in cold and insoluble in chloroform, benzene, ligrom, carbon disulphide, acetone, and ether. Its formation and structure are shown in the following equation : H-C-CO V H-C-CONHCONH a II ^0 + NH 3 CONH 2 = || H-C-CCT H-C-COOH No difficulty was expected in the formation of the imide of maleic acid, either from the maleiiric acid, or by the direct heating of maleic anhydride and urea; for, in the cases already studied,* the reaction ran with great smoothness and ease. But, for some unknown reason, the imide of maleic acid could not be obtained at least in quantities for complete identification. A great many experiments were carried out under varying conditions, but only under one set of conditions could any product be isolated, and then unfortunately the yield was so distressingly small that the study of the compound had to be abandoned. When equal molecules of maleic anhydride and urea are heated in a boiling-flask, at first slowly, and finally at the full heat of a Bunsen burner, a vigorous evolution of carbon dioxide and ammonia took place, and a small amount of a dark-colored distillate was obtained, which on cooling solidified and gave a melting-point of 130.5 after crystallization from anhydrous acetone. This same crystalline * Loc. cit. 44 THE ACTION OF UREA AND PRIMARY product was also obtained on distilling the maleiiric acid. In both cases a large amount of carbonaceous residue remained in the boiling-flask. Although the method of formation of this compound is exactly parallel to that employed in the preparation of succinimide,* yet the identity of this product cannot be regarded as established until sufficient quantities of it have been prepared to subject it to elementary analysis. The Action of Primary Amines on Maleic Anhydride. Anschiitzf in 1889 published a method by which anilic acids could be very easily prepared by dissolving molecular quantities of aniline and the anhydride of dibasic acids in dry chloroform, when after standing for a short tune the anilic acid crystallized out. Following out this line and method of formation, suggested by Anschiitz's work, some derivatives of maleic acid have been prepared and studied. p-Tolylmaleamic Acid. Molecular proportions of maleic anhydride and ^>-toluidine were dissolved in dry chloroform, and upon mixing these two solutions, there immediately separated out a light yellow precipitate. After standing for a time to insure the complete precipitation of the product, it was filtered off and purified by crystallization from alcohol, from which it separated in lemon-yellow needles. The pure product melted at 201, with evolution of gas. It is readily soluble in ether, acetone, and hot alcohol, but insoluble in benzene, ligrom, carbon disulphide, chloroform and water. Upon analysis the f ollowing results were obtained : 0.2072 gram substance gave 0.4881 gram C0 2 , and 0.1068 gram H 2 0. Calculated for v ^ C U H U S N. Found ' C 64.39 64.24 H 5.37 5.73 * Loc. cit. t Ber. Dtsch. chem. Ges., xx, 3214 AMINES ON MALEIC ANHYDRIDE. 45 The structure and mode of formation of this ^>-tolylmaleamic acid may be seen from the following equation : H-C-CO V X CH 8 H~C-CONHC 6 H 4 CH 8 (p) C 6 H 4 N H-C-C(r NH 2 (p)H-C-COOH o-Totylmaledmic Acid. This acid was prepared similarly to the ^-tolylmaleamic acid, using molecular proportions of maleic anhydride and 0-toluidine. The product was purified by crystallization from alcohol, from which it separated in bunches of thick light yellow prisms. When pure it melted at 117.5-118. It is readily soluble hi acetone, and in alcohol; sparingly soluble in chloroform, and insoluble in ligroin, carbon disulphide, benzene, water, and ether. Analy- sis gave the following results : 0.2124 gram substance gave 0.4993 gram C0 2 , and 0.1146 gram H 2 0. C 64.39 64.12 H 5.37 5.99 It has the structure represented by the formula H - C - CONHC 6 H 4 CH 8 (o) II H-C-COOH p-Naphthylmaleamic Acid. When molecular quantities of ^-naphthylamine and maleic anhydride were mixed in dry chloroform solution they reacted and united as readily as did the o- and ^>-toluidines and the anhydride. Almost immediately a yellow crystalline precipitate separated out. This, after standing a short time, was filtered off and crystallized from alcohol. It separated from alcohol in small bright yellow needles, which, when pure, melted at 200 with evolution of gas. It is soluble in acetone and alcohol, but insoluble in ether, carbon disulphide, chloroform, benzene, ligroin, and water. Analysis gave the following results : 46 ACTION OF UREA AND PRIMARY AMINES. 0.2033 gram substance gave 0.5226 gram C0 2 and 0.0948 gram H 2 0. Calculated for v , C 14 H U 3 N. FoTmd ' C 69.71 70.11 H 4.56 5.18 This addition-product is obviously formed as follows : H _ C - CO V H-C- CONHC 10 H 7 + C 10 H 7 NH 2 = H-C-CCT H-C-COOH a-Naphthylamine appears also to be added readily to maleic anhydride, but the product has not been submitted to analysis. VIII THE SEPARATION OF ALUMINUM AND BERYL- LIUM BY THE ACTION OF HYDROCHLORIC ACID. BY FRANKE S. HAVENS* IN a former paper f a method was described for the determi- nation of aluminum in the presence of iron, based upon the fact that the hydrous aluminum chloride A1C1 3 . 6H 2 O is practically insoluble in a mixture of strong hydrochloric acid and anhydrous ether saturated with hydrochloric acid gas, while the ferric chloride is entirely soluble in that medium. The work to be described in this paper is an extension of this process to cover the separation of aluminum from beryllium, with the subsequent determination of the beryllium by weighing as the oxide after conversion to the nitrate and ignition. The aluminum chloride solution was prepared by dissolving the so-called pure aluminum chloride of commerce in as little water as possible, precipitating and washing free from iron with strong hydrochloric acid, dissolving the chloride thus obtained in water, precipitating the hydroxide by ammonia, washing the precipitate free from all alkalies, and redissolving it in hot hydrochloric acid. From this solution, after cool- ing, gaseous hydrochloric acid precipitated the pure hydrous chloride. This prepared chloride was dissolved in water and the solution standardized by precipitating with ammonia the hydroxide from weighed portions and weighing as the oxide. The solution of beryllium used was made by dis- * From Am. Jour. Sci., iv, 111. t Gooch and Havens, Am. Jour. Sci., ii, 416. This volume, p. 20. 48 SEPARATION OF ALUMINUM AND BERYLLIUM solving in water beryllium chloride found to be free from iron by the sulphocyanate test, and giving no precipitate when tested by the gaseous hydrochloric acid process to be described later on. This was standardized by precipitating with ammonia the hydroxide from weighed portions and weighing the ignited oxide in the usual manner. In the experiments of Table I, weighed portions of the aluminum solution were mixed with portions of the beryllium chloride solution representing from 0.01 gram to 0.10 gram of the oxide, an equal volume of a mixture of strong hydro- chloric acid and ether (taken in equal parts) was added to the solution of the mixed chlorides, and the whole was completely saturated with gaseous hydrochloric acid while kept at a temperature of about 15 C. by immersing the receptacle TABLE I. Bxp. A1 2 O 8 taken in solution as the chloride. found. Final volume. Error. (1) grm. 0.1046 gnn. 0.1044 cms 12 grm. 0.0002- (2) 0.1046 0.1038 12 0.0008- (3) 0.1067 0.1066 12 0.0001- (4) 0.1071 0.1063 12 0.0008- (5) 0.1059 0.1064 30 0.0005- in running water. Ether was added, equal in volume to the aqueous aluminum and beryllium solutions originally taken, and the current of gas again turned on until satura- tion was complete. By this treatment there is present at the end of the saturation a volume of ether equal to that of the aqueous hydrochloric acid introduced and generated. The finely crystalline precipitate of aluminum chloride was caught on asbestos in a filter crucible washed with a pre- viously prepared mixture of hydrochloric acid and ether in equal parts saturated at 15 C. with hydrochloric acid gas, and dried for half an hour at a temperature of 150 C. It was next covered with a layer of pure mercuric oxide, which had been tested and found to leave no residue on volatilizing, BY THE ACTION OF HYDROCHLORIC ACID. 49 and the crucible was gently heated over a low flame under a ventilating hood and finally ignited over the blast. From these results it is obvious that the aluminum chloride may be determined in the presence of beryllium chloride with reasonable accuracy. The beryllium may be recovered in the filtrate from the aluminum chloride by precipitation with ammonia after nearly complete evaporation of the acid. It was found, however, upon trial that the conversion of the chloride to the oxide without precipitation and filtration may be easily accomplished by treatment with nitric acid and ignition. The results of Table II indicate this clearly. In these experiments weighed TABLE II. BeO taken in Exp. solution as the BeO found. Error. chloride. grin. gnu. grm. 0) 0.0483 0.0481 0.0002- 2) 0.0483 0.0483 0.0000 (3) 0.1076 0.1085 0.0009+ portions of the beryllium solution were evaporated just to dryness on a radiator, care being taken not to heat to the volatilizing point of the beryllium chloride, a few drops of strong nitric acid were added, the liquid was evaporated, and the residue heated at first gently, to break up the nitrate safely and finally on the blast. It was found that this conversion of the beryllium to the nitrate can be carried on in platinum without attacking that metal appreciably, providing care be taken to remove thoroughly all excess of hydrochloric acid before the nitric acid is added to the dry residue. In Table III, (l)-(9), are given the results of experiments in which both the aluminum and the beryllium were deter- mined the former by precipitation as the hydrous chloride and weighing as the oxide after igniting with mercuric oxide : the latter by the conversion of the chloride, through the VOL. II. 4 50 SEPARATION OF ALUMINUM AND BERYLLIUM nitrate, into the oxide. In experiment (10) (made to get a comparison of the methods) the beryllium was recovered by precipitating the hydroxide with ammonia from the par- tially evaporated solution of the chloride after removing the aluminum. In experiments (1) to (5), inclusive, the aluminum was determined exactly as previously described ; in (6) and (7) the solutions (being originally larger) were concentrated by evaporation previous to the addition of the ether and hydrochloric acid mixture. In experiments (8), (9) and (10), the treatment was varied advantageously by saturating the aqueous solution directly with hydrochloric acid gas before adding an equal volume of ether, and completing the saturation. TABLE III. A1 2 3 taken BeO taken Exp. in solution as the AM>, Error. Final volume. in solution as the BeO found. Error. chloride. chloride. grin. grm. grm* cm 3 gnu* grin. grm. (1) 0.1059 0.1058 0.0001- 12 0.0198 0.0204 0.0006+ (2) 0.1053 0.1044 0.0009- 12 0.0194 0.0196 0.0002+ (3) 0.1065 0.1059 0.0006- 12 0.0197 0.0205 0.0008+ (4 0.1068 0.1060 0.0008- 12 0.0199 0.0207 0.0008+ (5) 0.1049 0.1047 0.0002- 12 0.0198 0.0208 0.0010+ (6) 0.1060 0.1057 0.0003- 12 0.0977 0.0969 0.0008- (7) 0.1064 0.1063 0.0001- 12 0.1085 0.1084 0.0001- (8) 0.1046 0.1038 0.0008- 30 0.1083 0.1087 0.0004+ (9) 0.1051 0.1048 0.0003- 30 0.1071 0.1078 0.0007+ (10) 0.1076 0.1075 0.0001- 30 0.1086 0.1094 0.0008+ These results are plainly very good. The manipulation of the process is not difficult. The gaseous hydrochloric acid is most conveniently produced by the well known method of treating with strong sulphuric acid in regulated current a mixture of strong aqueous hydrochloric acid and common salt. A platinum dish hung in an inverted bell-jar, provided with inlet and outlet tubes through which the current of water for cooling is passed, makes the best container for the solution to be saturated with the gas. It is advantageous to arrange the filtration BY THE ACTION OF HYDROCHLORIC ACID. 51 upon asbestos so that the filtrate and washings may be caught directly in the crucible (placed under the bell-jar of the filter pump) in which the subsequent evaporation is to be effected. The heating of the strong acid solution must be gradual and conducted with care to prevent mechanical loss by a too violent evolution of the gaseous acid. IX THE TITRATION OF SODIUM THIOSULPHATE BY IODIC ACID. BY CLAUDE F. WALKER .* THIS investigation was undertaken to determine the nature and limitations of the reaction between iodic acid and thiosul- phuric acid, and to show the expediency of employing iodic acid in standard solution for the direct titration of sodium thiosulphate. Rieglert states that iodic acid is readily ob- tained in the pure state, that it may be accurately weighed out, and that a solution of it may be exactly made up to a desired strength and kept for a long tune unaltered. He further states that when a solution of sodium thiosulphate is titrated with iodic acid the reaction takes place according to the equation, 6Na 2 S 2 3 + 6HI0 3 = 3Na 2 S<0 6 + 5NaI0 3 + Nal + 3H 2 0, under which circumstances no free iodine will be evolved until all the sodium thiosulphate has been oxidized to tetra- thionate ; the first drop of iodic acid in excess, however, will react with the sodium iodide that has been formed, and sepa- rate iodine, as shown by the equation, 5NaI + 6HI0 3 = 5NaI0 8 + 3H 2 + 3I 2 , thus furnishing an accurate means for determining the end point. A careful repetition of the work of Riegler has shown that his conclusions are in a large measure erroneous. Thus, it has been found that the ordinary " chemically pure " iodic acid, purchased from reliable manufacturers, is likely to con- * From the Amer. Jour. Sci., iv, 235. t Riegler, Zeitschr. anal. Chem., xxxr, 308. TITRATION OF SODIUM THIOSULPHATE, ETC. 53 tain more than the theoretical amount of iodine, due probably to the presence of the anhydride, although iodic acid can be safely employed for standardizing when it is made in the laboratory by dissolving the purified anhydride, crystallizing out the acid, and drying over sulphuric acid. Such a care- fully prepared product, if used immediately, will be found to contain the theoretical amount of iodine. Riegler's proposed method of titration depends on two different reactions, and to insure the accuracy of the process these must be definite, com- plete and non-reversible under the conditions of analysis. Thus one molecule out of every six of iodic acid should be reduced by six molecules of thiosulphate, with the formation of a neutral mixture of iodide and iodate, free from other oxidizing or reducing substances. Under these circumstances it might be expected that iodine will be liberated by the first trace of iodic acid hi excess. It has been found by investiga- tion, however, that although the main reaction between iodic acid and sodium thiosulphate may result in the formation of sodium tetrathionate in the proportions given, there is never- theless striking evidence of some other obscure action of the thiosulphate, which influences the reduction of the iodic acid in such a way as to make it impossible to calculate the analy- ses according to Riegler's reaction. Moreover, a peculiar " after-coloration " which invariably follows the first formation of the starch blue during the titration of one solution against the other, seems to point to the possibility that the reaction between the iodide and iodic acid is dependent, under these circumstances, on conditions of time and mass for its com- pleteness. It is not impossible that some third compound of iodine unstable in its nature, may be formed as an interme- diate product and thus delay the liberation of iodine. In con- sideration of the results that have been obtained it appears that Riegler's proposed process for standardizing sodium thio- sulphate, as well as his related method for the analysis of iodides,* must remain impracticable unless they can be modi- fied so as to do away with a number of sources of error. * Riegler, Zeitschr. anal. Chem., xxxv, 305. 54 TITRATION OF SODIUM THIOSULPHATE The analyses of solutions of iodic acid, during the entire course of the work, was invariably performed by adding to the portion of the solution to be analyzed an excess of potas- sium iodide, acidifying with 5 cm 3 of dilute (1 : 3) sulphuric acid, and recovering the liberated iodine by directly titrating the acid solution with sodium thiosulphate, or by neutralizing with potassium bicarbonate in excess, and directly titrating the alkaline solution with arsenious acid. In the latter case the neutralization was performed in a trapped Drexel wash- ing bottle such as has been described in connection with the analysis of iodides.* In either case one-sixth of the iodine recovered was calculated to iodic acid, according to the terms of the equation, SHI + HI0 3 = 31, + 3H 2 0. It follows from these proportions that to bring the analyses within the range of the decinormal solutions ordinarily employed, the iodic acid taken for analysis must be restricted to comparatively small amounts. In the present work it was found convenient to analyze the iodic acid in quantities not much exceeding one-tenth of a gram, in which case the varia- tion in the results in the same series is found to be almost inap- preciable. In both variations of the process one or two blank analyses were invariably made, by performing the operation as detailed, except that no iodic acid was employed, and the cor- rection of one drop of iodine thereby shown to be necessary to bring out the starch blue was uniformly applied in the ana- lytical work. To determine whether or not the purity of the ordinary iodic acid is sufficient to admit of its direct application in standard solutions, a series of experiments was made. Two different samples of " chemically pure " iodic acid were used. The first was in coarse granular crystals, and the second was in the form of fine powder. Quantities of both of these were dried in a desiccator over sulphuric acid to constant weight. Neither sample lost weight appreciably when left * Gooch and Walker, Am. Jour. Sci. iii, 293. This volume, p. 33. BY IODIC ACID. 55 for a considerable time on the scale pan. A third sample of iodic acid was prepared by dissolving a quantity of the purest obtainable iodic anhydride in water, and evaporating at ordinary temperature. The resulting crystalline mass was dried over sulphuric acid in a desiccator for one week, until it ceased to lose weight, when it was presumed to consist of the pure normal acid. Two presumably decinormal solutions of each of the first two samples, and one such solution of the third sample of iodic acid were made by weighing out 17.585 grms. and dissolving in exactly one liter of water at 15 C. Convenient portions of each of these solutions were analyzed in the manner described, with results shown in the following table, averaged from many determinations. TABLE I. ANALYSES OF APPROXIMATELY ^ IODIC ACID. Solution analyzed. Sample used. mO 3 taken. HIO S found. Error. grin. grm. grm. I A 0.1055 0.1066 0.0011+ II A 0.1055 0.1062 0.0007+ III B 0.1055 0.1065 0.0010+ IV B 0.1055 0.1073 0.0018+ V C 0.1055 0.1053 0.0002- These results show that while the deviation from the theoretical strength of the solution in the case of the acid prepared from the anhydride is hardly appreciable, and will not affect the accuracy of any work in which the solution may be applied as a means of standardization, the solutions made from the purchased product, on the other hand, contain a very appreciable amount of iodine in excess of the theoretical. That iodic acid is somewhat unstable at 30-40 C., gradually losing water with the formation of the anhydride,* is well known, and it is quite possible that to some such gradual change as this must be attributed the fact that the ordinary iodic acid cannot be safely employed as a means of standard- ization unless its purity be directly determined by analysis. * Dammer, Anorg. Chem., i, 564. 56 TITRATION OF SODIUM THIOSULPHATE To determine whether a solution of iodic acid, once prepared and standardized, will retain its strength for a long period of time, two such solutions were kept for four months (in the dark) and then again analyzed. The results (averages of several determinations), given in Table II, substantiate the observation of Riegler that a solution of iodic acid will remain of constant strength. TABLE II. CONSTANCY OF STRENGTH OF IODIC ACID SOLUTIONS. Iodic acid solution. First analysis. HI0 3 found. Second analysis (after four months). HIO 3 found. Variation. I II grm. 0.1073 0.1049 grm. 0.1072 0.1046 grm. 0.0001- 0.0003- An approximately one-twentieth normal solution of " chemi- cally pure" sodium thiosulphate was made and its exact strength ascertained by titrating it with standardized iodine. A series of analyses made by oxidizing the sodium thiosulphate to sulphate, and precipitating and weighing as barium sulphate, gave results identical with those obtained with iodine, proving that all the sulphur present in the solution was in the form of thiosulphate. According to Riegler's equation, sodium thiosulphate and iodic acid react molecule for molecule, and solutions of these substances should therefore require for their mutual saturation volumes inversely proportional to their concentration. It was found, however, that when the one-twentieth normal solution of sodium thiosulphate that has been described was titrated in the presence of starch emulsion with an approximately decinormal solution of iodic acid, prepared from the anhydride, a distinctly blue color was produced long before the theoretical amount of iodic acid had been added. It was further noticed that the end-point of the reaction was far from distinct, a faint tinge of blue at first being visible, then suddenly becoming deeper, and immediately reappearing when bleached with sodium thiosulphate. The BY IODIC ACID. 57 deficiency in the amount of iodic acid actually required to produce the blue color was not lessened by introducing only three-fourths of the theoretical amount of iodic acid, and estimating the residual thiosulphate with iodine. It was found, however, that the addition of a considerable quantity of potassium iodide to the solution, either before or during the titration, had the marked effect of making the reaction sharp and distinct, entirely preventing the " after separation " of iodine, at the same time postponing the appearance of the starch blue until a quantity of iodic acid had been added considerably in excess of the theoretical. These experiments were repeated with entirely different reagents, and under varied conditions of concentration, the results in every case exactly confirming those already observed. For the purpose of more particularly investigating this subject, there were prepared and standardized an approxi- mately decinormal solution of sodium thiosulphate, and an approximately one-fiftieth normal solution of iodic acid. Measured portions of the sodium thiosulphate solution were titrated with the iodic acid in the presence of starch emulsion under varying conditions of mass, time and dilution. To determine the variability of the end-point of the reaction when the titration was conducted as directed by Riegler, a series of experiments was made. Measured amounts of the sodium thiosulphate 'solution were drawn from a burette into an Erlenmeyer beaker of suitable capacity, the sides of the beaker were carefully washed down with a small amount of water, 5 cm 3 of starch emulsion were added, and the iodic acid was slowly dropped into the small bulk of acid and starch, with constant agitation of the mixture, until the first tint of blue coloration appeared. The results obtained are given in Table III. These experiments indicate that the constancy of the end reaction in different titrations of equal volumes of the same solution depends to a certain degree on the volume of sodium thiosulphate taken. The results in the case of the maximum amounts vary within a range of 1.04 cm 3 , which corresponds 58 TITRATION OF SODIUM THIOSULPHATE TABLE III. VARIATION OF THE END REACTION BETWEEN ~ SODIUM THIOSULPHATE AND lODIC ACID, IN THE ABSENCE OP POTASSIUM IODIDE. Eip. NaAO, taken. HIO, introduced. Mean value. Variation. 1) cm 8 6 cm 8 28.131 cm 8 cm 8 0.19- 2) 6 27.79 0.53- 3) 6 28.03 0.29- 1 6 6 28.32 28.32 28.32 0.00 0.00 6) 6 28.71 0.39+ 7) 6 28.83 0.51+ 8) 6 28.43 0.11+ (9) 4 18.94 1 0.26+ (10) (11) 4 4 18.67 1 18.60 f 18.68 0.01- 0.18 (12) 4 18.60 J 0.08- to 0.0035 grm. of iodic acid, while the average variation is 0.25 cm 3 , corresponding to 0.0009 grm. The variation in the analyses of the smaller amounts is less, the range being 0.44 cm 3 , corresponding to 0.0015 grm., and the average variation being 0.13 cm 3 , or 0.0005 grm. The probable error which these irregularities would introduce in any series of practical analyses by this method is obviously greater than can ordinarily be permitted in iodometric work. TABLE IV. VARIATION OP THE END REACTION BETWEEN ^ SODIUM THIOSULPHATE AND lODIC ACID, IN THE PRESENCE OF POTASSIUM IODIDE. Eip. Na 2 S 2 3 taken. HIO S introduced. Mean value. Variation. (1) cm 8 6 cm 8 32.53 ] cm 8 cm 8 \ 0.05+ (2) 6 32.45 0.03- (3) (4) 6 6 32.67 32.37 32.48 0.19+ 0.11- (5) 6 32.36 0.12- (6) 6 32.50 0.02+ (7) 4 22.30 ' 0.11+ (8) (9) 4 4 21.98 22.17 22.19 i 0.21- 0.02- (10) 4 22.30 J 0.11+ BY IODIC ACID. 59 The experiments detailed in Table IV were performed exactly similarly to those of the last series except that two grams of potassium iodide were added to the sodium thiosul- phate before the titration was commenced. These experiments indicate plainly that in the presence of potassium iodide the end reaction of different titrations of equal volumes of the same solution is practically independent of the amount taken for analysis. The results in the case of the maximum amounts vary within a range of 0.31 cm 3 , or 0.0011 grm. of iodic acid, while the average variation is 0.09 cm 3 , corresponding to 0.0003 grm. The variation in the analyses of the smaller amounts is practically the same as that of the larger, the range being 0.32 cm 3 , corresponding to 0.0011 grm., and the average variation being 0.11 cm 3 , or 0.0004 grm. It is therefore evident that the presence of potassium iodide in the sodium thiosulphate to be titrated will bring the variation of the formation of the reading tint within permissible limits. A series of experiments was made to determine the nature and effect of the " after coloration" observed to take place when a solution of sodium thiosulphate, free from potassium iodide, was titrated with iodic acid to blue coloration, and then bleached with sodium thiosulphate. The titrations were performed in the usual manner except that the volume was adjusted just before the addition of the iodic acid, and the iodine that was set free after the formation of the first reading tint was destroyed at fixed intervals with measured amounts of sodium thiosulphate. The results are given in Table V. In the experiments with small volumes the evolution of iodine in any considerable quantity ceased after two or three hours, although the solution would become recolored as often as it was bleached for a number of days. The traces of iodine thus set free, however, were seldom equivalent to more than one or two drops of sodium thiosulphate. The larger volumes, however, continued to separate iodine in abundance for a very- long time. The amount of iodine thus liberated after the first coloration evidently varies with the amount of iodic acid required for the titration, although not strictly proportional to 60 TITRATION OF SODIUM THIOSULPHATE TABLE V. EFFECT OF DILUTION AND LAPSE OF TIME ON THE "AFTER COLORATION." ^03 HIO 8 intwi Na,S 2 O s introduced. taken. intro- duced. 15m. 45m. lh.45m. 2h.45in. 20 h. Total. Volume. cm 3 cm 8 cm 3 cm 3 cm 3 cm 3 cm 3 cm 8 cm 3 (1) 6 27.68 0.25 0.13 0.08 0.00 0.03 0.49 50 (2) 6 27.70 0.20 0.10 0.03 0.03 0.03 0.39 50 '3) 6 28.17 0.16 0.10 0.03 0.01 none. 0.30 50 (4) 6 27.03 0.60 0.26 0.09 0.03 none. 0.98 150 (5) 6 27.60 0.93 0.28 0.06 0.04 0.04 1.35 150 6) 6 28.60 1.34 0.46 0.17 0.03 0.14 2.14 200 7) 6 28.85 1.20 0.50 0.28 0.06 0.27 231 200 8) 6 31.63 1.46 0.74 0.10 0.21 0.23 2.74 250 9) 6 29.90 1.04 0.60 0.23 0.15 0.46 2.48 250 ( 0) 6 36.09 1.60 1.23 0.63 0.34 0.18 3.98 300 11) 6 37.59 1.65 1.33 0.72 0.27 0.10 4.07 300 (12) 6 37.23 1.92 1.05 0.64 0.33 * 300 it. Both of these quantities increase at a regular rate with the volume of the solution. To show with what accuracy the reaction between sodium thiosulphate and iodic acid may be applied to the direct estimation of one of these substances by the other, the averaged results of a large number of titrations are compared in Table VI. The operations were conducted as directed by Riegler, equal measured volumes of standardized sodium thiosulphate being titrated with iodic acid of known strength, in the presence of starch and under different conditions of time, dilution, and mass, the volume of iodic acid required to produced the blue coloration being in each case compared with the volume theoretically required by the terms of Riegler's equation. These results show plainly that the amount of iodic acid required to decompose a given amount of sodium thiosulphate may be considerably above or below that required by the terms of Riegler's equation. Thus, with small volumes, and in the absence of potassium iodide, the thiosulphate is destroyed and the separation of iodine commences when only 93 per cent of the theoretical amount of acid has been titrated. At higher * No observation. BY IODIC ACID. 61 TABLE VI. TITRATION OP SODIUM THIOSULPHATB WITH IODIC ACID. Exp. NaAO. taken. HIO S used. HI0 8 required by theory. Error. Error. KI present. Volume. cm 3 cm 3 cm 3 cm 8 per cent. grin. cm 3 (1) 4 f 18.68 20.32 1.64- 8.0- 50 (2) 6 28.32 30.48 2.16- 7.0- 60 (3) 6 27.32 30.48 3.16- 10.0- 150 (4) 6 3fc 28.73 30.48 1.75- 6.0- 200 (5) 6 30.77 30.48 0.29+ 0.01+ 250 (6) 6 36.97 30.48 6.49+ 21.0+ 300 (7) 6 27.46 30.48 3.02- 10.0- 50 8 6 26.15 30.48 4.33- 14.0- 150 (9) (10) 6 6 t" 26.50 28.16 30.48 30.48 3.98- 2.32- 13.0- 8.0- 200 250 (11) 6 1 32.93 30.48 2.45+ 8.0+ 300 (12) 4 .i 22.19 20.32 1.87+ 9.0+ 2.6 50 (13) 6 * I 32.48 30.48 2.00+ 7.0+ 2.0 60 dilutions the action is retarded, so that at 250 cm 3 very nearly the theoretical amount of acid is required to produce the first blue color, and at 300 cm 3 an excess of 21 per cent over the theoretical amount must be added. If the " after separa- tion " of iodine is considered to be a measure of the excess of iodic acid, and if its amount is accordingly applied as a correction, it appears that for all volumes below 300 cm 3 the original thiosulphate is completely destroyed when about 90 per cent of the theoretical amount of iodic acid has been added. The presence of potassium iodide in the system retards the action, so that at small volumes an excess of about 8 per cent of iodic acid must be added to completely destroy the thiosul- phate and commence the separation of iodine. It is obvious from the preceding experiments that the reaction between iodic acid and sodium thiosulphate is so indefinite in its nature, and so dependent for completeness on conditions of time, dilution, and mass, that its direct application as a means of standardizing solutions must remain impracticable. * HI0 8 added to first blue color. t Calculated by subtracting from the amount of iodic acid originally introduced, the volume of thiosulphate of equal strength required to bleach the solution after standing twenty hours. THE COMBUSTION OF ORGANIC SUBSTANCES IN THE WET WAY. BY I. K. PHELPS . IN a former paper f I have shown that carbon dioxide may be estimated iodometrically with a fair degree of accuracy. Inasmuch as this method is not dependent upon the rate of flow or rapidity of generation of the carbon dioxide, it seemed possible that some advantage might follow its application to the determination of organic carbon, oxidized by liquid reagents. Method of Oxidation by Potassium Permanganate. The first experimental test in this direction was made with oxalic acid, which was oxidized according to the well-known reaction of potassium permanganate in the presence of sulphuric acid. The apparatus used was the same as that previously described in the iodometric process, referred to above. It consisted, in the main, of an evolution flask, and an absorption flask, properly connected. As an evolution flask, a wide- mouthed flask of about 75 cm 3 capacity was used. This was closed by a doubly perforated rubber stopper, carrying a separating funnel for the introduction of liquid into the flask and a glass tube of 0.7 cm. internal diameter, which was expanded to a small bulb just above the stopper, to carry off the gas. This exit tube was joined by means of a rubber connector to a tube which passed through the rubber stopper of the absorption flask, which was an ordinary round-bottom flask of 250 cm 3 capacity. This tube ended in a valve of the * From Am. Jour. Sci., iv, 372. t Am. Jour. Sci., vol. ii, p. 70. Volume I, p. 369. COMBUSTION IN THE WET WAY. 63 Kreider pattern,* which was enclosed in a larger tube, reaching nearly to the bottom of the absorption flask. The second hole of the stopper of this absorption flask, was filled by a glass tube closed by a rubber connector and screw pinch-cock. The barium hydroxide solution for use in the determination of the carbon dioxide was prepared by filtering a cold saturated solution of the commercial salt into a large bottle, which was connected with a self-feeding burette. The solution was standardized in the manner described in my former paper by boiling with an excess of decinormal iodine solution in an ether wash bottle. The short tube of the glass ground stopper of the bottle was sealed to a Will and Varrentrapp absorption apparatus, which was charged during the operation with a solution of potassium iodide to prevent the loss of elementary iodine in the boiling ; the long tube of the bottle was used as an inlet tube and was closed externally by a rubber cap during the boiling. After cooling, the excess of iodine used was determined by titration with decinormal arsenious acid solution and the iodine lost calculated on barium hydroxide molecule for molecule. Potassium permanganate was prepared for use by dissolving the commercial salt in water, and boiling this solution, made acid with sulphuric acid, until free from carbon dioxide. Water was also prepared free from carbon dioxide by boiling distilled water until one-third had been driven off hi steam and was kept until used in full-stoppered flasks. For the first determinations of carbon, crystallized ammo- nium oxalate was weighed out and introduced into the boiling flask with 10-15 cm 3 of pure water and the flasks connected as described above with an appropriate amount of barium hydroxide solution (3-5 cm 8 in excess of the amount required to precipitate the carbon dioxide to be determined) in the absorption flask. The whole system was then evacuated with the water pump to a pressure of 200-225 mm. and the oxalate solution in the boiling flask warmed. An excess of potassium permanganate solution was then run in through * Am. Jour. Sci., 1, p. 132. Volume I, p. 307. 64 COMBUSTION OF ORGANIC SUBSTANCES the funnel tube and the mixture warmed again, when the oxidation of the oxalate was shown by the carbon dioxide evolved. The carbon dioxide was completely set free by the introduction of 10 cm 3 of sulphuric acid (1 : 4) and was driven completely to the absorption flask by boiling for five minutes. During the passage of the gas into the absorption flask, it was shaken frequently and was kept cool by standing in a dish of water and by pouring cold water over it from time to tune. If, during the boiling, any fears are enter- tained as to the strength of the vacuum in the flasks, they may be easily allayed by opening momentarily the stop-cock of the funnel tube and noting the direction of the flow of water, contained in the funnel. After the boiling was ended, the atmospheric pressure was restored by allowing air, purified from carbon dioxide by passage through potash bulbs, to enter through the funnel tube of the boiling flask. Then the flasks were disconnected and the stopper of the absorp- tion flask with its attachments was removed, the valve and its tube being carefully washed free from barium hydroxide. A second stopper, which was provided with a separating funnel, and a Will and Varrentrapp absorption apparatus, containing water to serve as a trap, was inserted into the mouth of the absorption flask and the emulsion brought to the boiling point. Decinormal iodine solution was then run in through the funnel tube in sufficient quantity to destroy the larger part of the excess of barium hydroxide and the emulsion brought to the boiling point again, after which iodine was again run hi but this time to the permanent red color of the excess of free iodine. After cooling, this excess of iodine was determined by titration with decinormal arsenious acid solution. Thus, the excess of barium hydroxide taken being determined by the iodine lost, the barium hydroxide used, now hi the form of carbonate, was known, from which the carbon dioxide which precipitated this carbonate, may be calculated. The following results were obtained by this procedure. IN THE WET WAY. TABLE I. 65 Exp. Ammonium oxalate taken. Ba0 2 H, taken. BaO 2 H 2 found. found. C0 2 calculated. Error on C0 2 . grm. grm. grm. grm. grm. grm. (1) 0.2522 0.7267 0.1170 0.1565 0.1561 0.0004+ 2 0.2542 0.7267 0.1113 0.1579 0.1574 0.0005+ 3) 0.5020 1.4535 0.2417 0.3110 0.3108 0.0002+ (4) 0.5058 1.3954 0.1753 0.3131 0.3131 O.OOOOi (5) 1.0033 2.6163 0.1955 0.6213 0.6211 0.0002+ (6) 1.0003 2.6951 0.1836 0.6189 0.6192 0.0003- (7) 1.0010 2.6163 0.2037 0.6192 0.6197 0.0005- In experiments (5) and (6), a few drops of ammonia were added to the oxalate solution before running in the permanganate ; in (3) and (7), the permanganate was treated to alkalinity with barium hydroxide ; in the remaining ex- periments, (1), (2), and (4), the permanganate was slightly acid with the sulphuric acid used in its purification from carbon dioxide, as already described. The results obtained are good and it is plain that the oxidation proceeded regularly, whether the first action of the permanganate was in the alkaline or slightly acid solution. Jones * has shown that formates may be determined volu- metrically by titration with potassium permanganate in alkaline solution. In an attempt to determine formates by the process outlined above, the pure barium salt was used. This was prepared by treating the aqueous solution of formic acid with pure barium carbonate to neutrality and crystalliz- ing the product. It was proved pure by ignition and weigh- ing in the form of carbonate. In making determinations of carbon in this formate, weighed portions were introduced into the boiling flask, together with sodium hydroxide solution, which was taken in such quantity as to more than neutralize the acid in the potassium permanganate. Naturally, the sodium hydroxide must be freed from carbonate and this was effected by treatment with an excess of barium hydroxide and filtering. An excess VOL. II. 5 * Amer. Chem. Jour., xvii, 539. 66 COMBUSTION OF ORGANIC SUBSTANCES of potassium permanganate is then run into the flask and the solution heated to boiling. An excess of dilute sulphuric acid is introduced into the mixture and the carbon dioxide, thus set free, completely driven over to the absorption flask and determined as before. Table II shows results obtained by the process. TABLE II. Exp. Barium formate taken. BaO,H 2 taken. BaO,H, found. CO. found. CO, calculated. Error on CO S . grm. gnu. grin. grm. grm. grm. ft (3) 0.5001 0.5033 1.0002 0.9302 0.9012 1.6861 0.1745 0.1402 0.1793 0.1939 0.1953 0.3867 0.1935 0.1947 0.3870 0.0004+ 0.0006+ 0.0003- (4) 1.0059 1.6279 0.1093 0.3897 0.3892 0.0005+ (5) 1.3750 2.2529 0.1820 0.5315 0.5320 0.0005- (6) 1.5028 2.4419 0.1754 0.5816 0.5814 0.0002+ These results show plainly that the carbon of formic acid may be determined accurately by the method outlined. It was found incidentally that ammonia cannot take the place of the sodium hydroxide in this process, probably because the ammonia volatilizes to the absorption flask dur- ing the boiling and is acted on by the iodine subsequently used and is thus registered as barium hydroxide. It is a well-known fact that tartrates are oxidized by permanganates. I have found, however, that when tartaric acid is treated under the conditions of analysis outlined above in acid solution, the oxidation is incomplete; but that oxidation is complete if the tartrate is heated in a solution alkaline with sodium hydroxide and then acidified with sulphuric acid. The tartrate used was a recrystallized tartar emetic, dried at 100 C. The following results were obtained with such a tartrate by this process. It seems possible to draw the general conclusion from the results recorded that organic substances which are oxidized completely by the permanganate may be determined by the process outlined above. It will also be seen that the use of IN THE WET WAY. TABLE IH. 67 Exp. Tartar emetic, taken. BaOjH, taken. BaO^ found. CO, found. calculated. Error on C0 2 . grlil. grm. gnu. grm. grm. grm, (1) 0.5051 1.2450 0.1709 0.2756 0.2751 0.0005+ (2) 0.5030 1.2226 0.1536 0.2743 0.2739 0.0004+ (3) 0.7509 1.7355 0.1401 0.4094 0.4091 0.000:3+ (4) 0.7541 1.7430 0.1410 0.4111 0.4107 0.0004+ (5) 1.0018 2.3456 0.2187 0.5458 0.5456 0.0002+ (6) 1.0005 2.2435 0.1196 0.5451 0.5450 0.0001+ the rubber stopper in the boiling flask, with due care to prevent its contact with the solution, does not introduce an appreciable error. Wanklyn and Cooper * and others have noted the fact that potassium permanganate, whether in acid or alkaline solution, will not oxidize all organic substances (acetates, for example), even at the boiling temperature. It is well known that a mixture of concentrated sulphuric and chromic acids has a much wider field of action in oxidizing organic compounds than the permanganate. With hopes of applying this reagent more widely to the determination of organic carbon, the experiments about to be recorded were tried. Method of Oxidation by Chromic Acid. A concentrated mixture of chromic and sulphuric acids, although a much more powerful oxidizer than potassium permanganate in aqueous solutions, fails to oxidize completely many organic compounds. Thus Cross and Higgin f have shown that carbohydrates are among the number of such organic substances ; and later Cross and Be van find that car- bohydrates and many other substances are oxidized to a mix- ture of carbon dioxide and monoxide. Messinger J has proved that carbon may be determined in organic compounds by pass- ing the mixed products, resulting from the oxidation with chromic and sulphuric acids, through a short combustion tube, * Phil. Mag. (5), vii, 138. t Jour. Chem. Soc., xli, 113. J Ber. Dtsch. chem. Ges., xxiii, 2756. 68 COMBUSTION OF ORGANIC SUBSTANCES filled with granular copper oxide and heated in a furnace all of which facts have been confirmed in my own experience. Ludwig * has observed that the contact of carbon monoxide with a mixture of chromic and sulphuric acids, especially when hot, results hi the oxidation of that gas to carbon dioxide. This fact suggested the idea of substituting for the apparatus described above a new form, adapted to retain the first products of oxidation in prolonged contact with the oxidizing mixture. This apparatus, shown in the ac- companying figure, by means of which, as the sequel shows, it has been found possible to extend the availability of the oxidizing mix- ture, is put together as follows: A thick-walled, round-bottomed flask of a liter's capacity, serving as an oxidizing chamber, is closed by a rubber stopper with two perforations, through one of which passes the tube of a separating funnel of about 100 cm 3 capacity. The tube of this funnel reaches nearly to the bottom of the flask and is drawn out at the lower end. A disc of platinum foil is hung in the neck of the flask, nearly closing it, and held in place by a platinum wire passing through the foil and tucked under the rubber stopper where the funnel tube enters. The second hole of the stopper is filled by the exit tube, a glass tube of 0.7 cm. internal diameter. This tube is expanded just above the stopper to a small bulb, which serves to prevent mechanical loss of the solid contents of the flask during the boiling, and is joined by means of a rubber connector (provided with a screw pinch-cock) to the inlet tube of the absorption flask, which is an ordinary 500 cm 3 round-bottomed flask. This flask is also closed by a rubber stopper with two perforations, through * Ann. Chem. (Liebig), clxii, 47. FIG. 19. IN THE WET WAY. 69 one of which passes the inlet tube described above and through the other the exit tube, which is also enlarged to a small bulb just above the stopper and is closed by a rubber connector and screw pinch-cock. The glass ground stopper of the funnel tube is carefully cleaned and lubricated with a thick solution of metaphosphoric acid. Instead of getting the vacuum by the water pump, it may be got almost as quickly and certainly more simply by boiling the water in the evolution flask and the barium hydroxide solution in the absorption flask at the same tune both flasks being connected, ready for making a determination. When steam issues in good quantity from the exit tube, the burner is removed from under the evolution flask, the attached pinch- cock closed, the burner under the absorption flask taken away, and the screw pinch-cock upon the exit tube quickly closed. The flasks are then allowed to cool. In making a determination, the organic substance is weighed out in a counterbalanced bulb, so thin that it may be easily broken later and made with a wide mouth for convenience in introducing the solid substance. After the substance is weighed, the mouth of the bulb is sealed by heating hi a small blow-pipe flame and the tube introduced into the evolu- tion flask, together with an amount of pure potassium dichro- mate, which is known to be hi excess of that required to oxidize the organic substance. The flasks are connected, as already described, with an appropriate amount of barium hydroxide solution in the absorption flask and 10 cm 3 of pure water in the evolution flask, and the vacuum is obtained (as described above) by boiling both flasks, the boiling being stopped when the water in the evolution flask has decreased to 2 or 3 cm 3 . Naturally, this boiling must be so regulated as not to allow loss of the solid material in either flask. The vacuum obtained, the tube containing the organic substance is broken by shaking the flask, and 20 cm 3 of concentrated sulphuric acid, previously purified from organic material by heating to the fuming point with a few crystals of potassium dichromate, are run in through the funnel tube, when reduc- 70 COMBUSTION OF ORGANIC SUBSTANCES tion of the chromic acid soon becomes evident. While still hot, the acid is shaken in the flask violently, the platinum foil hung in the neck serving to protect the rubber stopper. The flask is warmed to approximately 105 C., the highest temper- ature to which, as shown by Cross and Be van,* a mixture of chromic and sulphuric acids may be safely heated without the disengagement of oxygen gas. Water is then run in until the crystals of chromic anhydride have disappeared and the danger of the evolution of oxygen is past. The solution is heated to its boiling point, care being taken that it shall not get under pressure, which can easily be observed by opening momentarily the stop-cock of the funnel tube and noting the direction of the flow of water, contained in the funnel. The flask is shaken and heated alternately for five minutes a period of tune which appears to be sufficient to bring about the oxidation of the small amount of carbon monoxide origi- nally produced. Then more water (60-70 cm 3 ) is introduced through the funnel, and the stop-cock between the boiling and absorption flasks is opened, when the carbon dioxide enters the TABLE IV. Exp. Substance taken. Ba0 2 Hj taken. Ba0 2 H, found. C0 2 found. CO, calculated. Error on C0 2 . ANALYSIS OF AMMONIUM OXALATB. (1) (2) (3) (4) (5) grm. 0.5009 0.5006 0.5005 1.0002 1.0010 grm. 1.3534 1.3400 1.3400 2.5460 2.5192 grm. 0.1469 0.1308 0.1343 0.1347 0.1094 grin. 0.3097 0.3103 0.3094 0.6188 0.6185 grm. 0.3101 0.3099 0.3098 0.6192 0.6197 1*111. 0.0004- 0.0004+ 0.0004- 0.0004- 0.0012- ANALYSIS OF CANE SUGAR. Pj % 0.2001 0.2000 0.2001 0.2014 1.3926 1.3926 1.3926 1.3400 0.1905 0.1936 0.1857 0.1279 0.3085 0.3077 0.3097 0.3111 0.3088 0.3086 0.3088 0.3108 0.0003- 0.0009- 0.0009+ 0.0003-f- absorption flask, which is kept cool and shaken as before. The contents of the evolution flask are then heated to boiling * Jour. Chem. Soc., liii, 889. IN THE WET WAY. 71 and a slow current of air, freed from carbon dioxide by passing through potash bulbs, is allowed to enter through the funnel tube to keep the liquid from undue bumping. The boiling is continued for fifteen minutes, after which the excess of barium hydroxide is determined iodometrically and thus the carbon dioxide present estimated as before. Table IV shows results obtained by the treatment of crystallized ammonium oxalate and cane sugar, recrystallized from dilute alcoholic solution, in this manner. The results are evidently very satisfactory. The Determination of the Oxygen required to Oxidize an Organic /Substance. Several different methods have been proposed for estimating the oxygen present in organic substances, depending, in gen- eral, upon the determination of the oxygen which must be supplied to burn the substance to a known amount of carbon dioxide and water thus discovering by difference the oxygen originally contained in the substance. Lavoisier is said to have measured directly the oxygen used in burning organic sub- stances ; Gay-Lussac and The'nard determined the oxygen used by measuring the amount of potassium chlorate reduced hi burning the organic compound; Baumhauer* determined the oxygen used by measuring the volume of oxygen entering the combustion furnace and subtracting the measure of the gas coming from the combustion tube, which was set up according to the well known method for determining carbon and hydro- gen ; Stromeyerf determined the amount of copper reduced by the ignition of the substance in copper oxide; LadenburgJ oxidized the substance by heating in a sealed tube with a known amount of iodic acid, determining at the end of the operation the amount of iodic acid left; Mitscherlich has estimated the oxygen in organic substances directly by decom- posing the substance by ignition hi a stream of chlorine gas, / * Ann. Chem. (Liebig), xc, 228. t Ann. Chera. (Liebig), cxvii, 247. \ Ann. Chem. (Liebig), cxxxv, 1. Ann. Phys. ccvi, 536 (1867). 72 COMBUSTION OF ORGANIC SUBSTANCES estimating the oxygen content by determining the resulting carbon dioxide and monoxide. As it has been shown in the work described that carbon may be determined in organic substances by oxidation with chromic and sulphuric acids without the evolution of oxygen gas, it would seem that the determination of the oxygen in the sub- stance might be effected by determining the amount of chromic acid used in the operation, taking into consideration the products of combustion. This can be readily accomplished by taking a weighed amount of pure potassium dichromate as the oxidizing agent and determining, at the end of the operation, by treatment of the residue with hydrochloric acid, absorp- tion of the chlorine evolved in an alkaline arsenite of known strength, and titration of the excess of that substance with decinormal iodine solution, the amount of chromic acid left. To test the accuracy of the determination of chromic acid under these conditions of analysis, weighed portions of pure fused potassium dichromate were introduced into a Voit flask, whose outlet tube was sealed to the inlet tube of a Drexel wash bottle, the outlet of which, in turn, was sealed to a Will and Varrentrapp absorption apparatus. An amount of hydrochloric acid, more than enough to completely reduce the chromate (15-40 cm 3 of the strongest acid), was added with 20 cm 3 of strong sulphuric acid and the total volume made up to 120-140 cm 3 of liquid. The sulphuric acid used here was purified from carbonaceous matter (as in the carbon determination above) by heating with a few crystals of potas- sium dichromate, the excess of which was destroyed by hold- ing the acid at the fuming point for about two hours, when a portion diluted with water gave no color with potassium iodide and starch paste. Pure arsenious oxide, in amount slightly in excess of that required to take up the oxygen to be given up by the chromate, was dissolved by the aid of heat in a solution of pure sodium hydroxide, taken in such quantity as to more than neutralize the arsenious acid and the hydrochloric acid used to reduce the chromate, and this solution was introduced into the Drexel wash bottle. The IN THE WET WAY. 73 flask was then connected with the wash bottle, using a thick solution of metaphosphoric acid to lute the joint between the flask and its stopper. The absorption apparatus was charged with a dilute solution of sodium hydroxide. Carbon dioxide was generated in a Kipp generator by the action of hydro- chloric acid on marble and purified from reducing matter by bubbling through a strong solution of iodine in potassium iodide and finally washed with a solution of potassium iodide alone. A slow stream of this purified carbon dioxide was allowed to enter the inlet tube of the Voit flask, the con- tents of which were then boiled. When concentration to a volume of 30-40 cm 3 was reached, the boiling was discon- tinued and, after cooling and disconnecting the flask, the contents of the receiver were made acid with sulphuric acid and then alkaline with acid potassium carbonate, and the excess of arsenite was determined by titration with deci- normal iodine solution. Sometimes during the reduction of the chromic acid, the red fumes of the chlorochromic an- hydride Volatilized to the receiver; but since the chromic acid thus produced is reduced later by the arsenite,* this transfer is of no account in the working of the process. The following results were thus obtained. TABLE V. Exp. KjCr 2 O 7 taken. As 2 3 taken. found. K 2 Cr 2 7 found. Error on K 2 Cr a 7 . gnu. grill. grin. grm. grm. (1) 5.0002 5.1025 0.1144 4.9447 0.0555- (2) 5.0018 5.0799 0.0526 4.9849 0.0169- (3) 5.0005 5.0801 0.0582 4.9782 0.0223- U) 5.0013 5.0706 0.0908 4.9365 0.0648- The cause of the error shown in these experiments was traced finally to too great concentration of the sulphuric acid in the process. When the boiling begins the chromate is reduced gradually and if the evaporation of the water is pushed too rapidly, the sulphuric acid may reach a strength * Browning Am. Jour. Sci., i, 35. Volume I, p. 344 . 74 COMBUSTION OF ORGANIC SUBSTANCES at which it begins to cause the reduction of the chromic acid with the evolution of oxygen instead of chlorine. The obvious remedy is to conduct the boiling operation more slowly. It was found that, if from 5-6 hours' time was taken for the proper concentration of the contents of the Voit flask, the presence of the sulphuric acid worked no harm, as will be seen from the following results. Experi- ments (1) and (5) were made with 5 cm 3 of sulphuric acid present and the others with 20 cm 3 , as used before. TABLE VI. Bxp. K 2 Cr 2 T taken. As 2 s taken. As 2 3 found. K*Cr 7 found. Error on K 2 Cr 2 0,- grm. grm. grin. grm. grm. (1) 1.0004 1.0500 0.0398 1.0014 0.0010+ (2) 1.0007 1.0531 0.0437 1.0006 0.0001- (3) 2.0013 2.0501 0.0299 2.0026 0.0013+ 4 2.0037 2.0727 0.0502 2.0049 0.0012+ (6) 5.0020 5.1002 0.0495 5.0068 0.0048+ (6) 5.0037 6.1018 0.0513 5.0066 0.0029+ In applying this method to the determination of oxygen used in the oxidation of an organic substance, the carbon determination was made as already described, the amount of water used being such as to leave 60-80 cm 3 of liquid in the boiling flask after the carbon dioxide had been driven to the absorption flask by boiling. This liquid was then washed into the Voit flask and the chromic acid remaining determined by a second distillation (this time with hydrochloric acid) in the manner described above. In each of the experiments recorded below, 20 cm 8 of purified sulphuric acid were used in the carbon determination and 35 cm 3 of hydrochloric acid (sp. gr. 1.2) in the chromic acid determination. The ammo- nium oxalate used was the pure crystallized salt ; the phthalic acid was recrystallized from its water solution and dried for a short time over sulphuric acid ; the cane sugar was selected crystals of rock candy, recrystallized from dilute alcoholic solution and dried for a long time over sulphuric acid; the IN THE WET WAY. 75 paper was ashless filter paper, dried to a constant weight over sulphuric acid; the tartar emetic was recrystallized from water solution and air dried; the barium formate was pre- pared by treating formic acid with an excess of pure bar- ium carbonate, filtering hot and allowing the product to crystallize. TABLE VII. Exp. Sub- staiice taken. found. Error on CO, K 2 Cr 2 7 taken. A-,0, taken. found. Oxygen used. Oxygen required by theory. Error on Oxygen. ANALYSIS OF AMMONIUM OXALATE. (1) (2) grin. 1.0122 1.0019 grin. 0.8265 0.6212 grm. 0.0001- 0.0010+ grm. 2.0009 2.0002 grin. 1.3002 1.3517 grm. 0.0000 0.0440 grm. 0.1160 0.1147 grm. 0.1139 0.1128 grm. 0.0021+ 0.0019+ ANALYSIS OF PHTHALIC ACID. (1) (2) 0.1002 0.1093 0.2138 0.2324 0.0014+ 0.0007+ 2.0012 2.0000 1.2004 1.1031 0.0814 0.0634 0.1456 0.1582 0.1448 0.1580 0.0008+ 0.0002+ ANALYSIS OF CANE SUGAR. (1) (2) 0.2025 0.4012 0.3117 0.6166 0.0008- 0.0024- 3.0000 5.0000 1.7002 2.3022 0.0796 0.0366 0.2275 0.4495 0.2273 0.4502 0.0002+ 0.0007- ANALYSIS OF PAPER. (1) (2) 0.3034 0.4523 0.4932 0.7334 0.0010- 0.0033- 3.5015 5.0035 1.4017 1.8000 0.0879 0.0710 0.3589 0.5368 0.3598 0.5358 0.0009- 0.0010+ ANALYSIS OF TARTAR EMETIC. (1) (2) 0.5057 1.0099 0.2671 0.5321 0.0009- 0.0030- 2.5018 3.5003 1.7000 1.7520 0.0766 0.0198 0.1459 0.2911 0.1462 0.2919 0.0003- 0.0008- ANALYSIS OF BARIUM FORMATE. (1) (2) 1.0079 1.5014 0.3906 0.5814 0.0006+ 0.0005+ 3.0026 3.0010 2.2002 1.8080 0.0496 0.0890 0.1423 0.2118 0.1422 0.2118 0.0001+ 0.0000 From these results, it will be seen that the process works with accuracy upon a great variety of organic substances. It was found impossible, however, to determine the elements in bodies which are at the same time volatile and hard to oxidize ; for instance, ether oxidizes easily to acetic acid but difficultly beyond that stage; although the liquid acid 76 COMBUSTION OF ORGANIC SUBSTANCES. is oxidized vigorously by chromic and sulphuric acids, the gaseous acid is hardly attacked at the temperature used; naphthaline was also found to be volatilized, and hence not attacked, to such an extent as to render its determination by this process valueless. XI THE ESTIMATION OF MANGANESE AS THE SULPHATE AND AS THE OXIDES. BY F. A. GOOCH AND MARTHA AUSTIN.* THE estimation of manganese by the conversion of salts of that element with volatile acids to the form of the anhydrous sulphate by the action of an excess of sulphuric acid, evapora- tion, and gentle heating was formerly a recognized procedure. On the authority of Rose,f however, this method was set aside on account of the supposed difficulty of removing the excess of acid without disturbing the composition of the normal salt. Thus, Oesten, working under Rose's direction, obtained, upon submitting the crystalline hydrous sulphate MnSO 4 . 5H 2 O to heat, results which may be summarized and compared with the results obtained by Rose's sulphide method (the ignition of the residue with sulphur in hydrogen) as follows : MnS0 4 . 5H 2 taken. MnS0 4 found. Theory. Error. MnS found. Theory. Error. grm. grm. gnu. grm. grm. grin. grm. 1.659 1.043J 1.037 0.006+ 0.597 0.595 0.002+ 1.023 0.014- 1.481 0.934} 0.926 0.008+ 0.905 0.021- 0.725H 0.201- 1.430 0.880 0.893 0.013- 0.509 0.512 0.003- The residues remaining after the gentle ignition of the sulphate weighed apparently several milligrams more than should have been the case if the salt had been reduced to the normal anhydrous sulphate. At higher temperatures the sulphate * From Am. Jour. Sci., v, 209. t Ann. Phys., clxxxvi, 125 (1860). t Ignited gently. Ignited at low red heat. || Ignited at strong red heat. 78 ESTIMATION OF MANGANESE AS THE turned brown and lost altogether too much weight. A comparison of the errors of the process in which the ignition was at low temperature with those of the sulphide process would seem to justify Rose's rejection of the former method for the latter. Upon recalculating these results, however, using atomic weights in use at present viz. : Mn = 55, S = 32.06, O = 16 it becomes plain that the errors of the two processes, as shown in Oesten's work, are not very different numerically, though with opposite signs. MnS0 4 . 5H 2 taken. MnS0 4 found. Theory. Error. MnS found. Theory. Error. grin. grm. grm. grm. grm. grill. grm. 1.659 1.043 1.039 0.004+ 0.597 0.599 0.002- 1.481 0.934 0.928 0.006+ 1.430 0.509 0.516 0.007- The most uncertain element in these experiments is the difficulty, well-recognized at present, of getting the hydrous manganous sulphate, upon which the experiments were made, in a perfectly definite condition of hydration. Volhard* subsequently studied the sulphate process, and showed that manganous sulphate may be dehydrated, separated from an excess of sulphuric acid, and brought into definite condition for weighing as the anhydrous salt by careful and protracted heating over a special device of his own a ring burner enclosed in a sheet-iron casing. Thus, on evaporating and dehydrating a solution of pure neutral manganous sul- phate, Volhard obtained the results recorded in the following statement : Residue of MnS0 4 left by evaporation and dehydration . 0.1635 " after treatment with 3 drops of H 2 S0 4 and heating 3 hours 0.1635 after heating 2 hours 0.1638 " after treatment with 4 drops of H 2 S0 4 and heating 2 hours 0.1635 " after heating 3 hours 0.1635 * Ann. Chem. (Liebig), cxcviii, 328. SULPHATE AND AS THE OXIDES. 79 Similar results were obtained on evaporating with sulphuric acid and igniting in like manner an aqueous solution of manganous chloride. Volhard's recommendation of the method has not secured for it the acceptance which its simplicity and exactness would seem to demand possibly because the periods of ignition appear to be considerable and the manner of heating special. In our own experiments with the sulphate process we have found that special apparatus is unnecessary, that the time of treatment may be short, and that the process is in every respect simple as well as very exact. We took for a starting point manganous chloride prepared in the manner to be detailed. An aqueous solution of the so-called pure manganous chloride of commerce was boiled with pure manganous carbonate (to throw out aluminum, iron, and chromium), filtered and precipitated with ammonium sulphide. The precipitate thus obtained was dissolved in a very slight excess of hydrochloric acid (to leave behind possible traces of nickel, cobalt, and copper), the solution was boiled to expel hydrogen sulphide and precipitated with sodium carbonate. The manganous carbonate thus thrown down was boiled repeatedly with successive portions of water, and washed until the washings were free from chloride. The greater part of this purified carbonate was dissolved in the least possible amount of pure hydrochloric acid, the reserved portion of the carbonate was added, the mixture was boiled, and the solution of the purified and neutral manganous chloride was filtered from the excess of undissolved carbonate. Definite portions of this solution were precipitated with silver nitrate, and from the weight of the silver chloride thus obtained the amount of manganous chloride present was calculated. Portions of the solution thus standardized were drawn, for our experiments, from a burette into a weighed platinum crucible, sulphuric acid was added in amount more than equivalent to the manganese, the solution was evaporated on the water-bath until the water was removed, and then, supported by means of a porcelain ring, or triangle, within a larger porcelain crucible used as a radiator so that 80 ESTIMATION OF MANGANESE AS THE the bottom and walls of the one were distant from the bottom and walls of the other by an interval of about 1 cm., the crucible was heated more strongly. The outer porcelain crucible may be heated over a good Bunsen flame to a red heat without risk of overheating the manganese sulphate within the inner crucible, and the ignition may proceed as rapidly as is consistent with the avoidance of mechanical loss by spattering. The results obtained by treatment of equal portions (50 cm 3 ) of the same solution are given, together with the results of standardizing the solution by precipitation with silver nitrate, hi columns A of the following table. In the other columns are given comparative results got in the treatment of equal portions of several other solutions employed subsequently in other work. MnSO 4 MnSO 4 calculated found by Exp. from AgCl found in 50 cm 3 of Exp. treatment of50cm3of solution A Exp. MnS0 4 found by treatment of 50 cm 8 of various solutions with H 2 SO 4 . solution A. withH 2 SO 4 . A. A. B. C. D. E. F. G. grm. grm. grm. grm. grm. grm. grm. grm. (1) 0.3518 (1) 0.3513 (1) 0.3100 0.3256 0.3534 0.3524 0.3355 0.5475 (2) 0.3512 (2) 0.3514 (2) 0.3104 0.3254 0.3543 0.3520 0.3357 0.5476 (3) 0.3518 (3) 0.3096 These results show plainly that the process of estimating manganese in the form of the anhydrous sulphate is both simple and accurate. The estimation of manganese as the manganese-manganic oxide Mn 3 O 4 , has been so frequently criticised unfavorably that the method may be said to have passed from very general use excepting in certain cases in which the directness of the process is a temptation to incur the risk of some uncertainty. The production of the other oxides of manganese in definite condition is thought to be even more uncertain. Manganese dioxide, MnO 2 , begins, as Wright and Menke have shown * to * Jour. Chem. Soc., xxx, 775. SULPHATE AND AS THE OXIDES. 81 lose oxygen at a temperature (about 210 C.) to which the hydrated oxide must be heated to free it from water, or very nearly that at which the nitrate is converted into the dioxide ; so that the chance of producing an undecomposed dioxide by the ignition of the hydrated dioxide (the form in which the dioxide generally appears in analytical processes), or of the nitrate, is small. Manganic oxide, Mn 2 O 8 , is produced, it is said, from the other oxides by ignition at a low red heat under the ordinary conditions. The manganoso-manganic oxide, Mn 3 O 4 , forms, presumably, when an oxide of manganese is submitted, under ordinary atmospheric conditions, to the high heat of the blast>lamp. If the proportion of oxygen in the surrounding atmosphere is reduced below the normal, the conversion of Mn 2 O 3 to Mn 3 O 4 goes on very easily, as Dittmar has shown,* at a temperature between the melting points of silver and aluminum, while if the proportion of oxygen in the surrounding atmosphere rises much above the normal, the reverse change, from Mn 3 O 4 to Mn 2 O 3 tends to take place at the same temperature. It is not surprising, in view of these phenomena, that the estimation of manganese as the oxide Mn 3 O 4 should have fallen into disrepute; and yet, if the condition most favorable to the production of that oxide a low proportion of oxygen in the surrounding air can be maintained during the ignition, it is not impossible that the indications of the process might prove to be, under the con- ditions, reasonably accurate. Now, this may be exactly the condition of affairs when the ignition takes place ordinarily ; for, if the products of combustion displace the ordinary air about the crucible, the proportion of oxygen about the oxide falls to a low limit. We have made the experiment of enclosing the ignited crucible within an inverted crucible, so that the products of combustion should be held immediately about and above the ignited oxide, but our experience has shown that the object in view is attained, apparently, quite as well when the ignition is so arranged that the crucible simply rests well within the upper part of the flame of a strong Bunsen * Jour. Chem. Soc., xvii, 294. VOL. ii. 6 82 ESTIMATION OF MANGANESE AS THE burner, or blast-lamp, in such manner that an oxidizing flame covers nearly the entire wall of the crucible. In the following experiments we have put to the test this matter of getting definitely the different oxides of manganese. We started with a known amount of pure anhydrous sulphate, prepared from the pure chloride in the manner previously described. This sulphate was converted by ignition into the oxide presumably the oxide Mn 3 O 4 the containing crucible being well within the upper flame of a powerful burner. In the next step, this oxide was further oxidized by moistening it with nitric acid and heating the residue gently until the evolution of fumes ceased, the containing crucible being placed well above a porcelain crucible used as a radiator and heated so that only the bottom showed a faint red heat. In this process the attempt was made to arrest the ignition at the point where the anhydrous dioxide was produced. As the table shows, and as would be expected, this attempt was only occasionally and partly successful. The residue of the last process was then submitted to a higher heat. The platinum crucible containing the oxide was placed within and touching the bottom of a larger porcelain crucible which was heated to redness. Under these conditions the temperature should not be too hot, and the products of combustion should naturally be thrown so far away from the oxide undergoing ignition that circumstances should be favorable for the formation of the oxide Mn 2 O 3 . The event proved that the attainment of the exact condition corresponding to the symbol Mn 2 O 3 is a matter of some uncertainty. Next, the oxide was subjected to the highest heat of a strong Bunsen burner (or in some cases, the broad flame of a blast lamp), the crucible being well surrounded by the products of combustion. The results of this treatment, it will be seen, agree, with a single exception out of ten experiments, reason- ably well with the theory for Mn 8 O 4 . By treating the final oxide with nitric acid and repeating the cycle of operations described, the observations of the phenomena were multiplied, until, finally, the oxide formed SULPHATE AND AS THE OXIDES. 83 I 4 o'oo CO CO CO ooo 4 I OOO II II oo oo li rH t~ . (2) 0.0580 0.0577 0.0003- 0.1034 _ t t (3) f . . 0.1019 0.1016 0.0003- t (4) , . . . . . 0.1010 0.1007 0.0003- . . , (5) . . . . 0.1100 0.1095 0.0005- (6) 0.0572 0.0572 0.0000 0.1014 0.1027 0.0013+ 0.0007- 12 (7) 0.0563 0.0550 0.0013- 0.1026 0.1038 0.0012+ 0.0008- 16 8) 0.0577 0.0576 0.0001- 0.1000 0.1014 0.0014+ 0.0006- 16 (9) 0.0559 0.0558 0.0001- 0.1020 01035 0.0015+ 0.0005- 16 (10) 0.0563 0.0556 0.0007- 0.2024 0.2046 0.0022+ 0.0002+ 20 (11) 0.1111 0.1107 0.0004- 0.2092 0.2116 0.0024+ 0.0004+ 20 In Table I, (6) to (11), are given the results of experiments in which both the aluminum and zinc were determined, the former, as described, by precipitating as the hydrous chloride and weighing as the oxide, and the latter by carefully evapo- rating the strongly acid filtrate (best with a small current of air playing on the surface of the liquid to avoid spattering due to the too violent evolution of the ether and gaseous acid) and finally converting the chloride through the nitrate into the oxide. It is, of course, absolutely necessary that the treatment with nitric acid shall be thorough, so that no zinc chloride may remain to volatilize when the residue is ignited. On account of the danger to platinum from the aqua regia generated by the action of nitric acid on zinc chloride, the evaporations of the filtrates from the aluminum chloride and the treatment BY HYDROCHLORIC ACID. 109 with nitric acid were carried on in porcelain and the residual nitrate was transferred to a small crucible for ignition. In this process the porcelain was evidently attacked somewhat, so that the residual nitrate was slightly contaminated with material from the large porcelain dish. This fact accounts for the high results given in the first column of errors. However, on introducing a correction (0.0020) found by carrying through the process in blank with the quantities of reagents employed in the regular process, the results on zinc, slightly deficient, agree closely with those obtained in (3)-(5), Table I, where the zinc nitrate was converted directly to the oxide without the previous evaporation in porcelain of a large volume of strongly acid liquid. The errors thus corrected stand in another column of the table. These results show clearly that aluminum and zinc may be separated from one another by the action of hydrochloric acid gas in aqueous ethereal solution with a reasonable degree of accuracy. Separation of Aluminum from Copper, Mercury, and Bismuth. The separation of aluminum from copper, mercury, and bismuth does not differ materially from the separation of TABLE II. Exp. A1 2 O 8 taken as the chloride. found. Error. CuO taken. CuO found. Error. HgCl, taken. Bi 2 3 taken. gj (3) 18 (6) gnu. 0.0576 0.0561 0.0570 0.0548 0.0565 0.0576 gnn. 0.0571 0.0557 0.0574 0.0557 0.0571 0.0577 gnn. 0.0005- 0.0004- 0.0004+ 0.0009+ 0.0006+ 0.0001+ gnn. 0.0500 0.0400 gnu. gnn. gTIH. 0.1000 0.1000 grin. o.iobo 0.2000 J3 (9) 10) 11) 12) 13) 0.0558 0.0538 0.0566 0.0577 0.0545 0.0536 0.0562 0.0575 0.0013- 0.0002- 0.0004- 0.0002- 0.0437 0.0359 0.0345 0.0319 0.0343 0.0337 0.0651 0.0432 0.0359 0.0340 0.0324 0.0356 0.0349 0.0644 0.0005- 0.0000 0.0005- 0.0005+ 0.0013+ 0.0012+ 0.0007- 110 FURTHER SEPARATION OF ALUMINUM, ETC. aluminum and zinc. Aluminum chloride is precipitated quantitatively in the presence of pure salts of these elements as shown in experiments of Table II. In determining the copper in the acid filtrate it was found advantageous to weigh as the oxide, but to arrive at that condition through the sulphate rather than through the nitrate (which was the transition salt in the case of zinc), as this process can be carried on safely in platinum. In Table II, (10)-(13), are given results of experiments in which the aluminum was determined as previously described by precipitation as the hydrous chloride and conversion to the oxide. The acid nitrate was evaporated in platinum and the copper determined by treating the residue with a few drops of strong sulphuric acid, heating gently to drive off the excess of sulphuric acid, and then igniting the sulphate to the oxide at a red heat. That the copper sulphate is converted to the oxide by ignition at a red heat over a Bunsen burner is shown in experiments (7) to (9) of Table II. XVI THE IODOMETRIC DETERMINATION OF MOLYBDENUM. BY F. A. GOOCH AND JOHN T. NORTON JR.* A PROCESS for the iodometric determination of molybdic acid, which consists in treating a soluble molybdate in a Bunsen distillation-apparatus with potassium iodide and hydrochloric acid, has been advocated by Friedheim and Euler.f Accord- ing to this process the molybdate, containing from 0.2 grm. to 0.3 grm. of molybdenum trioxide, is treated with from 0.5 grm. to 0.75 grm. of potassium iodide and enough hydro- chloric acid, of sp. gr. 1.12, to fill two-thirds of the flask of the apparatus. The liquid is warmed until heavy vapors of iodine fill the flask and then boiled until iodine vapor is no longer visible and the color of the liquid residue is a clear green. The iodine liberated is collected in the distillate and titrated with sodium thiosulphate, every atom of iodine found indicating presumably the reduction of a molecule of molybdic acid to the condition of the pentoxide Mo 2 O 5 . It was pointed out in a former article from this laboratory, t that greater precaution than was taken by Friedheim and Euler is necessary in order that the reduction may proceed according to theory, and that the iodine collected may serve as a reliable measure of the molybdic acid. It was found that the green color of the liquid comes gradually and that it may develop distinctly before the molybdic acid is fully reduced. It was found, also, that since even a trace of oxygen liberates iodine from the hot mixture of potassium * From Am. Jour. Sci., vi, 168. t Ber. Dtsch. chem. Ges., xxviii, 2066. t Gooch and Fairbanks, Am. Jour. Sci., ii, 156. Volume I, p. 375. 112 THE IODOMETRIC DETERMINATION iodide and hydrochloric acid of the strength employed, it is not sufficient to rely upon the volatilization of iodine to expel the air originally in the apparatus, but that it is essential to conduct the distillation in an atmosphere devoid of oxygen. The suggestion was made therefore that the operation should be carried on in a current of carbon dioxide and that the mixture, constituted definitely, should be boiled between stated limits of concentration which were determined by experiment. It was found that when amounts of a soluble molybdate containing less than 0.3 grm. of molybdenum trioxide are treated with potassium iodide, not exceeding the theoretical proportion by more than 0.5 grm., and 40 cm 3 of a mixture of the strongest hydrochloric acid and water in equal parts, the reduction proceeds with a fair degree of regularity and is practically complete when the volume has diminished to 25 cm 3 . If the operation is properly conducted in an atmosphere of carbon dioxide, it was shown that the iodine in the distillate may be trusted to indicate the molybdic acid within reasonable limits of accuracy. It appeared, how- ever, that too great an excess of potassium iodide tends to induce excessive reduction, and that the same tendency shows when the liquid is concentrated to too low a limit. To this criticism Friedheim took exception * and contrasted, to their disadvantage, our results by the modified method with those of Friedheim and Euler by the original method. It became necessary, therefore, to point out f the fact that of the results published by Friedheim and Euler, upon which reliance was placed to prove the reliability of their method, five out of seven in one series and one out of five in another series had been calculated incorrectly from data given. Another series of six determinations was, however, apparently faultless in this respect. More recently f Euler has explained that the errors were not really arithmetical. Two of them may be presumed, inferentially, to be due to careless copying * Ber. Dtsch. chem. Ges., xxix, 2981. t Gooch, Am. Jour. Sci., iii, 237. J Zeit. anorg. Chem., xv, 454. OF MOLYBDENUM. 113 or proof-reading; and four, we are told by Euler, were introduced into the series by mistake, and actually represent (as Prof. Friedheim kindly informs him) the analysis of a sample of ammonium molybdate of undetermined constitution : that is to say, the figures now given by Euler represent the original percentages of molybdenum trioxide which had been changed by some unconscious process from 80.62 per cent to 81.85 per cent. 80.71 " 81.69 80.63 " " 81.67 " 80.78 " " 81.78 Curiously enough, Euler's corrected figures, as given here, are still affected by trifling arithmetical errors of from one to four units in the second decimal place. The agreement of these results among themselves is no proof of the correctness of the process of analysis. The great variation between the average percentage of molybdenum trioxide in ammonium molybdate as found by Euler in a molybdate of known con- stitution and the percentage of the trioxide as found by Friedheim (if we understand Euler aright) may be due con- ceivably to either or both of two causes, viz. : the change of material analyzed, and the change of operator or conduct of the operation. We shall show in the following account of our work that the exact control of the conditions of treatment, along the lines laid down formerly, is actually essential to the reduction of molybdic acid according to the theory of the process. Our experiments were made with ammonium molybdate twice recrystallized from the presumably pure salt. The con- stitution of the preparation was determined by careful ignition per se, and, for greater security, with sodium tungstate free from carbonate. It contained 81.83 per cent of molybdenum trioxide. The potassium iodide which we used was prepared by act- ing with re-sublimed iodine upon iron wire, and precipitating by potassium carbonate the proportions of iodine and iron VOL. II. 8 114 THE IODOMETRIC DETERMINATION having been adjusted to secure the formation of the hydrous magnetic oxide of iron. The filtrate from the iron hydroxide gave on evaporation and crystallization potassium iodide which was free from iodate. The hydrochloric acid was taken of sp. gr. 1.12, because this is the strength used by Friedheim and Euler. The sodium thiosulphate employed was taken in nearly decinormal solution, and was standardized by running it into an approximately decinormal solution of iodine which had been determined by comparison with decinormal arsenious acid made from carefully re-sublimed arsenious oxide. We chose this method of standardizing the introduction of the thiosulphate into the iodine rather than the reverse opera- tion, in order that the conditions of the actual analysis might be followed in the standardization. The distillation apparatus was constructed with sealed or ground joints of glass wherever contact with iodine was a possibility. It was made by sealing together a separating funnel A, a 100 cm 3 Voit flask B, a Drexel wash-bottle C, and a bulbed trap g, as shown in the figure. Upon the side of the distillation- flask B was pasted a gradu- ated scale by means of which the volume of the liquid within the flask might be known at any time. Carbon dioxide, generated in a Kipp apparatus by the action of dilute hydrochloric acid (carrying in solution cuprous chloride to take up free oxygen) upon marble previously boiled in water, was passed through the apparatus before and during the operation, so that it was possible to interrupt the process of boiling at any point of concentration, to remove the receiver by easy manipulation, to replace the receiver, and to continue FIG. 20. OF MOLYBDENUM. 115 the distillation without danger of admitting air to the distilla- tion flask. In experiments to be described (1) to (5) of the table the proportions of potassium iodide and molybdic acid, and the strength of the hydrochloric acid recommended by Fried- heim and Euler were retained. The essential change of condition is the removal of atmospheric air from the distillation flask before the acid is admitted to contact with the other reagents. Potassium iodide (3 grm.) and water (200 cm 3 ) were put into the receiver C, and a little of this solution was allowed to flow into the trap g. Ammonium molybdate care- fully weighed (0.3 grm.) and potassium iodide (0.5 grm. to 0.75 grm.) were introduced into the distillation flask B, the apparatus was connected as shown in the figure, and carbon dioxide was passed freely through the whole apparatus for some minutes. The stop-cock d, between the bulb of the funnel A and the flask B, was closed, and hydrochloric acid (40 cm 3 , sp. gr. 1.12) was poured into the funnel ; the air above the liquid in the funnel was displaced by carbon dioxide through the space between the neck of the funnel and the loosely adjusted stopper carrying the inlet tube ; the connec- tion between the funnel and inlet tube was tightened, the stop-cock opened, and the acid, under the pressure of carbon dioxide, was permitted to flow into the flask. In this way the acid, iodide, and molybdate were made to interact with little danger of the presence of oxygen. The flask was heated by the Bunsen burner, and the iodine evolved, passing over quietly in the slow current of carbon dioxide, collected in the receiver. The liquid was boiled until fumes of iodine were no longer visible above the liquid in the flask and connecting tubes backed by a ground of white, and then a full minute more. At this stage, the green color of the liquid having developed fully, the apparatus was permitted to cool, the current of carbon dioxide was increased, the cap of the receiver was loosened at /, the contents of the trap were washed back into the receiver, the rest of the apparatus was lifted bodily from the receiver, the liquid adhering to the inlet tube was 116 THE IODOMETRIC DETERMINATION washed off into the receiver, and the end of the tube was dipped immediately into a solution of potassium iodide. The constant flow of carbon dioxide prevented reflux of air during the transfer, and as soon as the end of the tube had been submerged in the solution of potassium iodide (which was employed not only as a water-seal, but to catch any iodine still carried in the gas), it was possible to reduce the rapidity of the current. After titrating the iodine in the distillate the receiver was again placed in the train and the process of distillation was resumed under the former conditions and continued until the volume of the liquid, as indicated upon the scale, had dimin- ished to 25 cm 3 , when the distillation was interrupted. The apparatus was manipulated as before to prevent access of air, and the iodine evolved in the second treatment determined. A third period of distillation served to show the iodine liberated during the concentration of the liquid from 25 cm 3 to 10 cm 3 . During the first period of distillation the liquid assumed the clear green color, which changed but slightly until the begin- ning of the third period, when the tint verged upon olive, and at the end of the operation the color of the liquid was an olive brown which grew browner on cooling. The addition of considerable hydrochloric acid to the residual liquid re- stored the clear green color, while water changed the olive brown to reddish yellow, the tint varying with the dilution. The results of these experiments are recorded in (1) to (5) of the accompanying table. In division A are given the weights of molybdenum trioxide corresponding to the amounts of iodine found in the three stages of distillation ; in division B, the molybdenum trioxide corresponding to the iodine evolved from the beginning of the process to the end of each stage. The mean error of the indications taken during the period of distillation advocated by Friedheim and Euler is 0.0045 grm. ; * that of the period of concentration from 40 cm 3 to * Even this figure does not disclose the full error, which is partly counter- balanced, as will appear later, by the effect of oxygen dissolved in the acid used in the process. OF MOLYBDENUM. 11T A. Exp. HCl Sp &i 12 Klin retort. MoO s taken as ammonium molybdate. Mo0 3 corresponding to iodine found. First stage 40cm 3 to 32cm 3 . Green color. Second stage 32cm3 to 25 cm 3 . Third stage 25cm 3 to 10cm 3 . (1) (2) (3) (4) (5) cm 8 40 40 40 40 40 grm. 0.5 0.5 0.5 0.75 0.75 grm. 0.2455 0.2455 0.2455 0.2455 0.2455 grm. 0.2399 0.2402 0.2414 0.2404 0.2431 grm. 0.0076 0.0053 0.0040 0.0061 0.0037 grm. 0.0004 0.0013 0.0004 0.0004 0.0004 (6) (7) 40 40 1 2 0.2455 0.2455 0.2404 0.0085 0.0019 B. Ezp. MoO 3 corre- sponding to iodine found during period of Friedheim and Euler. (1st stage.) Error. MoO 3 corre- sponding to iodine found in concentrating from 40cm 3 to 25cm 3 . Error. Mo0 3 corre- sponding to iodine found in concentrating from 40cm 3 to 10 cm 3 . Error. (1) (2) (3) (4) (5) grm. 0.2399 0.2402 0.2414 0.2404 0.2431 grin. 0.0056- 0.0053- 0.0041- 0.0051- 0.0024- grin. 0.2475 0.2455 0.2454 0.2465 0.2468 gnu. 0.0020+ 0.0000 0.0001- 0.0010+ 0.0013+ grm. 0.2479 0.2468 0.2458 0.2469 0.2472 grm. 0.0024+ 0.0013+ 0.0003+ 0.0014+ 0.0017+ (6) (7) 0.2404 0.0051- 0.2489 0.0034+ 0.2508 ( 0.2495 \ 0.2529* 0.0053+ 0.0040+ 0.0074+ 25 cm 3 is 0.0008 grm. + ; and that of the full period of distilla- tion is 0.0014 +. It is plain beyond a peradventure that in the process as conducted by Friedheim and Euler, except- ing the protection against atmospheric action the theoretical reduction of the molybdic acid does not take place. The best results are obtained when the distillation is prolonged until the original volume of 40 cm 3 has been diminished to 25 cm 8 . Concentration beyond the limit of 25 cm 3 tends to develop a tendency toward over-reduction, especially when the amount of potassium iodide is increased beyond about 0.5 grm. in * On repeating distillation with a fresh charge of acid. 118 THE IODOMETRIC DETERMINATION excess of that theoretically required. This is shown in exper- iments (6) and (7), conducted otherwise similarly to those described above, in which the amount of potassium iodide was increased to 1 grm. and 2 grms. The error after distilling from 40 cm 3 to 10 cm 3 , the lowest limit of the preceding ex- periments, was 0.0053 grm. + and 0.0040 grm. -f, and the latter error was increased to 0.0074 grm. -f on repeating the distillation with a fresh portion (30 cm 3 ) of the acid. It is interesting to note incidentally that in the experiment in which the largest amount of iodide (2 grms.) was used the solution did not take the green color at any stage of the dis- tillation, probably because the large excess of iodide held the free iodine and so masked the color until the degree of con- centration was reached at which the olive-brown color dis- places the green. The possibility of the interaction of atmospheric oxygen and gaseous hydriodic in the analytical process, even to the extent of producing errors of from one to three per cent reckoned as molybdenum trioxide, was recognized by Fried- heim and Euter ; and it was to obviate this difficulty that the recommendation was made by them to warm very gradually the distillation flask filled two-thirds with the mixture of iodide, molybdate, and acid, and to raise the liquid to actual boiling only when the space above the liquid in the retort and in the connecting tube is filled as completely as possible with iodine vapor, while the liquid in the receiver begins to rise in the tube. The action of atmospheric oxygen upon the solution of hydri- odic acid must, however, be also taken into account. It is a familiar fact that when a considerable excess of strong hydro- chloric acid is allowed to act in contact with air upon potassium iodide (free from iodate) dissolved in a little water, the mix- ture is colored by free iodine. The amount of iodine liberated by atmospheric action is insignificant when the acid is very di- lute, but is considerable when the acid is strong, and increases with tune and rise in temperature, as shown in the experiments recorded in the accompanying table. OF MOLYBDENUM. 119 Per- KI taken. Volume. centage of HC1 in aqueous Time in minutes. Temperature. Centigrade. MoO s equivalent to iodine found. Remarks. acid. gnu. cm 8 gnu. 1 66 2 1 23 None. 1 66 2 10 23 0.0001 1 66 24* 10 23 0.0017 ^ Diluted to 500 cm 3 1 1 66 66 24* 24* 4 10 ( From 23 to J the boiling ( point. 0.0067 [ 0.0121 ) before titrating with Na 2 S 2 8 . Even the precaution to conduct the operation in an atmos- phere of carbon dioxide does not eliminate all chance of error of this sort unless the liquid of the mixture the hydro- chloric acid is free from air. The experiments of the fol- lowing statement, which were conducted in the apparatus and manner previously described, show this point clearly. Thus, Per- TUff\f\^ KI taken. VoL centage of HC1 in aqueous acid. Concentration by boiling. Jjfl.OU o equivalent to iodine found. Remarks. grm. cm 8 cm 8 . grm. ( 40-30 0.0013 Iodine determined 1 40 24 } in distillate. (30-20 0.0002 1 grm. of KI added to retort at the beginning of the 2d stage. 1 40 20 40-25 0.0005 The acid taken, sp. gr. 1.1, was freshly boiled and introduced at once upon KI in retort in CO 2 . 40 cm 3 of unboiled acid, sp. gr. 1.12, introduced enough air into the apparatus to cause an error of 0.0013 grm. reckoned in terms of molybdenum trioxide, while the iodine set free by the action of the residual acid of this experiment upon another gram of potassium iodide introduced without admission of air corresponded to only 0.0002 grm. in terms of molybdenum * This corresponds nearly to sp. gr. 1.12. 120 IODOMETRIC DETERMINATION OF MOLYBDENUM. trioxide. The use of acid of sp. gr. 1.1, freshly boiled in the air, obviously reduces the error due to the unboiled acid, but even in this case the effect of included oxygen was not wholly obviated. It is obvious that the procedure recommended by Friedheim and Euler can by no possibility eliminate the effect of atmos- pheric action upon the mixture of acid and iodide. The extent of such action must depend upon such conditions as the size of the apparatus, the time of exposure, the body of air above and dissolved in the liquid, and the rate of displacement of the air. How great the error due to atmospheric action actually was in the process as conducted by Friedheim and Euler we, of course, have no means of knowing. It is to be hoped, however, that it was sufficiently great to counterbalance that other inevitable error (of about five milligrams) which exists by reason of the incompleteness with which molybdic acid is reduced under the conditions which these investigators prescribe ; for, the value of Euler's work upon the vanadio- molybdates rests upon the chance that these two very considerable and indisputable tendencies to error may have neutralized one another. It has been shown clearly that our former criticism of the procedure of Friedheim and Euler is justified in every par- ticular. We have no change to make in the recommendation made therein as to necessary modifications. If the conditions seem difficult, there is an alternative in the method proposed in the former article,* according to which the molybdate is reduced by the acid and iodide in an Erlenmeyer beaker (trapped loosely by means of a short bulbed tube hung in the neck) and the molybdenum pentoxide, freed from iodine by boiling, is reoxidized by standard iodine in alkaline solution. * Am. Jour. Sci., ii, 156. Volume I, p. 375. XVII ON THE DETERMINATION OF MANGANESE AS THE PYROPHOSPHATE. BY F. A. GOOCH AND MAKTHA AUSTIN.* FOR the estimation of manganese in a gravimetric way when accuracy is a consideration, recourse is usually taken to the excellent method of Wolcott Gibbs. f This method consists in the precipitation of a manganous salt by an alkaline phosphate, the conversion of the tribasic phosphate into the ammonium manganese phosphate, and the weighing of the product of ignition as the pyrophosphate. By Gibbs' original method the orthophosphate of manganese was precipitated by hydrogen disodium phosphate in large excess above the quantity required to cause the precipitation. The flocky white precipitate was dissolved either in sulphuric or hydrochloric acid, and precipitated again at the boiling temperature by ammonia in excess. This semi-gelatinous precipitate on boiling or long standing even in the cold becomes crystalline, the crystals forming beautiful talcose scales which have a pearly luster and a pale rose color. The precipitate was filtered off, washed with hot water, dried and ignited. The results obtained by Gibbs' students for the pyrophosphate accord closely with the theory. Fresenius J showed subsequently that ammonium manganese phosphate dissolves in cold water, in hot water, and in an aqueous solution of ammonium chloride [1 : 70] to the extent of 1 part in 32,000, 1 part in 20,000, and 1 part in 18,000, respectively. It is clear, however, that the solubility of this * From Am. Jour. Sci., vi, 233. t Am. Jour. Sci., xliv, 216. J Zeitschr. anal. Chem., vi, 415. 122 DETERMINATION OF MANGANESE precipitate is not indicated necessarily by the proportions given so long as an excess of the precipitant is present during the washing, though Fresenius did find in the filtrate traces of manganese which to his mind were sufficient to account for losses indicated by his test analyses, viz., one to three milligrams of oxide, or from two to six milligrams of phosphate. Another mode of manipulation has been advocated by Blair * in order that the precipitate may be obtained more easily in crystalline condition. According to this method dilute ammonia is added drop by drop to the hot acid solution until the precipi- tate begins to form, the boiling and stirring are continued until the small amount of flocky precipitate is converted completely to crystalline condition, and the process of adding ammonia drop by drop is repeated until the manganese is all down in crystalline condition. The dilute ammonia is added in excess and the liquid filtered after cooling in ice water. In discussing these methods of precipitation, McKenna f points out that both give good and accordant results, and that the process may be carried on in glass as well as in platinum, if the time of crystallization is made short enough. When a manganous salt is precipitated in the cold by an excess of an alkaline phosphate, it falls, as Heintz f has shown, in the form of the trimanganous phosphate of the formula Mn 3 P 2 O 8 . This same phosphate constitutes, as we have found, the greater part of the precipitate which forms when a man- ganous salt reacts in the cold in the presence of ammonium chloride with microcosmic salt and ammonia in slight excess. Boiling or even subsequent standing may, as is well known, effect a more or less complete conversion of the manganese phosphate to the ammonium manganese phosphate. Thus, in one experiment hi which an amount of manganous chloride enough to produce 0.2214 gram of the pyrophosphate was precipitated in the cold by 5 cm 3 of a saturated solution of microcosmic salt, with the subsequent addition of ammonia * The Chemical Analysis of Iron, 106. t Jour. Anal. Chem., v, 141. i Ann. Phys., cl, 449. AS THE PYROPHOSPHATE. 123 in excess, in a volume of 200 cm 3 containing also 5 grams of ammonium chloride, the residue after ignition weighed 0.1904 gram. Presuming this residue to consist entirely of the pyrophosphate and the trimanganous orthophosphate, the pro- portion of the former to the latter calculated from the relation of symbols, and the weights taken and found, is nearly one to six. That is to say, about six-sevenths of the precipitate fell in this experiment in the form of the tribasic orthophosphate. In another experiment made exactly similarly, excepting that the .liquid was heated to boiling, the proportion of the manganese pyrophosphate to the trimanganous orthophosphate in the only partially crystallized precipitate proved to be two to one. That is, in this case, two-thirds of the precipitate was in the form of the pyrophosphate. In the former of the experiments a small amount of manganese was found in the filtrate, but not enough to change materially the ratio recorded. The slight solubility appears to be connected with the incomplete conversion of the trimanganous phosphate to the ammonium manganese phosphate, for as will appear later, the manganese found in the nitrate, when the conversion is known to be nearly complete, is inappreciable unless extraordinary amounts of the ammonium salt are present. The success of the analytical process under discussion turns, therefore, upon the change of the trimanganous phosphate Mn 3 P 2 O 8 to the ammonium man- ganese phosphate NH 4 MnPO 4 . In the work to be described the attempt was made to learn the conditions under which this conversion may be best and most completely accomplished. The conversion of a molecule of trimanganous phosphate to the ammonium manganese phosphate might be due, con- ceivably, either to the action of free ammonia or to the action of a salt of ammonium. The action of ammonia could only take place at the expense of a partial loss of manganese from the phosphate and its appearance as a hydroxide, two-thirds of the manganese going into two molecules of the ammonium manganese phosphate. In the presence of ammonium salts it is possible that the manganese oxide thus replaced might enter into union with the acid radical of the ammonium salt 124 DETERMINATION OF MANGANESE setting free ammonia ; but if the ammonium salt present were the phosphate, or if an alkaline phosphate were present with other suitable ammonium salts, it is conceivable that the replaced manganese might appear as a constituent of a third molecule of ammonium manganese phosphate. In any event, it would be the ammonium salt and not the free ammonia which would determine the formation of the third molecule of the ammonium manganese phosphate. Plainly, too, the ammonium salt by itself, if it were a phosphate, or if a soluble phosphate were also present, might accomplish the conversion without the intermediate action of free ammonia. Unless, therefore, free ammonia favors the insolubility of the ammo- nium manganese phosphate, its presence would be unnecessary and might even be an actual disadvantage if the hydroxide naturally formed by its action upon the manganese phosphate were to fail to reunite fully with a phosphoric acid radical. It is plain, too, that the action of free ammonia might not stop with the replacement of one out of the three of the manganese atoms present in the molecule, but might even proceed under favorable conditions to the formation of phos- phate richer in ammonium and to the separation of more manganese from its union with the acid radical. As a matter of fact Munroe * has shown that the prolonged action of hot ammonia upon the precipitate produced by the interaction of a manganous salt and an alkaline phosphate does actually produce a hydroxide which blackens as it takes oxygen from the air. Our attention has been given, therefore, more especially to a study of the conditions of action under which a salt of ammonium the chloride may bring about the conversion of the precipitate first thrown down by an alkaline phosphate to the form of the ammonium manganese phosphate. Experiments were made upon solutions of pure manganous chloride prepared and standardized by means of the sulphate method, as described in a former paper, f to show the effect of varying amounts of ammonium chloride on the condition * Amer. Chemist, 1877. t Am. Jour. ScL, v, 209. This volume, p. 77. AS THE PYROPHOSPHATE. 125 of the precipitate and upon the solubility of the precipitate when once formed. The ammonium chloride for this work was prepared pure by boiling the chemically pure salt of commerce with a faint excess of ammonium hydrate and filtering to free it from traces of iron, silica and alumina. In the first series of experiments dilute ammonia was added slowly to the hot faintly acidulated solution containing the manganous chloride and more than enough, theoretically, of a saturated solution of microcosmic salt to precipitate the manganese present. The liquid was heated and stirred until the flocky mass was changed to a crystalline condition. The addition of ammonia drop by drop, with constant stirring and heating, was continued until the manganese was all precipi- tated in crystalline form. A slight excess of ammonia was added and the liquid with the precipitate was allowed to stand for a half hour, cooling gradually or chilled in ice water. The precipitate was filtered off on asbestos under pressure, washed carefully in water made faintly ammoniacal, dried and ignited. The filtrates were tested for manganese by treatment with bromine and heating. The results of these experiments are given in the following table. TABLE I. Mn 2 P 2 O 7 equivalent to MnCl,. Error in terms of Mn 2 P 2 7 . Error in terms of Manganese. Saturated solution of HNH 4 NaPO 4 . 4H 2 0. Total volume. Manganese in nitrate. Taken. Found. grm. grm. gnn. grm. cm cm 3 0.4033 0.3769 0.0264- 0.0102- 5 60* None. 0.4033 0.3728 0.0305- 0.0118- 5 60* None. 0.3770 0.3530 0.0240- 0.0090- 5 60 None. 0.3770 0.3620 0.0150- 0.0058- 5 60 None. 0.4033 0.3751 0.0282- 0.0109- 10 60 None. 0.4033 0.3774 0.0259- 0.0100- 10 60 None. 0.4033 0.3871 0.0162- 0.0062- 5 200 None. 0.3226 03066 0.0160- 0.0062- 5 200 None. In this method of precipitation of the manganese in a pure solution of a manganous salt the results are all wrong. The * Chilled in ice-water. 126 DETERMINATION OF MANGANESE proportion of the trimanganous phosphate to the pyrophos- phate in the residue, calculated from the symbols and the weights taken and found, is in the average two to five. That is. to say, five-sevenths of the trimanganous phosphate has been converted to the form of the ammonium manganous phosphate. The precipitate obtained in this manner is white and granu- lar but not silky, and after ignition it shows the same dead white color, and is powdery. Evidently the regulation of the volume in which the precipitation is made is not essential, and the chilling of the liquid is of no importance in changing the manganese to the ammonium manganese salt under the given conditions. It is plain, moreover, that the assumption of a crystalline condition cannot serve as an indication that the composition of the salt is ideal. It is to be noted, however, that the conditions obtaining here are essentially different from those hi common practice ; for, ordinarily, when man- ganese is to be determined ammonium salts are abundantly present as the result of previous steps in analysis. In the experiments of the next series the conditions are varied simply in this respect, that ammonium salts are in- troduced before the precipitation. The precipitate was less granular and more silky. After ignition the mass was white with a faint rose color. In the experiments of section A of the table the precipitate first thrown down was redissolved, reprecipitated and filtered after cooling; in those of section B, the precipitate was filtered after cooling without re-solution and without reprecipitation ; and hi those of section C, the first precipitate was filtered at once while the solution was still hot. The length of digestion before filtering and the indications of manganese in the filtrate are recorded in the table. It was observed in these experiments that when the amount of ammonium chloride is present in considerable quantity a fine crystalline condition is got much more readily than when the amount of that salt is small: with maximum amounts of ammonium chloride the change from the flocky to the crystal- AS THE PYROPHOSPHATE. TABLE IL 127 Mn 2 P,O T equiva- lent to the MnClj. Error in terms of Mn,P,O 7 . Error in terms of Mangan- ese. Saturated solution of HNaNH 4 P0 4 . 4H,0. NH 4 C1. Total volume. Time of stand- a. Mangan- ese in the filtrate. Taken. Found. A. grm. grm. grm. grm. cm 3 grm. cm 3 hrs. 0.1542 0.1520 0.0022- 0.0008- 5 5 200 15 None. 0.1542 0.1540 0.0002- 0.0000 5 10 200 15 None. 0.1542 0.1536 0.0006- 0.0002 5 10 100 5 None. 0.1542 0.1535 0.0007- 0.0002- 5 20 200 21 None. 0.3770 0.3712 0.0058- 0.0022 5 20 200 i None. 0.3770 0.3724 0.0046- 0.0018- 5 20 200 1 None. 0.3084 0.3069 0.0015- 0.0006- 5 40 200 1 None. 0.3084 0.3060 0.0024- 0.0009- 5 40 200 1 None. 0.3084 0.3059 0.0025- 0.0009- 5 40 200 15 Trace. 0.3084 0.3057 0.0027- 0.0010- 5 60 200 15 None. B. 0.1542 0.1521 0.0021- 0.0008- 5 10 100 40 None. 0.1542 0.1512 0.0030- 0.0010- 5 10 200 40 None. 0.1542 0.1532 0.0010- 0.0003- 6 20 200 15 None. 0.1542 0.1531 0.0011- 0.0004- 5 20 100 15 None. 0.3770 0.3720 0.0050- 0.0019- 5 20 200 i None. 0.3770 0.3745 0.0035- 0.0014- 5 20 200 None. C. 0.1542 0.1519 0.0023- 0.0009- 5 16 200 None. 0.1542 0.1530 0.0012- 0.0004- 5 20 200 None. 0.1542 0.1525 0.0017- 0.0007- 5 30 200 None. 0.3084 0.3020 0.0064- 0.0025- 5 10 200 None. 0.3084 0.3053 0.0031- 0.0012- 5 20 200 None. 0.3084 0.3033 0.0051- 0.0020- 5 20 200 None. 0.3084 0.3039 0.0045- 0.0017- 5 60 200 Trace. line condition is almost immediate ; even in the cold the change takes place to a marked extent in a few seconds. No manganese was found in the nitrate by boiling with bromine and ammonia a test which is capable of indicating 0.0001 grm. of manganous sulphate in 500 cm 3 of water containing 60 grm. of ammonium chloride until the ammonium chlo- ride amounted to 20 per cent of the mass, or to 40 grm. in 200 cm 3 of the liquid, and even then but once in three trials : even when the proportion was 30 per cent 60 grm. in 200 cm 3 the solvent action of the ammonium chloride upon the 128 DETERMINATION OF MANGANESE manganese salt was trifling. The pyrophosphate residues obtained in these experiments, as well as in all those recorded in this paper, were dissolved in nitric acid and tested for contamination by a chloride ; in no single case did silver nitrate produce more than an inappreciable opalescence in the solution. It is plain, therefore, that the variations of the results from theory are occasioned by variation in the degree of conversion of the trimanganese phosphate to the ammonium manganese phosphate, and that, while the ammonium chloride shows no appreciable solvent action on the precipitate in the presence of the precipitant, its effect in the process of conversion is plainly evident. For the smaller amounts of the manganese salts (equivalent to 0.1542 grm. of the pyrophosphate) the effect of the ammonium chloride reaches a maximum when that salt amounts to 10 per cent of the solution ; for twice that amount of manganese salt, the best results were obtained by doubling the amounts of ammonium chloride. Either line of treatment yields under the most favorable conditions, results which are passably good, but the advantage inclines slightly to the first method in which the first precipitate was dissolved and reprecipitated while the liquid was cooled before filtering. TABLE IH. Mn 2 P 2 O 7 equivalent Error in terms of MN 2 P 2 O 7 . Error in terms of Manganese. Saturatecl solution of HNaNH 4 P0 4 . 4H 2 0. NH 4 C1. Total volume. Manganese in the nitrate. Taken. Found. grin. grm. grin. grin. cm grm. cm 3 0.2214 0.2202 0.0012- 0.0005- 5 20 200 None. 0.2214 0.2202 0.0012- 0.0005- 5 20 200 None. 0.2214 0.2191 0.0023- 0.0009- 5 20 200 None. 0.2214 0.2191 0.0023- 0.0009- 5 20 300 None. 0.2214 0.2191 0.0023- 0.0009- 5 20 300 None. 0.2214 0.2185 0.0029- 0.0011- 10 20 200 None. 0.2214 0.2186 0.0028- 0.0010- 20 20 300 None. 0.2214 0.2192 0.0022- 0.0009- 20 20 300 None. In Table III are recorded results obtained by precipitating the cold acid solution of the manganese salt and the microcos- mic salt with a strong excess of ammonia. The mixture was AS THE PYROPHOSPHATE. 129 heated to boiling for from five to ten minutes and filtered hot. In this series of determinations the amount of ammonium chloride present was constant while the volume of the liquid present was varied and the amounts of the microcosmic salt. These results are possibly a trifle less satisfactory than those obtained for the smaller amounts of manganese by the method of Table II, it may be because the prolonged boiling tends to form a trifling amount of free oxide ; but the fact is disclosed that an increase of the microcosmic salt is without influence and that a variation in volume from 200 cm 8 to 300 cm 8 is the occasion of little change in the indications of the process. In another series of experiments the solution of manganous chloride was added drop by drop to the mixture of microcosmic salt and ammonium chloride made alkaline with ammonia. The precipitate which fell in the cold was crystallized by boiling the mixture a few minutes. The results are given below: TABLE IV. ^oVct VaIent Error Saturated Error. in terms of manganese. solution of HNaNH 4 P0 4 . NH 4 C1. Total volume. in the filtrate. Taken. Found. grm. grm. grm. grm. cms grm. cm.3 0.1542 0.1521 0.0021- 0.0008- 5 5 200 None. 0.2214 0.2203 0.0011- 0.0004- 5 10 275 None. 0.2214 0.2192 0.0022- 0.0009- 5 15 275 None. 0.2214 0.2197 0.0017- 0.0007- 5 20 275 None. 0.2214 0.2223 0.0009+ 0.0003+ 5 20 200 None. 0.1542 0.1528 0.0014- 0.0005- 5 30 275 None. The experience of this series of experiments demonstrated again that the ease with which the flocky precipitate is converted to the crystalline ammonium manganese phosphate is proportioned to the ammonium chloride present, and the mean error of the results for the phosphate when the ammonium chloride reached 20 grams (0.0007 grm.) is considerably less than the mean error (0.0018 grm.) when the amount of the ammonium salt was less than 20 grms. Experiments were also made according to the modifications VOL. II. 9 130 DETERMINATION OF MANGANESE suggested by Monroe,* viz., the boiling of the manganous salt with an excess of microcosmic salt until the precipitate becomes crystalline and just neutralizing with dilute ammonia ; but we have been unable to find the conditions of this treatment by which uniform results may be obtained in even moderate agreement with the theory. We have tried also the effect of substituting ammonium nitrate for ammonium chloride in the conversion process ; but, so far as our experience goes, the nitrate is not so effective weight for weight in producing the change of the trimanganous phosphate to the ammonium manganese phosphates, while the solubility of the product in the solution of the ammonium nitrate becomes appreciable more rapidly with the increase of the amount present than is the case when the ammonium salt is the chloride. In the light of the experiments described it would seem to be reasonable to expect the best results from the phosphate method for determining manganese when the conditions are so arranged that precipitation may take place in the cold solution in the presence of but little free ammonia, and of enough ammonium chloride to bring about the rapid conversion of the precipitate to the crystalline condition. Under such circumstances it should be possible to secure the conversion of the phosphate to the ideal constitution as completely as possible without danger of subsequent decomposition by the prolonged action of the hot free ammonia. In carrying out this idea, the solution of manganese chloride was treated as before with microcosmic salt and a large amount of ammonium chloride, the precipitate first formed was redissolved in hydrochloric acid and precipitation again brought about by the very careful addition of dilute ammonia in slight but distinct excess. The mixture was heated only until the precipitate became silky and crystalline, when it was allowed to stand and cool for a half hour. The precipitate was filtered off upon asbestos in a perforated platinum crucible under pressure, ignited and weighed. Table V comprises the results of experiments made * Loc. cit. AS THE PYROPHOSPHATE. 131 in this manner. In those of section A the precipitation was made in platinum vessels ; in those of section B the treatment was in glass. TABLE V. A. IN PLATINUM. Mn 2 P 2 7 equivalent to MnO 2 . Error in terms of Mn,P 2 7 . Error In terms of Mil 11 T<1 11686. Saturated solution of HNaNH 4 P0 4 . NH 4 C1. Total volume. Manganese in the filtrate. Taken. Found. grm. 0.1885 0.1885 0.1885 0.1885 0.3770 0.3770 0.3770 0.3770 gmi. 0.1903 0.1910 0.1913 0.1911 0.3776 0.3773 0.3778 0.3783 grin. 0.0018+ 0.0025+ 0.0028+ 0.0026+ 0.0006+ 0.0003+ 0.0008+ 0.0013+ grm. 0.0007+ 0.0010+ 0.0011+ 0.0010+ 0.0002+ 0.0001+ 0.0003+ 0.0005+ cm 8 5 5 5 5 5 5 5 5 grm. 20 20 20 20 20 20 20 20 cm 200 200 200 200 200 200 200 200 None. None. None. None. None. None. None. None. B. IN GLASS. 0.1885 0.1885 0.3770 0.3770 0.1904 0.1898 0.3767 0.3784 0.0019+ 0.0013+ 0.0003- 0.0014+ 0.0007+ 0.0005+ 0.0001- 0.0005+ 6 6 5 5 20 20 20 20 200 200 200 200 None. None. None. None. In this series of experiments the mean indication is, for the first time, in excess of the theory. Previously the error has been one of deficiency, and that in proportion to the amount of manganese handled, no doubt because the amount of uncon- verted trimanganese phosphate is proportioned to the entire amount of the phosphate. The positive error which is devel- oped in this last series of determinations is probably due to the appearance of the natural error of all precipitation pro- cesses viz., the tendency on the part of the precipitate to include matter in solution. In the previous experiments this effect was doubtless obscured by the incompleteness of the conversion of the trimanganous phosphate to the ammonium manganese phosphate. Indeed it is quite possible that even in the last determinations the conversion is not absolute, and that this is so suggested by the fact that the errors of excess are larger in the case of the smaller amounts of manganese for 132 DETERMINATION OF MANGANESE which the conversion throughout the entire work has appeared to be more complete. From the consideration of the results tabulated and described it would seem to be obvious that not only is the presence of ammonium chloride not objectionable in this analytical process, which depends upon obtaining the ammonium manganese phosphate from the trimanganese phos- phate precipitated from a pure solution of manganese, but that its presence in not too small amount, or that of a substitute, is absolutely essential to make this conversion complete. For a given amount of manganese and a given volume of solution it seems essential that the amount of ammonium chloride should reach a certain limit. According to our experience the proportion of ammonium chloride to the pyrophosphate should be at least 50 : 1 ; or, speaking approximately, more than 200 molecules of ammonium chloride must be present in the liquid (100 cm 3 or 200 cm 3 ) to every molecule of the ammonium manganese phosphate to be formed. However, the quantity of the ammonium salt may be increased almost to the point of saturation of the liquid without causing more than a trifling solubility of the ammonium manganese phos- phate in the presence of an excess of the precipitant. The statement of Fresenius and Munroe that ammonium manganese phosphate is soluble in ammonium chloride does not hold if there is an abundance of the soluble precipitating phosphate present. Further, our experience goes to show that the pre- cipitate may be washed with perfect safety with pure water as well as with slightly ammoniacal water, or with ammoniacal water containing ammonium nitrate, if the nitration is per- formed rapidly and the precipitate is gathered in small space, as is the case when the phosphate is collected on asbestos in a perforated crucible. The finely granular precipitate which may be obtained by slow action of dilute ammonia added gradually to the hot solution of the manganese salt apparently includes a portion of unconverted phosphate which resists the replacement of the manganese by ammonium. On the other hand, the precipitate of flocky condition thrown down hi the cold passes easily to the silky and crystalline condition when heated with the proper amount of ammonium salt and pos- AS THE PYROPHOSPHATE. 133 sesses a constitution approaching the ideal under such condi- tions. The conversion of the flocky manganous phosphate is so rapid that the precipitation may be carried on safely in glass vessels. If the ammonium chloride in the solution were to be included in the precipitate it would volatilize entirely during the ignition, leaving no trace unless, possibly, a por- tion of its chlorine were to combine with the manganese. Tests for chlorine in the residue of pyrophosphate resulted negatively no more than a mere trace being found in any case, so that the contaminating effect of the ammonium chlor- ide proves to be insignificant and the responsibility for the increase in weight above the theory must apparently rest with the included microcosmic salt. In the practical determination of manganese by the phos- phate method of Gibbs, therefore, we advocate strongly the presence of large amounts of ammonium chloride. Good results may be obtained by the method of precipitation origi- nally laid down by Gibbs, or by the modification proposed by Blair, if the ammonium salt is present in sufficient quantity. On the whole trustworthy results are obtained most easily and surely, according to our experience, by following the method of the experiments of Table V. The slightly acid solution, containing in a volume of 200 cm 3 (in platinum or glass) an amount of manganese not more than enough to make 0.4 grm. of the pyrophosphate, 20 grm. of ammonium chloride and 5 to 10 cm 3 of a cold saturated solution of microcosmic salt, is precipitated in the cold by the careful addition of dilute ammonia in only slight excess. The mixture is heated until the precipitate becomes silky and crystalline, the whole is allowed to stand and cool a half hour, the precipitate is col- lected upon asbestos in a perforated platinum crucible, washed (best with slightly ammoniacal water), dried at gentle heat and ignited as usual. By this process determinations of the larger amounts of manganese 0.4 grm. of the pyrophosphate approximate rather more closely to the theoretical values than do those of the smaller amounts 0.15 grm. In either case the average error should not exceed 0.0010 grm. in terms of manganese. XVIII ON THE DETECTION OF SULPHIDES, SULPH- ATES, SULPHITES AND THIOSULPHATES IN THE PRESENCE OF EACH OTHER. BY PHILIP E. BROWNING AND ERNEST HOWE * SOME three years ago R. Greig Smith f published a method for the detection of sulphates, sulphites and thiosulphates in the presence of each other, which promised much toward the solution of this most difficult problem. The method may best be described in the author's own language : To a solution of the salts of the above mentioned acids "barium chloride is added in excess, together with a good quantity of ammonium chloride, which, like many salts of ammonium, potassium and calcium, acts as a flocculent or coagulant, and facilitates the filtration of the barium sulphate. Hydrochloric acid is next added, drop by drop, until it is evident that there is no further solution of barium sulphite and thiosulphate, and that only the sulphate remains undissolved; the solution is then filtered through a moistened double filter paper, which should be free from 'pin holes.' The filtrate will probably be clear, but if not it should be returned to the filter for a second filtration. When too much thiosulphuric acid is present, the clear filtrate will visibly become clouded, or from being whitish will become more opaque; if this occurs the solution should be thrown out, and a fresh portion made more dilute. A solution of iodine is added to half of the filtrate until the color is of a permanent yellow tinge ; a white precipitate indicates the presence of a sulphite which has been oxidized by the iodine * From Am. Jour. Sci., vi, 317. t Chem. News, Ixxii, 39. DETECTION OF SULPHIDES, SULPHATES, ETC. 135 to sulphate. In the absence of a decided precipitate traces of sulphite may easily be detected by comparing the treated and untreated halves of the filtrate a procedure which very often saves a good deal of time, as it is unnecessary to wait until a clear filtrate is obtained. The two halves are mixed, and if the yellow color disappears more iodine is added, the solution filtered and the filtrate divided into two halves as before. With a slight turbidity filtration may be omitted. Bromine water is added to one of the halves when any thiosulphate in the original solution shows itself as a white precipitate of barium sulphate, readily seen on comparing the two test-tubes. The thiosulphate is by iodine converted to tetrathionate, which is oxidized by bromine water to sulphate." Three objections to this method as described will readily occur to the reader : first, the readiness with which the thiosulphate is decomposed by free hydrochloric acid; second, the comparatively large amount of acid necessary to effect the complete solution of the barium sulphite and thiosulphate when precipitated with the sulphate as compared with the amount required to prevent the precipitation ; third, the lack of delicacy necessitated by a comparison of portions of a colored solution in looking for small precipitates. The work to be described was undertaken to overcome these difficulties and to test the accuracy of a modified method. Solutions of potassium sulphite and sodium thiosulphate were made approximately decinorrnal and stand- ardized in the usual manner against an iodine solution of known value. It was found that by making a solution containing sulphates, sulphites, and thiosulphates very faintly acid, the sulphites and thiosulphates were held completely in solution when the barium sulphate was precipitated. The extreme sensitiveness of a thiosulphate to the decomposing action of free hydrochloric acid suggested the possible substi- tution of acetic acid to hold the sulphites and thiosulphates in solution. This being a weaker acid, we hoped to avoid the decomposition of the thiosulphate into sulphur and sulphurous acid, or at least to delay the decomposing action. The results of these experiments appear in the following table : 136 DETECTION OF SULPHIDES, SULPHATES, ETC., TABLE I. TTWI,, , a Hydro- E bcp. \ oiume cm 3 of water. chloric acid (1:4). Acetic acid. NaAO, taken. Result. drops. drops. grin. 1) 10 2 . . 0.01 No sulphur in 20 minutes. 10 2 . . 0.1 Sulphur in 45 seconds. 3) 100 3 . . 0.1 Sulphur in 15 minutes. 4) 10 . 8 0.01 No sulphur in 20 minutes. 5) 10 . 8 0.1 Sulphur in 90 seconds. (6) 100 , 10 0.1 No sulphur in 20 minutes. (7) 100 , 10 0.25 Sulphur in 15 minutes. i (8) (9 100 100 ' 10 10 0.6 1.0 Sulphur in 60 seconds. Sulphur in 30 seconds. From these results it would seem that the decomposition of a thiosulphate is more rapid in presence of hydrochloric acid than hi presence of a much larger amount of acetic acid. Our next experiments were directed toward a determination of the effect of adding stannous chloride to bleach the color of the free iodine and bromine used in the oxidation and of acidifying with acetic acid, before treating with barium chloride. That is to say, the process as we used it, consisted in acidifying the solution to be tested with acetic acid, adding barium chloride, filtering to remove precipitated sulphate (always present in the sulphite), adding iodine to the nitrate until the color was permanent, bleaching with stannous chloride, filtering off the sulphate which represents the sul- phite originally present, adding bromine in excess to the filtrate and again bleaching with stannous chloride to increase the visibility of the sulphate which now represents the thiosulphate originally present. The details of experiments in which the sulphite was taken alone and oxidized with iodine are given in Table II. A corresponding series of experiments was made in which hydrochloric acid was substituted for acetic acid and essentially the same results were obtained. A similar series of experiments was made to test the effect of treating the thiosulphate in an acidified solution, first with IN THE PRESENCE OF EACH OTHER. 137 TABLE H. Brp. KJO. taken. Volume of water. BaS0 4 precipitated after oxidation with iodine. Remarks. (1) (2) (3) (4) (5) gnu* 0.1 0.01 0.001 0.0005 0.0001 cm 8 10 10 10 10 10 Very abundant. Abundant. Distinct. Fair. Faint. Plainly visible before adding SnCl 2 . Plainly visible before adding SnCl 2 . More distinct after adding SnCl 2 . Hardly visible before adding SnCl 2 . Invisible before adding SnCLj. iodine and then after filtration (if a precipitate had formed) with bromine. In the experiments of division A hydrochloric acid (a few drops) was added before treating with barium chloride, and in those of division B acetic acid was used similarly. Stannous chloride was employed to bleach the excess of iodine and bromine. TABLE m. BaS0 4 pre- BaSO 4 precip- Exp. Na 2 S 2 O s taken. Volume of water. cipitated by action itated by action Remarks. of iodine. of bromine. A. grm. cm 3 (1) 0.1 10 Distinct. Abundant. Sulphur separated in 30 seconds. (2) 0.01 10 Faint. Abundant. No sulphur in 90 seconds. (3) 0.001 10 None. Distinct. No sulphur in several minutes. (4) 0.0005 10 None. Faint. No sulphur ; SnCl 2 necessary. (5) 0.0001 10 None. Very faint. No sulphur ; SnCl 2 necessary. B. (1) 0.1 10 Faint. Abundant. No sulphur separated in 1 minute. (2) 0.01 10 None. Abundant. No sulphur separated in several minutes. (3) 0.001 10 None. Distinct. No sulphur. (4) 0.0005 10 None. Faint. No sulphur ; SnCl 2 necessary. (5) 0.0001 10 None. Very faint. No sulphur ; SnCl 2 necessary. From these experiments the advantage of the use of acetic acid becomes apparent, as does also the use of stannous chloride in increasing the delicacy of this indication, so that a small fraction of a milligram may easily be detected. 138 DETECTION OF SULPHIDES, SULPHATES, ETC., If relatively large amounts of thiosulphate are present with small amounts of sulphite, we have sometimes found it advantageous to manipulate so that even the slow decom- position of the thiosulphate by acetic acid may be avoided by first attempting precipitation with barium chloride in a dilute ammoniacal solution. By this method the barium sulphate and sulphite are separated from the thiosulphate and identified the sulphate by its insolubility in dilute hydrochloric acid, and the sulphite by the action of iodine upon the acid filtrate from the barium sulphate. After filtering, the thiosulphate may be detected in the filtrate by the use of iodine and bromine as described above. Table IV gives some results by this treatment. TABLE IV. Exp. Na 2 S 3 8 taken. BaS0 4 pre- cipitated by iodine. BaS0 4 pre- cipitated by bromine. Remarks. grm. (1) 0.1 None. Abundant. (2) 0.01 None. Good. (3) 0.001 None. Fair. SnCl 2 necessary. (4) (5) 0.0005 0.0001 None. None. Faint. None. SnCl 2 necessary. As will be seen, the test for the thiosulphate by this method of treatment is not so delicate, probably on account of mechanical holding of the barium thiosulphate by the pre- cipitated sulphate and sulphite. Having determined the limits of accuracy of the method as applied to the sulphite and thiosulphate taken separately, our next experiments were directed toward an investigation of the working of the method when these two acids are found together in solution. Sulphates, almost invariably present with sulphites, are of course quite easily separated by filtration and treating with the barium salt in acid solution. Sulphides if present in the solution would seriously interfere with the working of this method if not removed, being readily oxidized by the iodine or bromine to sulphite, sulphate, or, should IN THE PRESENCE OF EACH OTHER. 139 sulphur also separate, to thiosulphate. We found in course of our work that in attempting to neutralize a mixture of freshly prepared alkaline sulphide together with a sulphite we often obtained a precipitate of sulphur. After the removal of the sulphide and sulphate, we were surprised to find on treating with iodine scarcely a trace of sulphite. On treating with bromine however an abundant indication of thiosulphate was obtained. It is well known of course that thiosulphate may be formed by boiling a sulphite with sulphur, but that this reaction should take place so readily and completely seemed to us rather unusual. For the removal of a sulphide before proceeding with the tests for sulphite and thiosulphate Greig Smith recommends the passing of carbon dioxide through the solution until the escaping gas gives no indication of hydrogen sulphide, but Bloxam* calls attention to the tedious and wholly unsatis- factory character of this method of removal and recommends a mixture of zinc chloride, cadmium chloride, ammonium chloride and ammonia. We have found that the addition of zinc acetate to a faintly alkaline solution accomplishes the same purpose in an entirely satisfactory manner. The sul- phide used in our work was freshly made by passing hydrogen sulphide through a dilute solution of sodium hydroxide. When portions of this solution, still alkaline, were treated with zinc acetate in excess, and the zinc hydroxide and sulphide removed by filtration, the filtrate gave no test for either sulphite or thiosulphate by the application of iodine and bromine as described, and the vapor evolved on boiling caused no darkening of lead paper. The following table shows the results of a few experiments in which tests were made for the sulphite and thiosulphate, after removing a con- siderable amount of the sulphide in the manner described, and of the sulphate by acidifying and adding barium chloride. The method as we have modified it may be summarized as follows : To about 0.1 grm. of the substance to be analyzed dissolved in 10 cm 8 of water or more, add sodium, potassium or * Chem. News, Ixxii, 63. 140 DETECTION OF SULPHIDES, SULPHATES, ETC. TABLE V. E 2 SO 3 taken. N^SjOj, taken. BaS0 4 precipitated after oxidation with iodine. BaS0 4 precipitated after oxidation with bromine. grin* 0.1 0.1 0.01 0.001 0.001 grin. 0.01 0.001 0.1 0.1 0.001 Abundant. Abundant. Good. Faint. Fair. Good. Distinct. Abundant. Abundant. Fair. ammonium hydroxide to distinct but faintly alkaline reaction. The solution should be neutral or alkaline rather than even faintly acid, owing to the readiness with which sulphur separates. To the alkaline solution add zinc acetate in distinct excess and filter. The precipitate may be tested for hydrogen sulphide, on acidifying, in the usual manner. To the filtrate add acetic acid, a few drops in excess of the amount necessary to neutralize, and barium chloride, and filter through a double filter. To the filtrate add iodine until the solution takes on a permanent yellow tinge, and then bleach with stannous chloride, best after adding a few drops of hydrochloric acid to prevent the possible precipitation of a basic salt of tin. A precipitate at this point indicates the sulphite. Filter, add bromine water in faint excess to the filtrate, bleaching again with stannous chloride. A pre- cipitate on adding bromine indicates a thiosulphate originally present. XIX ON THE SEPARATION OF NICKEL AND COBALT BY HYDROCHLORIC ACID. BYFRANKE STUART HAVENS. A QUANTITATIVE separation of nickel and cobalt by a process analogous to that published from this laboratory for the separation of aluminum and ironf has been put forward in a recent paper by E. Pinerua.J The process may be described briefly as follows: The hydrous chlorides of nickel and cobalt (0.3-0.4 grms.) are dissolved in a little water and to the solution are added 10 to 12 cm 3 of aqueous hydrochloric acid and 10 cm 3 of ether, and the whole, contained in a little beaker surrounded with water, and ice, is saturated with gaseous hydrochloric acid. The cobalt, which remains in solution, is decanted off and the yellow insoluble nickel chloride washed with a previously prepared solution of ether saturated with hydrochloric acid gas at a low temperature. The nickel is determined by known methods, preferably as the sulphate. The author claims very precise results for the process, but gives no experimental proof of his work. Previous to the appearance of this paper my experiments upon the solubility of nickel chloride in an ether-hydrochloric acid solution, such as used in our process for the separation of aluminum and iron, which is practically the same in proportions as that used by Pinerua to effect precipitation, had shown that, while nickel chloride is somewhat insoluble in such a mixture, the degree of insolubility is not sufficient for a quantitative separation. Since the appearance of Pinerua's work I have * From Am. Jour. Sci., vi, 396. t Gooch and Havens, Am. Jour. Sci., ii, 416. This volume, p. 20. t Gaz. chim. ital., xxvii, 56. 142 ON THE SEPARATION OF NICKEL AND been over the ground again and have reached the same conclusions as before, as shown in the following experiments. When a solution of 0.02 grm. of nickel chloride (free from iron and cobalt) in 7 cm 3 of aqueous hydrochloric acid, was saturated with hydrochloric acid gas at a temperature of 2 C. (obtained by immersing the container in a mixture of ice and salt) no precipitation resulted. When, however, an equal volume of ether was added and the whole was again saturated with hydrochloric acid gas a yellow precipitate formed, while the supernatant liquid still remained of a deep green color. The solution was filtered quickly through asbestos in a filter crucible, and the clear filtrate after evaporation with sulphuric acid was electrolyzed. The metallic deposit of 0.0020 grm. proved to be pure nickel ; for when dissolved in nitric acid it gave no test for iron with potassium sulphocyanide or ferro- cyanide, and neither the apple-green hydroxide nor the black sulphide, prepared by the usual methods, showed any trace of cobalt in the borax bead. It is obvious, therefore, that nickel chloride is not fully precipitated under these conditions and that the green color of the solution is due to nickel in solution and not to traces of iron, as Pinerua has supposed.* A second experiment similar to the first showed a solubility of the nickel chloride represented by 0.0018 grm. of metallic nickel. It is evident, then, that the solubility of nickel chloride in this mixture of aqueous hydrochloric acid and ether thoroughly saturated with hydrochloric acid gas is not far from an amount represented by 0.0020 grm. of metallic nickel for every 14 cm 8 of solution. Still another experiment, in which nickel chloride repre- senting 0.0020 grm. of metallic nickel was treated with 14 cm 3 of the ether-hydrochloric acid solution and the whole saturated for one hour at a low temperature with hydrochloric acid gas without precipitation, showed the same thing. When the nickel chloride remaining on the asbestos was washed with about 40 cm 8 of a mixture of equal parts ether and aqueous hydrochloric acid saturated with hydrochloric acid gas, * Loc. cit. ' COBALT BY HYDROCHLORIC ACID. 143 the washings evaporated with sulphuric acid and treated by the battery gave a deposit of metallic nickel weighing 0.0027 grm. an amount proportionately less than that found in the nitrate proper. Although employing a mixture of aqueous hydrochloric acid and ether saturated with gaseous hydrochloric acid for the precipitation, Pineriia has advised the use of pure ether saturated with gaseous hydrochloric acid for the washing. In my experiments with such a mixture I find that in it the hydrous nickel chloride is practically insoluble and that 30 cm 3 of the washings of the precipitated chloride with such a mixture gave no deposit of nickel by the battery. It seemed possible, therefore, that by reducing the water present to the lowest possible amount necessary to dissolve the chlorides to be treated the precipitation of the nickel might be made more complete. The experiments of the following table were made to put this idea to the test. Solutions of the pure chlorides of nickel and cobalt, carefully purified and freed from other metals and each other, were, after conversion to the form of the sulphate, standardized by the battery. Weighed portions of these solutions were taken in a small beaker, evaporated to dryness, the dry salts dissolved in as little water as possible (about 1 cm 3 ), 10 to 15 cm 3 of ether added, and the whole saturated with hydrochloric acid gas, the beaker being meanwhile immersed in running water and cooled to about 15 C. When saturation was complete the precipitated chloride was caught on asbestos in a filter crucible, washed thoroughly with a previously saturated solution of ether, dissolved in water, evaporated with sulphuric acid and determined as metallic nickel by the battery. The cobalt in the filtrate was recovered by evaporation and electrolysis in like manner. Experiments (1), (2), and (3) of the accompanying table show that by this process the nickel is thrown down quantita- tively, and experiments (2) and (3) show that in the presence of a few milligrams of the cobalt salt the separation of a small amount of nickel is sharp. The residue of nickel in these 144 SEPARATION OF NICKEL AND COBALT. experiments gave no test for cobalt with the borax bead. When, however, the cobalt is present to the amount of a few centigrams as in (4), (5), and (6), the precipitated nickel chloride, which forms a hard mass, includes the cobalt salt so that even a large quantity of washing solution (100 cm 3 was used in experiment 6) cannot remove it. Exp. Nickel taken as the hydrous chloride. Nickel found. Error. Cobalt taken as the hydrous chloride. Cobalt found. Error. (1) (2) (3) (4) (5) (6) giro. 0.0068 0.0090 0.0090 0.0469 0.0468 0.0472 grlu. 0.0066 0.0090 0.0091 0.0490 0.0503 0.0493 grm. 0.0002- 0.0000 0.0001+ 0.0021+ 0.0035+ 0.0021+ grm. 0.0030 0.0123 0.0700 0.0700 0.0700 grin. 0.0127 grin. 0.0004+ From the experiments described it is obvious that the pro- cess as proposed by Pinerua will not give a complete precipi- tation of the nickel chloride. Nickel chloride is, however, practically insoluble in pure ether saturated with hydrochloric acid gas and can be separated from small quantities of the solu- ble cobalt salt in that medium. In the presence of even a few centigrams of the cobalt chloride, however, the process is not practicable on account of the inclusion of the cobalt by the massive nickel chloride. It is possible that by repeated solu- tions and reprecipitations the nickel salt might be sufficiently freed from the cobalt, but the process must naturally be long and tedious. XX THE ETHERS OF TOLUQUINONEOXIME AND THEIR BEARING ON THE SPACE ISOMER- ISM OF NITROGEN. BY JOHN L. BRIDGE AND WILLIAM CONGER MORGAN * IN an article on the ethers of quinoneoxine (isonitrosophenol) published by one of us,f it was stated, that when boiled with alcohol, the benzoyl ether of quinoneoxime dichloride gave gave two monochlor substitution-products. Naturally it was supposed that the chlorine atom occupied, in the one, an ortho position, and, in the other, a meta position to the radical NOR, the reaction being : 000 + + 2HC1. Hk /"HC1 HL JH Hi JCl N N N O 00 R R R It was also found that these same isomers were formed when monochlorquinone was treated with hydroxylamine hydro- chloride, and the sodium salt of the chlorquinoneoxime thus formed, treated with benzoyl chloride. The preceding explanation regarding the splitting off of hydrochloric acid from the dichloride is not in accord with re- sults of work done by Kehrmann,J published in his article on the influence of radicals in the action of hydroxylamine on * From Am. Chem. Jour., xx, 761. t Ann. Chem. (Liebig), cclxvii, 79. t Ber. Dtsch. chem. Ges., xxi, 3315 ; Jour, prakt. Chem. [2], xl, 268. VOL. II. 10 146 THE ETHERS OF TOLUQUINONEOXIME. quinones, in which he generalizes the results of his observa- tions in the statement that the presence of a radical attached to the ring so much lessens the replaceability of the quinone oxygen atom neighboring to the radical that the principal part if not the whole of the resulting product, is a meta-sub- stituted quinoneoxime. The work of his former article has been repeated by Kehrmann,* who finds the same isomeric benzoyl ethers of monochlorquinoneoxime ; but believing that observations of their behavior indicate the substantiation of his rule, he states that both these ethers have the chlorine atom in the meta position to the oximido group, and attributes their difference to space isomerism of nitrogen, writing the re- action thus : + 2HC1. Kehrmann designates the compound represented by symbol I. as " chlorquinonemetaantioxime ether " and II. as " chlorquinone- metasynoxime ether." The question as to whether the chlor- ine atom occupies the same position in both compounds could be definitely settled if ortho- and metachlorphenols could be changed to the corresponding chlorquinoneoximes or so-called isonitrosophenols. This transformation has, however, unfor- tunately not yet been accomplished, and repeated efforts to obtain the corresponding bromquinoneoximes have resulted in failure, orthobromphenol not being attacked by nitrous acid or amyl nitrite. The toluquinoneoximes, obtained by the action of nitrous acid on ortho- and metacresol, we have taken up for study as being the most closely analogous compounds in which the position of the side groups is definitely known, believing that * Ann. Chem. (Liebig), cclxxix, 27. THE ETHERS OF TOLUQUINONEOXIME. 147 by an examination of these bodies, light may be thrown upon the nature of the others. When orthocresol is acted upon by nitrous acid, a toluqui- nonemetaoxine is formed according to the following reaction: H O + HONO = H H The metacresol forms similarly a corresponding orthooxime : H O O H r XXx i H + HONO == + H 2 0. H LJ CH 8 H N O H The benzoyl ethers of the toluquinoneoximes prove to be exceedingly well fitted to characterize these bodies since they are formed from the oximes in quantitative proportions, are easily crystallized, and readily distinguished from each other. The benzoyl ether of toluquinoneorthooxime produced from metacresol, crystallizes in light brownish-yellow crystals melting at 177 C. and appearing under the microscope as long rectangular blades, suggesting the orthorhombic system. The benzoyl ether of toluquinonemetaoxime produced from orthocresol is obtained in the form of yellow crystals which begin to soften at about 150 and do not melt entirely until at about 190. Subjected to fractional crystallization, a portion of the metaoxime is readily obtained consisting of branching needle-like crystals, melting at 193 C. These, as 148 THE ETHERS OF TOLUQUINONEOXIME. well as lower-melting fractions, can be readily distinguished from the orthooxime by their less regular appearance under the lens, as well as by their crystal habit, tending to produce curved forms. The marked difference in the form and habit of the crystals of the benzoyl ethers of the ortho- and metatoluquinone oximes makes it possible to study carefully the product of the action of toluquinone on hydroxylamine. Goldschmidt and Schmidt * have shown that the principal product of this reac- tion is toluquinonemetaoxime. This they demonstrated by oxidizing to a dinitro compound; but as some of the theo- retically possible nitro-derivatives of the cresols are not known, this method can scarcely be considered to prove con- clusively the absence of toluquinoneorthooxime. We have, therefore, studied further the product of the action of hydroxyl- amine on toluquinone by the aid of the benzoyl ethers. When the sodium salt of the oxime thus formed was treated in alco- holic solution with benzoyl chloride, and the benzoyl ether obtained was subjected to fractional crystallization, not a trace of the orthooxime ether, crystallizing in blades and melting at 177, was found. This proves that when tolu- quinone is treated with hydroxylamine the whole of the re- sulting product is toluquinonemetaoxime. The inference is plain that Kehrmann's rule concerning the influence of side- chains, attached to the ring in quinones, upon the entering oximido radical is quantitatively true in this case ; and similar indications furnished by the corresponding methyl ethers, as will appear later, strengthen this conclusion. As has been previously shown, the benzoyl ether of the metaoxime does not consist of a single compound, but is a mixture. The body melting at 193 is readily separated in considerable quantities, while the remainder consists of a very intimate mixture of this body with one of much lower melting-point, separated, if it can be separated at all, only with the greatest difficulty. By concentration of the mother- liquor from which the high-melting fractions have been * Ber. Dtsch. chem. Ges., xvii, 2063. THE ETHERS OF TOLUQUINONEOXIME. 149 obtained, and carefully crystallizing, fractions may be separated melting almost completely at temperatures, varying with the fraction, between 130 and 150. Repeated crystallizations separated each fraction into portions melting, on the one hand, always considerably higher, and, on the other hand, often somewhat lower, and ordinarily no definite body of low melting-point could be isolated. Twice, however, a nearly complete separation seemed to have been obtained. Thus, from fractions melting between 140 and 150, a few short, thick prisms melting once at 142 and again at 144 sepa- rated, leaving the compound which melts at 193. Of these crystals there was not enough for a combustion, but an analysis of a fraction melting almost completely at 137 gave figures which show without doubt that this portion had the same percentage composition as the body melting at 193. If we are to regard the product as it is first obtained as a mixture of two ethers, and consider that 142 -144 is the melting-point of the lower body, we can account for the melting of fractions below 142 on the ground that mixtures may have a lower melting-point than either of the component substances. These results were obtained repeatedly from three separate portions of Kahlbaum's C. P. orthocresol, purchased at different times, as well as from toluquinone melting at 67 C. and wholly volatile when exposed to the air. When the silver salt of toluquinonemetaoxime acts upon methyl iodide, the product is, likewise, not a single ether but a mixture of ethers, softening at 55 and not melting completely below 70. From it a body melting at 73 -74 can easily be separated, but no other compound of definite melting-point or different crystal form could be obtained. The acetyl compound of toluquinonemetaoxime presents phenomena similar to those of the benzoyl ethers, but in rather more marked degree. The product, as first obtained by the action of acetyl chloride on the silver salt of the oxime, or by acetic anhydride on the oxime itself, begins to soften at 90 and melts completely at 110. Upon the first recrystallization a distinction in crystal form appears, and a 150 THE ETHERS OF TOLUQUINONEOXIME. second crystallization of the separated portions gave short, thick prisms melting at 112-113, and some smaller, spheri- cally grouped crystals, melting at 85 -87. The extreme difficulty of preparing the acetyl compound prevented further investigation. In like manner, the benzoyl ether of monobromtoluquinone- metaoxime seems to be a mixture of isomeric bodies. Thus, it was found possible to add two atoms of bromine to the ethers of toluquinoneoxime forming colorless dibromides correspond- ing to the dibrom addition-products of quinoneoxime, and these dibromides, when boiled with dilute alcohol, split off hydrobromic acid with the formation of colored monobrom substitution-products. The benzoyl ether of monobromtolu- quinonemetaoxime thus formed shows a variation in melting- point similar to that of the ethers previously discussed. The foregoing facts speak in favor of Kehrmann's theory of space isomerism in the oximes so far as the metaoximes are concerned. On the other hand, there is no evidence to show the presence of isomers in the ethers of toluquinoneorthooxime : the methyl, acetyl, and benzoyl ethers all act as simple substances, each product melting completely at a definite temperature. It is difficult to understand why isomerism should be so much more evident in the ethers of toluquinone- metaoxime than in the ethers of the orthooximes, unless, possibly, the closer proximity of the side-chain to the oximido-group prevents the formation of a space isomer. There is, however, a remote possibility that isomeric bodies may exist, so similar in properties that they cannot be detected by the ordinary methods. EXPERIMENTAL PART. Preparation of the Oximes and their Salts. The toluquinone-, ortho-, and metaoximes used in the experiments to be described were made hi the following manner : To a solution of 10 grams of cresol and 8 grams of potassium nitrite in 900 cm 3 of water, a solution of 6 grams THE ETHERS OF TOLUQUINONE OXIME. 151 of concentrated sulphuric acid in 100 cm 3 of water was added in small portions during the course of half an hour, care being taken that both original solutions should be between 5 and 10, and that this temperature be maintained during the mixing. Nearly all of the oxime separates out on standing in ice-water for one to two hours, and after filtering and washing with 200-300 cm 3 of ice-water, the amount of oxime obtained by extracting the filtered solution with ether is so small that it may be disregarded. The substances were purified by dissolving in a saturated solution of sodium carbonate and filtering into dilute sulphuric acid, cooled with ice. At this stage it is generally ready for use, but if further purification is desired, it may be accomplished by dissolving the oxime in ether and shaking with animal charcoal. Upon filtering and evaporating, the oxime crystallizes in long, slightly colored needles. The yield is large in both cases, that of the orthooxime being nearly theoretical. As given by Beilstein, toluquinone-ra-oxime melts at 134 C. Toluquin- one-o-oxime melts at 155 C. ; Beilstein gives 145-150. The silver salt of the metaoxime was made in the following manner : 5 grams of toluquinone-7?M)xime were dissolved in a solution of sodium hydroxide, a little less than the quantity calculated to form the sodium salt, and this solution was filtered into 500 cm 3 of water containing 1J times the calcu- lated amount of silver nitrate. The precipitate comes down in a flocky, gelatinous condition, but goes over into a granular form on heating to 50 in a water-bath. When dissolved in the least possible amount of warm dilute ammonia, and the solution precipitated with hydrochloric acid, 0.1006 gram of the substance, dried over H 2 S0 4 , gave 0.0590 gram AgCl. Calculated for V^^A C 7 H 6 N0 2 Ag. Ag 44.23 44.14 The salt is light reddish-brown when first formed, but turns darker on standing or heating. It decomposes when heated to 100 and, when thoroughly dry, is spontaneously 152 THE ETHERS OF TOLUQUINONEOXIME. inflammable at a temperature above 60. It is a rather un- stable body and cannot be kept long when at all impure. The silver salt of toluquinone-o-oxime was made in the manner described for the preparation of the silver salt of the meta form. The salt falls in the cold as reddish-brown crystals, tending to darken when exposed to light or heat. 0.2560 gram, dried over H 2 S0 4 , gave 0.1498 gram AgCl. Calculated for ,, , C 7 H 6 N0 2 Ag. Found ' Ag 44.23 44.06 Although very similar in all its properties to the silver salt of toluquinone-w-oxime, this salt is like all the ethers of the ortho form, much more stable than its corresponding meta isomer. Toluquinone-m-oxime Methyl Ether. From o- CresoL Of the silver salt of toluquinone-w-oxime, 2 or 3 grams were suspended in 10-15 cm 3 of ligroin and twice the calculated quantity of methyl iodide added. After standing for an hour with frequent shaking, the liquid was filtered off and the residue extracted with a little hot ligroin. The united ligroin solutions were allowed to evaporate spon- taneously, and the methyl ether came out in large, dark- yellow, hexagonal prisms. A little more may be obtained by allowing the residue to stand for a week with methyl iodide. The yield in any case is small, the best results apparently being obtained by using not more than 2 or 3 grams of the silver salt at one tune. After purifying with animal charcoal and recrystallizing from ligroin, the product obtained softens at 55 yet does not melt completely below 70. Portions melting at 73-74 C. were separated by fractional crystalli- zation, and, on analysis, 0.1101 gram of this body, dried over H 2 S0 4 , gave 0.2582 gram C0 2 and 0.0599 gram H 2 0. 0.0881 gram of the substance gave 7.1 cm 8 N at 15 C. and 772 mm. pressure. THE ETHERS OF TOLUQUINONEOXIME. 153 Calculated for v> nm <{ CgHoNO,. C 63.53 63.96 H 6.00 6.05 C 9.29 9.59 Although fractions were often obtained melting from 55- 60, no other compound of very definite melting-point could be separated. The methyl ether is very soluble in all organic reagents ; in hot ligroin it is much more soluble than in cold, from which it crystallizes in small bright-yellow prisms. From Toluquinone. To a solution of 2 grams of tolu- quinone in 800 cm 3 of water, the calculated amount of methoxylamine hydrochloride was added. Yellow crystals began to precipitate in the course of two hours, and at the end of twelve hours the reaction was completed. The liquid was filtered off and extracted with ether, which, upon evapo- ration, left behind a yellow crystalline mass. The two portions were united and recrystallized from ligroin. The yield was very good, being 75 per cent of the theory. Even after boiling in ligroin with animal charcoal and recrystallizing several tunes, the substance acts like a mixture, softening at 58 and melting at 70. A portion, less soluble than any other, was easily separated, which melted at 73-74 C. and was identical in all respects with the methyl ether obtained from the silver salt of toluquinone-m-oxime made from 0-cresol. 0.1263 gram, dried over H 2 S0 4 , gave 0.2962 gram C0 2 and 0.0672 gram H 2 0. 0.2952 gram gave 25 cm 8 N at 15 C. and 772 mm. pressure. C 63.53 63.96 H 6.00 5.91 N 9.29 10.00 Toluquinonemetaoxime Acetyl Ether. This ether can be made in two ways : By adding the cal- culated amount of acetyl chloride, drop by drop, to 2-3 154 THE ETHERS OF TOLUQUINONEOXIME. grams of the silver salt suspended in 15-20 cm 8 of ligroin or absolute ether, kept cool by ice-water, evaporating at once, and extracting the residue with hot ligroin; or by heating 1 molecule of the oxime on the water-bath for an hour with 1.5 molecules of acetic anhydride, adding cold water, filtering off the tarry mass which separates, and extracting it with hot ligroin. The yield by either method is extremely poor, and sometimes after purifying by boiling with animal charcoal, the total product consisted of a few small crystals. By frac- tional crystallization two portions were separated, the less soluble composed of thick irregular prisms melting at 112- 113, and a much smaller fraction of minute, spherically grouped crystals, melting at 85-87. In analyzing the original product unf ractioned : 0.0953 gram, dried over H 2 S0 4 , gave 0.2090 gram C0 2 and 0.0444 gram H 2 0. 0.0848 gram gave 5.8 cm 8 N at 15 C. and 760 mm. pressure. Calculated for - , C 9 H 9 N0 3 . Found. C 60.30 59.91 H 5.06 5.18 N 7.84 8.01 The acetyl ether is very soluble in alcohol and ether, much less in ligroin, and very little soluble in water. Toluquinone-m-oxime Benzoyl Ether. From o-Cresol. This ether can be made from the silver salt suspended in absolute ether, or, better, from the sodium salt in alcohol solution. Slightly less than the amount of sodium calculated to form a sodium oxime is dissolved in 100 cm 3 of alcohol, 2-5 grams of the oxime added, and into the filtered solution slightly more than the theoretical quantity of benzoyl chloride is dropped slowly, the solution being kept cool. The benzoyl ether begins to separate immediately, and after a few moments the alcohol can be filtered off and rejected, as it contains little of the substance. After boiling THE ETHERS OF TOLUQUINONEOXIME. 155 in alcohol with animal charcoal, when the solution is submitted to fractional crystallization, three-fourths of the crude product can be readily separated in the form of bright-yellow needles melting at 193. Upon concentrating the mother-liquor to a small volume and cooling, nearly the theoretical quantity of the benzoyl ether can be recovered. Nothing further was ever obtained save a few flakes of benzoic acid formed by the saponifying action of hydrochloric acid, produced by the slight excess of benzoyl chloride acting on alcohol. When the portion obtained upon concentration was repeatedly fractioned, it could be separated into portions melting approxi- mately at 193, and others melting almost completely from as low as 135 to 155. There seemed to be a tendency, however, for these lower fractions to liquefy at 142 -144, and once, from an alcoholic solution of a fraction melting at 140-150, that evaporated at ordinary temperature, short, thick, prismatic crystals separated from the curved needles of the higher melting (193) fraction. These few prisms melted at 144 without decomposition. An analysis of the body liquefying at 193 C. gave the following figures: 0.2277 gram, dried over H 2 S0 4 , gave 0.5785 gram C0 2 and 0.0945 gram H 2 0. 0.2001 gram gave 10 cm 8 N at 15 C. and 760 mm. pressure. C 69.68 69.30 H 4.60 4.61 N 5.82 5.85 An analysis of a fraction melting from 145 -165 gave these percentages: 0.1421 gram, dried over H 2 S0 4 , gave 0.3635 gram C0 2 and 0.0565 gram H 2 0. 0.6726 gram gave 34.1 cm 8 N at 15 C. and 770 mm. pressure. Calculated for uv*,,^ n TT icrk iround. C 14 H n NO 3 . C 69.68 69.74 H 4.60 4.43 N 5.82 6.02 156 THE ETHERS OF TOLUQUINONEOXIME. The benzoyl ether is not at all soluble in water or ligroin, is slightly soluble in cold alcohol, but dissolves readily in ether, chloroform, and glacial acetic acid. The low-melting body seems to be more soluble in alcohol than its higher-melting isomer, but, as nine-tenths of the total product appeared to be the body melting at 193, the small proportion of the low- melting compound may account partially for the idea that the latter is more soluble. From Toluquinone. The oxime was made according to the method of Goldschmidt and Schmidt,* by treating toluquinone in aqueous solution with an excess of hydroxylamine hydro- chloride, extracting with ether, and purifying with animal charcoal. From the sodium salt of the oxime thus formed the benzoyl ether was made in the manner previously described. The product is identical with the benzoyl ether made from o-cresol. The body melting at 193 C. was readily isolated and analyzed: 0.1002 gram, dried over H 2 S0 4 , gave 0.2565 gram C0 2 and 0.0452 gram H 2 0. 0.3581 gram gave 17.8 cm 8 N at 15 C. and 760 mm. pressure. Calculated for v , C M H U N0 8 . Found. C 69.68 69.81 H 4.60 5.01 N 5.82 5.77 By fractional crystallization, low-melting portions were separated, exactly as hi the case of the benzoyl ether made from o-cresol, except that some fractions were obtained melt- ing partially as low as 129. In order to be certain that benzoic acid (m. p., 120 C.) was not unduly lowering these melting-points, these fractions were boiled with water and filtered hot. The filtrate was not acid to litmus and contained only a trace of organic matter. The ethers, when dried, gave the same melting-point as before boiling with water, and, upon recrystallizing from a little alcohol, did not exhibit any * Ber. Dtsch. chem. Ges., xvii, 2063. THE ETHERS OF TOLUQUINONEOXIME. 157 change. Furthermore, the benzoyl ether obtained by the usual method, when only 75 per cent, of the theoretical quantity of benzoyl chloride was used, gave fractions begin- ning to melt at 129. Obviously benzoic acid could not be present in these instances. By slow, spontaneous evaporation of the alcoholic solution of a fraction liquefying at 140-150, a few crystals melting at 142, similar to those obtained in the same manner from the oxime made from 0-cresol, separated from the body melting at 193. Since there was not enough of this com- pound for a combustion, an analysis was made of a fraction melting at 137 C. with the following results : 0.1131 gram, dried over HgSO^ gave 0.2901 gram C0 2 and 0.0480 gram H 2 0. Calculated for _ , C U H U N0 8 . Found - C 69.68 69.96 H 4.60 4.72 Fractions were frequently obtained melting at about 177 C., and from these attempts were repeatedly made to isolate some of the ortho isomer. None was ever detected under the lens, and fractional crystallization always separated such portions, principally into the body melting at 190, and a small fraction melting much lower. Since the benzoyl ether will decompose into a dark-brown liquid with the evolution of brown fumes of nitrogen oxides when heated above 160, it is only by rapidly heating that 193 can be observed as a melting-point. This applies also to the bromine addition-products of the benzoates of both the ortho- and metaoximes. Dibromtoluquinone-m-oxime Benzoyl Ether. The benzoyl ether was dissolved in chloroform and cooled while the theoretical quantity of bromine was added in small portions. After standing for an hour, the chloroform was evaporated spontaneously, and the light-brown residue re- 158 THE ETHERS OF TOLUQUINONEOXIME. crystallized from glacial acetic acid. It can also be purified by dissolving in fuming nitric acid and pouring into water. On analysis : 0.1437 gram, dried over H 2 S0 4 , gave 0.2176 gram C0 2 and 0.0339 gram H 2 0. 0.1215 gram gave 0.1140 gram AgBr. Calculated for ,, , C M H u Br 2 N0 8 . Found - C 41.90 41.30 H 2.76 2.62 Br 39.87 39.93 The dibromide is insoluble in water, somewhat soluble in cold alcohol, and readily dissolves in chloroform and glacial acetic acid, from which it crystallizes in white prisms melting at 165 C. with decomposition. Monobromtoluquinone-m-oxime Benzoyl Ether. When the dibromide is boiled with alcohol hydrobromic acid splits off, two hours being required to complete the process, during which little or no saponification takes place. After recrystallizing from alcohol, a mixture of monobrom compounds is obtained, melting from 155 -170. A portion melting with decomposition at 174 C., was separated and analyzed : 0.1208 gram, dried over H 2 S0 4 , gave 0.2319 gram C0 2 and 0.0345 gram H 2 0. 0.0735 gram gave 0.0425 gram AgBr. Calculated for Fminrl C 14 H 10 BrN0 8 . C 52.49 52.36 H 3.15 3.17 Br 24.98 24.63 It is very similar in its properties to toluquinone-ra-oxime benzoyl ether, crystallizing from alcohol in bright>yellow needles. THE ETHERS OF TOLUQUINONEOXIME. 159 Toluquinone-o-oxime Methyl Ether. The methyl ether of toluquinone-o-oxime was made from the silver salt and methyl iodide in the same manner as its meta isomer, the yield being somewhat better. Once recrys- tallized from ligroin: 0.1247 gram of the ether, dried over H 2 S0 4 , gave 0.2931 gram C0 2 and 0.0675 gram H 2 0. 0.3484 gram gave 27 cm 8 N at 15 C. and 772 mm. pressure. Calculated for , , Found ' C 63.53 64.10 H 6.00 6.01 N 9.29 9.22 Its properties are almost identical with the methyl ether of the meta form. It crystallizes from ligroin in long yellow needles, every portion of which, obtained by fractional crys- tallization, melts at 69 C. Dibromtoluquinone-o-oxime Methyl Ether. The methyl ether was dissolved in chloroform, cooled, and the calculated quantity of bromine added. The reaction is completed in twenty minutes, and, on evaporation of the chloroform, the dibromide is left behind as a dirty white mass. Once recrystallized from ligroin, the substance is ready for analysis: 0.1690 gram, dried over H 2 S0 4 , gave 0.1877 gram C0 2 and 0.0456 gram H 2 O. 0.0744 grain gave 0.0909 gram AgBr. Calculated for &*,,** C 8 H 9 Br 2 N0 2 . C 30.87 30.29 H 2.92 3.00 Br 51.41 52.00 It is insoluble hi water but quite soluble in most organic reagents. From ligroin it crystallizes in white prisms, melt- ing at 112 C. 160 THE ETHERS OF TOLUQUINONEOXIME. Toluquinone-o-oxime Acetyl Ether. This ether was made by Wurster and Riedel * from tolu- quinone-0-oxime and acetic anhydride. It can be prepared also from the silver salt and acetyl chloride, the yield being very poor, only a trifle better than its meta isomer, which it closely resembles in its properties. 0.1194 gram, dried over H 2 SO 4 , gave 0.2666 gram C0 2 and 0.0545 gram H 2 O. 0.2389 gram gave 15 cm 8 N. at 15 C. and 760 mm. pressure. Calculated for - , C 9 H 9 N0 8 . Found - C 60.30 60.89 H 5.06 5.07 N 7.84 7.35 From ligroin it crystallizes in irregular yellow prisms, melting at 92 C. Toluquinone-o-oxime Benzoyl Ether. This ether was made in alcohol solution from the sodium salt in the manner already described for the benzoyl ether of the meta form. The yield is practically theoretical, and, after one recrystallization from alcohol, the product is ready for analysis. 0.1589 gram, dried over H 2 S0 4 , gave 0.4074 gram C0 2 and 0.0686 gram H 2 0. 0.5101 gram gave 26.1 cm 8 N at 15 C. and 7.66 mm. pressure. Calculated for im,! C U H U N0 3 . C 69.68 69.92 H 4.60 4.80 N 5.82 6.04 Although submitted to the most careful fractional crystal- lization, every particle obtained melted sharply at 177 C., with slight decomposition. From alcohol it crystallizes in light, brownish-yellow blades, which have all the properties * Ber. Dtsch. chem. Ges., xii, 1799. THE ETHERS OF TOLUQUINONEOXIME. 161 of the metaoxime benzoyl ether except that it is a little more soluble in organic reagents. Dibromtoluquinone-o-oxime Benzoyl Ether. The benzoyl ether was dissolved in chloroform, bromine added to the cooled solution, and the product recrystallized from glacial acetic acid. 0.2111 gram, dried over H 2 S0 4 , gave 0.3208 gram C0 2 and 0.0564 gram H 2 0. 0.1003 gram gave 0.0924 gram AgBr. Calculated for Found C 41.90 41.44 H 2.76 2.97 Br 39.87 39.21 It is insoluble in ligroin and water and but little soluble in cold alcohol. Fuming nitric acid dissolves it readily, and from the solution it is precipitated unchanged by water. It crystallizes from glacial acetic acid hi short, thick, ortho- rhombic prisms melting at 159 C. with decomposition. VOL. II. 11 XXI THE APPLICATION OF IODINE IN THE ANALYSIS OF ALKALIES AND ACIDS. BY CLAUDE F. WALKER AND DAVID H. M. GILLESPIE* IT is well known that when a free mineral acid is added to a neutral mixture of metallic iodate and iodide, the iodate is reduced and iodine is liberated according to the equation : EI0 8 + 5EI + 3H 2 S0 4 = 31, + 3E 2 S0 4 + 3H 2 0. This reaction is complete and non-reversible under the conditions of analysis, and it may therefore be applied to the estimation of amounts of iodate, iodide or mineral acid present in an unknown solution. A solution of iodate to be analyzed is mixed with an excess of iodide and mineral acid, the resulting free iodine estimated by directly titrating with sodium thiosulphate or arsenious acid, and one-sixth of the amount found taken as equivalent to the iodate originally present.f Similarly, a solution of iodide to be analyzed is mixed with an excess of iodate and mineral acid, the resulting free iodine estimated by directly titrating in alkaline solution with arsenious acid, and five-sixths of its amount taken as equivalent to the iodide originally present.^ A solution of mineral acid to be analyzed is mixed with an excess of iodate and iodide, the resulting free iodine estimated by directly titrating with sodium thiosulphate, and its entire amount taken as equivalent to the amount of mineral acid originally present. Groger has applied the last mentioned method to * From Am. Jour. Sci., vi, 455. t Rammelsberg, Fogg. Ann., cxxxv, 493 ; Walker, Am. Jour. Sci., iv, 235. This volume, p. 52. J Gooch and Walker, Am. Jour. Sci., iii, 293. This volume, p. 33. Kjeldahl, Zeitschr. anal. Chem., xxii, 366 ; Furry, Am. Chem. Jour., vi, 341 ; Groger, Zeitschr. angew. Chem., 1894, 52. IODINE IN ANALYSIS OF ALKALIES, ETC. 163 the direct analysis of various mineral acids, and has obtained results manifestly better than those afforded by the use of vegetable indicators. Groger has also indirectly analyzed solutions of alkali hydroxides and carbonates by adding the solution to be analyzed to a measured volume of mineral acid, previously standardized by the above method, and estimating the small excess of free mineral acid that finally remains by the same method. The only difficulty with the Groger process lies in the fact that under the conditions present the end-point of the final reaction between iodine and sodium thiosulphate is somewhat obscured by a peculiar back-play of color due to a continuous slow liberation of iodine in the system. When a solution of a metallic hydroxide is acted on by iodine at a temperature high enough to decompose the small amounts of hypoiodites that might otherwise be present, the final action results in the formation of an exactly neutral mixture of iodate and iodide, according to the equation : 6EOH + 3I 2 = BI0 8 + 5RI + 3H 2 0. Phelps * has shown that in the case of barium hydroxide at least this reaction is regular and complete under the conditions of analysis, and is independent of the excess of iodine which remains hi the neutral mixture unacted upon and may be estimated by directly titrating with arsenious acid. Phelps not only applies this principle of action to the standardization of solutions of barium hydroxide by boiling with an excess of iodine in a trapped flask, but also bases thereon a differential method for determining carbon-dioxide, in which the liberated gas is run into a measured amount of barium hydroxide, the final excess of which is estimated by treating with iodine in the presence of the precipitated barium carbonate. The good result obtained by Phelps with barium hydroxide suggested that the attempt be made to analyze alkali hydroxides, and possibly carbonates, by a method, simpler than that devised by Groger, based on the direct treatment of these compounds * Am. Jour. Sci., ii, 70. Volume I, p. 369. 164 APPLICATION OF IODINE IN THE with iodine in hot solution. It also seemed possible to apply the differential method not only to carbon dioxide but to any acid or other compound that will act definitely and completely with the metallic hydroxide employed, provided the soluble or insoluble product formed will not be attacked when heated in the presence of iodine. It was decided to modify the Phelps process, however, in order to obviate the necessity of handling large measured amounts of iodine in a flask trapped to prevent mechanical loss by heating. The flask was therefore dispensed with altogether, and the hydroxide solution to be analyzed was mixed with an approximately measured excess of iodine solution, in an Erlenmeyer beaker, the mouth of which was lightly closed with a little trap to prevent loss by spattering. The excess of iodine was then completely removed by boiling, and the cooled colorless solution remaining, which contained a neutral mixture of iodate and iodide, was acidified with a mineral acid and the liberated iodine titrated with sodium thiosulphate, the amount found being equivalent to the amount of hydroxide taken for analysis. The present investigation was undertaken to study the limitations and possible applications in analysis of the reactions between iodine on the one hand, and barium hydroxide, potassium hydroxide and sodium carbonate on the other. It was soon found that the reaction in the case of sodium carbonate is entirely dependent on conditions of time, mass, and temperature, and cannot be pushed to completion except under conditions that make its application in analysis impossible. In the case of barium and potassium hydroxides both the original procedure of Phelps and the modification above described were employed. The modified method was found to be the more convenient and speedy of the two. The results obtained in both cases agreed with one another, but were invariably lower by a small nearly constant amount than those obtained by both the gravimetric and the Groger processes. This error of the Phelps process and its modification is possibly due to the action of atmospheric carbon dioxide on the hydroxide solution during the short tune it is ANALYSIS OF ALKALIES AND ACIDS. 165 exposed. While it will affect the value of the method as a means of accurately determining the absolute amount of hydroxide present in a given volume of solution, it cannot so affect the accuracy of any differential method founded on the original Phelps process or its modification. This is demonstrated by the work of Phelps in the case of carbon dioxide, and by the present investigation in the case of hydrochloric and sulphuric acid. Analyses of these two acids were made by adding the solution to be analyzed to a measured volume of barium or potassium hydroxide, previously standardized by the modified Phelps method. The small excess of hydroxide remaining was then estimated by the same method, the results agreeing with those already obtained by both the gravimetric and the Gro'ger processes. It seems probable that other acids and compounds for which there is now no rapid iodometric method may be analyzed by a method similar to this, which has given good results with carbonic, hydrochloric, and sulphuric acids. Decinormal solutions of the alkali hydroxides were prepared, and kept with great care in trapped bottles, from which por- tions for analysis were measured by means of a self-feeding burette, which was also fitted with a trap. All vessels and water used were made as free as possible from carbon dioxide, and the operations were conducted as rapidly as possible. In the analyses by the Phelps method a carefully measured excess of decinormal iodine was drawn into a small ether wash- bottle, and the desired amount of alkali was rapidly run into it. The stopper, to which had been sealed a Will and Varren- trapp absorption bulb was placed in the bottle and the bulb was charged with a 5 per cent solution of potassium iodide to catch any escaping vapors of iodine. The apparatus was placed over a low flame and the contents heated to boiling or slightly longer, and then cooled hi a stream of water. The contents of the bulb and connecting tubes were then washed into the flask, and the excess of free iodine remaining was titrated with arsenious acid, in the presence of 5 cm 3 of starch emulsion. Blank analyses were made to insure against me- chanical loss of iodine during boiling and to prevent any error 166 APPLICATION OF IODINE IN THE on account of the presence of carbonate or other impurity in the solutions employed. Some of the results obtained with barium hydroxide are given in Table I. The variation in different analyses of the same series is not large, and the results are independent of the amount taken for analysis and of the excess of iodine employed. TABLE I. ANALYSES OP * BARIUM HYDROXIDE SOLUTION. (By boiling in a trapped flask with an excess of iodine.) Exp. Ba(OH) a taken. Iodine taken. Iodine absorbed by Ba(OH) 3 found. Mean. Variation. Ba(OH) 2 . cm 8 grm. grm. grm. grm. grm (1) 10 0.13 0.1054 0.0712 0.0699 0.0013+ (2) 10 0.14 0.1028 0.0692 0.0699 0.0007- (3) 20 0.23 0.2072 0.1399 0.1398 0.0001+ (4) 20 0.25 0.2074 0.1401 0.1398 0.0003+ (5) 40 0.44 0.4143 0.2798 0.2796 0.0002+ (6) 40 0.44 0.4148 0.2802 0.2796 0.0006+ (7) 40 0.48 0.4160 0.2809 0.2796 0.0013+ (8) 40 0.48 0.4126 0.2786 0.2796 0.0010- (9) 40 0.51 0.4115 0.2779 0.2796 0.0017- (10) 40 0.51 0.4136 0.2793 0.2796 0.0003- The analyses of potassium hydroxide were made in the same way as were those of barium hydroxide, and gave quite similar results. They follow in Table II. TABLE H. ANALYSIS OP J POTASSIUM HYDROXIDE SOLUTION. (By boiling in a trapped flask with an excess of iodine.) Exp. KOH taken. Iodine taken. Iodine absorbed by KOH. KOH found. Mean. Variation. cm> grm. grm. grin* grm. grm. (1) 10 0.20 0.1621 0.0716 0.0717 0.0001- (2) 10 0.23 0.1613 0.0715 0.0717 0.0002- ( 3 ) 15 0.30 0.2404 0.1063 0.1076 0.0013- 4) 15 0.30 0.2429 0.1074 0.1076 0.0002- 5 16 0.34 0.2431 0.1076 0.1076 0.0001- (6) 25 0.51 0.4089 0.1808 0.1792 0.0016+ (7) 25 0.51 0.4058 0.1794 0.1792 O.OOOSH- ANALYSIS OF ALKALIES AND ACIDS. 167 The analyses by the modification of the Phelps method were made by drawing into an Erlenmeyer beaker of convenient size an approximately measured excess of decinormal iodine, and rapidly running the desired amount of alkali into it. The neck of the beaker was then closed by a little trap, made of one of the halves of a double end calcium chloride drying tube, to prevent appreciable loss by spattering. The beaker was then placed over a low flame, and the contents boiled until the last trace of the excess of iodine had volatilized from the solution and the trap. The volume was carefully regulated before and during the boiling, being kept as small as possible, usually amounting to about 100 cm 3 at the start and 35 cm 3 at the close. In the case of barium hydroxide care had to be taken to keep the dilution sufficient to prevent the separation the crystalline barium iodate, which is soluble with difficulty. To steady the ebullition a little spiral of platinum was intro- duced into the beaker. After the boiling had ceased, the colorless solution, containing a neutral mixture of iodate and iodide, was cooled hi running water and treated with 10 cm 3 of dilute (1 : 3) hydrochloric acid or (1 : 3) sulphuric acid. The liberated iodine was titrated directly with sodium thiosul- phate, hi the presence of 5 cm 3 of starch emulsion. In the case of barium hydroxide the iodine was liberated with dilute (1 : 3) hydrochloric acid to save the inconvenience of working in the presence of precipitated barium sulphate ; with potas- sium hydroxide, however, dilute (1 : 3) sulphuric acid was employed. In view of a statement by Pickering* that titra- tions with sodium thiosulphate in the presence of acid involve an error, a series of blank analyses was made which showed con- clusively that no such error exists under the conditions which obtain in the process under consideration. Care was also taken, as in a former case, to guard against the possible presence of carbonates or other impurities in the reagents employed. In Table III are given the results of a series of analyses of barium hydroxide by the modified method just described. They agree fairly well with those of Table I. * Jour. Chem. Soc., xxxvii, 134. 168 APPLICATION OF IODINE IN THE ANALYSES OF TABLE III. BAKIUM HYDROXIDE SOLUTION. (By boiling with excess of iodine in an open beaker to decoloration, and acidifying the residue.) Exp. Ba(OH), taken. Iodine taken. Iodine absorbed by Ba(OH),. Ba(OH), found. Mean. Variation. cm** grni. grin. gnu. griii. grin. (1) 10 0.13 0.1023 0.0691 0.0695 0.0004- 2) 10 0.16 0.1020 0.0689 0.0695 0.0006- 3 16 0.18 0.1548 0.1046 0.1043 0.0003+ 4 15 0.20 0.1546 0.1045 0.1043 0.0002+ (5) 20 0.23 0.2049 0.1384 0.1390 0.0006- (6) 20 0.25 0.2058 0.1390 0.1390 0.0000 (7) 20 0.32 0.2065 0.1394 0.1390 0.0004+ (8) 25 0.29 0.2567 0.1734 0.1738 0.0004- (9) 25 0.32 0.2562 0.1730 0.1738 0.0008- (10) 40 0.47 0.4120 0.2783 0.2780 0.0003+ (11) 40 0.48 0.4119 0.2782 0.2780 0.0002+ (12) 40 0.48 0.4152 0.2804 0.2780 0.0024+ (13) 40 0.49 0.4109 0.2775 0.2780 0.0005- The analyses of potassium hydroxide by the modified method are given in Table IV, and are found to agree well with those of Table II. TABLE IV. ANALYSES OF ^ POTASSIUM HYDROXIDE SOLUTION. (By boiling with excess of iodine in an open beaker to decoloration, and acidifying the residue.) Eip. KOH taken. Iodine taken. Iodine absorbed by KOH. Ba(OH) 2 found. Mean. Variation. cm grm. gnu. grm. grm. grm. (1) 10 0.20 0.1624 0.0718 0.0721 0.0003- (2) 10 0.23 0.1618 0.0715 0.0721 0.0006- (3) 10 0.25 0.1622 0.0717 0.0721 0.0004- (4) 15 0.30 0.2459 0.1087 0.1082 0.0005+ (5 15 0.34 0.2473 0.1093 0.1082 0.0011+ (6) 15 0.38 0.2441 0.1079 0.1082 0.0003- (7) 20 0.41 0.3274 0.1447 0.1442 0.0005+ (8) 20 0.46 0.3259 0.1441 0.1442 0.0001- (9) 20 0.51 0.3269 0.1445 0.1442 0.0003+ (10 25 0.51 0.4052 0.1791 0.1803 0.0012- (11) 25 0.57 0.4082 0.1805 0.1803 0.0002+ (12) 25 0.63 0.4080 0.1804 0.1803 0.0001+ ANALYSIS OF ALKALIES AND ACIDS. 169 A gravimetric analysis of the barium hydroxide solution in which the barium was weighed as the sulphate, gave as a result 0.1411 grm. Ba(OH) 2 for each 20 cm 3 taken. An anal- ysis of the same solution by the Groger process gave for the same volume 0.1420 grm. The result by the Phelps process, however, was 0.1398 grm., and by the modified process 0.1390 grm. That the difference of 2 mg. between the results by the gravimetric and the Groger processes on one hand, and the Phelps process and its modification on the other, may be due to atmospheric carbon dioxide, has already been pointed out. A gravimetric analysis of the potassium hydroxide solution by evaporating and weighing as KC1 gave 0.1111 grm. KOH for each 20 cm 3 taken, agreeing with 0.1106 grm. obtained by the Groger process. The analyses by the Phelps process and its modification for the same solution gave 0.1076 grm. and 0.1082 grm. respectively. These results are strikingly in accord with those obtained with barium hydroxide. In the application of the modification of the Phelps process to the indirect analysis of hydrochloric and sulphuric acids the procedure was essentially the same as that detailed for the analysis of barium and potassium hydroxides in Tables III and IV. The acid solution to be analyzed was drawn into an Erlenmeyer beaker, a measured excess of standardized alkali TABLE V. ANALYSES OF ^ HYDKOCHLOKIC ACID SOLUTION. (By adding to excess of ^ Ba(OH) 2 , boiling with excess of iodine to decolora- tion and acidifying the residue.) Exp. HCl taken. Ba(OH), taken. Ba(OH) 2 neutralized HCl found. Mean. Variation. by HCl. cm 3 grill. grm. grm. grm. grm. (1) 15 0.17 0.1128 0.0480 0.0476 0.0004+ 2 15 0.17 0.1118 0.0475 0.0476 0.0001- (3) 16 0.17 0.1112 0.0473 0.0476 0.0003- (4) 25 0.26 0.1860 0.0791 0.0794 0.0003- 5) 25 0.26 0.1866 0.0794 0.0794 0.0000 6 35 0.34 0.2634 0.1120 0.1111 0.0009+ (7) 36 0.34 0.2603 0.1107 0.1111 0.0004- 170 APPLICATION OF IODINE IN THE added, and the operation completed as described. It was found that barium hydroxide and potassium hydroxide may be applied with equal accuracy to the analysis of both hydrochloric and sulphuric acids. Some of the results obtained are given in Tables V, VI and VII. TABLE VI. ANALYSES OP ^ HYDROCHLORIC ACID SOLUTION. (By adding to excess of ^ KOH, boiling with excess of iodine to decoloration, and acidifying the residue.) Exp. HC1 taken. KOH taken. KOH neutralized by HC1. HC1 found. Mean. Variation. cm 8 . grm. grm. grm. grm. grm. (1) 20 0.14 0.0972 0.0633 0.0633 0.0000 (2) 20 0.14 0.0975 0.0634 0.0633 0.0001+ (3) 25 0.14 0.1222 0.0795 0.0791 0.0004+ (4) 25 0.14 0.1207 0.0785 0.0791 0.0006- TABLE VII. ANALYSES OF J SULPHURIC ACID SOLUTION. (By adding to excess of ^ Ba(OH) 2 , boiling with excess of iodine to decolora- tion, and acidifying the residue.) Eip. as BaOH 2 taken. Ba(OH), neutralized by H 2 80 4 . H 2 80 4 found. Mean. Variation. cm 8 grm. gTIH. grm. grm. grm. (1) 10 0.21 0.0884 0.0506 0.0498 0.0008+ (2) 10 0.21 0.0880 0.0503 0.0498 0.0005+ (3) 15 0.30 0.1328 0.0754 0.0748 0.0006+ (4) 15 0.30 0.1313 0.0751 0.0748 0.0003+ (5) 25 0.43 0.2168 0.1239 0.1246 0.0007- (6) 30 0.43 0.2600 0.1481 0.1495 0.0014- An analysis of the hydrochloric acid solution by the Grb'ger method, which was found to agree in every case with the gravimetric determination, gave for each 25 cm 8 0.0801 grm. of HC1, agreeing with 0.0794 grm. and 0.0791 grm. obtained by the new method. An analysis of the sulphuric acid solution ANALYSIS OF ALKALIES AND ACIDS. 171 by the Groger method gave for each 25 cm 3 0.1241 grm. of H 2 SO 4 agreeing with 0.1246 grm. obtained by the new method. This investigation shows that the reaction between iodine and hydroxides of the alkalies and alkaline earths in hot solution is regular and complete under analytical conditions, not being appreciably affected by the mass action of considerable excesses of iodine. The reaction is best applied in analysis by titrating the alkali with an excess of iodine, removing this excess by boiling, and estimating the iodine in the residue. While certain mechanical difficulties may effect the extreme accuracy of the process as a direct means for analyzing alkalies, the action is at all times regular and may be indirectly applied with fair accuracy to the analysis of various acids and possibly to other compounds. The reaction between iodine and alkali carbonates on the contrary is irregular and cannot be made the basis of any analytical process. XXII THE ESTIMATION OF BORIC ACID. BY F. A. GOOCH AND LOUIS CLEVELAND JOKES.* THE estimation of boric acid by treating the salts of that acid with sulphuric acid, distilling with methyl alcohol, evaporating the distillate over magnesium oxide, igniting and weighing, was proposed by Rosenbladt-t A little later, and without knowledge of Rosenbladt's experience, a somewhat similar process, J which consisted in the treating of the compound of boric acid with acetic acid or nitric acid, distillation with methyl alcohol, evaporation of the distillate over calcium oxide, and ignition of the residue, was described by one of us. In the course of the development of this process, it transpired that the insolubility of magnesium oxide retards the absorption of boric acid by that substance, and that the more soluble calcium oxide retains boric acid more actively and is therefore to be preferred. Points in the treatment upon which special emphasis was laid in the original description of this process were the choice of a suitable apparatus for the distillation, the employment of a loosely stoppered receiver for the reception of the distillate upon slaked lime, the careful removal of water from the substance in the retort before acidifying and treating with the methyl alcohol, regulated use of acid, and care in the evapora- tion and ignition. The attainment of good results in this process depends upon attention to details. Modifications have been suggested by several investigators. Thus, instead of igniting the calcium oxide in a large platinum crucible, transferring it to the * From Am. Jour. Sci., vii, 34. t Zeitschr. anal. Chem., xxvi, 21. J Am. Chem. Jour., ix, 23. ESTIMATION OF BORIC ACID. 173 receiver to hold the boric acid, and returning the calcium oxide with the distillate to the same crucible for subsequent ignition of the residue, as was originally proposed, Penfield* prefers to ignite the calcium oxide in a small crucible, to collect the distillate hi ammoniacal water, to evaporate the latter over the calcium oxide in a large platinum dish, and to transfer this residue back to the small crucible for the final evaporation and ignition. Kraut f suggests a modification of form in the apparatus with no other essential change hi conditions. MoissanJ has suggested changes in the apparatus and avoids a transfer of the calcium oxide collecting the distillate by itself in a closed receiver, trapped with an ammonia bulb to prevent the escape of the boric acid from the distillate; furthermore, Moissan's process calls for the use of an amount of calcium oxide from fifteen to twenty tunes greater than that theoretically required. From our experience it seems obvious that the demand for this amount of calcium oxide arises from an excessive use of nitric acid in the retort and the consequent modification of conditions in the distillate. For- tunately this difficulty may be avoided by the use of a little phenolphthalein as an indicator in the retort and care to limit the addition of nitric acid to the amount required to produce distinct acidity. The addition of a drop of the acid and another of the indicator should be repeated once or twice during the distillation to insure the return of the volatilized acid to the salt slightly decomposed in the process. The effect of much nitric acid is bad, not only because it neu- tralizes the calcium oxide when it passes to the distillate, but because when it is used a tendency is developed on the part of the dried mixture of calcium hydroxide and borate to puff explosively if the ignition is begun as soon as the residue is dry. If the residue is heated gradually and as strongly as possible over a radiator before the flame is actually applied to the crucible, no such action takes place ; we are disposed to attribute it to the effect of the nitrate and nitrite, produced by * Am. Jour. Sci., xxxiv, 222. t Zeitschr. anal. Chem., xxxvi, 3. t Comp. rend., cxvi, 1084. 174 ESTIMATION OF BORIC ACID. the absorption of nitrous fumes in the lime, upon the alcohol or other organic matter retained by the lime hi the evaporation and drying unless the latter process is prolonged at high temperature. That good results may be obtained with small amounts of calcium oxide, provided care as to the use of nitric acid and the conditions of ignition be taken, is shown by the figures of the original description and by the following experiments, in which phenolphthalein was employed as an indicator and the residue heated strongly over the radiator before actual ignition. CaO taken. B 2 O 8 taken. B 2 3 found. Error. grin. 2.3405 1.7620 2.1757 2.5656 grm. 0.1788 0.1790 0.1824 0.1788 grm. 0.1792 0.1785 0.1840 0.1786 grm. 0.0004+ 0.0005- 0.0016+ 0.0002- These results are accurate within reasonable limits. On the other hand, without care to ignite gradually we have noted errors of from 0.0030 grm. to 0.0060 grm. in the process otherwise conducted similarly. Doubtless the use of large amounts of calcium oxide as suggested by Moissan may serve the purpose of diffusing the explosive mixture through a mass of inert matter sufficient to prevent violent puffing, but care to heat over the radiator as strongly as possible before opening the flame directly to the crucible answers the same end. The difficulty does not exist when acetic acid is used in place of nitric acid, though even in this case it is safer to use the radiator in the first stages of heating, thus avoiding the danger of mechanical loss by too rapid ignition. Following are determinations made by this method with the use of acetic acid. The results of these experiments, as well as those of the investigators mentioned, are a sufficient answer to the criticism of Reischle,* that acetic acid and nitric acid do not liberate boric acid in the distillation pro- * Zeitschr. anal. Chem., xxvi, 512. ESTIMATION OF BORIC ACID. 175 CaO taken. B 2 3 taken. B,O 3 found. Error. gnu. 0.9977 1.0220 1.3717 1.1310 gnu. 0.2065 0.2067 0.2077 0.1791 gnu. 0.2062 0.2070 0.2075 0.1795 gnu. 0.0003- 0.0003+ 0.0002- 0.0004-1- cess so that good results may be obtained. Moreover, it has been shown by one of us * that even carbonic acid is strong enough to bring about complete volatility of boric acid with methyl alcohol. The use of Calcium Oxide as a Retainer. Quite recently Thaddeeff f has advocated the abandonment of calcium oxide as an agent for holding boric acid in the evaporation of alcoholic and aqueous solutions, on account of the hygroscopic nature of the oxide and the consequent difficulty of securing it in definite conditions for weighing, and proposes, instead of using calcium oxide, to retain and estimate boric acid in solution by converting it into the form of potassium borofluoride. In the final modification of ThaddeefFs method the proposal is made to liberate the boric acid from its compounds by sulphuric acid, to volatilize it in methyl alcohol with the aid of a current of dry air, to catch the distillate in potassium hydroxide, to treat the mixture of hydroxide and borate with hydrofluoric acid in excess and evaporate on the steam bath, to digest the residue of fluoride and borofluoride at normal temperatures for two hours with 50 cm 3 of a potassium acetate solution (sp. gr. 1.14) and for twelve hours more after adding 100 cm 3 of ethyl alcohol (sp. gr. 0.805), to filter on paper, wash the residue with 62-72 cm 3 of alcohol (sp. gr. 0.805), dry at 100 and weigh as potassium borofluoride, after which the borofluoride is to be dissolved in boiling water and tested with calcium chloride for possible contamination * Jones, Am. Jour. Sci., v, 442. This volume, p. 100. t Zeitschr. anal. Chem., xxxvi, 568. 176 ESTIMATION OF BORIC ACID. by the presence of a fluoride. Plainly ThaddeefFs procedure presents at the outset difficulties; for besides the incon- venience of conducting long digestions with reagents of regulated strength, the difficulty of procuring hydrofluoric acid free from silica, which if present (as it usually is hi the so-called chemically pure hydrofluoric acid of commerce) would be retained in the borofluoride as potassium fluosilicate, the inaccuracy of the dried paper filter, and the obvious uncertainty of success in an attempt to wash a mixture of acid potassium fluoride and potassium borofluoride in potas- sium acetate and alcohol so that the one shall be rendered entirely soluble while the other remains sensibly unaffected, besides these objections, there is the theoretical probability that boric acid mnst be lost by volatilization during the evaporation of the solution of the mixed salts in the presence of free hydrofluoric acid. This last point was put to the proof by submitting to distillation in a platinum retort a mixture of equal quantities of borax and potassium hydroxide with an excess of hydrofluoric acid, collecting the distillate in potassium hydroxide, evaporating it to dryness and testing it for the presence of boric acid. When this residue from the evaporated distillate was treated with sulphuric acid and methyl alcohol, the burning alcohol vapor gave plainly the green flame of boric acid. Another portion showed clearly the presence of boric acid when acidulated with hydrochloric acid tested with turmeric paper. No boric acid could be detected in any of the reagents used. It is plain, therefore, that boric acid does volatilize upon the evapora- tion of a mixture of potassium fluoride and borofluoride in acid solution. The amount of such loss is disclosed hi the record of the following experiment. Portions of a standard solution of boric acid, prepared by dissolving a known weight of anhydrous boric oxide in a liter of water, were mixed with a solution of potassium hydroxide (free from silica and standardized by conversion to the chloride) in the proportions to form the potassium borofluoride, and an excess of hydro- fluoric acid was added. The mixture was evaporated and ESTIMATION OF BORIC ACID. 177 the residue was dried and weighed at 100, the whole opera- tion being conducted in platinum. Exp. HKP, equivalent to KOH taken. taken. KPBFs theoretical weight. KFBF 8 found. Error in terms of KFBF 8 . Error in terms of 8,03. grm. grin. grm. grin. grm. grm. (1) 0.3531 0.1582 0.5701 0.5580 0.0121- 0.0033- (2) 0.3192 0.1430 0.5154 0.5100 0.0054- 0.0015- (3) 0.3192 0.1430 0.5154 0.5030 0.0124- 0.0034- (4) 0.3192 0.1430 0.5154 0.5088 0.0066- 0.0018- (6) 0.3192 0.1430 0.5154 0.5114 0.0040- 0.0011- In experiments (1) to (3) the volume of the solution evaporated was about 50 cm 3 . In experiment (4) this volume was reduced about one-half before acidifying with hydrofluoric acid, while in experiment (5) the solution was diluted about one-half before adding the hydrofluoric acid. It is plain, therefore, that in this single step of Thaddeeff's process there is a considerable error of deficiency. On the other hand, the errors for the full process as laid down by Thaddeeff have been in our experience invariably differences of excess presumably because the loss due to volatilization of boric acid has been overbalanced by the inaccuracy in washing. It is plain that the process can give true indica- tions only by the balancing of considerable errors. If we take into consideration, therefore, the inevitable inaccuracy and inconvenience of Thaddeeff's proposal, it cannot be regarded as a desirable substitute for the process according to which boric acid is absorbed and retained for weighing with calcium oxide, especially since the difficulties in the way of getting constant weights of that substance are by no means insuperable. Thus the following table shows the series of weights taken in several experiments in bringing calcium oxide to a constant weight in a 50 cm 3 platinum crucible ignited over a blast lamp, as well as the weight taken after adding a known amount of standard boric acid solution to the slaked oxide, evaporating, and igniting. The results recorded are those of experiments VOL. II. 12 178 ESTIMATION OF BORIC ACID. made on days not moist beyond the average and with the greatest care to approach the limit of accuracy with which calcium oxide and the boric acid held thereby can be weighed under ordinarily favorable conditions. The first weight of calcium oxide recorded under each experiment was taken after a strong ignition over the blast lamp for about one-half hour. The succeeding weights were taken after similar ignitions of five minutes. In all cases the crucible was left to stand a definite period in a sulphuric acid desiccator, and, after the approximate value had once been obtained, the weights of the preceding weighing were replaced on the balance before the crucible was taken from the desiccator. The average of the weights bracketed is the weight taken as constant for the calculations. Exp. CaO taken. B,0 8 taken. CaO + B 2 8 taken. CaO + B 2 O 8 found. Error. (1) grm. 0.9505 0.9493 I ft 0400 0.9493 f Uyu4b 0.0002+ Obviously calcium oxide may be weighed with accuracy, with or without boric acid ; but the fact remains that a less hygroscopic absorbent one requiring less care in the handling, is to be desired. The use of Sodium Tungstate as a Retainer. In searching for a suitable material of less hygroscopicity to replace calcium oxide as a retainer for boric acid, we have found that sodium tungstate, fused with a slight excess of ESTIMATION OF BORIC ACID. 179 tungstic acid over that contained in the normal tungstate (to insure its freedom from carbonate), answers this purpose excellently. This substance is definite in weight, not hygro- scopic, soluble in water, and recoverable in its original weight after evaporation and ignition. To test its value as a retainer for boric acid, portions of it 4 to 7 grm. were fused and weighed in a 50 cm 3 crucible, the tungstate was dissolved in water and to it was added a known amount of a standard solution of boric acid. After diluting, mixing, evaporating, and fusing the residue, the increase in weight should represent the boric anhydride held by the tungstate. The results of the accompanying table show how accurately the boric acid is retained under these conditions. In experiments (3) to (7) the tungstate, after its first weighing, was dissolved, transferred to a larger platinum dish and mixed therein with the boric acid. After evaporation to a suitable volume this solution of tungstate and boric acid was transferred to the original crucible for final evaporation and ignition. Exp. NajWOi + W0 3 taken. B 2 8 taken. B,0 3 found. Error in B 2 3 . grm. grm. grm. grm. (1) 6.6416 0.1784 0.1771 0.0013- (2) 7.3134 0.1786 0.1773 0.0013- (3) 5.5003 0.0950 0.0952 0.0002-f- (4) 4.1394 0.0944 0.0944 0.0000 (5) 7.5037 0.2148 0.2149 0.0001+ (6) 4.7744 0.2718 0.2702 0.0016- (7) 6.6470 0.2503 0.2487 0.0016- It is plain that though the sodium tungstate does not hold the boric acid with absolute accuracy the errors are not unreasonable 0.0008 grm. in the mean. Upon substituting the tungstate for calcium oxide as a retainer in the distillation process, the results were likewise highly favorable. We used by preference the apparatus originally proposed, excepting that the Erlenmeyer flask used as a receiver is fitted tightly to the condenser and trapped with water bulbs. The retort is made very easily from a 150 cm 3 pipette and has 180 ESTIMATION OF BORIC ACID. the special advantage that particles of the residue spattering during distillation are easily washed from the walls of the vessel by a slight rotary motion of the retort. It was found that special care should be taken to give the tungstate ample time for contact with the distillate before exposing the latter to atmospheric evaporation. The distilkte was received, therefore, in a dilute solution of sodium tungstate placed in the receiver, cooled by ice and trapped with water, and the mixture was well stirred, allowed to stand one half-hour, evaporated to small volume in a large dish, and transferred to the crucible in which the tungstate had been originally weighed. After thorough drying the residue was ignited to fusion and weighed. When acetic acid was employed in the retort, care was taken in the ignition to expose the fused mass freely to the air (by causing it to flow upon the sides of the crucible) until the color of the cooled tungstate was white, in order that the reducing effect of the acetate might be eliminated. In the experiments recorded in the following table the tungstate Na,W0 4 + WO S B,0 3 taken. B 2 3 found. Error. WITH NITRIC ACID. gnu. 8.5516 4.9639 8.0033 grin. 0.1582 0.1329 0.1267 grm. 0.1572 0.1323 0.1256 grm. 0.0010- 0.0006- 0.0011- WITH ACETIC ACID. 4.9658 6.0289 4.6797 4.0013 0.1434 0.1431 0.1589 0.1433 0.1418 0.1433 0.1587 0.1422 0.0016- 0.0002+ 0.0002- 0.0011- WITH SULPHURIC ACID. 6.3439 8.8227 10.1516 6.5738 0.1582 0.1582 0.1265 0.1392 0.1579 0.1577 0.1264 0.1390 0.0003- 0.0005- 0.0001- 0.0002- was used in the receiver to retain the boric acid distilled as usual with methyl alcohol, from the borates treated with acetic ESTIMATION OF BORIC ACID. 181 acid, nitric acid or sulphuric acid, in amounts regulated by the use of phenolphthalein as an indicator. Excessive use of acid is disadvantageous, and this is especially true in the case of sulphuric acid; for, if this acid is carried over with the methyl alcohol, as it is at 100 if present in appreciable excess, a part of it, at least, is held permanently by the tungstate to increase the apparent weight of the boric acid to be estimated. The manipulation of the tungstate presents no difficulties, and the results obtained by its use are reasonably accurate. XXIII A VOLUMETRIC METHOD FOR THE ESTIMATION OF BORIC ACID. BY LOUIS CLEVELAND JONES.* WHEN boric acid and mannite are mixed in solution a peculiar compound of strongly acid properties is the result. This com- pound decomposes carbonates, and its acid taste is comparable to that of citric acid, much stronger than that of boric acid alone. Magnaninit has found that the product of such a mixture of boric acid and mannite solutions shows greater electrical conductivity and a lower freezing point than a simi- lar molecular solution of either substance alone. Other poly- atomic alcohols (but all to a less degree than mannite) and some organic acids show this peculiar property of combining chemically with boric acid to increase its acid qualities.^: Of this reaction between boric acid and the polyatomic alcohols, Thomson, Barthe,|| and Jb'rgensen^" have taken advantage to develop methods for the volumetric estimation of boric acid. Glycerine is used to form a combination with boric acid, suffi- ciently acidic to give an acid reaction when used with a sensi- tive indicator and make possible its titration with an alkaline solution. Honig and Spitz** show that in the method of Jb'rgensen a very large amount of glycerine must be used to prevent the appearance of the indication of alkalinity with phe- nolphthalein before all the boric acid is neutralized according to the following equation, 2NaOH+B 2 O 3 = 2NaOBO+H 2 O ; * From Am. Jour. Sci., vii, 147. t Gaz. Chim., xx, 428-440; xxi, 134-145. t Klein, Jour. Pharm. Chim., 4, vol. xxviii ; Lambert, Comp. rend., cviii, 1016-1017. Jour. Soc. Chem. Ind., xv, 432. || Jour. Pharm. Chim., xxix, 163. IT Zeitschr. angew. Chem., 1897, 5. ** Zeitschr. angew. Chem. (1896), 649. ESTIMATION OF BORIC ACID. 183 that in the presence of carbonates the solution must be boiled to decompose bicarbonates and the escape of boric acid by vol- atilization prevented by the use of a return condenser; and that silica must be removed by the process of Berzelius, and the solution then neutralized by the use of methylorange before a titration of the boric acid can be made. Vadam,* for the estimation of boric acid in butter makes use of mannite, which, as he finds, gives sharper indication with litmus than glycerine. According to this process, the solution to be analyzed for boric acid is neutralized by the use of litmus and a solution of sodium hydroxide. Mannite (1-2 grm.) is then added, bringing about an acid reaction with the boric acid present in free condition. The solution is then titrated to alkalinity by sodium hydroxide. None of the above methods with glycerine have, in my experience, given anything but comparatively crude results. The weak acidic properties of boric acid, the interference (and difficulty of removal) of carbon dioxide with all organic indi- cators sufficiently delicate to be used with boric acid, and indeed, the procuring of a standard alkali containing no car- bonate, together with the supposed detrimental influence of silica and the lack of a convenient method for its removal, have made the process of Gooch,f which involves distillation and weighing with calcium oxide, the only means (though requiring long time and exceeding care) in use for the accu- rate separation and estimation of boric acid. Recently sodium tungstate has been recommended from this laboratory^ as a substitute for calcium oxide to retain the distilled boric acid. The entire process, however, is one of the most exacting in analytical chemistry, and for this reason a convenient, rapid and at the same time accurate method for the estimation of boron is especially desirable. The first step toward the development of such a process must be the convenient prepa- * Jour. Pharm. Chim. (6), viii, 109-111. t Am. Chem. Jour, ix, 23-33; Moissan, Comp. rend., cxri, 1087; Kraut, Zeitschr. anal. Chem. xxxvi, 165 ; Montemartini, Gaz. Chim. Ital., xxviii, 1, 344. t Gooch and Jones, Am. Jour. Sci., rii, 34. This volume, p. 172. 184 A VOLUMETRIC METHOD FOR THE ration and the accurate estimation of the standard solution of alkali to be used for neutralizing the boric acid. This has been found to be easily accomplished by the process recommended by Kusler.* This observer, in an extensive investigation of the analytical methods for the volumetric estimation of alkalies and alkali carbonates in solution, finds that both phenolphthal- ein and methylorange are appreciably sensitive to carbonic acid, but when this interfering agent is removed by precipitation with barium chloride according to the process of Winkler,f the remaining free alkali may be estimated with great accuracy by phenolphthalein and decinormal hydrochloric acid. Obviously, if the difficulties dependent upon the action of carbon dioxide can be obviated, and if the acidity of the boric acid can be increased to such an extent that a sufficiently sen- sitive indicator will give with accuracy the neutralization point with free alkali, and if the alkali and stronger acid can be combined while boric acid alone remains free, then it should be possible to estimate boric acid volumetrically. Ex- periment has shown that barium chloride removes carbon dioxide present in carbonates, and that mannite makes a com- bination with boric acid strongly acidic to phenolphthalein. To obtain the boric acid alone in free condition many attempts have been made. Gladding,f Thaddeeff and Rosen- bladt || have isolated the boric acid by distillation with methyl- alcohol and a non-volatile acid. Many indicators theoretically insensible to free boric acid have been used to indicate the neutralization of the stronger acids. Honig and Spitz,*[f and Thomson,** use methylorange, Morse and Burton,|t tropae- olin 00, while Vadam JJ makes use of litmus. All these indicators, however, have been found by experiment to be more or less affected by boric acid in solution. On the other hand, I have found in the well known reaction according * Zeitschr. anorg. Chem., xiii, 124-150. t Massanalyse. t Jour. Am. Chem. Soc., iv, 568. Zeitschr. anal. Chem., xxxvi (9), 568. || Zeitschr. anal. Chem., xxvi, 18. f Zeitschr. anorg. Chem. (18), 549. ** Jour. Soc. Chem. Ind., xv, 432. ft Am. Chem. Jour., x, 154. |J Jour. Pharm. Chim. (6) viii, 109-111. ESTIMATION OF BORIC ACID. 185 to which a stronger acid liberates regularly iodine from a mixture of iodide and iodate, the solution of this difficulty. If both the iodide and iodate are in excess of the acid the entire amount of free acid will be neutralized and the cor- responding amount of iodine liberated according to the following equation: 5KI + KI03 + 6HC1 = 6KC1 + 3H 2 + 3I 2 . This liberated iodine may be removed by sodium thiosulphate and a solution obtained which is absolutely neutral containing only neutral salts, potassium iodide, iodate, and tetrathionate. The statements made by P. George vie* and Furry, f that boric acid present in moderate amount in solution has not the slightest action on a mixture of iodide and iodate, have been experimentally verified. Therefore when this acid is liberated by an excess of a stronger acid, and the iodine set free destroyed by thiosulphate, it remains free in solution to be titrated in any convenient manner possible. Following along the lines suggested by the above reac- tions, a volumetric process for the estimation of boric acid has been developed. For a basis of the investigations, a standard solution of boric acid was prepared by dissolving in a liter of water about eight grams of carefully weighed anhydrous boric oxide. This anhydrous boric oxide was prepared from the several times recrystallized hydrous boric acid, by long-continued fusion over a blast lamp. A solution of approximately \ sodium hydroxide was prepared from the ordinary sodium hydroxide of the laboratory. The free alkali in this solution was estimated by the process of Winkler recommended by Ktisler. The acid used to make this estima- tion was hydrochloric, standardized by silver nitrate. The full method for the estimation of boric acid as finally elaborated is as follows : The solution is made clearly acid to litmus by hydrochloric acid and 5 cm 3 of a solution (10%) of barium chloride added. An amount of iodate and iodide of potassium sufficient to liberate an amount of iodine at * Jour, prakt. Chem., xxxviii, 118. t Am. Chem. Jour., vi, 341. 186 A VOLUMETRIC METHOD FOR THE least equivalent to the excess of hydrochloric acid in the acidified solution is mixed with starch in a separate beaker, and the iodine which is usually thrown out by this mixture, is just bleached by a dilute solution of thiosulphate. To the now neutral solution of iodide and iodate a single drop of the solution to be analyzed is transferred by a glass rod. If a blue coloration is developed, the solution is acidic with hydrochloric acid, and all the boric acid is in free condition. The amount of iodide and iodate used depends upon the acidity of the solution containing boric acid. Usually 10 cm 3 of a 25 per cent solution of iodide and the same amount of a saturated solution of iodate is sufficient. Any larger excess of hydrochloric acid should be neutralized by sodium hydroxide before the iodide and iodate mixture is added. After the addition of the iodide and iodate solution, containing starch, to the boric acid solution, the liberated iodine should be carefully bleached by thiosulphate. Any excess of thiosulphate in reasonable amount does not seem to be detrimental, but in practice the starch iodide color is clearly bleached, and no more then added. Carbonates pre- vent a definite indication of the neutral point by thiosulphate and starch iodide, therefore the barium chloride (about 5 cm 3 ) should be added before this point in the process. The mixture of iodide and iodate is not added to the solution to be analyzed until after it is made acidic, for the reason that when the neutral point is approached by the addition of hydrochloric acid the starch iodide is thrown out locally by the acid, and the small amount of sodium borate remaining undecomposed does not again bleach the coloration produced thus obscuring the neutral point which must be obtained before titrating for boric acid. The solution after the bleaching of the iodine by thiosul- phate is colorless and contains only starch, neutral chloride, potassium tetrathionate, iodide and iodate, and all the boric acid present in uncombined condition. The carbonate lies out of the sphere of action in insoluble form as barium carbonate. A few drops of the indicator, phenolphthalein, are now added, ESTIMATION OF BORIC ACID. 187 and the alkaline solution run in until a strong red coloration is produced. A pinch of mannite is then added, which bleaches the phenolphthalein coloration, and the alkali solution again run in to a faint indication, which if permanent on the addition of more mannite, may be taken as the reading point. About 1-2 grm. of mannite are necessary for a determination. The boro-mannite compound is sufficiently acidic to liberate iodine abundantly, but it appears to be a time reaction, and at the end of six hours only about 95 per cent of the theoretical amount (considering B 2 O 3 as a bivalent acid) has been thrown out. The combination of boric acid and mannite liberates in the presence of iodide and iodate immediately only about one-half the iodine required on the theory that B 2 O 8 under these conditions acts as a bivalent acid, or with the neutralizing power of metaboric acid, HOBO. When no mannite is present phenolphthalein gives an alkaline indication when only about one-half the amount of alkali theoretically necessary to form the metaborate, NaOBO, has been added. Obviously, then, the starch iodide coloration will not appear at all on the addition of mannite, if about one-half the free boric acid is first neutralized by the solution of alkali, and the remainder of the alkali immediately added to complete neutralization. The point at which the danger of the appearance of the iodide coloration on the addition of mannite has been passed, is roughly indicated before the mannite has been added by the appearance of the strong alkaline indication of phenolphthalein. This indicator would not need to be added at all, if the boromannite compound quickly and regularly liberated iodine from the iodide and iodate. The fact, however, that this compound of boric acid and mannite as has been ascertained by experiment liberates, on standing twelve hours, about 99 per cent of the theoretical amount of iodine, places the strength of this acid above that of citric or tartaric acid as investigated by Furry.* With phenolphthalein, however, the end reaction is sharp and the small amount of carbonate present in the standard solution of alkali is precipitated by the barium * Am. Chem. Jour., vi, 341. 188 A VOLUMETRIC METHOD FOR THE chloride already in the solution. The calculation must therefore be based on the amount of free hydroxide in the standard solution of alkali used, according to the following representation : B 2 8 + 2NaOH = 2NaOBO + H 2 0. The best results and the most definite indications are obtained in cold solution of a volume not greater than 50 cm 3 . This fact accords with the observations of Magnanini * that the relative electrical conductivity of the boro-mannite solution is decreased by dilution and elevation of the temperature. When silicates are present in solution, the silicondioxide is liberated by the excess of hydrochloric acid, and this oxide, whether in hydrous or anhydrous condition, neither affects the indication with iodine nor phenolphthalein, nor does it form with mannite a compound of acidic proper- ties. Ammonium salts interfere with the indication given by phenolphthalein and may be removed by boiling with potassium hydroxide in excess, or an indicator used not affected by them. To test the action of fluorides in the process, several experiments were made in which hydrofluoric acid (10 cm 3 of T^ solution) was introduced into the solution containing salts of sodium, free hydrochloric and boric acids. Barium chloride was then added and the analysis for boric acid completed in the usual way without the accuracy of the results being in any way interfered with by the presence of hydrofluoric acid. Table I contains the results of a series of analyses in which the boric acid was first drawn into a excess of sodium hy- droxide, then estimated according to the method described. The standard solutions of boric acid used contained, in A, 7.153 grm., and in B, 7.706 grm. per liter. The solution of free sodium hydroxide was 0.21427 normal. Practical tests of the method upon specimens of crude calcium borate and colemanite are recorded in Table II and Table III. * Gaz. chim. ital., xx, 428, and xxi, 134. ESTIMATION OF BORIC ACID. TABLE I. 189 Exp. B,0 8 SoL taken. NaOH SoL required. SSL B,0, found. Errors on B 2 S . A f(l) j 2) 1(3) ClUg 21.95 20.68 20.73 cm s . 21.02 19.65 19.63 grin* 0.1671 0.1479 0.1483 grin. 0.1577 0.1474 0.1473 grm. 0.0006+ 0.0005- 0.0010- B - (4) (5) (6) (7) (8) ,(9) 23.05 23.10 22.76 24.08 22.00 20.78 23.71 23.80 23.35 24.78 22.60 21.28 0.1776 0.1780 0.1754 0.1855 0.1695 ' 0.1601 0.1777 0.1783 0.1750 0.1857 0.1686 0.1695 0.0001+ 0.0003+ 0.0004- 0.0002+ 0.0009- 0.0006- TABLE n. ANALYSIS OP CRUDE BORATE OF LIME. Bxp. Ca borate taken. B 2 O 8 found. B 2 0, *% grm. grm. (1) 0.4016 0.2289 56.99 (2) 0.4044 0.2302 56.92 (3) 0.4000 0.2285 57.11 TABLE m. ANALYSIS OP COLEMANITE. Bxp. Mineral taken. B 2 O 8 found. %B,0 8 . Average. grm. grm. (1) 0.4034 0.2064 61.15] (2) 0.4070 0.2069 60.80 (3 0.6004 0.3054 50.86 ff\ QQOf (4 0.6006 0.3056 50.89 ou.yy% (5 0.5059 0.2592 61.24 6 0.5092 0.2592 60.89 J The finely-ground minerals were dissolved in hydrochloric acid and the analyses proceeded with as described above. An analysis for boric acid by this process can be completed in five minutes and the results are obviously accurate within the limits of ordinary analysis. The usually interfering substances, fluorine, silica, and carbon dioxide, have no detrimental influence on the results of this process. XXIV THE CONSTITUTION OF THE AMMONIUM MAGNESIUM PHOSPHATE OF ANALYSIS. BY F. A. GOOCH AND MARTHA AUSTIN.* IN a recent paper from this laboratory f it has been shown that the presence of ammonium chloride or other ammonium salt is necessary in the precipitation of manganese as the ammonium manganese phosphate by microcosmic salt in order that the precipitate may have the ideal constitution represented by the symbol NH 4 MnPO 4 . It was also shown that the solvent effect of the ammonium chloride upon the precipitated ammonium manganese phosphate is not marked when an excess of the precipitant is present in solution. The relations disclosed in this paper suggest that the chemical constitution of the precipitate rather than mechanical contamination and varying solubility the explanations gene- rally accepted, and, indeed, advocated by one of us in a former paper J may be responsible for observed variations in the weight of the residue derived by the ignition of the similar salt of magnesium, the ammonium magnesium phosphate, precipitated by an excess of a soluble phosphate from the solution of a magnesium salt, or from the solution of a soluble phosphate by an excess of a magnesium salt. Precipitation by Excess of the Soluble Phosphate. The precipitation of the magnesium salt by an excess of the soluble phosphate was first studied. For this work a solution of pure magnesium nitrate was prepared by dissolving the * From Am. Jour. Sci., vii, 187. t Am. Jour. Sci., vi, 233. This volume, p. 121. J Am. Chem. Jour., i, 391. AMMONIUM MAGNESIUM PHOSPHATE. 191 pure magnesium oxide of commerce in a slight excess of pure hydrochloric acid and boiling with more magnesium oxide. After filtering off the excess of magnesium oxide and any trace of iron or members of the higher groups, the solution was precipitated by ammonium carbonate, the precipitate was washed by repeated boilings and nitrations until silver nitrate gave no precipitate in the solution acidified with nitric acid. This precipitated carbonate was nearly dissolved in nitric acid and the solution was boiled with an excess of the carbonate (for the purpose of removing traces of barium, strontium, and calcium) filtered, and diluted to definite volume. The evaporation of a definite volume of the solution and strong ignition of the residue would be a most natural method of establishing a standard of the solution, were it not for the fact, pointed out by Richards and Rogers,* that the oxide of magnesium retains on ignition occluded nitrogen and oxygen enough to increase its weight sensibly. For this reason the nitrate was converted to the sulphate and weighed as such either by evaporating to dryness in a weighed platinum crucible a definite volume of the solution, igniting as oxide, and changing to the sulphate by heating with sulphuric acid ; or, by evaporating the magnesium nitrate directly with an excess of sulphuric acid of half strength. In this treatment the excess of acid was removed by heating the platinum crucible upon a porcelain ring or triangle so placed within a porcelain crucible that the bottom and walls of the inner crucible were distant about one centimeter from the bottom TABLE I. MgS0 4 obtained by converting ignited MgO into the sulphate. MgSO 4 obtained directly from 50 cm 3 Mg(N0 3 ) s . Theoretical amount of MgO in MgS0 4 . grm. 0.5748 0.6739 grm. 0.5741 0.5750 grin. 0.1924 0.1923 0.1922 0.1925 * Amer. Chem. Jour., xvi, 567. 192 CONSTITUTION OF THE AMMONIUM and walls of the outer crucible. The excess of acid is easily removed in this way, and the outer crucible may be heated to redness without danger of breaking up the magnesium sulphate. The results of this work, taking O = 16, Mg = 24.3, N = 14.03, S = 32.06, are given in the accompanying table. The magnesium oxide obtained by direct ignition of the nitrate weighed on the average about 0.0010 grm. more than the oxide theoretically present in the weighed sulphate from equal portions of the solution. Before proceeding to study possible chemical effects of ammonium chloride in determining the constitution of the ammonium magnesium phosphate, it is obviously necessary to define the extent to which the ammonium salt may exert a solvent action in presence of the precipitant. Fresenius estimated that ammonium magnesium phosphate is soluble in 15293 parts of cold water, but the method of investigation employed did not entirely preclude the possibility of counting as ammonium magnesium phosphate soluble material included and held in the original precipitate.* According to Kissel f the phosphate, which dissolves in a mixture of ammonia and water in the proportion of 0.0040 grams to the liter and in the proportion of 0.0110 grams to the liter in a similar mixture containing also 18 grm. of ammonium chloride, is practically insoluble in the latter mixture if an excess of magnesia mixture be added; and HeintzJ showed that the effect of adding an excess of sodium phosphate in the solution is similar. So far as appears, no quantitative experiments have been recorded hi which the behavior of a mixture of ammonium chloride and magnesium salt and an insoluble phosphate in a solution only slightly ammoniacal has been tested, though as a matter of convenience the use of faintly ammoniacal solutions and faintly ammoniacal washwater is to be preferred to the mixture of strong ammonia and water [1 : 3] ordinarily employed. As a preliminary step, therefore, in the work to be * Fresenius, 6te Aufl., p. 805. t Zeitschr. anal. Chem., viii, 173. J Zeitschr. anal. Chem., ix, 16. MAGNESIUM PHOSPHATE OF ANALYSIS. 193 described, experiments were made to find how small an amount of magnesium could be detected in solution by precipitating with microcosmic salt, either alone or in presence of ammonium chloride in faintly ammoniacal solutions. The ammonium chloride used for these tests (as well as in the similar quanti- tative work following) was purified by boiling with a faint excess of ammonia, filtering, digesting twelve hours with microcosmic salt, and filtering again. The results are given in Table II. TABLE II. Weight of MgO taken as the nitrate. H(NH 4 )NaP0 4 . 4H,0 taken. Volume. NH 4 C1 taken. Opalescent precipitation. grm. grm. cm 8 grm. ( 0.0003 1.75 100 Marked. 1 0.0003 1.75 500 Marked. (0.0003 1.75 100 10 Marked. } 0.0003 1.75 500 10 Marked. ( 0.0003 1.75 500 30 Faint. 0.0001 1.75 100 . Marked. ( 0.0001 1.75 100 10 Marked. } 0.0001 1.75 500 10 Faint. ( 0.0001 1.75 500 60 Faint. The results of these tests show that even so little as 0.0001 grm. of magnesium oxide may be detected in five hundred cubic centimeters of faintly ammoniacal water containing as much as sixty grams of ammonium chloride, f It is plain that strongly ammoniacal liquids are entirely unnecessary in the precipitation of the ammonium magnesium phosphate under the conditions. In nearly all the experiments to be detailed use was made, therefore, of faintly ammoniacal solutions and wash-water. In Table III are given the results obtained in a study of the effects of varying proportions of ammonium chloride and the soluble phosphate upon the constitution of the precipitate. All precipitates were gathered upon asbestos in the filtering t It was found also, incidentally, that the presence of reasonable amounts of ammonium oxalate (100 cm 8 of the saturated solution) does not interfere with the precipitation of the ammonium magnesium phosphate by microcosmic salt. VOL. XI. 13 194 CONSTITUTION OF THE AMMONIUM crucible, washed in faintly ammoniacal water, and ignited as usual. In every case the precipitation was practically com- plete ; for, upon allowing the filtrates with the wash-water to stand for several days after further addition of microcosmic salt, nothing but insignificant traces of a precipitate not exceeding 0.0001 grm. ever appeared. In the experiments of section A precipitations were made in the cold by the action of microcosmic salt in considerable excess upon the solutions of magnesium nitrate containing varying amounts of ammonium chloride. In experiments (1) to (5) the liquid was made faintly ammoniacal after the addition of the precipitant and the precipitate was filtered off immediately after complete subsidence; in experiments (6) to (10) the precipitate first TABLE in. Exp. M g2 P 2 7 corresponding to Mg(N0 3 ) a Mg 2 P 2 7 found. Error in terms of Mg 2 P 2 7 . Error in terms of MgO. NH 4 C1 present. HNaNH 4 P0 4 . 4H 2 O used. Volume. A. grm. grm. grin. grm. grm. grm. cm 3 . (1) 0.5311 0.5418 0.0107+ 0.0038+ 2.5 150 2) 0.5311 0.5462 0.0151+ 0.0057+ 2 2.5 150 8 0.5311 0.5408 0.0097+ 0.0035+ 2 2.5 150 (4) 0.5311 0.5500 0.0189+ 0.0068+ 60 2.5 250 (5) 0.5311 0.5520 0.0209+ 0.0075+ 60 2.5 250 (6) 0.5311 0.5345 0.0034+ 0.0012+ 2.5 150 (7 0.5311 0.5371 0.0060 f 0.0022+ 2.5 150 (8) 0.5311 0.5384 0.0073+ 0.0026+ 2.5 150 (9) 0.5311 0.5386 0.0075+ 0.0027+ 2.5 150 (10) 0.5311 0.5415 0.0104+ 0.0037+ 2.5 150 B. (11) 0.5311 0.5312 0.0001+ 0.0000 # 2.5 150,100 (12) 0.5311 0.5311 0.0000 0.0000 # 2.5 150,100 (13) 0.5311 0.5346 0.0035+ 0.0013+ 2 + 2 2.5 150,100 (14) 0.5311 0.5348 0.0037+ 0.0014+ 2+ 2 2.5 150,100 (15) 0.5311 0.5383 0.0072+ 0.0026+ 5+ 6 2.5 150,100 (16) 0.5311 0.5368 0.0057+ 0.0021+ 5+ 5 2.5 150,100 (17) 0.5311 0.5376 0.0065+ 0.0023+ 10 + 10 2.5 200,100 (18) 0.5311 0.5395 0.0084+ 0.0030+ 10 + 10 2.5 200,100 (19) 0.5311 0.5396 0.0085+ 0.0031+ 60+ 5 2.5 250,100 (20) 0.5311 0.5389 0.0078+ 0.0028+ 60+ 6 2.6 250,100 * Probably less than 1 grm. MAGNESIUM PHOSPHATE OF ANALYSIS. 195 thrown down was redissolved in a very little hydrochloric acid and reprecipitated by dilute ammonia (the operation being repeated several times) with a view to improving the crystalline condition of the precipitate, and this treatment introduced, of course, a small amount of ammonium chloride, probably less than a gram. It will be observed that errors of excess appear in all of these determinations, those being the greatest in the experiments in which the largest amounts of the ammonium salt were present. In the experiments of section B the manipulation was so changed that the supernatant liquid was poured off (through the filtering crucible which was to be used subsequently to collect the phosphate) after the precipitate had subsided and the insoluble phosphate was dissolved in hydrochloric acid and brought down again, after dilution, by the addition of a faint excess of dilute ammonia. By thus removing the supernatant liquid after the first precipitation, the excess of the precipitant and the amounts of ammonium chloride orginally present were reduced to relatively low limits, so that their effects in the reprecipitation were at a minimum, and by adding varying amounts of ammonium chloride, or none at all, before the reprecipitation, it became possible to demonstrate the individual effect of that reagent apart from that of an excess of the microcosmic salt. It will be noted that in experiments (11) and (12), in which no ammonium salt was added after the decantation from the first precipitate, the results are ideal, and that the errors of excess advance as the amounts of ammonium salt present in the final precipitation increase. The quantity of the ammonium salt present during the first precipitation does not influence the error in the final precipitation unless it is so large that a simple decantation of the supernatant liquid would naturally leave an appreciable amount of it to act when the second precipitation takes place. It is plain that the errors of excess which appear when either the ammonium chloride or the soluble phosphate is present in considerable amount, must be due either to mechan- ical inclusion on the part of the highly crystalline precipitate, 196 CONSTITUTION OF THE AMMONIUM or to variation in the ammonium magnesium phosphate from the ideal constitution toward a condition represented by a phosphate richer in ammonia and correspondingly deficient in magnesium. If any appreciable amount of the ammonium chloride present were held by the precipitate, it would natu- rally be represented by magnesium chloride after ignition, but, in no one of these experiments, even in those dealing with sixty grams of ammonium chloride, did the residue, after dis- solving in nitric acid, give with silver nitrate evidence of the presence of more than a mere unweighable trace of chloride. A special experiment, moreover, in which an attempt was made to determine the silver chloride precipitated from the solution hi nitric acid of an unignited precipitate thrown down by microcosmic salt in presence of sixty grams of ammonium chloride, confirms this conclusion : 'the precipitate was unweigh- able. If ammonium chloride present in the solution to so great an amount is not included in the precipitate in signifi- cant quantity, it would seem to be unnatural that the micro- cosmic salt should be included mechanically in any very great amount. But unless the microcosmic salt was mechanically included, the increase in weight must be due to the chemical influence of the reagents that is, to the production of a phosphate rich in ammonium and deficient in magnesium. Berzelius* recognized the existence of such a phosphate of magnesium ; but Wachf in following the work of Berzelius, failed to find it. It would be natural to expect its formation, if ever, when the precipitating phosphate is in excess and ammonium salts are present in abundance, with free ammonia. Obviously the natural effects of all these reagents would be toward the production of a salt holding more ammonia and more phosphoric pentoxide for a given amount of magnesium. The results of the table seem to point strongly to such ten- dencies, and, by inference, toward the existence of such a compound. Thus in experiments (11) and (12), in which the * Berzelius, Jahresbericht, 3. Jahrgang (1824), iibersetzt von C. G. Gmelin, 8.92. t Schweigger, 1830, Band 29, s. 265. MAGNESIUM PHOSPHATE OF ANALYSIS. 197 greater part of this excess of microcosmic salt was removed by decantation before the second precipitation, while no am- monium chloride was present excepting the small amount made by the solution and reprecipitation of the first precipi- tate, the error is practically nothing. In experiments (13) and (14), (15) and (16), (17) and (18), all similar to (11) and (12) excepting that ammonium chloride was present, the aver- age errors (+0.0036 grm. in terms of magnesium phosphate, +0.0064 grm., +0.0074, respectively) increase as the ammonium chloride is increased in the final precipitation. In experiments (19) and (20), in which the ammonium chloride amounted to sixty grams in the first precipitation and to five grams in the second in addition to the amount that would naturally remain after decanting the strong solution of the former precipita- tion, the similarity of this error (+0.0082 in the mean) to that of the experiments in which smaller amounts of the ammonium chloride were used throughout goes to show that only the amount of ammonium salt present in the final pre- cipitation counts. Further, a comparison of corresponding experiments of A and B shows very plainly that the treatment which involves the removal of the large part of the micro- cosmic salt, the solution of the precipitate, and reprecipitation tends to reduce the higher indications. Thus, for example the error in (2) and (3) is +0.0124 gram in terms of magnesium pyrophosphate, while in (13) and (14), similarly carried out except the decantation of the excess of the precipitant, solu- tion and reprecipitation, the error is +0.0036 grm. The special influence of free ammonia during precipitation, was investigated in the following experiments. Definite volumes of the magnesium nitrate solution were drawn from a burette into a platinum dish, ammonium chloride 10 grm. was added, the magnesium was brought down by dilute ammonia in presence of microcosmic salt, and strong ammonia equal to one-third the volume of the solution was added. The solutions, after standing, were filtered off on asbestos under pressure in a perforated crucible, and the precipitates were washed with ammonia diluted to the proportion of three parts 198 CONSTITUTION OF THE AMMONIUM of water to one of ammonia, dried after moistening with a drop of saturated solution of ammonium nitrate, ignited and weighed. The results are given in experiments (1) and (2) of Table IV. In these determinations the mean error reaches +0.0193 gnn. in terms of magnesium pyrophosphate ; while in experiments (3) and (4), made similarly excepting that the supernatant liquid was decanted from the precipitate first thrown down, the precipitate dissolved in hydrochloric acid, and after dilution reprecipitated by dilute ammonia imme- diately supplemented by enough strong ammonia to make one- fourth the volume of the entire solution, the error amounts in the mean to +0.0061 in terms of the pyrophosphate. TABLE IV. Exp. Mg 2 P 2 7 corresponding to Mg(NO 3 ) a Mg 2 P 2 7 found. Error in terms of M g2 P 2 7 . Error in terms of MgO. NH 4 C1 present. HNaNH 4 P0 4 . 4H 2 used. Volume. (1) gnn. 0.5311 grm. 0.5503 grm. 0.0192+ gnn. 0.0069+ grm. 10 grm. 2.5 cms 200 (2) 0.5311 0.5505 0.0194+ 0.0070+ 10 2.5 200 (3) 0.6311 0.5393 0.0082+ 0.0029+ 10, 2.5 200,100 (4) 0.5311 0.5351 0.0040+ 0.0017+ 10,- 2.5 200,100 In experiments (1) and (2) the precipitate was influenced by an excess of microcosmic salt, ammonium chloride, and free ammonia in large amount; in experiments (3) and (4), by decanting in the manner previously described, by dissolving the precipitate, and reprecipitating, the effects of an excess of microcosmic salt and ammonium chloride are reduced to a mini- mum, and, in a comparison of the results with those of experi- ments (11) and (12) of Table III the tendency of the free ammonia comes to view. The results discussed seem certainly to point to a general tendency on the part of free ammonia, ammonium choride and excess of the phosphate to produce a salt rich in ammonia and deficient in magnesium, which for a definite amount of magnesia precipitated must leave upon igni- tion a residue weighing more than the normal phosphate. If it be assumed that a salt of the symbol (NH 4 ) 4 Mg(PO 4 ) 2 MAGNESIUM PHOSPHATE OF ANALYSIS. 199 the next natural step to the normal salt, NH 4 MgPO 4 is present in the precipitate, the residue which such a salt would leave upon ignition would be the metaphosphate Mg(PO 3 ) 2 . From the relations of the symbols for magnesium pyrophosphate and magnesium metaphosphate the weight of the residue obtained, and the weight of the pyrophosphate theoretically derivable from the weight of magnesium salt used, it is possible, of course, to calculate the proportionate amounts of pyrophosphate and metaphosphate present in any ignited residue. Proceeding in this manner, it appears that, in order to account for the variations noted, it is necessary to assume the presence in many cases of very considerable amounts of the metaphosphate. Thus, in the case of those results obtained according to the usually accepted method of precipitating and washing with strongly ammoniacal liquids, viz., in experiments (1) and (2) of Table IV, the proportion of metaphosphate needed to account for the observed error reaches ten per cent. Precipitation ~by Excess of the Magnesium Salt. The relations which obtain in the reverse process of pre- cipitation the action of an excess of the magnesium salt upon a soluble phosphate were studied in experiments to be described. A solution of pure hydrogen disodium phos- phate was prepared by carefully recrystallizing the pure salt of commerce five tunes from distilled water in a platinum dish, dissolving the crystals, and diluting to definite volume. The standard of the solution was established by evaporating to dryness in a weighed platinum crucible known volumes of the solution, igniting the residue and weighing the sodium pyrophosphate. Magnesia mixture, the precipitant, was pre- pared by dissolving fifty-five grams of magnesium chloride in as little water as possible and filtering, mixing with this solution twenty-eight grams of ammonium chloride purified by treating it in strong solution with bromine water and a slight excess of ammonia, filtering, diluting to one liter, and, after standing for some hours, filtering again. 200 CONSTITUTION OF THE AMMONIUM The tests of the following table show that the precipitation of a soluble phosphate by the magnesia mixture is practically complete in faintly ammoniacal solutions even when very dilute and charged with large amounts of ammonium chloride, provided the magnesia mixture is present in sufficiently large excess. TABLE V. P 2 O, in HNa,PO 4 taken. Magnesia mixture. Volume. NHtCl. Precipitation visible. grm. cms cm 8 grm. 0.0005 10 100 t At once throughout 0.0005 50 100 . the liquid. 0.0005 10 100 10 it 0.0005 10 200 60 tt 0.0001 50 250 60 u 0.0001 10 100 0.0001 10 100 10 tl 0.0001 50 200 10 0.0001 10 250 60 0.0001 0.0001 50 50 300 500 60 60 After settling out. This conclusion was further substantiated by an actual test (by the molybdate method) of the ignited residue, obtained by evaporating a filtrate from ammonium magnesium phos- phate (equivalent to 0.8614 grm. of the pyrophosphate) precipitated by a faintly ammoniacal solution of magnesia mixture in presence of 60 grm. of ammonium chloride, which gave a precipitate of ammonium phosphomolybdate yielding 0.0002 grm. of magnesium pyrophosphate. It is evident, therefore, that any considerable deficiencies of weight of the magnesium phosphate obtained by precipitating equal amounts of a soluble phosphate by magnesia mixture in presence of varying amounts of ammonium chloride, cannot be attributed to varying solubility of the magnesium phosphate under changing proportions of the ammonium chloride. The results recorded in section A of Table VI were obtained by treating definite volumes of the pure solution of hydrogen disodium phosphate with magnesia mixture, in slight excess above the amount required to bring down the phosphate, and MAGNESIUM PHOSPHATE OF ANALYSIS. 201 making the solution distinctly ammoniacal. After thorough subsidence, the precipitate was filtered off on asbestos under pressure in a perforated platinum crucible, washed hi water faintly ammoniacal, dried, ignited and weighed. In experi- ments (1), (5) and (6), only the ammonium chloride present in the magnesia mixture was used ; in the other cases weighed portions were added. In the experiments of section B, the precipitate was dissolved in hydrochloric acid after filtering off the supernatant liquid, brought down again in dilute solution by ammonia in distinct excess, and thereafter treated as in the experiments of section A. The experiments of section C were conducted similarly to (1), (5) and (6) of A excepting that the magnesium mixture was introduced into the am- TABLE VL Exp. Mg 2 P 2 7 corre- sponding to taken. found. Error in terms of M gl P,0 7 . Error in terms of P. Volume. NH 4 C1 in mag- nesia mix- ture. NH^l added. MgCl, 6H 2 O in mag- nesia mix- ture. A (1) (2) (3) (4) (5) (6) 7) 8) 9) grm. 0.8615 0.8615 0.8615 0.8615 0.0852 0.0852 0.0852 0.0852 0.0852 gnu. 0.8613 0.8615 0.8602 0.8561 0.0862 0.0866 0.0847 0.0830 0.0811 grm. 0.0002- 0.0000 0.0013- 0.0054- 0.0010+ 0.0014+ 0.0005- 0.0022- 0.0041- grm. 0.00005- 0.00000 0.00036- 0.00151- 0.00028+ 0.00039+ 0.00014- 0.00062- 0.00115- cm 3 150 200 200 300 100 100 200 200 300 griii. 1.68 1.68 1.68 1.68 0.28 0.28 0.28 0.28 0.28 griii. *20' 20 60 *20' 20 60 grm. 3.3 3.3 3.3 3.3 0.55 0.55 0.55 0.55 0.55 B (10) (11) (12) (13) (14) (15) (16) (17) 0.8111 0.8615 0.8615 0.8615 0.0852 0.0852 0.0852 0.0852 0.8114 0.8613 0.8578 0.8487 0.0855 0.0656 0.0853 0.0819 0.0003+ 0.0002- 0.0037- 0.0128- 0.0003+ 0.0004+ 0.0001+ 0.0033- 0.00008+ 0.00006- 0.00103- 0.00358- 0.00008+ 0.00011+ 0.00003+ 0.00092- 150,100 150,000 200,100 200,100 100,100 100,100 150,100 200,100 1.68 1.68 1.68 1.68 0.28 0.28 0.28 0.28 ! ,20 .,60 io] ! 10,10 20,20 3.3 3.3 3.3 3.3 0.55 0.55 0.55 0.55 C (18) (19) 0.8111 0.8111 0.8071 0.8052 0.0040- 0.0059- 0.00112- 0.00165- 120 120 1.4 1.4 2.75 2.75 202 CONSTITUTION OF THE AMMONIUM moniacal solution of the phosphate drop by drop from a burette. The precipitations in A, B, and C were proved to be practically complete ; for by treatment of the filtrates with more magnesia mixture and standing, no more than a trace 0.0001 grm. at the most of the phosphate was found. The ignited residues never contained more than a mere trace of chlorine. While the results are not entirely regular, the tendency of the ammonium salt to produce errors of deficiency in propor- tion to its amount is plain if we compare among themselves the experiments of A upon similar amounts of phosphate, and then those of B upon similar amounts of phosphate among themselves ; and by a comparison of corresponding results in A and B it is clearly shown that the presence of an excess of magnesia mixture tends to counteract more or less completely errors of deficiency due to the action of the ammonium chloride. These facts are quite in harmony with the hypothesis that the ammonium salt tends to produce an ammonium magnesium phosphate richer in ammonia and phosphoric acid and poorer in magnesia than the normal salt NH 4 MgPO 4 ; for, though the production of such a salt in presence of an excess of the soluble phosphate compels the combination of a definite amount of magnesium with more than the normal amounts of phosphoric acid and ammonia (as was the case in the former series of experiments), when the supply of the soluble phosphate is limited the amount of magnesium associated with it must fall below the normal (as is the case in the present series of experiments). Moreover, the behavior of the precipitant is quite in accord with the hypothesis ; for, though the influence of an excess of the soluble phosphate would naturally tend (as was observed) in the same direction as that of the ammonium salt and free ammonia, viz., to the production of the phosphate deficient in magnesium, the tendency of an excess of the magnesium salt must obviously be to increase the amount of magnesium in the phosphate, as was observed in the experiments of Table VI. The hypothesis fits the facts, therefore, on both sides ; and, if precipitation is practically MAGNESIUM PHOSPHATE OF ANALYSIS. 203 complete (as was shown to be the case throughout) the argument for the existence of an ammonium magnesium phosphate poorer than the normal salt in magnesium possibly the salt (NH^Mg^O^a seems to be strong. The Practical Determination of Magnesium and Phosphoric Acid. In determining magnesium by the procedure in ordinary use, the tendency isstrong as is shown in experiments (1) and(2) of Table IV toward high plus errors, and the error is due to the combined effects of excesses of the precipitant, the ammonium salt, and free ammonia. The experiments (11) and (12) of B, Table III, show conclusively that such tendencies to error may be counteracted effectively by pouring off the supernatant liquid (through the filter to be used subsequently to collect the precipitate) as soon as the precipi- tate subsides, dissolving the phosphate in the least amount of hydrochloric acid, bringing it down again, after dilution, by a faint excess of ammonia, filtering (best, we think, on asbestos, under pressure), washing with faintly ammoniacal water, and igniting as usual. Many years ago* a method of precipitating the ammonium magnesium phosphate was advocated by Professor Wolcott Gibbs, which consists, essentially, in boiling the solution of the magnesium salt with microcosmic salt and adding ammonia after cooling, and by which most exact analytical results were obtained. Our experience confirms completely that of Gibbs, and we desire to direct attention again to a procedure the advantage of which has, unfortunately, not been broadly known and accepted. Even in the presence of considerable amounts of ammonium chloride this process yields a phosphate of nearly ideal constitution if only the boiling be prolonged from three to five minutes. The greater part of the ammonium magnesium phosphate about 90 per cent forms in this process before free ammonia is added, and the ammonium which enters the phosphate thus formed is derived from the * Am. Jour. Sci. [3], v, 114. 204 CONSTITUTION OF THE AMMONIUM microcosmic salt, which must become correspondingly acidic. Under these conditions, the tendency to form an insoluble ammonium magnesium phosphate richer in ammonia and poorer in magnesia than the normal salt, does not develop. In the process of Gibbs, as well as in the modified precipitation process in the cold, the use of the faintly ammoniacal solution and wash-water is sufficient and advantageous. In the precipitation of a soluble phosphate by magnesia mixture the tendency of the precipitant and that of the ammonium salt are antagonistic, so that the effect of the latter salt is somewhat masked, though manifest. This opposition of effects has been noted by Mahon,* who, though regarding the actual attainment of an exact balance as uncertain, ventures the opinion that accurate results should be attainable by the careful relative adjustment of the proportions of the precipitant and ammonium salt. Mahon claims to get the best results by a very gradual addition of magnesia mixture to the ammoniacal solution of the phosphate containing about sixteen per cent of ammonium chloride, strong ammonia being added subsequently. From our observations, however, recorded in section C of Table VI, it appears that the method of introducing the magnesia mixture gradually into the ammoniacal phosphate (taken in quantity sufficiently large to give unmistakable indications) produces a precipitate deficient in magnesium and so leads to errors of deficiency in the phosphorus indicated. The use of strong ammonia, moreover, we have shown to be both unnec- essary and disadvantageous. Our experiments go to show that good results may be expected when the solution of the phosphate containing a moderate excess of the magnesium salt and not more than five to ten per cent of ammonium chloride is precipitated by making it slightly ammoniacal, the precipitate being washed in slightly ammoniacal wash-water. In general, however, and especially when more ammonium chloride than this proportion, or more magnesium salt than twice the amount theoretically necessary, is present, it is safer to decant the supernatant liquid from the precipitate (through the filter to * Jour. Am. Chem. Soc., xx, 445. MAGNESIUM PHOSPHATE OF ANALYSIS. 205 be used subsequently to hold the phosphate), to dissolve the precipitate in a little hydrochloric acid, and reprecipitate by dilute ammonia, washing with faintly ammoniacal wash-water. Since our first publication of the work described above, Neubauer* has called attention to the fact that the influence of ammonia and ammonium salts upon the constitution of the ammonium magnesium phosphate obtained in determining phosphoric acid had been previously pointed out by him in a paper | discussing methods for the estimation of that acid. We take pleasure, therefore, in conceding to Neubauer full priority in the observation of the effect which we have endeavored to overcome in the determination of magnesium and of phosphoric acid. * Zeitschr. anorg. Chem., xxii, 162. t Zeitschr. anorg. Chem., ii, 45. XXV THE INFLUENCE OF HYDEOCHLOEIC ACID IN TITEATIONS BY SODIUM THIOSULPHATE WITH SPECIAL EEFEEENCE TO THE ESTIMA- TION OF SELENIOUS ACIDS. BY JOHN T. NORTON JR.* IN the method of Norris and Fay f for the iodometric deter- mination of selenious acid, advantage is taken of a direct and unique action of sodium thiosulphate upon selenium dioxide in the presence of hydrochloric acid. Most excellent results are claimed for this method; but the explicit state- ment of the originators of the method, that the amount of hydrochloric acid present does not influence the result, pro- vided the titration is made at the temperature of melting ice, is so extraordinary in view of generally accepted ideas in regard to the interaction of hydrochloric acid and sodium thiosulphate, as to, suggest the necessity of careful investiga- tion of this point. Pickering, J in his investigation of the reaction between iodine and sodium thiosulphate, has shown that more iodine is required to oxidize the thiosulphate as the proportion of hydrochloric acid increases. He ascribed this effect to the formation of a sulphate, apparently, by the increased activity of the iodine, but the more rational explanation is that, although some sulphate is ultimately formed, the thiosul- phate is first partially decomposed into free sulphur and * From Am. Jour., Sci. vii, 287. t Am. Chem. Jour., vol. xviii, p. 703. t Jour. Chem. Soc., vol. xxxvii, p. 135. HYDROCHLORIC ACID IN TITRATIONS. 207 sulphur dioxide. Finkener* and Mohrf also mention the decomposing effect of free acid upon sodium thiosulphate. The sodium thiosulphate used in the following experiments was taken in nearly decinormal solution and was standardized by running it into an approximately decinormal solution of iodine, the value of which had been determined by comparison with decinormal arsenious acid made from carefully resub- limed arsenious oxide. In the experiments of Table I the solutions were stirred continuously and kept at a temperature of from to 5 C., while the thiosulphate ran into the acidified liquid. The volume of the solution, though fixed at the beginning as given in the table was considerably increased during the operation by the melting of the ice. Titrations TABLE I. Volume of liquid at beginning of titration. Na^Og approx- imately ^ taken. Volume of ^ iodine used in titration. HC1 = none. =: 1 cms. = 5 cm 3 . = 10 cm 3 . cm 3 cm 3 cm cm 3 cm 3 cm 8 100 30 30.251 30.75 30.76 31.20 200 30 30.22 30.21 30.56 31.40 300 30 30.20 mean = on 99 30.22 31.03 30.90 400 30 30.21 OvwDS 30.20 30.20 30.55 500 30 30.20 30.20 30.21 30.55 100 25 25.29' 25.32 25.98 25.70 200 25 25.28 25.34 25.40 25.45 300 25 25.29 mean = fc"*C OT 25.41 25.38 25.83 400 25 25.27 25.27 25.24 25.30 25.63 500 25 25.22 . 25.23 25.40 25.30 100 20 20.15 20.17 20.33 20.23 200 20 20.20 20.13 20.27 20.23 300 20 20.21 mean =: 9O 1 1\ 20.15 20.20 20.17 400 20 20.20 Zv.lO 20.10 20.27 20.07 500 20 20.10 20.10 20.17 20.13 were conducted as rapidly as possible to avoid the separation of sulphur, which is likely to occur, especially when the acid and thiosulphate are present in large quantities. A perusal of the table shows that the influence of the hydrochloric acid upon the thiosulphate depends chiefly upon the amount of * Anal. Chem., 6. Aufl., p. 620. t Titrirmethode, 6. Aufl., p. 279. 208 INFLUENCE OF HYDROCHLORIC ACID IN the thiosulphate present and afterwards upon the degree of dilution and its own absolute quantity. Thus when 30 cm 3 of sodium thiosulphate were employed the effect of 10 cm 3 of acid is marked at all dilutions within the range of the experiments; the effect of 5 cm 3 of acid is inappreciable only at a dilution of from 400 to 500 cm 3 , and when 1 cm 3 of acid is employed the effect is only perceptible at a volume of 100 cm 3 . When 25 cm 3 of the thiosulphate were used the influence of the acid is less marked; for at a dilution of 500 cm 3 the effect of 10 cm 8 of acid is not seen, and 20 cm 3 of the thio- sulphate may be present at any dilution down to 100 cm 3 in the presence of as much as 10 cm 3 of the acid, and even considerably more, as experiments not included in the table indicated. The slight discrepancies which appear occasionally in the table were due, no doubt, to unavoidable differences in the tune of action. This influence of time upon the reaction between sodium thiosulphate, iodine, and hydrochloric acid comes out clearly in the following series of experiments, in which the thiosulphate was run into the acidified water, cooled to a temperature of from to 5 C. by means of ice, the solution being allowed to stand 5, 10, and 15 minutes. Sulphur was thrown down in nearly every case. TABLE IL Volume of the liquid at beginning of titration. HC1 (sp. gr. 1.12) present. Na 2 S,0 8 approxi- mately. taken. Volume of ^ iodine used in titration after standing. 5 minutes. 10 minutes. 15 minutes. cm 8 cm 8 cm 8 cm 8 cm 8 cm 8 200 10 30 30.80 31.30 32.32 200 10 25 25.50 26.00 26.30 200 10 20 20.30 20.70 20.68 The results of the table emphasize sufficiently the necessity of proceeding as rapidly as possible with the titration of sodium thiosulphate by iodine in presence of hydrochloric TITRATIONS BY SODIUM THIOSULPHATE. 209 acid, when the thiosulphate is present in considerable amount. As would be expected, the effect of temperature upon the reaction is also marked. In the following experiments the sodium thiosulphate was run into the acidified water, pre- viously heated to the temperature indicated, and then titrated with iodine. TABLE m. Volume of Volume of ^ liquid at beginning of HCl (sp. gr. 1.12) Temperature Centigrade. approxi- mately. iodine used in titratious at different titration. is temperatures. cm 3 cm C. cm 8 cm 3 400 10 6 25 23.52 400 10 22 25 23.73 400 10 34 25 24.35 400 10 42 25 24.5 400 10 54 25 25 400 10 64 25 26.1 From these results it is plain that the conditions under which considerable amounts of sodium thiosulphate are titrated in presence of hydrochloric acid must be carefully guarded when accuracy is a consideration. It is also apparent that in all cases the temperature should be reduced as nearly to C. as possible and rapidity of titration by the iodine is an essential. So long as the thiosulphate present does not exceed 20 cm 3 of the T ^ solution, rapid titration in cold solution proceeds with fair regularity in presence of hydro- chloric acid up to 10 cm 3 of the acid of sp. gr. 1.12. When, however, the amount of thiosulphate is greater than 20cm 3 of the -fy solution, care as to the restriction of the acid and dilution of the solution becomes a necessity. Fortunately, in most analytical processes involving the use of the thio- sulphate it is possible to add that reagent from the burette to the solution to be acted upon, so that it is destroyed normally as fast as it is introduced and the danger of inter- action with the acid does not occur. In the process of Norris and Fay, however, the method involves the addition of an excess of the thiosulphate to the solution of selenious and VOL. II. 14 210 INFLUENCE OF HYDROCHLORIC ACID IN hydrochloric acids, and thus the conditions prevail which demand care as to the relation of the acid, the thiosulphate and the degree of dilution. I have experimented, therefore, with this process under varying conditions. The process of Norris and Fay for the iodometric determi- nation of selenious acid consists briefly in treating the solution of that acid in ice water, in the presence of hydrochloric acid, with an excess of a T * ff solution of sodium thiosulphate and titrating back the excess of the thiosulphate with iodine. Four molecules of sodium thiosulphate act, apparently, upon one molecule of selenious acid according to a reaction which the authors propose to study. The selenium dioxide used was made by dissolving presum- ably pure selenium in nitric acid and evaporating to dryness. The residue was then treated with water, and a little barium hydroxide was added to remove any sulphate which might be present. The solution was then filtered and the filtrate evaporated to dryness. The residue was mixed with four or five times its volume of dried pulverized pyrolusite, and the whole was put into a porcelain crucible and heated. The sublimate of selenium dioxide was carefully collected on a dry watch-glass and put into a drying bottle as quickly as possible. The pyrolusite prevents any reduction of the selenium dioxide to selenium and the product consisted of beautiful long white needles. This method of preparing the selenium dioxide, which has been used for some time in this laboratory, avoids contamination of the selenium dioxide by nitric acid or water, resulting from the decomposition of the latter, which would be possible in case this reagent were employed in the final sublimation, as is recommended by Norris and Fay. The hydrochloric acid used was of a sp. gr. 1.12, as recommended by the originators of the process. For the experiments of Table IV the dilution at the beginning was fixed at 400 cm 3 , and this was increased in every case by the melting of the ice used to cool the liquid. A glance at the preceding part of this paper will show that at this degree of dilution the hydrochloric acid present has the least effect. TITRATIONS BY SODIUM THIOSULPHATE. 211 TABLE IV. Exp. Amount SeO, taken. HCl(sp. gr. 1.12) taken. Volume at begin- ning of titration. Excess Na,S 2 3 employed. SeO, found. Error. grin. cm 3 cm 8 cm 3 grm. grm. (1) 0.0616 10 400 2.28 0.0625 0.0009+ (2) 0.0628 10 400 7.11 0.0631 0.0003+ (3) 0.0508 10 400 11.4 0.0511 0.0003+ (4) 0.0587 10 400 12.8 0.0594 0.0007+ mean. (5) (6) 0.0807 0.0633 10 10 400 400 15.3 20.85 0.0813 0.0638 0.0006+ 0.00005+ 0.0005+ (7) 0.0682 25 400 1.11 0.0685 0.0003+ (8) 0.0779 25 400 1.35 0.0788 0.0009+ (9) 0.0465 25 400 18.93 0.0469 0.0004+ These results, while not so good as those obtained by Norris and Fay, are satisfactory and show that at this degree of dilution the process is accurate. These results accord closely with those contained in Table I. At a dilution of 400 cm 3 or in the presence of only 20 cm 3 of sodium thiosulphate in excess the hydrochloric acid present had no perceptible effect. Of course, it must be kept in mind that the hydrochloric acid acts only upon the excess of thiosulphate which is not taken up by the selenium dioxide. The slight constant plus error in these results cannot be accounted for by errors in the standards; they were all carefully determined. Another preparation of selenium dioxide was made, starting with pure selenium carefully precipitated by sulphurous acid, before putting it through the course of treatment previously described, and the results obtained by the action of the sodium thiosulphate recorded in Table V agree closely with those of the preceding table. TABLE V. Exp. Amount SeO, taken. HC1 (sp. gr. 1.12.) H ? Oat beginning. NajSjOg in excess. SeO, found. Error. grm. cm* cm 8 cm 8 grm. grm. (1) 0.0562 10 400 9.52 0.0566 0.0004+ (2) 0.0651 25 400 11.20 0.0655 0.0004+ The next step was to determine the effect of diminishing the 212 INFLUENCE OF HYDROCHLORIC ACID IN dilution and of varying the strength of acid. The following table gives the results of my experiments. TABLE VI. Exp. Amount of SeO 2 taken. Volume of H 2 O at begin- ning. HC1 (sp. gr. i:i2). Excess of NajSjOj. SeO, taken. Error. grin. cm 3 cm 3 cm* grm. grm. (1) 0.1042 200 5 24.16 0.1041 0.0001- (2) 0.0611 200 10 13.3 0.0611 0.0000 (3) 0.0850 200 10 21.9 0.0828 0.0022- (4) 0.0757 200 25 13.07 0.0749 0.0008- (5) 0.0540 200 25 21.02 0.0522 0.0018- (6) 0.0674 300 5 10.04 0.0679 0.0005-1- (7) 0.2416 400 5 15.9 0.2424 0.0008+ It is apparent that at the dilution of 200 cm 3 we run into difficulties, and the greater the excess of thiosulphate present the greater is the error. When the amount of sodium thiosulphate exceeds 20 cm 3 a reduction in the amount of acid to 5 cm 3 is plainly of advantage, as is shown in a comparison of Exps. (1), (3), and (5), and is not disadvantageous at larger dilutions and with smaller amounts of the thiosulphate, as shown in Exps. (6) and (7). The necessity of placing some limits on the method of Norris and Fay has now, I think, been established. The excess of the thiosulphate must be carefully regulated, as well as the temperature. If one has knowledge of the approximate amount of selenious acid in solution, this is not a matter of great difficulty, and things should be so arranged that no more than 20 cm 3 of the ^ thiosulphate should ever be present in excess of the amount necessary to reduce the selenious acid. If this limit amounting to 0.0400 cm 8 of SeO 2 is placed upon the thiosulphate, so much as 10 cm 3 of hydrochloric acid (sp. gr. 1.12) may be present without endangering the accuracy of the process, provided the solution is diluted to 400 cm 3 at the outset; if only 5 cm 8 of hydrochloric acid are present, the volume at the beginning may be reduced with safety to 200 cm 3 . At all events, 5 cm 3 of the hydrochloric acid are TITRATIONS BY SODIUM THIOSULPHATE. 213 amply sufficient to bring about the reaction between the thiosulphate and the selenium at any dilution within the range of my experiments. With these precautions taken, the process of Norris and Fay is simple, rapid, and accurate ; without them, as the experimental results indicate, errors of considerable amount may enter. According to the method of Muthmann and Shafer,* the determination of selenious acid is effected by the simple addition of potassium iodide to the acidulated solution of selenious acid, and the iodine set free is titrated with sodium thiosulphate. In this procedure the thiosulphate is taken up by the iodine as it is added to the solution, so that the danger of any action between the thiosulphate and the acid is out of the question. It was shown in a former paper from this laboratory f that this simple method is inaccurate on account of the incompleteness of reduction in the cold and in presence of the iodine evolved. In a later article also from this laboratory J it was shown that selenium may be completely precipitated and determined with accuracy gravimetrically provided the amount of potassium iodide employed is enor- mously in excess of that theoretically required. This suggests naturally the trial of very large excesses of potassium iodide in the procees of Muthmann and Shafer. The details of experiments made in this manner are given in the following table. TABLE VII Vol. of HCl Exp. sa KI. solu- tion. (sp. gr. 1.12). found. Error. grm. grlii. cm 3 cm 3 grui. grm. (1) 0.0553 10 150 10 0.0558 0.0005+ (2) 0.0574 5 150 10 0.0567 0.0007- (3) 0.0683 5 150 10 0.0683 0.0000 (4) 0.0487 5 150 10 0.0484 0.0003- (5) 0.2617 10 150 10 0.2589 0.0028- * Ber. Dtsch. chem. Ges., xxyi, 1008. t Gooch and Reynolds, Am. Jour. Sci., 1, 254. Volume I, p. 310. t Peirce, Am. Jour. Sci., i, 1896, 416. Volume I, p. 365. 214 HYDROCHLORIC ACID IN TITRATIONS. It is obvious that for small quantities of selenium dioxide the accuracy of the process is very much increased by the use of large amounts of iodide, though, of course, the difficulty in reading the end reaction due to the presence of precipitated red selenium still remains ; but the process is still inaccurate when large amounts of selenium dioxide are employed. XXVI THE VOLATILIZATION OF THE IRON CHLOR- IDES IN ANALYSIS AND THE SEPARATION OF THE OXIDES OF IRON AND ALUMINUM. BY F. A. GOOCH AND FRANKE STUART HAVENS * IT is well known that metallic iron is easily acted upon by an excess of chlorine at moderately elevated temperatures with the formation of ferric chloride, and that the product of the action of hydrochloric acid gas upon the metal is ferrous chloride. Out of contact with air, or moisture, both chlorides may be volatilized at appropriate temperatures the ferric chloride below 200 C. ; the ferrous chloride at a bright red heat. If water vapor, or oxygen, or air be present during the heating, both chlorides are partially decomposed with the formation of non-volatile residues, ferric oxide or ferric oxy-chloride. Analytical processes involving the volatilization of iron at temperatures more or less elevated, in an atmosphere of chlo- rine or hydrochloric acid, have been the object of considerable at- tention. Thus, Fresenius,f Drown and Shimer,J and Watts, have heated crude iron in chlorine to remove the metal and leave the non-volatile constitutents ; and Sainte-Claire Deville || has employed hydrochloric acid to volatilize iron from mix- tures of that metal with alumina (obtained by heating the mixed oxides of iron and aluminum in hydrogen according to Rivot),^[ exposing the mixture, contained in a porcelain boat and placed within a porcelain tube, to the bright red heat of a charcoal furnace an operation which was bettered by * From Am. Jour. Sci., yii, 370. t Zeitschr. anal. Chem., iv, 72. t Jour. Inst. Min. Eng., viii, 613. Chem. News, xlv, 279. || Ann. China. [3], xxxviii, 23. IT Ann. Chim. [3], xxx, 188. 216 THE VOLATILIZATION OF THE Cooke's* use of a tube of platinum instead of the porcelain tube and a gas blowpipe in place of the charcoal furnace. Sainte-Claire Devillef showed, further, that ferric oxide may be converted to ferric chloride and volatilized at the heat of the charcoal furnace if the current of hydrochloric acid is sufficiently rapid ; but the curious effect was observed that in a sufficiently limited current of the acid no chloride whatever was volatilized, while the amorphous oxide was converted to the highly crystalline oxide of the same composition a phe- nomenon which gave rise to a theory of the natural formation of specular iron in volcanic regions. Quite recently, Mover has made record of an unsuc- cessful attempt (in the course of experimentation upon the volatility of certain chlorides at comparatively low tempera- tures) to convert ferric oxide completely to ferric chloride by the action of gaseous hydrochloric acid at about 200 C. At this temperature the greater part of the iron sublimed, but a residue remained, which, volatilizing neither on long heating at 200 nor upon considerable elevation of the temperature, proved upon examination to be ferrous chloride. In the experiments to be described we have acted with gaseous hydrochloric acid upon ferric oxide made by igniting the nitrate prepared from pure iron deposited electrolytically by high currents passing between electrodes of platinum in a strong solution of ammonio-ferrous sulphate. The oxide, contained in a porcelain boat, was heated within a roomy glass tube over a small combustion furnace. The hydrochloric acid (generated by dropping sulphuric acid into a mixture of strong hydrochloric acid and salt, and dried by calcium chlor- ide) entered one end of the tube and passed out at the other through a water trap. In early experiments a high-reading thermometer was inserted through the stopper in the exit end of the tube so that its bulb was above and immediately adjacent to the boat carrying the oxide. In this way the actual temperatures of the vapors about the boat were fixed * Am. Jour. Sci., xlii, 78. t Compt. rend., Hi, 1264. J Jour. Am. Chem. Soc., xviii, 1029. IRON CHLORIDES IN ANALYSIS. 217 with considerable accuracy ; later, after a little experience in gauging the effect of the burners, it was found that the tem- peratures could be regulated very closely without actually depending upon the thermometer. We found, as did Moyer, that ferric oxide, submitted to the action of dry hydrochloric acid gas, volatilizes partially as ferric chloride at low tempera- tures 180 to 200 C. leaving ultimately a crystalline residue which does not change visibly when heated for an hour or two at 200, or even at 500, in the pure dry acid. According to our experience, this residue is generally slightly reddish or salmon-colored ; but sometimes, especially after a second heating, the boat having been withdrawn from the tube or exposed to the atmosphere (and so to moisture), the residue is white. When it is white it dissolves in water, yields 'the characteristic reaction for a ferrous salt with potas- sium ferricyanide, gives no reaction with potassium sulphocy- anide, and upon treatment in weighed amount with potassium permanganate destroys the amount of that reagent theoretically required for its oxidation upon the supposition that it is fer- rous chloride. The slightly colored residue when treated with water yields a solution showing the reaction of a ferrous salt only, but when treated with hydrochloric acid and then tested shows the presence of a trace of iron hi the ferric condition. Doubtless the coloration of the residue is due to an included trace of ferric oxide or oxychloride, which after exposure of the containing crystals to slight atmospheric action, is more easily reached in the second heating by the gaseous acid. The amount of residue is somewhat variable, but approximates under the conditions of our work to from five to ten per cent of the oxide taken : thus, in one typical experiment 0.1 grm. of ferric oxide left a residue which (withdrawn after cooling) weighed 0.0115 grm. The greater portion of the ferric oxide volatilizes when submitted to the action of the gaseous acid at 200 quickly and abundantly in the form of the greenish vapor of ferric chloride, and if the operation is interrupted at this stage the residue which remains is nearly black, insoluble in water, 218 THE VOLATILIZATION OF THE slightly soluble in cold hydrochloric acid, and readily soluble in hot hydrochloric acid with the formation of ferric chloride. It is probably something analogous to the oxychloride which Rousseau* identifies as the product of the action of water upon ferric chloride at 275 to 300. This dark residue yields to the action of the hydrochloric acid at 180 to 200 only slowly, but ultimately only the residue which is essen- tially ferrous chloride remains ; thereafter little volatilization occurs within the range of temperature of our experimentation 200 to 500. It is obvious that a reduction of iron in the ferric condition to iron in the ferrous condition takes place under the con- ditions of our work, and it in difficult to see how this can occur otherwise than by the direct dissociation of ferric chloride under the low partial pressure conditioned by the brisk current of hydrochloric acid gas. The temperature of formation, 180 to 200, is far below that at which such dissociation is supposed to begin. Thus, Gruenewald and Meyer f found, after cooling, no evidence of the dissociation of ferric chloride which had been heated in the Victor Meyer vapor-density apparatus to 448 in contact or partial mixture with nitrogen; but ten per cent of the residue obtained by heating to 518 was in the ferrous condition. Friedel and Crafts,! however, did see crystals of ferrous chloride at 440 on the walls of a Dumas container filled with the vapor of ferric chloride and nitrogen, the former exerting a partial pressure of 0.75; while ferric chloride volatilized into an atmosphere of chlorine without evidence of dissociation. It seems rather surprising, therefore, to find so large a percen- tage of dissociation as that shown in our experiments at a temperature so low 180 to 200. Curiously, too, we find, on repeating the experiment of heating ferric oxide in gaseous hydrochloric acid, that if the temperature of the oxide is 450 to 500 when the brisk current of acid begins to act, the whole mass of oxide is converted and volatilizes without * Compt. rend., cxvi, 118. t Ber. Dtsch. chem. Ges., xxi, 687. | Compt. rend., cvii, 301. IRON CHLORIDES IN ANALYSIS. 219 residue. It is hardly to be supposed that the degree of dissociation at 450 to 500 can be less than that at 180 to 200, and a test of the sublimate, after cooling, shows that it contains a ferrous salt. Plainly, ferrous chloride (formed by dissociation) has volatilized, and inasmuch as the ferrous chloride constituting the residue formed at 180 to 200 does not volatilize in the hydrochloric acid even at 500 it is plain that the volatility of the former is not determined by the presence of the latter. Apparently, the cause of the com- pleteness of volatilization must be sought in its rapidity ; and this is not an unreasonable hypothesis, if one considers that an action sufficiently rapid to keep above the boat an atmos- phere of ferric chloride and its products of partial dissociation, might naturally provide the very condition which would be effective in counteracting the tendency of the residue to dissociate before it volatilizes. If this hypothesis is correct, it is plain that the introduction of chlorine gas, the active product of dissociation, into the atmosphere of hydrochloric acid ought to bring about the volatilization of the residue of ferrous chloride, formed at 180 to 200, which refuses to volatilize in the acid alone. As a matter of fact, we find by experiment that if a little manganese dioxide is added to the contents of the generator, so that the hydrochloric acid may carry with it a little chlorine, every trace of ferric oxide is volatilized from the boat at 180 to 200; and the residue of ferrous chloride found at 180 to 200 when the hydrochloric acid is used alone is likewise volatilized at the same temperature, when the admixture of chlorine is made. These facts, that ferric oxide is completely volatile hi hydrochloric acid gas applied at once at a temperature of 450 to 500 C., and at 180 to 200 if the acid carries a little chlorine, open the way to many analytical separations of iron from substances not volatile under these conditions. In the experiments of the following table we have applied these methods to the separation of intermixed iron and aluminum oxides. The ferric oxide employed was made, as 220 THE VOLATILIZATION OF THE before, by ignition of the nitrate prepared from iron deposited electrolytically by a strong current passing between platinum electrodes in a solution of ammonio-ferrous sulphate.* The aluminum oxide was made by igniting to a constant weight the carefully washed hydroxide precipitated by ammonia from a pure hydrous chloride thrown down from the solution of a commercially pure chloride by hydrochloric acid.f The hydrochloric acid gas was made by dropping sulphuric acid into strong hydrochloric acid mixed with salt, and a little manganese dioxide was added when the mixture with chlorine was desired. The experimental details are given in the table. Feo0 3 taken. A1 2 3 taken. found. Error. Time. Temperature. Atmosphere. grm. 0.1000 0.2000 0.1020 0.2145 gnu. o!ibi5 0.1006 grm. o!ibi5 0.1008 grm* 0.0000 0.0000 0.0000 0.0002+ hrs. 1 C. 450-600 450-500 450-500 450-500 HC1 HC1 HC1 HC1 0.1000 0.1000 0.1072 0.2045 0.1050 0.2008 o!l032 0.1013 0.1032 0.1023 0.1007 0.1087 o!l032 0.1015 0.1033 0.1019 0.1006 0.1087 0.0000 0.0000 0.0002+ 0.0001+ 0.0004- 0.0001- 0.0000 1 H 1 180-200 180-200 180-200 180-200 450-500 450-500 450-500 HC1 + CL. HC1 + C1 2 . HC1 + C1 2 . HCi + C1 2 . HC1 + CL. HCI + C1 2 . HCI + C1 2 . The residual alumina tested in several experiments by fusion with sodium carbonate, solution in hydrochloric acid, and addition of potassium sulphocyanide gave no indication of the presence of iron. The separation of the iron is obviously complete at 450 to 500 when the mixed oxides are submitted at once to the action of hydrochloric acid gas, or at 180 to 200 when * The use of an anode of commercially pure iron wire naturally facilitates the operation, but in our experience the deposit thus obtained is likely to carry traces of impurity. In an attempt, too, to prepare pure ferric oxide from the oxalate thrown down out of ferrous sulphate with all precautions, the material obtained still held traces of silica, and possibly alumina, amount- ing to 0.0004 grm. in 0.1 grm. of the oxide. t From Am. Jour. Sci., ii, 416. This volume, p. 20. IRON CHLORIDES IN ANALYSIS. 221 chlorine is mixed with the hydrochloric acid. Plainly, the extremely high temperatures employed by Deville are un- necessary if the mixed oxides are submitted at once to the action of hydrochloric acid at 450 to 500 without previous gentle heating in the acid atmosphere. We prefer, however, to use the mixture of chlorine and hydrochloric acid, not only because the temperature of the reaction is lower, but because it needs no regulation, while the danger of error arising from the liability of ferric chloride to dissociate, or from deficiency of oxidation in the oxide treated, or from mechanical loss due to rapid volatilization, is avoided. XXVII THE TITRATION OF OXALIC ACID BY POTAS- SIUM PERMANGANATE IN PRESENCE OF HYDROCHLORIC ACID. BY F. A. GOOCH AND C. A. PETEES.* LOWENTHAL and LENSSEN f were the first to show that the titration of a ferrous salt by potassium permanganate in the presence of hydrochloric acid, according to the process of Margueritte J is vitiated by the evolution of chloride outside the main reaction, and to point out that a remedy for the difficulty is to be found in the titration of the ferrous salt in divided portions, other equal volumes of the ferrous solution being added to the liquid in which the first titration is accomplished until the amount of iron indicated by successive titrations becomes constant. Kessler showed the restraining influence of certain sulphates, of manganous sulphate in particular, upon the irregular and undesirable interaction of the permanganate and hydrochloric acid, and Zimrnermann, || in apparent ignorance of Kessler's forgotten proposal, advocated the introduction of a manganous salt, best the sulphate, into the ferrous salt to be determined, thus accomplishing the purpose of the empirical procedure of Lowenthal and Lenssen. The tendency toward evolution of chlorine in the oxidation of a ferrous salt by permanganate, as compared with the absence of such tendency in the similar oxidation of oxalic acid, in presence of hydrochloric acid, was explained by * From Am. Jour. Sci., vii, 461. t Zeitschr. anal. Chem., i, 329. t Ann. Chim. Phys. [3], xviii, 244. Ann. Phys., cxciv, 48 (1863) ; CXCT, 225 (1863). |i Ann. Chem., ccxiii, 302. TITRATION OF OXALIC ACID, ETC. 223 Zimmermann on the hypothesis that an oxide of iron higher than ferric oxide is formed as an intermediate product, and that this unstable oxide is sufficiently active to break up hydrochloric acid as well as to oxidize more of the ferrous salt. Quite recently, Wagner * finds explanation of the sensitiveness of the hydrochloric acid solution of the ferrous salt in the probable formation of chlor-f errous acid (analogous to chlor-platinic and chlor-auric acids), which suffers oxidation more readily than hydrochloric acid under the action of the permanganate. The protective influence of the manganous salt turns apparently, as Zimmermann suggested, upon the initiation of Guyard's reaction, according to which the per- manganate and manganous salt interact to form a higher oxide of manganese of a constitution approaching the dioxide more or less closely this oxide being capable of oxidizing the ferrous salt, but slow to act upon the hydrochloric acid, or the chlor-ferrous acid of Wagner. According to Volhard,f the reaction of Guyard is favored and hastened by heat and concentration of the solution, while it is delayed by acidity and dilution ; but even in solutions containing very little manganous salt and a considerable quantity of free acid the faint rose color developed by the careful addition of perman- ganate ultimately vanishes until every trace of the manganous salt is precipitated. When a considerable amount of the salt is present interaction follows immediately the introduction of the permanganate. Zimmermann advocates the use of 4 grams of manganous sulphate uniformly in titrations of a ferrous salt by permanganate, a procedure to which Wagner gives acquiescence, though pointing out that a ninth of that amount is all that he finds to be necessary. The excess of the manganous salt can do no harm so long as the higher oxide, the product of interaction of the manganous salt and the permanganate, is immediately reduced by even traces of a ferrous salt, and this appears to be the case at least within the limits proposed by Zimmermann and Wagner. Thus we * Massanalytische Studien, Habilitationsschrift, Leipzig, 1898. t Ann. Chem., cxcviii, 318, 1879. 224 TITRATION OF OXALIC ACID find, as shown in results of the accompanying table, that so much as five grams of the sulphate may be present in 135 cm 3 of the liquid, containing about 5 cm 3 of hydrochloric acid of full strength, without interfering with the regularity of the titration ; and the effect is trivial even when the amount of manganous sulphate reaches ten grams. We find also practical regularity of working when manganous chloride is substituted for the sulphate, and in this respect our results accord with those of Zimmermann and differ from those of Wagner.* Total volume at beginning of titration. HCl (sp.gr. 1.09). FeCl,. KMn0 4 ^. MnSO 4 . 5H 2 O. MnCl 2 .4H,O. cm 3 cm 8 cm 8 cm 8 grams. grams. 135 10 25 21.70 1 135 10 25 21.70 3 ( 135 10 25 21.70 5 135 10 25 21.75 7 135 10 25 21.75 10 . 145 20 25 21.75 , 10 p 175 50 25 21.75 10 135 10 25 21.70 t 1 135 10 25 21.70 m 2 145 20 25 21.70 t 2 155 30 25 21.76 3 165 40 25 21.70 4 In all cases, however, in which the larger amounts of manga- nous salt are present, the end reaction is marked by the advent of a brownish-red precipitate rather than the clear pink of the soluble permanganate, and it is obvious that in case the solutions to be oxidized were not active enough to act with rapidity upon the product of the Guyard reaction, difficulty might follow the failure to adjust the conditions more particularly. It has been stated by Fleischerf and Zimmermann J that hydrochloric acid interferes in no way with the titration of oxalic acid by potassium permanganate. This statement, however, is not in accord with our experience; for we find that in such titrations there is a small though real waste of * Loc. cit, p. 104. t Volumetric Analysis ; trans, by Muir, p. 71. J Loc. cit. BY POTASSIUM PERMANGANATE. 225 permanganate proportionate to the amount of hydrochloric acid present. This fact is brought out clearly in the comparison of experiments of section A in the following table, in which no hydrochloric acid was present, with experiments B in which hydrochloric acid was present. Temperature at beginning, about 80 C. Approximate volume at beginning of titration. H,80 4 1 : 1. HCl (sp. gr. 1.09.) Ammonium oxalate * KMn0 4 . Variation from mean of A taken as standard. A.. 200 200 200 200 200 200 cm 8 5 5 10 10 25 25 cm cm 3 50 50 50 50 50 50 cm 47.50 47.50 47.50 47.50 47.50 47.50 cm 3 0.00 0.00 0.00 0.00 0.00 0.00 ] B. f 150 I 150 1 150 1 150 (500 }500 (500 10 10 10 10 5 10 10 2.5 2.5 5.0 10.0 ib.b 10.0 25 25 25 25 25 25 25 23.80 23.90 23.90 24.00 23.80 24.00 24.10 0.05+ 0.15+ 0.15+ 0.25+ 0.05+ 0.25+ 0.35+ From these results it is evident that, though the error intro- duced by the presence of the hydrochloric acid during the action of the permanganate upon the oxalic acid is small, it is plainly appreciable. The questions arise, therefore, first, as to whether the secondary action of the permanganate upon the hydrochloric acid may be prevented by the presence of a suitable amount of a manganous salt, and, secondly, as to whether in this event the reducing agent the oxalic acid is sufficiently active, like the ferrous salt, to prevent the premature establishment of an end color due to the Guyard reaction. The latter question must naturally be settled before the former can be taken up. In the accompanying table are recorded the effects of varying amounts of manganous salt in VOL. IX. 15 226 TITRATION OF OXALIC ACID presence of different amounts of sulphuric acid in the reaction of permanganate upon oxalic acid. Temperature at beginning, about 80 C. Volume at beginning. HjSO 4 1 : 1. Ammonium oxalate ff TV' MnS0 4 . 5H,O. KMnO 4 . Variation from standard. cm* cm' cm* grm. cm 3 f 130 5 25 23.75 0.00 130 6 25 0.0008 23.75 0.00 130 5 25 0.0032 23.76 0.00 130 5 25 0.0160 23.75 0.00 . 130 5 25 1 23.70 0.05- 130 5 25 2 23.75 0.00 130 5 25 2.5 23.60 0.15- 130 5 25 3.0 23.40 0.25- 130 5 25 4.0 23.60 0.15- 500 5 25 23.80 0.05+ 500 6 25 0.0008 23.80 0.05-f 500 5 25 0.0032 23.80 0.05+ . 500 5 25 1 23.70 0.05- 500 5 25 2 23.40 0.35- 500 5 25 3 23.50 0.25- 500 5 25 4 23.30 0.45- 130 no 25 1 23.80 0.05+ 130 ho 25 2 23.75 0.00 130 jio 25 3 23.65 0.10- 130 1 10 25 4 23.50 0.25- 130 (15 25 2 23.75 0.00 " 130 )l6 25 4 23.70 0.05- 130 1 16 25 5 23.50 0.25- 130 (30 25 2 23.75 0.00 130 ?30 25 4 23.70 0.05- 130 <30 25 5 23.75 0.00 From the results given it is evident that the persistence of the Giiyard reaction is liable to interfere with the end reaction of oxidation of oxalic acid unless an adjustment is made between the quantity of the manganous salt, the amount of acid, and the dilution. In hot solutions of a total volume of 130 cm 3 at the beginning, no more than 2 grms. of the manganous sulphate should accompany 5 to 10 cm 3 of the 1 : 1 sulphuric acid ; when the total volume at the beginning reaches 500 cm 3 no more than a single gram of the salt should be present with 5 cm 8 of the 1 : 1 sulphuric acid. The amount of manganous salt may, however, be increased considerably if the quantity of acid is increased. OF THE UNIVERSITY BY POTASSIUM PERMANGANATE. 227 As Kessier has noted, a sufficiency of the manganous salt, acting no doubt as the medium of transfer of oxygen, may bring about interaction between the permanganate and the oxalic acid at atmospheric temperatures without the tedious delay ordinarily encountered in the attempt to consummate that action in cold solutions. It would seem natural that the manganic hydroxide formed in the Guyard reaction at low temperatures should yield more readily to the reducing action of the oxalic acid than the more anhydrous form to be expected in hot solutions, so that at such temperatures the limits as to proportions of manganous salt, acid, and dilution, within which favorable action may take place, should be wider; moreover, the undesirable action of the permanganate upon Temperature 20-26 C. Number of experi- BMOfc* Volume at begin- ning of titration. W HCl Ammo- nium oxalate KMn0 4 . MnSO 4 . 5H 2 0. MnCl 2 . 4HjO. Variation from standard. cm 3 cm' cm cm grin. grin. cm (ij 130 130 10 10 25 25 23.90 23.90 0.0040 0.0120 0.15+ 0.15+ (3) 130 10 25 23.80 0.0250 0.05+ (4) 130 10 25 23.75 0.0400 0.00 (5) 130 10 25 23.76 0.0500 0.01+ (6) 130 10 25 23.70 0.1000 0.05- (7) 130 10 25 23.75 0.2000 0.00 (8) 130 10 25 24.20 0.0200 0.45+ (9) (10) 130 130 10 10 25 25 23.95 23.80 0.0200 0.0400 0.20+ 0.05+ (11) 130 20 25 23.75 0.0400 0.00 (12) 130 30 25 23.75 0.0400 0.00 (13) 130 10 25 23.75 1.0000 t 0.00 (14) 130 10 25 23.75 2.0000 > 0.00 (15) 130 10 25 23.75 3.0000 0.00 (16) 130 1 25 23.72 1.0000 0.03- (17) 130 1 25 23.74 2.0000 0.01- (18) 130 1 25 23.72 3.0000 0.03- (19) 130 2 25 23.70 0.5000 0.05- (20) 130 3 25 23.76 0.5000 0.00 Temperature about 80. (21) 145 10 10 25 23.90 0.5000 0.15+ (22) 145 10 10 25 23.70 1.0000 0.05- (23) 500 10 10 25 23.75 1.0000 0.00 (24) 500 m 10 25 28.70 1.0000 0.05- (25) 500 10 25 24.10 0.5000 0.35+ 228 TITRATION OF OXALIC ACID hydrochloric acid, when that acid is present, should be less appreciable at lower temperatures. In our experiments, therefore, upon the oxidation of oxalic acid by potassium permanganate in presence of hydrochloric acid, we have studied the effect of varying the proportions of the manganous salt both at atmospheric temperatures and the higher temperatures generally employed. From these results it appears that the presence of a suit- able amount of manganous salt either the sulphate (4-7), (13-15), (22-24) or the chloride (10-12), (16-20) is capable, either in cold solution (1-20) or in hot solution (22-24) of preventing the action of the permanganate upon the hydrochloric acid. It appears, also, that, for a given dilution and strength of acid, less manganous salt is needed in the cold solution (4-7) than in the hot solutions (22-24). Thus, in the hot solution, at a dilution at 145 cm 3 to 500 cm 3 1 grm. of manganous sulphate must be present with 5 cm 3 . of strong hydrochloric acid, with or without sulphuric acid ; while in the cold solution 0.04 grm. of either the sulphate or chloride is enough to secure adequate protective effect. Experience showed, however, that 0.5 grm. or 1.0 grm. of the manganous salt should be present in order to push the re- action with reasonable speed in cold solutions. Wagner* has made record of the increased evolution of chlorine in oxidations of ferrous chloride by potassium permanganate in presence of various salts, of which barium chloride was the most active. We have made some experi- ments, therefore, to determine whether such action would appear in the oxidation of oxalic acid in cold solutions containing certain salts, and, if so, whether it would be preventable by the presence of the manganous salt under our conditions of working. From the results given in the accompanying table, it is plain that the evolution of chlorine in cold solutions is less in the presence of these salts than when hydrochloric acid is used without them, and that such evolution may be entirely prevented (within the proportions * Loc. cit. BY POTASSIUM PERMANGANATE. 229 of our work) by the presence of 0.5 grm. to 1 grm. of manganous chloride. Finally, it appears as the result of an investigation, that the titration of oxalic acid by potassium permanganate in presence of hydrochloric acid is ordinarily attended with some inaccu- racy due to liberation of chlorine from the hydrochloric acid ; that this tendency may be overcome by the presence of a manganous salt either the sulphate or chloride; that 1 grm. of the manganous salt is enough to so affect the conditions of equilibrium that titrations in moderate volumes (100 cm 3 to 500 cm 3 ) and in presence of hydrochloric acid (5 cm 3 to 15 cm 3 of the strong acid) may be conducted with safety and reasonable rapidity, either with or without sulphuric acid, at the ordinary atmospheric temperature. Volume at Beginning of Titration = 140 cm 8 . Temperature = 20-24 C. Ammonium oxalate. HCl. strongest. MnCl 2 . 4H 2 0. BaClo. SrCl,. CaCla- MgCl,. KMn0 4 used. Error. cm 8 25 cm 8 5 grm. 0.5 grm. grm. grm. grm. C1113 26.05 cm 3 0.00 25 25 25 25 25 5 5 5 5 5 27.45 26.50 26.53 26.36 26.13 1.40+ 0.45+ 0.48+ 0.35+ 0.08+ 25 25 25 25 5 5 6 5 0.5 0.5 0.5 0.5 2 26.05 26.10 26.10 26.05 0.00 0.05+ 0.05+ 0.00 25 25 25 25 10 10 10 10 1.0 1.0 1.0 1.0 2 V .. V V 26.10 26.05 26.06 26.11 +0.05 0.00 0.01+ 0.06+ XXVIII THE ESTIMATION OF IKON IN THE FERRIC STATE BY REDUCTION WITH SODIUM THIOSULPHATE AND TITRATION WITH IODINE. BY JOHN T. NORTON, JB.* THE action of sodium thiosulphate on ferric iron has long been known and depends upon the following reaction: 2FeCl 3 + 2Na2S 2 O 8 = SFeCL, + Na 2 S 4 O 6 + 2NaCl. As early as 1859 Schererf proposed a method for the estimation of ferric iron depending on the above reaction. Scherer's method of procedure was to act upon a solution of ferric chloride with sodium thiosulphate until the purple color produced by the interaction of these two salts just vanished. Mohr's $ experimental tests of this process were not successful. A year or two later Kremer and Landolt, after a careful investigation of Scherer's process, recommended it with the modification that any free hydrochloric acid pres- ent should be neutralized by sodium acetate until the solution assumed a red color, just enough hydrochloric acid added to destroy this red color, and sodium thiosulphate run into the solution in slight excess. When the liquid became perfectly colorless and gave no reaction for ferric iron with potassium sulphocyanide, the excess of sodium thiosulphate was titrated back with iodine and starch. The authors also state that the ferric iron should not be present in concentrated solution. Very good results were claimed for this process, but it apparently gained but slight recognition. * From Am. Jour. Sci., viii, 25. t Gel. Anzeig. k. Bayrisch. Acad., Aug. 31, 1859. $ Ann. Chem. Pharm., cxiii, 260. Zeitschr. anal. Chem., i, 214. ESTIMATION OF IRON IN THE FERRIC STATE. 231 Oudemanns,* who was the next to study the action of ferric iron and sodium thiosulphate, claimed that the addition of a small quantity of cupric salt to the iron solution hastened the reducing action of the sodium thiosulphate. Mohr,f however, condemned this method also as unreliable, both because the sodium thiosulphate acted upon the copper as well as the iron and also because the potassium sulphocyanide, added as an indicator of the completeness of the reduction, produced a precipitate of cupric sulphocyanide which interfered with the reaction. In a second paper Oudemanns J reiterated his former statement as to the accuracy of his method but advised the use of a smaller quantity of the cupric salt. An improvement on Oudemanns' process was proposed by Haswell, who mixed the moderately acid solution of ferric chloride in the presence of a cupric salt with a few drops of sodium salicylate and then reduced with sodium thiosulphate previously standardized upon a known quantity of iron by the same process and estimated the excess by potassium dichromate. Bruel || modified this process by operating without the copper solution, relying merely on the discharge of the violet color in a boiling solution by sodium thiosul- phate standardized on a ferric solution of known strength. Although considerable work has been done on the reaction between ferric iron and sodium thiosulphate, no process depending upon this reaction has obtained acceptance. In view, therefore, of previous work on the action of hydro- chloric acid upon sodium thiosulphate If and with the idea that a careful control of the dilution and quantity of acid present might greatly better the accuracy of the method, it has seemed to me to be desirable to study this process again in detail. The ferric oxide employed in the experiments was prepared with great care by the ignition of ferrous oxalate obtained by * Zeitschr. anal. Chem., vi, 129. t Titrirmethode, 5 et Aufl., 294. J Zeitschr. anal. Chem., ix, 362. Kepertorium der analytischen Chem., i, 179. || Compt. rend., xcvii, 954. 1 Am. Jour. Sci., vii, 287. This volume, p. 206. 232 ESTIMATION OF IRON IN THE FERRIC STATE acting with oxalic acid on pure ammonium ferrous sulphate. To ascertain, however, if this oxide contained any impurity, about 0.5 of a grm. was put into a porcelain boat and sub- mitted to the action of a current of hydrochloric acid gas and chlorine at a temperature of about 280 C. (according to a process recently described from this laboratory *) until all the ferric salt is volatilized hi the form of ferric chloride. A residue of 0.0010 grm. for every 0.5 of a grm. of the oxide was found, and this correction, small for the amounts generally used, has been applied in the following determina- tions. The sodium thiosulphate used was taken in nearly ^ solution and was standardized against an approximately decinormal solution of iodine which had been determined by comparison with decinormal arsenious acid made from care- fully resublimed arsenious oxide. In those experiments which deal with amounts of ferric oxide not exceeding 0.2 of a grm., measured portions of a solution of ferric chloride made of known strength by dis- solving about 2 grms. of the pure carefully weighed ferric oxide in 20 cm 3 of strong hydrochloric acid and diluting to one liter, were drawn from a burette. In the case of the krger quantities of ferric oxide the salt was weighed out, dissolved in hydrochloric acid and brought to the required dilution. The ferric chloride, either drawn from the burette or prepared directly from the weighed oxide, was diluted with water, a drop of potassium sulphocyanide added to serve as an indicator and an excess of sodium thiosulphate was run in until, after standing for a few minutes, the solution became perfectly colorless, and the excess of sodium thiosulphate was then titrated back with decinormal iodine after the addition of starch. Several sources of error are, plainly, possible in the process : incompleteness in the reduction of the ferric salt ; decomposition of the thiosulphate by the acid, resulting in the subsequent over-run of iodine; the possible tendency of the ferric salt under concentration to oxide the thiosulphate to the condition * Gooch and Havens, Am. Jour. Sci., vii, 370. This volume, p. 215. BY DEDUCTION WITH SODIUM THIOSULPHATE. 233 of the sulphate rather than to that of the tetrathionate ; and finally the oxidizing action of the air, which may tend to keep up progressive oxidation of the iron salt and excessive expenditure of thiosulphate. The first three sources of diffi- culty tend to produce errors of deficiency; the fourth an error of excess. The first step in the experimental study of the process was to determine the effect of varying dilution upon the estimation of a given quantity of iron reduced by sodium thiosulphate, taken in practically uniform excess above the amount theo- retically required, in the presence of 1 cm 3 of hydrochloric acid. TABLE I. Exp. Fe 2 3 taken. Fe 2 3 corrected. Dilution. HC1. NaAO, in excess. Fe 2 8 found. Error. grm. grm. cm* cm 8 cm grm. grm. (1) 0.1000 0.0998 100 1 18.08 0.0957 0.0041- 2 0.1000 0.0998 200 20 0.0966 0.0032- (3) 0.1000 0.0998 300 17.56 0.0995 0.0003- (4) 0.1000 0.0998 400 17.16 0.0998 0.0000 M 0.1000 0.0998 600 17.76 0.0996 0.0002- (6) 0.1000 0.0998 800 17.65 0.0993 0.0005- (7) 0.1000 0.0998 1000 18.02 0.0988 0.0010- (8) 0.1000 0.0998 1200 17.95 0.0977 0.0021- (9) 0.1000 0.0998 1400 17.99 0.0965 0.0033- (10) 0.1000 0.0998 1600 1 18.01 0.0947 0.0051- (11) 0.2001 0.1997 400 2 27.05 0.2029 0.0032+ (12) 0.2001 0.1997 800 2 15.95 0.1998 0.0001+ (13) 0.4998 0.4988 1000 2 22.36 0.5104 0.0116+ (14) 0.5051 0.5041 1800 4 15.27 0.5026 0.0015 (15) 0.4002 0.-3994 1500 4 27.29 0.3996 0.0002+ (16) 0.7502 0.7487 1000 1 9.73 0.7572 0.0085+ (17) 0.7029 0.7015 2000 4 12.67 0.7004 0.0011- This table shows plainly that with quantities of ferric oxide present up to 0.1 grm. the dilution can vary from 400 cm 3 to 1000 cm 3 for each cm 3 of strong hydrochloric acid and still give excellent results. At a dilution greater than 1000 cm 3 the action of the thiosulphate is evidently incomplete, and at a smaller dilution than 400 cm 3 the decomposing action of the acid on the thiosulphate becomes noticeable. When larger quantities of iron oxide are dealt with, it appears that the 234 ESTIMATION OF IRON IN THE FERRIC STATE dilution ought to be increased proportionally with the quantity of ferric oxide present as well as with that of the acid. This is illustrated in experiments 9-15 of the table. On this account it seems necessary, assuming that the quantity of acid present is always kept within the maximum strength mentioned, 1 cm 3 to 400 cm 3 , to regulate the dilution from the approximate quantity of the iron so that not less than 400 cm 3 of water shall be used to every 0.1 grm. of iron oxide present. Under properly regulated conditions of dilution as regards acid and the iron salt, the reduction is completed in from five to ten minutes. Great excesses of acid, however, contrary to the statement of Kremer,* retard the reduction greatly, and, hi spite of the tendency of the thiosulphate to decomposition and the pro- duction of errors of deficiency under such circumstances, plus errors due to partial oxidation come to light. This fact appears in the following table, which records the results of processes lasting many hours. TABLE II Exp. Fe 2 s taken. Fe 2 8 corrected. DUution. HC1. Na 2 S 2 0, in excess. Fe 2 8 found. Error. grm. grm. cm 8 cm* cm 3 grm. grm. (18) 0.5012 0.6002 1700 10 25.99 5308 0.0306+ (19) 0.7512 0.7497 1200 16 57.8 7685 0.0188+ (20) 0.7520 0.7505 2000 16 56.4 7983 0.0478+ (21) 0.7520 0.7505 1700 15 27.2 7627 0.0122+ As to the temperature at which the reduction should be made, my experience, contrary to that of Kremer, goes to show that no elevation above atmospheric conditions is necessary ; under the conditions of acidity and dilution laid down, the process of reduction is complete within ten minutes after the introduction of the thiosulphate ; moreover, former experience f shows clearly the danger of submitting mixtures of sodium thiosulphate and acid to temperatures much above the ordinary. On the other hand, artificial reduction of temperature tends * Zeitschr. anal. Chem., i, 214. t Am. Jour. Sci., vol. vii, 287. This volume, p. 206. EY REDUCTION WITH SODIUM THIOSULPHATE. 235 to retard the action to an impossible degree. Thus, in an experiment it took five minutes to reduce 0.0500 of ferric oxide at 21 C. completely at a dilution of 200 cm 3 and in the presence of |- cm 3 of hydrochloric acid ; under conditions otherwise precisely similar excepting that the temperature was lowered to C., the action lingered forty-five minutes. Lastly, the question as to the excess of thiosulphate necessary to complete the reduction within a reasonable time must be considered. In nearly all previously recorded experiments the excess of thiosulphate was not less than 15 cm 3 of the ^ solution. The following table shows the effect of dimin- ishing this excess. TABLE HI. Kxp. Fe,0 8 taken. Fe,0 s corrected. Dilution. HC1. NaAO, in excess. aa Error. grin* grm. cm 3 cm 3 cm s grm. grm. (22) 0.0250 0.0250 400 12.2 0.0241 0.0009- (23) 0.0500 0.0499 400 12.2 0.0495 0.0004- (24) 0.0500 0.0499 400 13.66 0.0493 0.0006- (25) 0.1000 0.0998 400 1 7.31 0.0984 0.0014- (26) 0.1000 0.0998 400 1 7.63 0.0972 0.0026- (27) 0.1001 0.0999 400 1 12.88 0.1007 0.0008+ (28) 0.1498 0.1495 600 1* 11.97 0.1475 0.0020- (29) 0.1996 0.1992 800 2 12.43 0.1980 0.0012- From the above experiments taken in connection with those of Table I it is clear that there should always be present an excess of at least 15 cm 3 of the T ^ solution of sodium thiosul- phate. If the quantity of hydrochloric acid is kept very low there is no reason why this excess of thiosulphate could not be considerable without producing any disturbing effect. Practically, however, the presence of an excess between the limits of 15 cm 8 and 35 cm 3 of the ^ solution has been found to give the most satisfactory results. To recapitulate, then, it has been shown that the dilution must be at least 400 cm 3 for each 0.1 of a grm. of iron oxide present, that the quantity of acid should never exceed 1 cm 3 of the strong acid to each 400 cm 3 of water, that the time of reduction must be short to avoid progressive oxidation, that 236 ESTIMATION OF IRON IN THE FERRIC STATE the temperature of the solution should be kept at the normal temperature of the atmosphere, and finally that the excess of sodium thiosulphate present should never be less than 15 cm 3 of the f solution. In the case of large dilution the use of freshly boiled water is recommended so as to avoid the reoxidizing effect of the air upon the reduced iron. In the experiments included in the following table, the above precautions were closely adhered to and manifestly satisfactory results were obtained. TABLE IV. Exp. Fe 2 3 taken. Fe 2 3 corrected. Dilution. HC i. Excess Na,S 2 3 found. Error. grin. grm. cm 8 cm 3 cm* grm. grm. (30) 0.0125 0.0125 200 23.5 0.0125 0.0000 (31) 0.0250 0.0250 400 21.98 0.0250 0.0000 (32) 0.0250 0.0250 400 17 0.0250 0.0000 (33) 0.0250 0.0250 400 17 0.0250 0.0000 (34) 0.0500 0.0499 400 24 0.0498 0.0001- (35) 0.0500 0.0499 400 19 0.0498 0.0001- (36) 0.0500 0.0499 400 15.1 0.0497 0.0002- (37) 0.0500 0.0499 400 19 0.0498 0.0001- (38) 0.1001 0.0999 400 23.1 0.0993 0.0006- (39) 0.1001 0.0999 400 17.93 0.0997 0.0002 (40) 0.1001 0.0999 400 22.92 0.0997 0.0002- (41) 0.1001 0.0999 400 18 0.0997 0.0002- (42) 0.1001 0.0999 400 16 0.0996 0.0003- (43) 0.1498 0.1495 600 j - 23.26 0.1493 0.0002- (44) 0.1498 0.1495 600 ^ - 16.66 0.1493 0.0002- (45) 0.1498 0.1495 600 - 26.87 0.1475 0.0020- (46) 0.1996 0.1992 800 2 22.38 0.1990 0.0002- (47) 0.1996 0.1992 800 2 17.29 0.1999 0.0007+ (48) 0.1996 0.1992 800 2 22.20 0.1991 0.0001- (49) 0.4045 0.4037 1600 4 16.03 0.4042 0.0005+ (50) 0.4045 0.4037 1600 4 16.2 0.4023 0.0014- (51) 0.4018 0.4010 1600 4 16.34 0.4007 0.0003- (52) 0.5051 0.5041 1800 4 15.27 0.5026 0.0015- As seen in the table this process is very accurate, especially in the use of small amounts of ferric oxide. The introduc- tion of cupric sulphate as recommended by Oudemanns, or of sodium salicylate according to Haswell's method, seems to be unnecessary and only complicates the process. In treating ferric oxide, the following method of procedure is recommended. Dissolve an amount not exceeding 0.2 grm. of the oxide in hydrochloric acid, evaporate to a pasty mass BY REDUCTION WITH SODIUM THIOSULPHATE. 237 dilute to about 800 cm 3 with freshly boiled water, add a drop of potassium sulphocyanide, and into this solution run 50 cm 3 of approximately ^ sodium thiosulphate; allow the liquid to stand until perfectly colorless and determine the excess of thiosulphate by ^ iodine and starch. For quantities of iron oxide up to 0.2 of a gram this process is quick and most accu- rate; when care is taken to preserve the relations of acidity and dilution, twice the amount of ferric oxide mentioned above may be handled. XXIX THE DETERMINATION OF TELLUROUS ACID IN PRESENCE OF HALOID SALTS. BY F. A. GOOCH AND C. A. PETEKS.* THE estimation of tellurous acid by oxidation with excess of potassium permanganate (either in acid or alkaline solution), destruction of the higher oxides of manganese or the manga- nate by standard oxalic acid in presence of sulphuric acid, and titration of the residual oxalic acid by more permanganate, has been shown by Braunerf to be feasible. The tendency of the permanganate to throw off too much oxygen when the oxida- tion is made in solutions strongly acidified with sulphuric acid (as must be the case if the tellurous oxide is to be held perma- nently in solution by sulphuric acid) necessitates the applica- tion of a considerable correction.:]: Fortunately, however, as has been shown in a former paper from this laboratory,! when the tellurous oxide is dissolved originally in an alkaline hydrox- ide and the solution made acid only to a limited degree with sulphuric acid either before or after oxidation by the perman- ganate, no correction appears to be necessary. Thus, when an excess of permanganate is added to the alkaline solution, followed by an excess of oxalic acid and sulphuric acid to an amount not exceeding 5 cm 3 of the [1 : 1] mixture with water, the titration of the residual oxalic acid by more permanganate (after heating to 80 C.) leads to results which give no indi- cation of over-decomposition of the permanganate; so also, when the process is similarly conducted excepting that before addition of the permanganate the original alkaline solution is acidified with sulphuric acid [1 : 1] to an amount 1 cm 3 in excess * From Am. Jour. Sci., viii, 122. t Jour. Chem. Soc., lix, 238. J Loc. cit, p. 249. Gooch and Danner, Am. Jour. Sci., xliv, 301. Volume I, p. 145. DETERMINATION OF TELLUROUS ACID, ETC. 239 of that necessary to redissolve the first precipitate, the results are theoretically accurate, and in close agreement with those obtained by the former procedure. In the presence of free hydrochloric acid the action of the permanganate upon tellurous acid has been shown by Brauner* to be irregular and excessive, and the irregularity cannot be corrected (as in the titration of ferrous salts in presence of hydrochloric acid) by the addition of a manganous salt accord- ing to the well-known procedure of Kesslerf and Zimmer- mann4 So far as appears, however, there should be nothing to prevent the accurate determination of tellurium in tellurous compounds in the presence of chlorides by the permanganate process providing the first oxidation is made in alkaline solu- tion, and the second oxidation carried out with such precau- tions as are necessary to a correct determination of oxalic acid by permanganate hi presence of hydrochloric acid ; for the special danger of over-action on the part of the permanganate cannot exist while the solution is alkaline, and has passed when the tellurite has become a tellurate and before the solu- tion is made acid. As to the proper conditions for the titra- tion of oxalic acid by permanganate we have shown recently that the presence of a manganous salt is necessary and suf- cient to secure regularity of action when a considerable amount of hydrochloric acid is in the solution; when the amount is small so much as would be formed in the decom- position of a gram or two of halogen salt of tellurium the disturbing effect under ordinary conditions of work is prob- ably inappreciable, but even in such a case it is better to work in the presence of a manganous salt for the reason that the titration of the oxalic acid may then be made at the ordinary atmospheric temperature. In the following table are gathered the results of experi- ments made with, and without, the addition of the manganous salt. * Loc. cit., p. 241. t Ann. Phys. cxviii, 48 ; cxix, 225, 226. | Ann. Chera. (Liebig), ccxiii, 302. Am. Jour. Sci., vii, p. 461. This volume, p. 222. 240 DETERMINATION OF TELLUROUS ACID TABLE I. O = 16, Te = 127.5. Volume at beginning, 150 cm 8 . Temperature of titration, 60-80 C. TeO, taken. NaCl. 5f2- MnCl 2 .4H 2 0. TeOg found. Error. grm. 0.1000 0.1000 0.1000 0.1000 0.0650 grm. 0.4 0.4 0.4 1.0 1.0 cm 8 5 5 5 5 5 grill. grm. 0.1003 0.1000 0.1004 0.1003 0.0653 grm. 0.0003+ 0.0000 0.0004+ 0.0003+ 0.0003+ B. Temperature of titration, 20-26 C. 0.0700 0.0700 0.0700 0.1000 0.4 0.4 0.4 0.4 5.7 5.7 5.7 5.7 1.0 1.0 0.5 0.5 0.0705 0.0698 0.0701 0.1008 0.0005+ 0.0002- 0.0001+ 0.0008+ The tellurium dioxide, made by the careful ignition of the crystallized basic nitrate obtained by oxidizing tellurium with nitric acid, was dissolved in a small amount of sodium hydrox- ide, the halogen salt was added to the amount shown, the per- manganate standardized against ammonium oxalate was run in until its characteristic color appeared, standard ammonium oxalate was added in excess of the quantity required to reduce the excess of permanganate, manganate, and higher oxides, and the solution was heated with enough sulphuric acid [1:1] to neutralize the alkaline hydroxide and have an excess of about 5 cm 3 . In the experiments of Section A the liquid was heated to 60 -80 C. to dissolve the oxides at the final titra- tion begun at that temperature ; in those of Section B, manganous chloride (0.5 to 1 gram) was added, so that the reduction of the higher oxides of manganese and the final titration of the excess of oxalic acid might take place at the ordinary temperature of the room. Plainly the presence of the chloride does not interfere materially in the determination of the tellurium by this process whether the titration is made at a high or low temperature. IN PRESENCE OF HALOID SALTS. 241 It appears, also, upon putting the matter to the test, that fairly good determinations of tellurous acid may be made similarly in the presence of a bromide, provided the titration is made at the atmospheric temperature in the presence of a sufficiency (0.5 gram to 1 gram) of a manganous salt and of an excess of sulphuric acid limited to about 5 cm 3 or less of the 12.5 per cent mixture. At the higher temperatures bromine is liberated at once from the acid solution by the permanganate. The experimental results are given in Table II. TABLE II. = 16, Te = 127.5. Volume at beginning, 150 cm 8 . Temperature of titration, 24-26 C. Te0 2 taken. NaCl KBr. Sfc MnCl 2 . 4H 2 O. TeO, found. Error. grm. grm. grm. cm s grm. gnn. grm. 0.1000 . . 0.5 20 1.0 0.1022 0.0022+ 0.3000 . . 1.5 25 1.0 0.3030 0.0030+ 0.0650 0.5 1 1.0 O.OG61 0.0011+ 0.0650 0.5 1 1.0 0.0647 0.0003- 0.1000 0.5 1 1.0 0.1002 0.0002+ 0.3000 . . 0.5 5 0.5 0.3010 0.0010+ 0.0650 0.5 0.5 1 1.0 0.0661 0.0011+ It is obvious, therefore, that tellurous acid may be deter- mined with a fair degree of accuracy by the permanganate method in the presence of chlorides and bromides, provided the first oxidation is made in alkaline solution and the final titration of the residual oxalic acid is made at ordinary temperatures in the presence of a manganous salt and restricted amounts of free sulphuric acid. In the presence of an iodide, however, the case is different. Upon acidifying the mixture of iodide and the higher oxygen compounds of manganese, produced in the action of the permanganate upon the solution, iodine is at once set free, and oxalic acid does not suffice to reconvert it. In the presence of an excess of potassium iodide the higher manganic compounds are completely reduced with rapidity and the iodine liberated is the measure of the excess of permanganate over that VOL. II. 16 242 DETERMINATION OF TELLUROUS ACID required to oxidize the tellurous acid ; the difference between the amount of permanganate thus indicated and that originally introduced should determine the amount of the tellurous acid. It is upon this basis that Norris and Fay * have founded their excellent iodometric determination of tellurous acid. This process consists in treating the alkaline solution of tellurous acid with standard permanganate until the meniscus of the liquid shows a deep pink color, then diluting the solution with ice-water, adding potassium iodide and sulphuric acid, and titrating with sodium thiosulphate. The results are excellent. It is plain that any agent capable of converting the iodine to hydriodic acid without at the same time reducing telluric acid should be capable of measuring the excess of the permanganate, and so the amount of tellurous acid originally present. We find that the standard arsenite made, as usual, by dissolving 4.95 grams of pure resublimed arsenious oxide to the liter of water containing potassium bicarbonate answers the purpose admirably, and possesses the further advantage of fixing at once the entire standard of the process, the strength of the permanganate (approximately -^ being determined by running a definite volume of its solution into water containing potassium iodide (1 gram) with 2 to 3 cm 3 of dilute sulphuric acid and titrating by the standard arsenite the iodine (set free by the action of the excess of permanganate and higher oxides) after neutralization with acid potassium bicarbonate. In this titration of iodine by the arsenite we find it best to dispense with the starch solution usually employed to secure the end reaction. The color of the free iodine itself is sufficiently definite, even at a dilution so much as 300 cm 8 , and its disappearance under the action of the arsenite is much sharper than that of the blue starch iodide. In Table III are recorded results obtained by adding the alkaline solution of tellurous oxide to 100 cm 8 of water containing 0.5 gram or 1 gram of potassium iodide, introducing the standardized potassium permanganate until the green color of the manganate appears (about 30 cm 3 of the ^ solution for * Am. Chem. Jour., xx, 278. IN PRESENCE OF HALOID SALTS. 243 every 0.1 gram of TeO 2 ), adding a few cubic centimeters of dilute sulphuric acid, followed, when the solution has cleared and separated iodine, by an excess of acid potassium carbonate, and titrating to the destruction of color with the standard solution of arsenic. It is essential, in order that oxygen may not go to waste in the breaking down of the oxides, that more than enough iodide should be present when the solution is acidified to complete the reduction of the manganese oxides, or else, that the arsenious acid should be present in suitable amount before the sulphuric acid is put in. This latter procedure may be used in case, for any reason, it is preferred not to introduce more iodide into the solution than may be present originally : when, for example, a direct determination of the iodine present is to follow. TABLE III. O - 16, Te = 127.5. Te0 2 taken. NaCL KBr. KI. Total volume NaOH present during TeO, found. Error. oxidation. griii* grin* grm. grm. cm 3 . grm. grm. grm. 0.1000 0.5 160 0.1 0.1005 0.0005+ 0.1000 0.5 160 0.1 0.1001 0.0001+ 0.1000 . 0.5 160 0.1 0.1003 0.0003+ 0.1000 1.0 250 0.1 0.1007 0.0007+ 0.2000 1.0 250 0.2 0.1997 0.0003+ 0.1000 0.6 0.5 0.5 250 0.1 0.1000 0.0000 0.2100 1.0 1.0 1.0 225 0.2 0.2105 0.0005+ 0.1000 0.5 160 1.0 0.1011 0.0011+ 0.2000 1.0 300 2.0 0.2009 0.0009+ These results are reasonably good. Like those of Table I they would be brought practically in the average to the figure demanded by theory if the value of the Committee of the German Chemical Society, Te = 127, were to be taken instead of Te = 127.5, the value of Clarke and of Richards. XXX AN IODOMETRIC METHOD FOR THE ESTIMA- TION OF BORIC ACID. BY LOUIS CLEVELAND JONES.* IN a recent article,! I have described a process for the alka- limetric estimation of boric acid, depending upon the forma- tion of a strongly acidic compound when boric acid and a polyatomic alcohol are placed together in solution. The method in brief consists in destroying the free mineral acid in a solution containing boric acid, by means of a mixture of potassium iodide and iodate, bleaching the liberated iodine by sodium thiosulphate, adding the indicator phenolphthalein and sufficient standard solution of caustic soda to give a faint alkaline coloration, bleaching by a small amount of mannite and adding caustic soda again to alkalinity, and thus alternating with mannite and alkali until the alkaline coloration produced is permanent. The amount of sodium hydroxide used represents the amount of acidity developed by the influence of the mannite upon the boric acid present, according to the hypothesis that the molecule B 2 O 8 acts as two molecules of a univalent acid, HOBO. On making further study of this reaction, I have found that the acid developed by the combination of boric acid and mannite is, under certain definite conditions, sufficiently strong to liberate, quantitatively, from a mixture of potassium iodide and iodate, the amount of iodine required on the supposition that each molecule of metaboric acid (HOBO) acts in a manner similar to a univalent mineral acid under the same conditions. (5KI + KIO 8 + 6HOBO = 3I 2 + * From Am. Jour. Sci., viii, 127. t Am. Jour. Sci., vii, 147. This volume, p. 182. THE ESTIMATION OF BORIC ACID. 245 6KOBO + 3H 2 O.) Obviously, this reaction depends upon the behavior of the acidic boromannite compound as a strong acid, stronger than acetic, tartaric, or citric acid; for these acids have been found by Furry * to be incapable of liberat- ing iodine regularly from a mixture of iodide and iodate. Conditions which tend to increase the acidic activity of this compound are concentrated solutions and moderately low temperatures.f Glycerine acts in general like mannite to produce acidic compounds with boric acid; and hi a preliminary way, the relative acidity of the products formed by these two poly- atomic alcohols with boric acid may be indicated by the results of two experiments in which the iodine liberated from a mixture of potassium iodide and iodate, proportionately to the time required for the liberation, is taken as a measure of the strengths of the acids developed. Equal amounts (10 cm 3 ) of a standard solution of boric acid, prepared from the anhydride, J were drawn into separate Erlenmeyer flasks and a neutral solution of iodide and iodate added to each in an amount sufficient to liberate iodine in quantities corresponding to the acid used. One solution was treated with glycerine enough to constitute one-half the entire volume of the liquid: mannite (about 5 grms.) was added to the other. The thiosulphate required immediately and after definite periods of tune, is shown for each solution in the following table. The solution of boric acid contained 7.706 grm. per liter. The thiosulphate was 0.0999 normal. According to theory, the amount of thiosulphate required for 10 cm 3 of the boric * Am. Chem. Jour., vi, 341. t Magnanini, Gaz. chim. Hal. xx, 428, xxi, 134 ; and Lambert, Compt. rend., CTiii, 1016, 1017. J The recrystallized hydrous boric acid should be fused in a platinum dish and, after cooling and breaking into small pieces, the desired amount placed in a small weighed platinum crucible and again fused until no more water escapes. After cooling and weighing, the boric oxide may be separated from the crucible, or with it placed in warm water, dissolved and made up to a definite volume. 246 AN IODOMETRIC METHOD FOR TABLE I. Bj0 8 solution (10 cm) with maimite. Time. B 2 S solution (10 cm 8 ) with glycerine. Thiosulphate required. Thiosulphate required. cm* cm' 18.60 21.30 Immediately. After 15 minutes. 8.48 10.50 22.00 After 30 minutes. 11.15 22.05 After 2 hours. 11.60 acid solution is 22.02 cm 3 . From these data we may observe that at the end of 30 minutes, in the solution containing mannite, practically the theoretical amount of thiosulphate had been used, while only about 50 per cent of that amount had been required to bleach the iodine liberated by the glycerine compound. Obviously, mannite forms with boric acid a more acidic compound than glycerine does, and, from the indication given in the above experiments, may be relied upon, under certain conditions, to liberate the theoretical amount of iodine. If, from the iodide and iodate used to destroy the excess of mineral acid already present, the boric acid, upon the addition of mannite does liberate iodine regu- larly as the previous experiments seem to indicate this liberated iodine should form a most convenient measure of the boric acid present. On studying the conditions requisite for the complete liberation of iodine according to theory, several important points have come to light. It has not been found possible under any conditions to rely upon the immediate liberation of the full amount of iodine: a certain period of time is required for the completion of the reaction. When the solution is of small volume and saturated with mannite, the reaction goes to the end most quickly sometimes almost immediately but there is this limitation, which must be made emphatic, viz.: that if the solution of boric acid is too concentrated near saturation the boric acid alone, when the iodate and iodide are added to destroy any other free acid present, throws out some iodine THE ESTIMATION OF BORIC ACID. 247 and on bleaching with thiosulphate a starting-point is ob- tained at which some of the boric acid has already entered into combination. The amount of iodine thus liberated by the boric acid is, however, not large, and if upon the addition of the iodide and iodate, the iodine thrown out by the free hydrochloric acid present is immediately bleached by thio- sulphate and the analysis proceeded with from this as the neutral point, even in concentrated solutions the error is almost inappreciable. If, however, considerable time inter- venes between the adding of the iodide and iodate and the determination of the neutral point by thiosulphate, as much as several milligrams of boric acid may have liberated its amount of iodine and is, therefore, not capable of being registered by thiosulphate after the addition of mannite. This difficulty was not met with in those experiments in which the iodide and iodate were added at a dilution little greater than that of the standard solution used (7.738 grm. per liter), but in an attempt to estimate the boric acid in colemanite, where the solution was kept as concentrated as possible, hoping in this way to decrease the time required for the complete liberation of iodine, low values were obtained; that is, a false starting point was used. The dilution found most convenient at the time of adding the iodide and iodate is not less than 25 cm 3 for each decigram of boric acid (B 2 O 3 ) present and should not be much greater than two or three times that amount. This limitation as regards volume is equally applicable, whether after obtaining the neutral point and treating with mannite, the boric acid is to be measured by a standard solution of alkali as before described or as here by the iodine liberated. As has been suggested, a large volume, even though saturated with mannite, prolongs the time of standing necessary and increases the effect of carbon dioxide upon the iodide and iodate present, for carbon dioxide, whether derived from the atmosphere or existing dissolved in the solution, upon standing, slowly liberates iodine. The amount, however, is small, and, in the time required for the completion of the process, has never been 248 AN IODOMETRIC METHOD FOR found equivalent to more than a single drop of the solution of thiosulphate used. Even if the material to be analyzed contains carbonates, after acidifying in concentrated solution and shaking vigorously, the small amount of uncombined carbon dioxide remaining has almost an inappreciable effect upon the results. The length of tune required for the liberation of the theoretical amount of iodine hi a solution of the volume suggested above, is 20 to 45 minutes, and at the end of 45 minutes standing in a solution saturated with mannite the reaction may be considered complete. During this period, how- ever, it is well to keep the solution cool at zero will do no harm and shake occasionally to insure thorough mixture. The free iodine would tend to escape upon standing unless kept in a closed flask, but it is more convenient, immediately after the addition of mannite, to treat with an excess of the standard solution of thiosulphate 8 or 10 cm 3 more than the amount required to bleach the iodine liberated, and at the expiration of 40 to 60 minutes titrate back with ^ iodine. The strength of the thiosulphate solution found most convenient is -", while the use of iodine of one-half this strength (^) enables the error of reading to be correspondingly diminished. In solutions of the volume recommended the addition of starch to give the indication with iodine is unnecessary and even detrimental, since a single drop of one-twentieth normal iodine in excess is sufficient to give a strong lemon coloration, while in the presence of starch an indefinite dirty red first appears and remains until the blue is brought out by the further addition of iodine. With these observations in mind, a series of experiments was made in which the standard solution of boric acid was drawn into an Erlenmeyer flask, containing a small amount of free hydrochloric acid and made up to a definite volume. To bring the conditions to those of an actual analysis about 1 grm. of crystalline calcium chloride in solution was also added. Potassium iodate (5-10 cm 3 of a 5 per cent solution) and iodide (3-5 cm 3 of a 40 per cent solution) were added, and the iodine liberated by the hydrochloric acid, barely bleached and THE ESTIMATION OF BORIC ACID. 249 again brought to coloration by iodine. Mannite was added to saturate the solution, an excess of standard thiosulphate put in, and the solution set aside for various periods of time, at the end of which the excess of thiosulphate was titrated by iodine and the amount of unrecovered thiosulphate taken as a measure of the boric acid present. The thiosulphate used was 0.198 normal and the iodine 0.0996 normal. The solution of boric acid contained 7.733 grm. per liter. TABLE II. B 3 8 taken. Thio. taken. Iodine taken. Time of standing. Volume. S&. B 2 3 found. Error. A cm 3 28.00 27.03 27.02 cm 3 32.00 32.00 31.97 cm 8 1.88 4.37 4.04 hrs. 0.30 0.27 1.00 cm 3 28 27 27 grm. 0.2165 0.2090 0.2089 gnu. 0.2168 0.2081 0.2090 gnu. 0.0003+ 0.0009- 0.0001+ B 27.06 27.02 27.04 32.04 32.02 31.72 3.88 4.40 3.39 1.00 1.00 1.00 50-60 50-60 50-60 0.2093 0.2089 0.2091 0.2101 0.2081 0.2096 0.0008+ 0.0058- 0.0005+ C 27.01 26.05 31.53 31.01 2.88 4.01 2.00 3.00 50-60 50-60 0.2089 0.2014 0.2100 0.2025 0.0011+ 0.0011+ D 27.00 27.00 26.01 27.03 27.05 26.07 27.00 31.00 32.00 32.02 31.01 31.89 31.02 32.04 2.12 4.05 6.20 2.21 3.81 4.14 4.30 0.30 0.30 0.30 0.48 0.45 0.40 0.40 50-60 50-60 50-60 50-60 50-60 50-60 60 0.2088 0.2088 0.2011 0.2090 0.2092 0.2016 0.2088 0.2089 0.2092 0.2018 0.2087 0.2093 0.2020 0.2086 0.0001+ 0.0004+ 0.0007+ 0.0003- 0.0001+ 0.0004+ 0.0002- These results are so regular that the method seems worthy of high commendation, and especially since the standard solutions, thiosulphate and iodine, upon which the process depends, are so easily prepared and generally at hand. The full method of procedure recommended is as follows : The borate is dissolved in as small volume and as little 250 AN IODOMETRIC METHOD FOR hydrochloric acid as possible, shaking well to remove free carbon dioxide and diluting so that, at the time of adding potassium iodide and iodate, there shall be approximately 25- 50 cm 3 of solution for each decigram of boric anhydride present. The greater part of the excess of hydrochloric acid in the solution is destroyed by sodium hydroxide and the use of litmus paper, leaving the solution distinctly acid in reaction. Potassium iodide (3-5 cm 3 of a 40 per cent solution), and iodate (5-10 cm 3 of a 5 per cent solution) are added in excess of that required to liberate iodine in an amount corresponding to the hydrochloric acid and the boric acid present. The iodine liberated by the free hydrochloric acid is bleached by a small amount of a strong solution of thiosulphate, and after agitating to insure thorough mixture, iodine is added to faint coloration. Sufficient mannite is now used to saturate the solution about 10-15 grm. for a volume of 50 cm 3 and sodium thiosulphate added in standard solution 8-10 cm 8 in excess of that required to bleach the iodine immediately thrown out by the mannite. The solution is again brought to saturation, if necessary, by mannite and after standing in a cool place for 40-60 minutes, titrated with decinormal iodine to determine the excess of thiosulphate present. In the manner described, specimens of crude calcium borate and crystals of colemanite were analyzed with the results given below. TABLE III. CALCIUM BORATE. Mineral. Thio. taken. Iodine taken. Time stand- ing. Volume of solu- tions. B 2 0. found. Per cent. grm. 0.4015 0.4010 cms 35.05 35.34 cm 8 4.75 5.23 hrs. 1.00 2.00 cm 8 40 45 grm. 0.2280 0.2283 56.92 56.94 COLEMANITE. 0.4002 0.2513 0.4007 32.00 32.01 33.03 5.50 7.36 7.72 1.30 1.00 0.50 50 40 65 0.2043 0.1279 0.2036 51.04 50.91 50.81 THE ESTIMATION OF BORIC ACID. 251 The solution of thiosulphate used was 0.19939 and the iodine 0.0996 normal. These results show little variation and in the case of colemanite correspond closely to the theory 50.97 per cent. The process is convenient, generally applicable, and accurate within the ordinary limits of analysis. XXXI THE DOUBLE AMMONIUM PHOSPHATES OP BERYLLIUM, ZINC, AND CADMIUM IN ANALYSIS. BY MARTHA AUSTIN.* IT has been shown f that the composition of the phosphate of manganese thrown down by microcosmic salt from the solution of a pure manganous salt contains more manganese than belongs to the ideal ammonium manganese phosphate NH 4 MnPO 4 ; and, further, that by acting with ammonium chloride in proper proportion the phosphate of manganese thrown down by microcosmic salt may be completely converted to the ideal ammonium manganese phosphate. Ammonium chloride, likewise, in the case of magnesium phosphate f tends to cause the replacement of the metal by ammonia. Indeed, the replacement here is readily carried so far beyond the point corresponding to the normal ammonium magnesium phosphate, NH 4 MgPO 4 , that the tendency to form a salt richer in ammonia and poorer in magnesium perhaps something like Mg(NH 4 ) 4 (PO 4 ) 2 must be recognized. These facts suggested an investigation into the constitution of certain other ammonium phosphates with reference to their utility in analytical processes. Of the elements of Mendele'eff's second group, beryllium, magnesium, zinc, cadmium, and mer- cury are capable of yielding double ammonium phosphates, while no such compounds of calcium, strontium and barium have been described. The solubility in ammonia of the double ammonium phosphates of the elements of the former category * From Am. Jour. Sci., viii, 206. t Am. Jour. Sci., vol. yi, 233. This volume, p. 121. } Am. Jour. Sci., vol. vii, 187. This volume, p. 190. DOUBLE AMMONIUM PHOSPHATES IN ANALYSIS. 253 appears to increase as the elements of which they are compounds are removed in the series from the beryllium, and, while the same is true of the simple phosphates of members of the latter category, the extent of such solvent action is slight comparatively. According to the work recorded in the literature, calcium, barium, and strontium form individu- ally a neutral tribasic phosphate or acid phosphates of greater or less degree of acidity according to the conditions of precipitation. In my experience where salts of these elements were precipitated either with ammonium phosphate or microcosmic salt hi presence of varying amounts of ammonium chloride, or ammonia, or both, only the recognized phosphates were obtained. The effect of ammonium salts in presence of ammonia seemed to promote the formation of the tribasic salt in the case of calcium and strontium; barium tends to form the barium acid phosphate almost exclusively even in the presence of ammonium salts and free ammonia. No double ammonium phosphate of either calcium, strontium, or barium was produced under any condition. As is well known, mercury does form an ammonium mercuiy phosphate, but the salt is soluble to so great a degree in ammonia, ammonium chloride, and even in the precipitant itself, that nothing of any value for analytical work seemed likely to come from its study. The Ammonium Beryllium Phosphate. The ammonium beryllium phosphate has beeen described by Roessler * as a crystalline salt produced by boiling some time in ammoniacal solution the phosphate precipitated by ammonium phosphate, though the best results of this treatment failed to yield the ideal constitution of this salt, NH 4 BePO 4 . This same precipitate cannot be obtained, Roessler further states, by using a sodium salt as the precipitant. In order to follow out this work of Roessler, a solution of berryllium chloride for use was prepared as follows : The pure beryllium chloride of commerce was dissolved in as little water as * Fresenius, Zeitschr. anal. Chem., xvii, 148. 254 DOUBLE AMMONIUM PHOSPHATES OF BERYLLIUM, possible and treated for the precipitation of aluminum by ethereal hydrochloric acid.* After filtering and evaporating from the filtrate the ether and a part of the hydrochloric acid, the beryllium was precipitated with ammonia, filtered to remove any members of the magnesium group, and washed free from ammonium chloride. The larger part of the precipitate was dissolved in hydrochloric acid in slight excess, and boiled with the reserved portion. After filtering, the solution was diluted to definite volume and standardized by precipitating measured portions of the solution with ammonia, filtering on asbestos under pressure in a perforated platinum crucible, igniting the residue and weighing as the oxide. The results recorded in section A of the following table were obtained by precipitating definite volumes of the pure solution of beryllium chloride with ammonium phosphate in a platinum dish, dissolving the precipitate in hydrochloric acid in faint excess, and while hot precipitating slowly with dilute ammonia, boiling (while the solution was kept distinctly ammoniacal) until the flocky precipitate was entirely converted to a fine, powdery, semi-crystalline, rapidly subsiding mass. A quarter to a half-hour is necessary under the most favorable conditions to cause this conversion. After cooling, the precipitate was filtered off on asbestos under pressure in a perforated platinum crucible, washed carefully with distilled water, dried, ignited and weighed. The filtrate was tested for beryllium by boiling with ammonia. None was found in these cases, nor in any of the following work. Faint traces of chloride were found in the residues after ignition after dissolving in nitric acid and testing with silver nitrate. The results are in every case in excess of the theory for the pyrophosphate derived by ignition of the ammonium beryl- lium phosphate, possibly because the ammonium chloride present may have a tendency to form a salt too rich in ammo- nium (as was shown to be the case with the magnesium salt), consequently giving too much phosphoric acid in the ignited residue ; or, because of inclusion of the chloride and phosphoric * Am. Jour. Sci., ir, 111. This volume, p. 111. ZINC, AND CADMIUM IN ANALYSIS. 255 acid. It might reasonably be expected that some phosphoric acid may be held, since a trace of chloride was found. Either or both of these substances may have been held mechanically, or in combination. It was found that on boiling for some tune the solution of beryllium chloride with microcosmic salt (6) section B of the table and precipitating in the same manner as when ammonium phosphate was used the same sort of powdery mass remained as was obtained by the ammonium phosphate. The residue being tested for sodium according to the method brought out by Kreider and Brecken^idge,* showed sodium present to the amount of 0.0062 grm. reckoned as sodium phosphate. It may be reasonably supposed that the presence of the sodium was due to one of two causes, inclusion of the soluble phosphate, or a tendency on the part of the beryl- lium to form an ammoniumf sodium beryllium phosphate or a sodium J beryllium phosphate, both of which are known to exist. Long boiling of the precipitates is tedious, and, unless great care is taken, may involve small losses of material ; hence if the same results could be obtained with less boiling such treatment would be decidedly advantageous. The results in section C of the table were obtained by adding microcosmic salt to the hot solutions of the chloride, boiling five minutes, cooling, filtering off on an ashless filter because of the flocky condition of the precipitate treating as usual before igniting the residue in a platinum crucible. The results compare well with those obtained by long boiling of the precipitated beryl- lium although all are in excess of the theory. That ammo- nium chloride here, as in cases above, has a marked effect in changing the constitution of the phosphate precipitated by microcosmic salt is not readily seen. It is obvious that the presence of an excess of the soluble phosphate is essential to precipitate the beryllium as the double ammonium phosphate from the results recorded in section D of the table, where, * Am. Jour. Sci., ii, 263. Volume I, p. 401. t Persoz, Ann. Chem. (Liebig), Ixv, 174 ; Atterberg, Bulletin Soc. Chim., xxiv, 358. J Scheffer, Ann. Chem. (Liebig), cix, 144. 256 DOUBLE AMMONIUM PHOSPHATES OF BERYLLIUM, after the precipitate of beryllium phosphate had subsided and the supernatant liquid had been poured off, the precipitate dis- solved in hydrochloric acid was brought down again at the boiling temperature with ammonia either alone or in presence of ammonium chloride. The results obtained show that the salt approaches the constitution of the tribasic phosphate, when it is precipitated in presence of a faint excess of phos- phoric acid, even though ammonium chloride in large amount be present. TABLE I. Exp. Be 2 P 2 O 7 corresponding to BeCl,. Be 3 P,O 8 corresponding to BeCl,. (NH 4 ) 3 P0 4 . NB^Cl. Taken. Found. Error. Taken. Found. Error. A. (1) (2) (3) (4) (5) grm. 0.3578 0.3578 0.3578 0.3578 0.3578 grin. 0.3613 0.3808 0.3707 0.3640 0.3680 grin. 0.0035+ 0.0230+ 0.0129+ 0.0062+ 0.0102+ grm grm. grm. grm. 2 2 2 2 2 grm ' 'so B. HNaNH 4 PO 4 .4H 2 6. (6) 0.3578 0.3697 | 0.0119+ | ... ... | ... C. (7) (8) (9) (10) 0.3578 0.3578 0.3578 0.3578 0.3618 0.3680 0.3729 0.3631 0.0040+ 0.0102+ 0.0151+ 0.0053+ 1.2 1.2 1.2 1.2 ' *10 60 D. (11) (12) (13) (14) 0.2700 0.2700 0.2700 0.2700 0.2589 0.2989 0.2936 0.2507 0.0111- 0.0289+ 0.0236+ 0.0193- 0.5 0.5 0.5 0.5 10 5-60 60 From the work described it is clear that the ammonium beryllium phosphate is not obtained in ideal condition by pre- cipitating a solution of the chloride with ammonium phosphate. Roessler's own results were likewise only approximately cor- rect, as he states. It is also plain that hydrogen sodium ammonium phosphate precipitates the ammonium beryllium ZINC, AND CADMIUM IN ANALYSIS. 257 phosphate in a condition as nearly ideal as does the ammonium phosphate, while the effect of the ammonium chloride in either case is not marked in producing a phosphate containing ammonia. Of most importance in obtaining the ammonium salt is an excess of the soluble phosphate, for when the amount of the precipitant is reduced to a little more than the theo- retical amount the condition of the phosphate coincides almost exactly with the theory for the tribasic phosphate, even though a large excess of ammonium chloride be present. When there is an abundance of the precipitant the results are all in excess of the theory, which may be accounted for on the supposition that foreign material is included the chloride of ammonia and the soluble phosphate to a greater or less extent by the pre- cipitate. The formation of a phosphate of beryllium contain- ing too much ammonia and phosphoric acid, or, in case of the precipitations by microcosmic salt, sodium by the formation of a sodium ammonium beryllium phosphate and sodium beryl- lium phosphate (known salts), is not definitely proved. The Ammonium Zinc Phosphate. Debray,* Bette f and Heintz J separately found that am- monium zinc phosphate is formed by boiling a solution of zinc sulphate with ammonium phosphate. This salt was investigated later by A. Guyard (Hugo Tamm), who found that if to a solution of a zinc salt of an organic or a mineral acid supersaturated with ammonia until all the zinc oxide is dissolved and made faintly acid with hydrochloric acid, sodium phosphate be added, a flocky precipitate resulted, which on being kept near the boiling point for some seconds was converted to crystalline zinc ammonium phosphate, which filtered readily and was washed free from impurities with the greatest facility. He found that all the zinc in solution was thrown down as the ammonium zinc phosphate, which on ignition yielded the zinc pyrophosphate. With care in handling this process to avoid an excess of the precipitant, * Compt. rend., lix, 40. t Ann. Chem. (Liebig), xv, 129. J Ann. Chem. (Liebig), cxliii, 156. Chem. News, xxir, 148. VOL. ix. 17 258 DOUBLE AMMONIUM PHOSPHATES OF BERYLLIUM, and the presence of sodium and potassium salts (on account of the danger of occlusion) the precipitation of the ammo- nium zinc phosphate, ignition, and weighing as the pyro- phosphate made, Guyard believed, an ideal process for the estimation of zinc. Although there was slight solubility of the salt, it made an insignificant loss when the process was handled properly. Acids present, or certain alkalies to any great extent, increased the solubility of the salt so much that the loss became appreciable. Another source of error was to Guyard's mind loss of zinc during the ignition of the zinc ammonium phosphate with the paper on which the precipitate had been collected. Garrigues* found, in estimating zinc in a practical way, that this process advocated by Guyard gives in solutions of zinc free from salts of all metals, even alkaline salts solutions that from previous steps in analysis, however, must have contained ammonium chloride in large amount as satisfactory results as Guyard claimed for it. Garrigues' method of procedure was to add acid diammonium phosphate to a warm solution of zinc exactly neutralized with either hydrochloric acid or ammonia, so that the weights of zinc ammonium phosphate and that of the diammonium phosphate added should be as one to five respectively, to heat until the flocky precipitate becomes crystalline and subsides, filtering off on asbestos, drying at 100 C. and weighing preferably, although the residue may be ignited without loss, since the filtration is made on asbestos in a perforated crucible. Langmuir f modifies the method by destroying with dilute acetic acid any free ammonia that may be left in the solution after boiling. In the work that follows, in which an attempt was made to show what precipitate is formed from a solution of zinc by the action of a soluble phosphate, also what effect ammonium chloride has upon the precipitate, a solution of zinc chloride prepared as detailed below was employed. The pure zinc chloride of commerce was treated with zinc carbonate, filtered and precipitated with ammonium sulphide. This precipitate * Jour. Am. Chem. Soc., xix, 936. t Jour. Am. Chem. Soc., xxi, 115. ZINC, AND CADMIUM IN ANALYSIS. 259 was boiled in a slight excess of hydrochloric acid until all the hydrogen sulphide was removed, and then was precipitated with sodium carbonate. After washing carefully until all the chloride was removed, the greater part of the carbonate was dissolved in sulphuric acid in slight excess, boiled with the remaining portion of the carbonate and filtered. This solution diluted to definite volume was standardized as sul- phate by evaporating the solution to dryness in a platinum crucible and heating the residue.* The heating is carried on safely by so placing the platinum crucible in a radiator (consisting of a crucible and a triangle) that the bottom of the platinum crucible was held about one centimeter above the bottom of the outside crucible. Constant weights were obtained in successive treatment with a few drops of sulphuric acid and heating over the radiator. The results obtained in this manner were a trifle higher, though in fair agreement (when the nature of the carbonate process is taken into consideration) with determinations of the zinc in the solu- tions as oxide after precipitating with sodium carbonate with the usual precautions, filtering off on asbestos under pressure in a perforated platinum crucible, washing with distilled water, drying, and igniting. Results are given in Table II showing the amount of zinc sulphate found in five different portions each of forty cubic centimeters of the solution of zinc sulphate, and, for comparison, the results of de- terminations as zinc oxide by the carbonate processes are included. TABLE II. ZnSO 4 found in 40 cm of solution. Mean value of ZnO corresponding to ZnSO 4 in 40 cms of solution. ZnO found in 40 cm 8 of solution by precipitation as the carbonate. grm. grim* gnu. 0.5386 ' 0.2712 0.2691 0.5385 0.2685 0.5387 0.2711 0.5387 t 0.5390 j Rose-Finkener, Analytische Chemie, 6te Auflage, vol. ii, 117. 260 DOUBLE AMMONIUM PHOSPHATES OF BERYLLIUM, Definite portions of the solution of zinc sulphate were carefully drawn from a burette into a platinum dish, heated and treated with ammonium phosphate until the solution turned red litmus paper blue. The whole was heated until the flocky precipitate became crystalline and fell to the bottom of the dish. The solution after standing as recorded in section A of the table was filtered off on asbestos under pressure in a perforated platinum crucible, and the precipitate was washed with distilled water, dried, ignited and weighed. The filtrate in each case, as in all following cases, was tested for zinc with sulphuretted hydrogen. The results recorded in section B of the table were obtained in the same manner as those of section A, with microcosmic salt substituted for the ammonium salt as the precipitant. The results are below the theory for the pyrophosphate, but no appreciable amount of zinc appeared in the filtrates. Neither ammonium phosphate nor ammonium sodium phosphate seems to precipitate the ideal ammonium zinc phosphate under these conditions; and the time of standing appears to be without effect. The results recorded in section C were obtained by precipi- tating the warm solution of the zinc in presence of large amounts of ammonium chloride by adding microcosmic salt until the solution was alkaline to litmus. From these results it seems that the presence of ammonium chloride is essential for the conversion of the zinc phosphate precipitated by hydrogen sodium ammonium phosphate to the ammonium zinc salt. As a matter of fact the solutions employed by Guyard and those in which estimations are made by practical workers do contain ammonium chloride formed in previous steps of the analysis. The proportion of zinc to phosphate suggested by Garrigues 1 : 5 is the amount of soluble phosphate neces- sary to turn red litmus blue after the zinc is precipitated. In order to find out whether the presence of so large an amount of the soluble phosphate is necessary in presence of ammonium chloride, the solution of zinc sulphate was precipitated in presence of the necessary amount of ammonium chloride by the microcosmic salt in small excess above the equivalent of ZINC, AND CADMIUM IN ANALYSIS. 261 the zinc salt, and the solution was made just ammoniacal to litmus with a few drops of dilute ammonia both before and after heating to convert the precipitate to crystalline condition. Experiment (15) shows that precipitation is not complete under these conditions. The zinc left in the solution was precipitated at once as sulphide, and estimated as the oxide, after dissolving in hydrochloric acid and precipitating TABLE III. Exp. Zn 2 P 2 O 7 corre- sponding to ZnS0 4 . Taken. Pound. Error. > Error in terms of Zinc. Zn 2 P 2 7 corre- sponding to Zn left in the filtrate. (NH 4 )^0 4 . NH 4 CL Time of stand- ing* A. $ (3) grm. 0.6355 0.6355 0.6355 grin. 0.6206 0.6254 0.6300 grin. 0.0149- 0.0101- 0.0055- grm. 0.0060- 0.0040- 0.0022- grm. Trace. Trace. Trace. grm. 3.13 3.13 3.13 grm. hrs. iJ* 16 B. (4) (5) 0.6355 0.6355 0.6271 0.6256 0.0084- 0.0099- 0.0034- 0.0040- Trace. None. HNaNH 4 P0 4 . 4H a O. 0.5 0.5 1 20 4.47 4.47 C. (6) | (9). (10) (11) (12) (13) (14) 0.6355 0.6355 0.6355 0.6355 0.6355 0.6355 0.6355 0.6355 0.6367 0.6285 0.6304 0.6295 0.6335 0.6381 0.6379 0.6386 0.6393 0.6355 0.0070- 0.0051- 0.0060- 0.0020- 0.0026+ 0.0024+ 0.0031+ 0.0038+ 0.0012+ 0.0028- 0.0020- 0.0024- 0.0008- 0.0010+ 0.0009+ 0.0012+ 0.0014+ 0.0005+ None. None. None. None. None. None. None. None. None. 4.47 4.47 4.47 4.47 4.47 4.47 4.47 4.47 4.47 10 10 10 10 20 20 20 20 30 16 ,> j D. (15) (16) 0.6355 0.6355 0.6172 10.6227 II 0.0040 0.0183- 0.0098- 0.0072- 0.0039- 0.0108 None. 0.894 10.894 II 3.576 20 20 3 * E. (17) (18) (19) 0.6355 0.6355 0.6355 0.6270 0.6125 0.6303 0.0085- 0.0230- 0.0052- 0.0034- 0.0093- 0.0021- None. 0.0148 0.0020 4.47 4.47 4.47 10 3 18 18 262 DOUBLE AMMONIUM PHOSPHATES OF BERYLLIUM, with sodium carbonate. In (16) of the table the first nitrate was treated with an excess of microcosmic salt, and boiled. Another portion of the ammonium zinc phosphate was precipitated, and was filtered off and estimated. No zinc was found by sulphuretted hydrogen in the second filtrate. From the results it seems obvious also that an excess of the soluble phosphate is necessary to complete the precipitation of the zinc as the ammonium zinc phosphate instead of partly ammonium zinc phosphate and partly tribasic phosphate. In section E of the table are recorded results where the precipitation was made in presence of an excess of the precipi- tant either alone or in presence of ammonium chloride, the solution being made faintly acid to litmus with acetic acid, according to the manner in which Langmuir recommends to conduct the precipitation. All the results by the method are low. The condition of the ammonium zinc phosphate most nearly approximating to the ideal is obtained as shown in ( 9) to (14) by precipitating in presence of ammonium chloride in large amount. Microcosmic salt is added until the solution containing the ammonium salt is alkaline and the whole is heated until the mass subsides in crystalline condition. The amount of ammonium chloride should be twenty grams if the filtration is to be made as soon as the solution cools. One-half the amount will do if the liquid stands a number of hours. Larger amounts tend to give a salt too rich in ammonia. The time of standing seems to be a less important factor than either the excess of microcosmic salt or ammonium chloride. The Ammonium Cadmium Phosphate. According to S. Drewsen* the cadmium ammonium phos- phate is precipitated by allowing a solution of cadmium sulphate to stand twenty-four hours with ammonium phosphate. It is very soluble both in acids and alkalies. No further preparation of this seems to have been recorded. For the work on this salt to be given below, done with reference to the constitution of the salt precipitated by hydrogen sodium * Gmelin-Kraut, 6te Auflage, iii, 74. ZINC, AND CADMIUM IN ANALYSIS. 263 ammonium phosphate, the effect of ammonium chloride in the precipitation, and the value of the salt for quantitative work, the solution of cadmium chloride employed was prepared as follows : A solution of cadmium sulphate acidulated with hydrochloric acid was precipitated with sulphuretted hydrogen, filtered and washed, and the precipitated sulphide was dissolved in hydrochloric acid and filtered from possible traces of copper and lead. The solution of the sulphide in hydrochloric acid was boiled until all the sulphuretted hydrogen was expelled, and filtered on asbestos in a perforated crucible of platinum under pressure. The cadmium in the filtrate precipitated with ammonium carbonate in excess was washed free from chloride, dissolved in hydrochloric acid and diluted to definite volume. It was standardized as oxide* after precipitating with sodium carbonate with the necessary precautions. The standard solution of cadmium chloride was drawn carefully from a burette into a platinum dish, and, while hot, was precipitated by adding hydrogen sodium ammonium phosphate until the solution was alkaline to litmus. After heating until the solution became crystalline, the whole stood three hours in case of (1) of the table and sixteen hours in case of (2) and (3), before filtering. In experiments (4) to (12), inclusive, recorded in the table, precipitation was made in the same manner as in (1) to (3) in presence of varying amounts of ammonium chloride, and the precipitates were filtered after standing as stated below in the table. It is clear from the results that the cadmium separates out completely on long standing only. Moreover, the ideal condition of the ammonium cadmium, phosphate is obtained only when an abundance of ammonium chloride is present; but large amounts of ammonium chloride dissolve this salt. In (14), where ammonia was added after precipitation was complete, the salt dissolved somewhat ; also in (15), where the solution was left faintly acid with acetic acid, a large part of the salt was dissolved. These weights of cadmium dissolved in the filtrate were obtained by treating the filtrates with sulphuretted * Browning, Am. Jour. Sci., xlvi, 280. Volume I, p. 226. 264 DOUBLE AMMONIUM PHOSPHATES OF BERYLLIUM. TABLE IV. Cd 2 P,0 7 Cd s P 2 7 corre- Time Exp. corre- sponding to CdCl,. Found. Error. Errorin terms of Cadmium. sponding toCd found in HNaNEUPO* .iHsO. NH 4 C1. of stand ing. Taken. the nitrate. griu. grm. grm. grm. grm. grm. grm. hra. (1) 0.6972 0.6201 0.0771- 0.0434- 0.0059 4.5 t 3 (2) 0.6972 0.6135 0.0837- 0.0471- None. 4.5 . . 16 (3) 0.6972 0.6134 0.0838- 0.0471- None. 4.5 . . 16 (4) 0.6972 0.6792 0.0180- 0.0101- Trace. 4.6 1 16 (5) 0.6972 0.6831 0.0141- 0.0079 0.0113 4.5 10 2 ( 6 ) 0.6972 0.6976 0.0004+ 0.0002-f Trace. 4.5 10 16 (7) 0.6972 0.6969 0.0003- 0.0002- Trace. 4.5 10 18 ( 8 ) 0.6972 0.6962 0.0010- 0.0006- Trace. 4.5 10 16 9 0.6972 0.6891 0.0081- 0.0045- 0.0191 4.5 20 16 (10) 0.6972 0.6972 0.0000 0.0000 Trace. 4.5 20 16 (11) 0.6972 0.6942 0.0030- 0.0016- Trace. 4.5 20 16 (12) 0.6972 0.6737 0.0235- 0.0132- 0.0304 4.5 30 16 (13) 0.6972 0.5655 0.1317- 0.0741- 0.1378 4.5 30 16 (14) 0.6972 0.6922 0.0050- 0.0023- 0.0088 4.5 10 16 (15) 0.6972 0.3209 0.3763- 0.2117- 0.2449 4.5 16 hydrogen, dissolving the sulphide in nitric acid, and weighing as oxide after precipitating with sodium carbonate. The ammonium cadmium phosphate is obtained in ideal condition by precipitating with microcosmic salt in presence of 10 grm. ammonium chloride in a total volume of 100 cm 3 to 150 cm 3 shown in (6), (7), and (8) filtering after standing some time. On drying and igniting the pyrophosphate is left. Very large amounts of ammonium chloride 30 grm. dissolve the salt, and seem to tend to cause the formation of a phosphate too rich in ammonia. Either acid or ammonia in small amount dissolves the salt, as is shown in (14) and (15). The results of this investigation as to the analytical application of the double ammonium phosphates of beryllium, zinc, and cadmium may be summarized briefly as follows : It is impossible to estimate beryllium with accuracy as the pyrophosphate obtained by igniting the double ammonium phosphate precipitated from beryllium solutions by microcosmic salt or ammonium phosphate in presence of ammonium chloride. In presence of the proper amount of ammonium chloride (10 grm. to 20 grm. in 100 cm 3 -200 cm 3 of liquid) zinc ammonium ZINC, AND CADMIUM IN ANALYSIS. 265 phosphate can be obtained in the ideal condition, which on ignition yields the pyrophosphate. This method may serve, therefore, for the accurate estimation of zinc. Cadmium may be estimated with accuracy as the pyrophos- phate if the precipitate by microcosmic salt in the nearly neutral solution containing ammonium chloride in the proportion of ten grams to one hundred cubic centimeters is allowed to stand several hours before filtering. In this way all cadmium separates out from the solution as a beautiful crystalline mass of cadmium ammonium phosphate of ideal constitution. The conditions, must, however, be preserved with care ; there must be no excess of ammonia, no free acid, and no excess of ammonium salt beyond the quantity indicated, while that amount is necessary. XXXII SEPARATION OF IRON FROM CHROMIUM, ZIRCONIUM, AND BERYLLIUM, BY THE ACTION OF GASEOUS HYDROCHLORIC ACID ON THE OXIDES. BY FRANKE STUART HAVENS AND ARTHUR FITCH WAY.* IT has been shown in a former paper from this laboratory f that iron oxide may be completely volatilized as chloride by a strong current of hydrochloric acid gas acting at a temperature of 450 -500, and also that the addition of a little free chlorine to the gaseous hydrochloric acid renders this action complete at lower temperatures, 180-200, without the danger of error arising from the liability of ferric chloride to dissociation, or from deficiency of oxidation in the oxide treated, or mechanical loss due to too rapid volatilization. It has also been shown that this reaction can be employed for the separation of iron and aluminum, taken as the oxides, and its application to the separation of iron from other metallic oxides has been suggested. The oxides of chromium, zirconium, and beryllium, like aluminum oxide, are not acted upon by a current of dry hydrochloric acid gas at the temperatures before mentioned, and these oxides also can be entirely freed from iron by this reaction, as the experiments to be described will show. The procedure was the same in each case and analogous to that employed for the separation of iron from aluminum. A mixture of a weighed portion of one of these oxides with a weighed portion of ferric oxide, contained in a porcelain boat and placed within a roomy glass tube supported in a small * From Am. Jour. Sci., viii, 217. t Gooch and Havens, Am. Jour. Sci., vii, 370. This volume, p. 215. SEPARATION OF IRON FROM CHROMIUM, ETC. 267 combustion furnace, was submitted to the action of a dry current of hydrochloric acid gas and chlorine generated by dropping sulphuric acid upon a mixture of strong hydrochloric acid, common salt, and a small amount of manganese dioxide. The gas was admitted at one end of the combustion tube and passed out at the other through a water trap, while the required temperature, from 200 -300, was maintained by regulating the various burners of the furnace. The time of action varies somewhat with the condition of the oxide to be volatilized, and the temperature; generally an hour's heating at 200, proves sufficient for the complete removal of 0.1 grm. of iron. At higher temperatures the action is more rapid ; but the lighter oxide, the beryllium especially, is liable to mechanical loss through the too rapid volatilization of the iron, as experiment (17), where a temperature of 500 was used, will show. It is better, therefore, to use lower temperatures, raising the heat for a few minutes when the action is apparently complete to ensure the removal of the last traces of iron. Tests showed Exp. Fe 2 3 taken. Cr,O 3 taken. Cr 2 O 8 found. Error. (1) (2) (8) (4) (5) (6) grm. 0.1007 0.1007 0.1010 0.1019 0.2007 grm. 0.1008 0.1006 0.1000 0.1005 0.1006 0.1003 grm. 0.1008 0.1006 0.1002 0.1003 0.1005 0.0999 grm. 0.0000 0.0000 0.0002+ 0.0002- 0.0001- 0.0004- ZrO, taken. ZrO, found. (7) (8) (9) (10) (11) 0.1053 0.1204 0.1236 0.2150 0.1516 0.1010 0.1519 0.1516 0.1517 0.1516 0.1010 0.1523 0.1517 0.1519 0.0000 0.0000 0.0004+ 0.0001+ 0.0002+ BeO taken. BeO found. (12) (13) (14) (16) (16) (17) 18) 0.0997 0.1045 0.1215 0.1510 0.0230 0.1309 0.1285 0.0456 0.1099 0.1080 0.1305 0.1081 0.1311 0.1285 0.0457 0.1099 0.1081 0.1290 0.1083 0.0002+ 0.0000 0.0001+ 0.0000 0.0001+ 0.0015- 0.0002+ 268 SEPARATION OF IRON FROM CHROMIUM, ETC. the residual oxides from which the ferric oxide had been removed in this manner to be entirely free from iron. The separation of iron from chromium, zirconium, and beryllium by this method is obviously complete within very satisfactory limits of error. XXXIII THE IODOMETRIC DETERMINATION OF GOLD. BY F. A GOOCH AND FREDERICK H. MORLEY.* IN a recent attempt to measure small amounts of gold in solution by titrating with sodium thiosulphate the iodine set free in the action of an excess of potassium iodide upon auric chloride, Petersonf has been led to conclude that, on the aver- age, one-half more thiosulphate is used up in changing the characteristic starch iodide blue to the faint rose color which precedes entire bleaching than is called for upon the theory that the thiosulphate is simply converted to the tetrathionate in the usual manner. Peterson explains the anomaly upon the hypothesis that, besides acting upon the free iodine, the thio- sulphate is used up coincidently by interaction with the aurous salt, formed in the reduction, with formation of a gold sodium thiosulphate on the type of the well-known silver sodium thio- sulphate. The reaction of this hypothesis is in the nature of things most improbable, since there is no reason to suppose that the soluble double thiosulphate could resist the action of the free iodine which is present to the end the appearance of the rose color, and our study of the reaction of sodium thio- sulphate upon the mixture of gold chloride and potassium iodide, the account of which follows, discloses no evidence of the consumption of more thiosulphate than is demanded by the usual theory, which postulates the simple formation of the tetrathionate by the interaction of the thiosulphate and free iodine. It appeared in the course of our preliminary experimenta- * From Am. Jour. Sci., viii, 261. t Zeitschr. anorg. Chem., xix, 63. 270 IODOMETRIC DETERMINATION OF GOLD. tion that, while practically similar results were obtained by adding the thiosulphate until the blue of the starch iodide had changed to rose, the indications were somewhat more con- cordant when the final rose color was developed by adding iodine to the solution from which the blue had been bleached to colorlessness by a slight excess of the thiosulphate. It appeared, also, that the reduction of the auric salt, with the consequent liberation of iodine, is conditioned by the vol- ume of the solution, the mass of the iodine present, and the time of action. The following statement, in which each result is the average of several titrations in close agreement, shows the effect upon the immediate evolution of iodine brought about by adding varying amounts of water to the gold solution before introduc- ing the iodide, and the effect of different amounts of iodide at different dilutions. Potassium iodide. Gold chloride. Volume before the addition of the thio- 0.01 grm. 0.02 grm. 0.05 grm. 0.1 grm. 0.2 grm. 0.00087 grm. sulphate. .sF . fO.81 0.81 0.81 0.82 0.84 0.00087 cm 3 15 If 1 ?"!. 0.77 0.74 0.78 0.72 0.80 0.78 0.81 0.79 0.81 0.80 0.00087 0.00087 25 60 11 1 S 0.61 0.61 0.68 0.76 0.79 0.00087 100 2 fl [0.45 0.49 0.60 0.72 0.75 0.00087 200 It is evident that for the smaller amounts of iodide the lib- eration of iodine decreases rapidly with the dilution. The larger amounts at the highest concentration show readings a trifle above the normal perhaps because the well-known effect of concentrated solutions of a soluble iodide upon the delicacy of the starch end-color begins to appear. At vol- umes lying between the limit of 25 cm 3 and 50 cm 3 0.1 grm. of potassium iodide is an appropriate amount to use; at a volume of 15 cm 3 , 0.01 grm. to 0.05 grm. of the iodide will do the work; and at lower dilutions, as will appear in IODOMETRIC DETERMINATION OF GOLD. 271 the tabular statements to follow, even less of the iodide is effective. In the series of experiments of which the details are given in Table I, use was made of a solution of pure gold chloride containing 0.8710 grm. to the liter as determined by careful precipitation in the usual manner by ferrous sulphate, and by an alkaline solution of formaldehyde according to the method of Vanino.* A nearly centinormal solution of iodine was pre- pared by diluting to a liter 100 cm 3 of nearly decinormal iodine in potassium iodide carefully standardized against exactly decinormal arsenious acid. A nearly centinormal solution of sodium thiosulphate (containing 1.7012 grm. of Na 2 S 2 O 8 to the liter) was made by diluting to a liter 100 cm 3 of a nearly decinormal solution of that reagent which had been standard- ized carefully against the standard iodine prepared as described. The solution of potassium iodide employed contained 10 grm. of that salt in the liter. In conducting the experiments, a convenient amount of the solution of gold chloride was drawn from a burette, potassium iodide was introduced in the amounts indicated (always several times the theoretical equivalent of the gold, and more than enough to dissolve the aurous iodide precipitated at first), a sufficiency of clear starch indicator was added, the starch blue was bleached by the thiosulphate, and the iodine was added until the liquid assumed a faint rose color. Upon the theory that potassium iodide sets free two atoms of iodine for every molecule of auric chloride (or every atom of gold) present, and that the thiosulphate acts only upon the free iodine to form the tetrathionate in the usual manner, every cubic centi- meter of the thiosulphate solution used in the reaction after deducting the amount equivalent to the iodine introduced to get the end-color, should represent 1Q7 3 X 0.0017012 = 0.001061 grm. of gold. 2(158.22) Ber. Dtsch. chem. Ges., xxi, 1763. 272 IODOMETRIC DETERMINATION OF GOLD. TABLE I. Gold chloride = 0.8710 to 1 liter. Sodium thiosulphate, nearly ~ = 1.7012 " " Iodine, nearly ^ =1.3697 " " Volume at beginning of titration, approximately 50 cm 8 . AuCl 3 taken. KI taken. Na 2 S 2 8 used. Gold found. Theory for gold. Error. Per cent. cm 3 grm. cm 3 grm. grm. grin. (1) 5 0.05 4.02 0.00426 0.00435 0.00009- 2.1 (2) 6 0.05 4.01 0.00425 0.00435 0.00010- 2.3 3) 5 0.05 4.06 0.00431 0.00435 0.00004- 0.9 4) 5 0.05 4.07 0.00432 0.00435 0.00003- 0.7 5) 5 0.05 4.04 0.00428 0.00435 0.00007- 1.6 6) 10 0.08 8.17 0.00867 0.00871 0.00004- 0.5 7) 10 0.08 8.15 0.00864 0.00871 0.00007- 0.8 8) 10 0.08 8.16 0.00865 0.00871 0.00006- 0.7 9 10 0.08 8.15 0.00864 0.00871 0.00007- 0.8 (10) 10 0.08 8.19 0.00869 0.00871 0.00002- 0.2 (11) 10 0.08 8.46 0.00897 0.00871 0.00026+ 3.0 (12) 10 0.08 8.24 0.00874 0.00871 0.00003+ 0.3 Plainly, these results accord reasonably with the theory that two molecules of the thiosulphate are the equivalent in this reaction of two atoms of iodine and one atom of gold. There is no evidence whatever of the excessive action affirmed by Peterson. The strength of the standard solutions used in the experi- ments described was such that an error of 0.01 cm 3 in reading the volumes used would correspond to an error of 0.00001 grm. of gold. It is not to be expected that such readings can be trusted ordinarily to a higher degree of accuracy than 0.02 cm 3 . In case all three solutions should be read to this limit of accuracy with the errors of all lying in the same direction, the summation of error would correspond to 0.00006 grm. of gold. In the following experiments, therefore, solutions obtained by properly diluting those of the previous series were em- ployed. The use of a more dilute solution of gold obviated the necessity for diluting the mixture of gold chloride and the iodide before titrating with the thiosulphate. It was found, IODOMETRIC DETERMINATION OF GOLD. 273 TABLE II. A. Gold chloride = 0.0871 to 1 liter. Sodium thiosulphate, nearly ^ = 1.7012 " " Iodine, nearly j =1.3697 " Solution of gold chloride not diluted before mixing with potassium iodide. Exp. AuCl s taken. Kl taken. JSra,BJO, used. Gold taken. Gold found. Error. cm 8 grin. cm grm. grm. grm. (1) 10 0.01 0.83 0.00087 0.00088 0.00001+ 2) 10 0.01 0.83 0.00087 0.00088 0.00001+ 3) 10 0.01 0.80 0.00087 0.00085 0.00002- 4) 10 0.02 0.84 0.00087 0.00089 0.00002+ 5) 10 0.02 0.88 0.00087 0.00093 0.00006+ 6) 10 0.02 0.82 0.00087 0.00087 0.00000 (7) 10 0.02 0.88 0.00087 0.00093 0.00006+ (8) 10 0.02 0.83 0.00087 0.00088 0.00001+ (9) 10 0.10 0.80 0.00087 0.00085 0.00002- (10) 10 0.10 0.82 0.00087 0.00087 0.00000 (11 10 0.01 0.83 0.00087 0.00088 0.00001+ (12) 9 0.01 0.73 0.00078 0.00077 0.00001- (13) 8 0.01 0.65 0.00070 0.00069 0.00001- (14) 7 0.01 0.58 0.00061 0.00061 0.00000 (15) 6 0.008 0.51 0.00052 0.00054 0.00002+ (16) 6 0.008 0.41 0.00043 0.00044 0.00001+ (17) (18) 4 3 0.005 0.005 0.35 0/24 0.00035 0.00026 0.00037 0.00026 0.00002+ 0.00000 (19) 2 0.003 0.21 0.00017 0.00022 0.00005+ (20) 1 0.003 0.10 0.00009 0.00011 0.00002+ B. Gold chloride =0.0871 to 1 liter. Sodium thiosulphate, nearly ~ = 0.17012 " " Iodine, nearly ^ =0.13697 " " (21) 10 0.01 8.39 0.000871 0.000890 0.000019+ (22) 9 0.01 7.45 0.000784 0.000790 0.000006+ (23) 8 0.01 6.30 0.000697 0.000668 0.000029- (24) 7 0.008 5.50 0.000610 0.000583 0.000027- (25) 6 0.008 5.12 0.000523 0.000543 0.000020+ (26) 5 0.005 4.23 0.000435 0.000449 0.000014+ (27) 4 0.005 3.38 0.000348 0.000358 0.000010+ (28) 3 0.003 2.55 0.000261 0.000270 0.000009+ (29) 2 0.003 1.71 0.000174 0.000181 0.000007+ (30) 1 0.003 0.90 0.000087 0.000095 0.000008+ however, that when the T ^- solution of iodine is employed a correction of 0.1 cm 3 for volumes not exceeding 30 cm 3 be- VOL. II. 18 274 IODOMETRIC DETERMINATION OF GOLD. comes necessary the amount required to bring out the rose color in blank tests containing no gold. After the introduc- tion of 0.1 cm 3 of 1 -j W5 iodine into a mixture of potassium iodide and starch indicator of volume not exceeding 30 cm 3 , a single drop of the gold solution equivalent to 0.000002 grm. of gold gave a distinct rose color : before such adjustment of the solution five drops equivalent to 0.000010 of gold were needed to develop the same color. These results run on the whole as regularly as could be expected, and the use of the dilute standard solutions is obviously of advantage. In the practical application of any such process for the determination of gold, the elementary form of that metal is the natural starting-point. To get the metal into solution with chlorine water or mixed hydrochloric and nitric acids is an easy matter, but the removal of the excess of the oxidizer by evaporation without reducing some auric chloride to the aurous form is difficult. We have found, however, that the free chlorine may be removed from a solution of auric chloride, without reducing the auric salt, by treatment of the solution with ammonia in excess, boiling gently, acidifying with hydrochloric acid and heating if necessary to redissolve the precipitate by ammonia, again treating with ammonia and heating, and once more acidifying. On the second addition of ammonia no precipitation usually takes place with the amounts of gold which we have thus handled, perhaps because enough ammonium chloride has been found to hold it up. The following table contains determinations made with such a solution of pure gold leaf tested gravimetrically as to purity. Obviously, this method, which rests upon the hypothesis that sodium thiosulphate acts in the normal manner only upon the iodine set free by the interaction of gold chloride and potassium iodide, offers trustworthy means for the determination of small amounts of gold. IODOMETRIC DETERMINATION OF GOLD. 275 TABLE III. Gold chloride made by dissolving 0.0104 grm. of pure gold in the manner described and diluting to 200 cm 8 . Sodium thiosulphate, nearly j^ = 0.17012 to 1 liter. Iodine, nearly ~ =0.13697 " " Potassium iodide = 10 grm. " " Portions were treated with the potassium iodide without previous dilution. Eip. AuCl. taken. KI taken. &' Gold taken. Gold found. Error. cm 3 grm. cm** grm. grm. gnn. (1) 1 0.005 0.55 0.000052 0.000058 0.000006+ (2) 1 0.005 0.55 0.000052 0.000058 0.000006+ (3) 2 0.005 1.06 0.000104 0.000112 0.000008+ (4) 2 0.005 1.08 0.000104 0.000114 0.000010+ (5) 5 0.01 2.45 0.000260 0.000260 0.000000 (6 5 0.01 2.50 0.000260 0.000265 0.000005+ 7) 5 0.01 2.45 0.000260 0.000260 0.000000 (8) 5 0.01 2.50 0.000260 0.000265 0.000005+ (9) 5 0.01 2.50 0.000260 0.000265 0.000005+ (10) 10 0.02 4.86 0.000520 0.000515 0.000005- (11) 10 0.02 4.85 0.000520 0.000517 0.000003- (12) 10 0.02 4.90 0.000520 0.000520 0.000000 (13) 10 0.02 4.80 0.000520 0.000512 0.000008- (14) 10 0.02 4.84 0.000520 0.000516 0.000004- XXXIV THE ACTION OF ACETYLENE ON THE OXIDES OF COPPER. BY F. A. GOOCH AND DEFOREST BALDWIN.* IN a recent paper by Erdmann and Kothnerf an account is given of the formation of a peculiar, light-brown, highly voluminous substance by the action of acetylene below 250 C. upon cuprous oxide, or even (though more slowly) upon copper. The product obtained by passing acetylene during eighteen hours over 1 grm. of cuprous oxide (prepared from copper sulphate, grape sugar, and sodium hydroxide) amounted to 7 grm. and filled a space of nearly 300 cm 3 . At higher temperatures a black carbonaceous mass is the result, and at red heat (400-500 C.) carbon is deposited in graphitic condition. The light-brown fluffy material yielded cuprous chloride to hydrochloric acid, a distillate from its mixture with zinc dust possessing the characteristics of naphthene or, at higher temperature and under rapid heating, aromatic compounds among which naphthalene and a kresol were indicated. Erdmann and Kothner classify this body as a very complex but non-explosive copper acetylene (acetylen- kupfer,), and from their analyses deduce the formula C4 4 H 64 Cu s . Apart from the unusual constitution of this symbol, its most striking peculiarity is that it implies a loss of carbon, rather than hydrogen, from the acetylene in the reaction with cuprous oxide a condition of affairs which would be most remarkable in the light of Campbell's experience,^ according to which acetylene passed over palladinized copper oxide * From Am. Jour. Sci., viii, 354. t Zeitschr. anorg. Chem., xviii, 49. t Amer. Chem. Jour., xvii, 690. ACTION OF ACETYLENE ON OXIDES OF COPPER. 277 yielded water at 225-230 and carbon dioxide only when the temperature rose to 315-320 with the formation of a black deposit. Upon scrutinizing the figures of Erdmann and Kbthner with care, however, it appears that the formula given by these investigators rests upon some oversight in calculation: the ratio of carbon atoms to hydrogen atoms proves to be actually, according to the data given, 6.45 : 5.70 ; which means, of course, that the new product is deficient, as would be expected, in hydrogen (not in carbon) as compared with acetylene. As to the content of the new substance in copper, the analytical data are unfortunately ambiguous; for we note the weights found of copper oxide converted into percentages of copper without preliminary reduction. If the fault is typographical and in the analytical data, the calculated percentages of copper being correct, the average percentage of copper amounts to 15.43 : if, on the other hand, the ana- lytical data are right, the error being in their reduction, the percentage of copper amounts to 12.92. In the one case the summation of the analysis leaves a deficiency of about 1.5 per cent, and in the other of about 4 per cent, which hi either case may really represent oxygen hi the substance. This condition of matters leaves the " acetylen-kupfer " of Erd- mann and Kbthner in uncertain standing. More than thirty years ago it was noticed by Berthelot * that acetylene is polymerized by heat or decomposed partially into carbon and hydrogen, and that such action takes place more readily and at lower temperatures in presence of metallic iron with production of carbon, hydrogen and compounds different from those formed by heat alone. Moissan and Moureu f have observed the incandescence of acetylene passed over finely divided iron, cobalt, nickel, or platinum at the ordinary temperature, with production of carbon, hydrogen, and pyrogenic compounds, and have found the occasion of such behavior in the porosity of the metals employed. * Ann. Chim. [4], ix, 448. t Compt. rend., cxxii, 1240. 278 THE ACTION OF ACETYLENE It would seem natural, however, that the presence of oxygen, free or combined, may also play a considerable part in such phenomena, just as appears to be the case in the peculiar action recorded by Gruner J of carbon monoxide upon iron reduced by hydrogen, which, as Moissan has shown, is produced pure only with the greatest precaution and generally carries a large proportion of ferrous oxide. The fact that the " acety- len-kupfer " of Erdmann and Kothner is produced more easily by the action of cuprous oxide upon acetylene than by the action of metallic copper upon acetylene, suggests that it may be the oxidizing power of the cuprous oxide which gives to this reagent its peculiar activity. The question arises, therefore, as to whether the copper is in reality an essential constituent of the compound of Erdmann and Kothner. In our experiments upon the action of acetylene upon the oxides of copper (and other elements) we have conducted the gas (made in the ordinary way by the action of water on calcium carbide, and kept over water) over the oxide contained in a porcelain boat placed within a glass tube, 2 cm. in diameter and 50 cm. long, which was heated over a small combustion furnace. The glass tube was fitted at each end with a rubber stopper, one carrying a smaller tube for the introduction of the acetylene and a high-temperature thermometer so held that its bulb rested horizontally immediately over the boat containing the oxide, while the other was fitted with a water-trap. In the preh'minary experiments no attempt was made to purify the acetylene employed other than to keep it over water, or, since water is a product of its action upon oxides, to dry it : in later experiments to secure products for careful analysis it was dried and purified with care. We found that 225 C. is the temperature most favorable for the formation of the voluminous product obtained by acting with acetylene upon cuprous oxide as described by Erdmann and Kothner. At this temperature the tube is choked rapidly with the fluffy product and water forms, but, as Campbell found in his experiments upon palladinized copper oxide, no t Ann. Chim. [4], xxvi, 6. Ann. Chim. [5], xxi, 199. ON THE OXIDES OF COPPER. 279 appreciable amount of carbon dioxide is produced. The content of the product in copper varies in the sample and in different experiments, our results lying between 1.54 per cent and 24.21 per cent of the substance taken for ignition. It appeared, also, that the action of acetylene upon cupric oxide is precisely similar to that upon cuprous oxide excepting the evident reduction of the former oxide early in the action. The amount of copper in the product of such action varied in our experiments from 6.53 per cent to 21.30 per cent. In one case the experiment of re-submitting to the action of acetylene a product containing 9.34 per cent of copper was made with the result that a new growth of the substance formed which on analysis yielded 3.87 per cent of copper. A roll of copper gauze carefully reduced in hydrogen and then oxidized at one end in the outer flame of a Bunsen burner gave, when acted upon by acetylene at 225-250 C., the characteristic deposit upon the oxidized end only, the unoxidized end being merely discolored. These results go to show that, while metallic copper may at comparatively high temperatures induce the polymerization of acetylene, it is an oxidiiing action which starts at moderately low temperatures the formation of the peculiar derivatives under consideration. Thus we find that ferric oxide heated in acetylene at temperatures varying from 150 to 360, accord- ing to circumstances, darkens, glows, and gathers with evolution of heat a dark carbonaceous deposit. In the products of such action we have found the content of iron varying from 2.80 per cent to 5.86 per cent. Silver oxide, too, acts upon acetylene: thus, in one experiment, action was evident at the ordinary atmospheric temperature, and a violent explosion, which completely shat- tered the boat and scattered metallic silver upon the sides of the glass tube, followed before the temperature reached 100. In the locally violent explosion of the last experiment we have evidence of the formation in the early stage of an acetylide which is decomposed later when the temperature of dissociation is reached. In the experiments with the oxides of 280 THE ACTION OF ACETYLENE copper and iron the temperature at which the acetylene begins to act is evidently above the point at which sensitive acetylides would naturally dissociate, and we have in the observed phenomena no evidence of the formation of such compounds of copper and iron under the conditions of experimentation. In experiments (1) to (3) of the following table are given the results of the analysis of several products obtained by conducting acetylene (purified by passing through a solution of mercuric chloride in hydrochloric acid and dried over caustic potash) over pure cuprous oxide. The temperature was kept in these experiments at 225, and in the course of a half-hour the tube was choked completely by material compacted by the pressure to (1) a spongy mass of light-brown color on the exterior next the walls of the tube, (2) darker within and (3) nearly black in the bottom of the boat, where the cuprous oxide lay originally. Ezp. Weight of sub- stance taken. Found. Calculated. CO, H 2 O CuO C H Cu Oby differ- ence. (1) (2) (3) grin. 0.1170 0.2247 0.1096 grnit 0.3978 0.7489 0.3678 gnu. 0.0673 0.0979 0.0488 gnu. 0.0022 o!o045 gnu. 0.1085 0.2042 0.1003 grm. 0.0075 0.0109 0.0054 grm. 0.0018 grm. 0.0008- 0.0036 0.0003 (4) (5) 0.1360 0.1188 0.4116 0.3098 0.0579 0.0461 0.0182 0.0317 0.1123 0.0845 0.0064 0.0051 0.0146 0.0253 0.0027 0.0039 Per cent of carbon Pet cent of hydrogen Per cent of copper Per cent of oxygen (1) 92.74 (2) 90.88 (3) 91.51 (4) 82.57 (6) 71.13 6.41 4.85 4.93 4.71 4.29 1.54 3.29 0.27 10.74 1.98 21.30 3.20 100.69 100.00 100.00 100.00 In experiments (4) and (5) the substances analyzed repre- sent the products of the action of acetylene (not specially purified) on cupric oxide. The oxygen present in these products is obviously pro- portional to the amount of copper and is never more than ON THE OXIDES OF COPPER. 281 enough to be completely accounted for upon the supposition that some of the original oxide taken stills holds its oxygen. So far as the analyses show, the product of lightest color (1) contains very little copper and no oxygen ; the darkest prod- uct (3) obtained from the cuprous oxide contains oxygen corresponding to a mixture of two parts of copper with three parts of cuprous oxide; the oxygen in the products of (4) and (5) obtained by acting upon cupric oxide is approximately enough to correspond to a mixture of cuprous and cupric oxides in equal proportions. This fact, taken in connection with the great range of variation in proportion and the minimum to which the copper falls in the product, which would be least likely to include contaminating metal or oxide, suggests very strongly the probability that the oxygen present is in union with copper and that the copper is held mechanically as metal or oxide and is not the essential constituent of an organic compound. Leaving out of con- sideration, therefore, the copper and copper oxides, and calculating the composition of the products assumed to consist essentially of carbon and hydrogen, we derive the following statement: (1) (2) (3) (4) (5) Per cent of carbon 93.54 94.93 94.88 94.60 94.31 Per cent of hydrogen 6.46 5.07 5.12 5.40 5.69 100.00 100.00 100.00 100.00 100.00 These figures correspond to symbols varying from Ci 2 H 10 to nearly Ci 6 Hi , with an average approximating CnHi , the symbol of anthracene or paranthracene. The analytical data of Erdmann and Kothner point in the average to a product corresponding more nearly to the first of these symbols than to either of the others. The product is doubtless variable with the temperature and the activity of oxidation. Thus, in one experiment in which acetylene was passed over ferric oxide the action began at 365 with incandescence, as de- scribed by Moissan and Moureu,* and the analysis of the * Loc. cit 282 ACTION OF ACETYLENE ON OXIDES OF COPPER. product (carbon 91.53, hydrogen = 1.36, Fe = 5.85, O = 1.26) indicates a proportion of carbon to hydrogen about four times as great as that of the average product of action at 225 on the oxides of copper. Finally, we find no evidence that the product of the action of acetylene on the oxides of copper under the conditions of our experimentation is other than a mixture of a hydro- carbon or hydrocarbons with metallic copper or an oxide of copper, and, probably, in the darker preparations, some free carbon. XXXV NOTES ON THE SPACE ISOMERISM OF THE TOLUQUINONEOXIME ETHERS. BY WM. CONGER MORGAN.* IN an article on the "Ethers of Toluquinoneoxime and their Bearing on the Space Isomerism of Nitrogen,"! pub- lished from this laboratory, it was stated that the methyl, acetyl, and benzoyl ethers of toluquinonemetaoxime, whether formed by the action of hydroxylamine on the quinone or by nitrous acid on the corresponding cresol, showed evidence of existing in isomeric forms. Of these bodies the benzoyl ether received the most careful investigation, and by fractional crystallization, from the crude reaction-product, two portions were obtained, one readily separating from an alcoholic solu- tion in the form of yellow crystals melting at 193, while a second body, melting approximately at 144, was never obtained in a state of purity. The fact that from this lower-melting fraction, on recrystallization, a portion of the higher-melting body was always obtained, suggested the possibility that the solvent might have a tendency to cause a transition from one isomer to the other. Since, however, after repeated crystallizations from boiling alcohol a low- melting fraction was obtained, and therefore such rearrange- ment was evidently incomplete, the action of alcohol under pressure on the different fractions was investigated. From a portion of the ether melting at 139, heated for three hours in a closed tube to 120, the product melting at 193 crystallized in a characteristic form and no crystals were obtained melting * From Am. Chem. Jour., xxii, 402. t Bridge and Morgan, Am. Chem. Jour., xx, 761. This volume, p. 145. 284 NOTES ON THE SPACE ISOMERISM lower than 180. On treating the higher-melting isomer in the same way, no action was observed until the tempera- ture was raised above 150, when complete decomposition ensued. In order to ascertain, if possible, a direct method of tran- sition from one form into the other, the action of alkalies, among other reagents, was tried. No change other than a hydrolytic cleavage was observed, and this was readily completed on warming. By saponifying a fraction melting completely at 142 and treating the isolated oxime in the usual way with benzoyl chloride, the benzoyl ether was again obtained, liquefying at 193, and, although there was a trifling irregularity in the melting-point of some of the crys- tals, no low-melting fraction was isolated. On similar treat- ment of a portion melting at 193 the original ether was obtained, but none of the low-melting isomer. From these facts it is obvious that the isomer existing in a much smaller proportion in the crude reaction-products, present as the principal constituent in the low-melting frac- tions, must be regarded as the labile form, tending to go over into the stable form under the influence of boiling alcohol or during the process of a chemical reaction. This action of alcohol increases the difficulty of isolating the labile form and naturally suggests the idea of using other solvents ; but imperfect as was the separation of the two isomers, better results were obtained from an alcoholic solution than by any other means. The phenomena described in the article to which reference was previously made, have been reproduced completely in the ethers made by the action of acid chlorides on the sodium salt of the oxime (to be described later) produced by the action of pure amyl nitrite on the sodium salt of the cresol. In the resulting metathesis, as hi the product of the reaction of hydroxylamine on the quinones, there is not the possibility of the formation of a nitro-body as there is in the action of free nitrous acid on the cresols. Consequently the interfering action of an admixture of such a body with the oxime ethers OF THE TOLUdUINONEOXIME ETHERS. 285 cannot influence the results as obtained, yet the presence of a low-melting body existing hi a larger proportion in this product than in the substances formed by the agency of free nitrous acid according to the type fraction, was indisputably evidenced. The identity of the observed phenomena under varying conditions and methods of formation, thus removing the probability of an admixture of an impurity and at the same tune of a structural difference, seems to establish the hypothesis of a space isomerism in the case of the ethers of toluquinone- metaoxime. Inasmuch as the monohalogen derivatives of the quinone- oxime ethers are beautifully crystalline bodies, it was thought advisable to prepare the monochlor- and monobrombenzoyl ethers of toluquinoneorthooxime in the hope of obtaining from these well-characterized products additional evidence as to the existence of isomeric phenomena in the orthooxime ethers. Although there was the possibility of both " space" and " place " isomerism neither body offered any indication of the presence of isomers of any kind, but each appeared to be an entirely homogeneous and simple substance, liquefying sharply at a definite melting-point. EXPERIMENTAL PART. The Sodium Salt of Toluquinone-m-oxime. The sodium salt was prepared according to the general method suggested by Walker.* To a molecule of sodium alcoholate freshly prepared by dissolving metallic sodium in as little alcohol as possible, a molecule of the orthocresol was added and the solution treated with slightly more than the theoretical quantity of amyl nitrite, the whole being thoroughly mixed together. On standing in a vacuum over sulphuric acid, the sodium salt separates in fine purple crystals, which, on washing carefully with ether to remove amyl alcohol and excess of nitrite, is ready for use. It may be recrystallized from dilute alcohol if further purification is desired. This * Walker, Ber. Dtsch. chem. Ges., xvii, 399. 286 NOTES ON THE SPACE ISOMERISM salt is extremely soluble in water, much less in alcohol, and insoluble in most other organic liquids. On standing in the air it tends to decompose, turning almost black. On analysis : 0.1231 gram, dried over H 2 S0 4 , gave 0.0541 gram Na 2 S0 4 . Calculated for Fmind C 7 He0 2 NNa. Found. Na 14.48 14.26 Monobromtoluquinone-o-oxime Benzoyl Ether. This ether was obtained from the dibromide previously described* by boiling with dilute 75 per cent alcohol, during which process hydrobromic acid is split off. Being much less soluble in alcohol than the dibromide, the monobrom-body separates from the solution as soon as formed and may be obtained as a yellow crystalline powder by filtration of the cooled liquid. On analysis, 0.1450 gram, dried over H 2 S0 4 , gave 0.2774 gram C0 2 , and 0.0415 gram H 2 0. 0.1102 gram gave 0.0643 gram AgBr. Calculated for w , C M H 10 BrN0 8 . C 52.49 52.17 H 3.15 3.18 Br 24.98 24.83 On crystallizing from alcohol the ether was readily obtained in two, apparently unlike, modifications, one being long prismatic crystals, the other appearing as broad, thick, mono- clinic plates. It was at first believed that this distinction in crystal form was due to the presence of isomeric bodies, but no difference in melting-point could be found since each portion liquefied sharply at 184. Under the lens, moreover, the plates are seen to be striated parallel to one edge and have all the appearance of consisting of a number of the simple crystals united to each other, since both forms are plainly of * Bridge and Morgan, Am. Chem. Jour., xx, 776. OF THE TOLUQUINONEOXIME ETHERS. 287 the same system. Slow cooling was found to be productive of the massed crystals. Toluquinone-o-oxime Benzoyl Ether Dichloride. The benzoyl ether was dissolved in a small amount of chloroform and dry chlorine gas passed into the solution. The action is very rapid and after fifteen minutes the liquid may be allowed to evaporate and the white product, crystallized once from glacial acetic acid, melts sharply to a colorless liquid at 149 without decomposition. Fractional crystalliza- tion does not change the melting-point, except as the action of the solvent causes a slight formation of the monochlor-body. On analysis, 0.1241 gram, dried over H 2 S0 4 , gave 0.2431 gram C0 2 , and 0.0424 gram H 2 O. 0.2105 gram gave 0.1930 gram AgCl. Calculated for ,, . Found - C 53.84 53.42 H 3.55 3.80 Cl 22.72 22.67 The dichloride, like the dibromide, is slightly soluble in alcohol, readily soluble in chloroform, glacial acetic acid, and fuming nitric acid. Water precipitates the ether entirely unchanged from the two last-mentioned solvents. In crystal form it resembles very closely the dibromide, separating from a boiling acetic acid solution in short, thick, colorless, almost microscopic prisms, suggesting the orthorhombic system. In point of stability this ether, as well as the monochlor-body, far surpasses the corresponding bromine compounds. Monochlortoluquinone-o-oxime Benzoyl Ether. Analogous to the dibromide, by the action of dilute alcohol on the dichloride, hydrochloric acid is split off and a yellow monochlor-substitution-product is formed, one hour being sufficient to complete the reaction. This ether closely resem- 288 ISOMERISM OF TOLUQUINONEOXIME ETHERS. bles the monobrom-derivative in properties and in crystal form. The product as obtained from an alcoholic, solution will decompose and melt at a temperature varying ordinarily from 185-193, depending on the rapidity with which the heat is applied. " Dipped " for fifteen seconds, the crystals melt without decomposition at 200. Although carefully fractioned from alcohol, ligroin, and benzol, the substance appeared to be entirely homogeneous, and no variation in the melting-point was observed. On analysis, 0.1209 gram, dried over H 2 S0 4 , gave 0.2688 gram C0 2 , and 0.0424 gram H 2 0. 0.1035 gram gave 0.0533 gram AgCl. Calculated for -p rt ,,j C M H 10 C1N0 8 . Found< C 60.97 60.63 H 3.66 3.90 Cl 12.87 12.75 XXXVI ON THE VOLUMETRIC ESTIMATION OF CERIUM. BY PHILIP E. BROWNING.* SOME forty years ago Bunsenf showed that the oxide obtained by the ignition of cerium oxalate might be estimated volumetrically by bringing it in contact with potassium iodide and strong hydrochloric acid and determining the iodine set free. This method may be briefly described by a translation of part of the original article : " The substance to be deter- mined is weighed out in a glass flask of from ten to fifteen cubic centimeters capacity, a few crystals of potassium iodide are added, and the neck of the flask is drawn out by the aid of a blowpipe to a narrow opening. The flask is filled almost to the narrowing of the neck with hydrochloric acid which is free from chlorine or iron chloride, and a little sodium carbonate is added in order to displace the last trace of air by carbon dioxide. The flask is then closed by sealing off the neck in the blowpipe and warmed in a water bath until the cerium compound is completely dissolved, and the quantity of iodine set free is determined by iodometric analysis." The anhydrous dioxide prepared by the ignition of the oxalate or hydroxide is very slowly acted on by acids, especially when pure.f For this reason the method which Bunsen described has remained the only one adapted to the satisfactory volumetric estimation of the ignited dioxide. Two portions of the dioxide were prepared by treating the crude cerium chloride in concentrated solution with gaseous * From Am. Jour. Sci., viii, 451. t Ann. Chem. Phar., cv, 49. $ Rose, Handbuch der analytischen Chemie, Band i, 219. VOL. II. 19 290 VOLUMETRIC ESTIMATION OF CERIUM. hydrochloric acid * to saturation to remove the iron. The cerium chloride was then dissolved hi water, potassium hydroxide added in excess and chlorine gas passed until the precipitate became distinctly orange in color and the solution gave a strong odor of chlorine.! This operation was repeated until a portion of the precipitate dissolved in acid showed no didymium absorption bands when examined before the spectroscope. The whole precipitate of the dioxide was then dissolved in hydrochloric acid and the oxalate precipitated by ammonium oxalate in large excess. The precipitated oxalate was then washed thoroughly with hot water until the washings gave no test for hydrochloric or oxalic acids and ignited to the dioxide. Another portion of the dioxide was later prepared by precipitating a solution of pure cerium chloride by means of ammonium oxalate, washing and igniting as described. The dioxide in all three cases was of a light chamois color, and uniform results were obtained from the three portions. A modification of the method of Bunsen (with G. A. FOKD and F. J. HALL). Weighed portions of the pure cerium dioxide were placed in small glass stoppered bottles of about 100 cm 3 capacity, together with a gram of potassium iodide free from iodate and a few drops of water to dissolve the iodide. A current of carbon dioxide was passed into the bottle for about five minutes to expel the air, 10 cm 3 of pure strong hydrochloric acid were added, the stopper inserted and the bottle heated gently upon a steam radiator for about one hour until the dioxide dissolved completely and the iodine was set free. After cooling the bottle, to prevent loss of iodine upon removing the stopper, the contents were carefully washed into about 400 cm 3 of water and titrated with standard sodium thiosulphate to determine the amount of iodine liberated according to the well known reaction 2Ce0 2 + 8HC1 + 2KI = 2CeCl s + 2KC1 + 4H 2 + I 2 . * Dennis and Magee, Zeitschr. anorg. Chem., iii, 260. t Mosander, Phil. Mag., xxviii, 241 ; Dennis, Zeitschr. anorg. Chem., yii, 252. VOLUMETRIC ESTIMATION OF CERIUM. 291 A few blank determinations were carried through in the bottles without the presence of the cerium dioxide to determine the amount of iodine set free under these conditions. The amount obtained was uniformly equal to 0.04 cm 3 of the ^ iodine solution which was taken as the correction and applied to all the determinations. The results follow in Table I. TABLE I. Exp. CeO, taken. CeOj found. Error. grm. grm. grm. (1) 0.1000 0.0994 0.0006- (2) 0.1032 0.1034 0.0002+ (3) 0.1016 0.1017 0.0001+ (4) 0.1054 0.1041 0.0013- (5) 0.2010 0.2021 0.0011+ (6) 0.1104 0.1109 0,0005+ (7) 0.1914 0.1907 0.0007- (8) 0.1604 0.1603 0.0001- (9) 0.2146 0.2145 0.0001- (10) 0.1108 0.1099 0.0009- (11) 0.1346 0.1347 0.0001+ (12) 0.1540 0.1534 0.0006- 13 0.1976 0.1968 0.0008- (14) 0.1230 0.1240 0.0010+ (15) 0.1199 0.1202 0.0003+ (16) 0.1524 0.1528 0.0004+ (17) 0.1212 0.1211 0.0001- (18) 0.1528 0.1543 0.0015+ In order to obtain a further check upon the accuracy of the method, portions of the cerium dioxide were weighed out and placed in a distillation apparatus previously employed for similar purposes and described in former articles from this laboratory, viz.: a Voit flask, serving as a retort, sealed to the inlet tube of a Drexel wash-bottle, used as a receiver, the outlet tube of which was trapped by sealing on Will and Varrentrapp absorption bulbs. In the retort the cerium dioxide together with 15 cm 3 of water, 1 grm. of potassium iodide and 10 cm 3 of pure strong hydrochloric acid were placed. In the receiver were 100 cm 3 of water and 2 to 3 grm. of potassium iodide, and in the bulbs a dilute solution of potas- sium iodide. Before adding the hydrochloric acid a current of carbon dioxide was passed through the apparatus for some 292 VOLUMETRIC ESTIMATION OF CERIUM. minutes. After adding the acid, the liquid was boiled in the current of carbon dioxide * to a volume of 15 cm 3 , when the free iodine had almost completely left the retort and passed into the receiver, and the apparatus was allowed to cool. The iodine in the receiver was titrated directly with sodium thiosulphate, and that in the retort after dilution of the residue to about 400 cm 3 , the later amount seldom exceeding the equivalent of a few drops of -^ iodine solution. The results f oUow in Table II. Here also blank determinations were made but no correction was found to be necessary. An attempt early in the work to titrate by an alkaline arse- nite the iodine liberated, after neutralizing the hydrochloric acid, brought out some curious results which seem worthy of mention. TABLE H. Exp. CeO, taken. CeOj found. Error. grm. grm. grm. 1) 0.1028 0.1013 0.0015- 2) 0.2060 0.2055 0.0005- 3) 0.2014 0.2012 0.0002- 4 0.1716 0.1711 0.0005- 5) 0.0974 0.0972 0.0002- (6) 0.1600 0.1587 0.0013- (7) 0.1268 0.1254 0.0014- . 8) 0.1276 0.1268 0.0008- 9 0.1620 0.1612 0.0008- (10) 0.1016 0.1011 0.0005- (H) 0.1648 0.1543 0.0005- (12) 0.1352 0.1342 0.0010- In these experiments the contents of the bottles after the cerium had dissolved were carefully washed into a Drexel wash bottle upon the inlet tube of which was fused a thistle tube with a stop-cock and to the outlet tube a Will and Var- rentrapp absorption trap. In the trap a solution of potassium iodide was placed and through the thistle tube a saturated * The carbon dioxide gas was furnished by a Kipp generator from marble and hydrochloric acid of one-half strength, both of which had been boiled previously to remove all air. VOLUMETRIC ESTIMATION OF CERIUM. 293 solution of potassium bicarbonate was added to complete neu- tralization of the acid. Any iodine carried mechanically by the carbon dioxide should be held by the potassium iodide solution in the trap. After neutralization the free iodine was titrated by standard arsenious oxide solution. The results appear in Table III. TABLE IIL Exp. CeO 2 taken. CeO, found. Error. grin. gnn. gnn. (1) 0.1000 0.0987 0.0013-] 2 (3) 0.1005 0.1030 0.0981 0.1009 0.0024-1 T 0.0021- [* (4) 0.1500 0.1475 0.0025- 1 (6) 0.1030 0.1005 0.0025-' (6 0.1010 0.0988 0.0022- 7 0.1510 0.1608 0.0002- TT (8) 0.1530 0.1485 0.0045- 11. (9) 0.2045 0.2011 0.0034- (10) 0.2000 0.1958 0.0042- (11) 0.1334 0.1302 0.0032- (12) 0.1354 0.1330 0.0024- (13) 0.1312 0.1294 0.0018- (14) (15) 0.1308 0.1060 0.1277 0.1042 0.0031- 0.0018- m. (16) 0.1602 0.1567 0.0035- (17) 0.1504 0.1488 0.0016- As will be seen by the table, an average error of about 2 per cent runs through the entire set. The natural conclusion would be that the cerium dioxide contained some impurity; but, as the first, second, and third samples, very carefully pre- pared, gave the same results, it seemed necessary to look else- where for an explanation. Two possible causes suggested themselves : first, mechanical loss during the process of neu- tralization, and second, the possible formation, under the condi- tions, of iodine chloride, which if formed would in the process of neutralization probably take the form of potassium chloride, iodide, and iodate, and thus some of the originally free iodine would be withdrawn from the amount titrated. To test these theories, portions of the ^ iodine solution roughly equiva- lent to the amounts of iodine set free by 0.1 and 0.2 grms. of CeO a , were drawn off into bottles previously rilled with carbon 294 VOLUMETRIC ESTIMATION OF CERIUM. dioxide, treated with the usual amount of strong hydrochloric acid (10 cm 3 ), and after standing from thirty to forty-five minutes, neutralized and titrated as already described. The results were most interesting and seemed to show a loss of iodine closely equivalent to that shown by the results of Table III, and proportional to the amount of iodine originally present. A few determinations were carried through in the same way except that the neutralization was omitted and dilution and titration with thiosulphate substituted. These showed a loss of iodine well within the limits of such a proc- ess. The results follow in Table IV. TABLE IV. WITH ARSENIOUS OXIDE. Exp. Iodine *>. taken. Iodine , found. Error. Equivalent error on Ce0 2 . (1) (2) (3) (4) 5) (6) (7) cm 3 5.22 6.09 5.10 6.66 5.10 10.22 10.21 cms 6.07 4.97 4.97 6.47 4.97 9.97 9.97 cm 8 0.15- 0.12- 0.13- 0.19- 0.13- 0.25- 0.24- grm. 0.0026- 0.0021- 0.0022- 0.0033- 0.0022- 0.0043- 0.0041- WITH SODIUM THIOSULPHATE. g| i 5 5 10 10.08 5.01 4.99 10.02 10.12 0.01+ 0.01- 0.02+ 0.04+ 0.0002+ 0.0002- 0.0003+ 0.0007+ The action of arsenious oxide upon Cerium Dioxide (with WM. D. GUTTER). The fact that cerium dioxide is reduced by hydriodic acid suggested the possibility of the application of arsenious acid in acid solution to the same end according to the reaction 4Ce0 2 + As 2 8 = 2Ce 2 3 -f As 2 6 . The extreme difficulty with which the ignited cerium dioxide when pure dissolves in acids has already been mentioned, and for this reason it was found practically impossible to obtain VOLUMETRIC ESTIMATION OF CERIUM. 295 any results by this method. Weighed portions of the dioxide were placed in Erlenmeyer beakers with an excess of a solu- tion of arsenious oxide , 10 cm 3 of (1 : 1) sulphuric acid were added, and the boiling continued until the fuming point of the acid was reached ; but even at this point only a partial solution of the dioxide had taken place. The dark brown powder obtained by igniting the carefully washed oxalates, precipitated in acid solution by treating a solution of crude cerium chloride with ammonium oxalate or oxalate acid is very fairly soluble in acids. Mengel* has recently shown that this product contains a dioxide of praseo- didymium which acts as does cerium dioxide toward reducing agents. This fact makes the results recorded in the treatment of this ignited mixture of oxides of no value analytically, but of interest in the comparative study of the two reducing agents, arsenious oxide and hydriodic acid. Two portions of this mixture of oxides gave the following results, which agree fairly well with those of Mengel. Ezp. Amount of substance taken. Ce0 2f -HPr0 2 ?) Calculated on 0.1000 grm. 8 grm. 0.1037 0.1034 grm. 0.0530 0.0538 gnu. 0.0511 0.0520 The average of these results was taken as a standard 0.0515 grm. CeO 2 , etc., to every 0.1000 grm. of material. Three carefully weighed portions of this same material were placed in Erlenmeyer beakers with 10 cm 3 of arsenious oxide solution and 10 cm 3 of dilute (1 : 4) sulphuric acid and boiled until complete solution had taken place. The liquid was then cooled, neutralized with potassium bicarbonate and titrated with standardized iodine to determine the amount of arsenious oxide remaining, and from it the amount used in the reduction of the dioxide according to the reaction given above. The results obtained follow. * Zeitschr. anorg. Chem., xix, 67. 296 VOLUMETRIC ESTIMATION OF CERIUM. Bxp. Amount) taken. Amount CeO s found. CeO, calculated for 0.1000 grm. gnu. grin. grm. (1) 0.1005 0.0493 0.0491 (2) (3) 0.1016 0.1005 0.0494 0.0486 0.0487 0.0484 As will be seen, the results obtained by this method fall about 0.0030 grm. below the standard as obtained by the dis- tillation method, which seems to show that the arsenious oxide does not effect the complete reduction of the cerium dioxide from CeO 2 to Ce 2 O 8 . In order to study this point a little more fully and upon the pure dioxide, definite portions of a standard solution of pure cerium chloride were precipitated by ammonia in the presence of hydrogen dioxide and boiled to reduce the CeO 8 formed to the conditions of CeO 2 . The precipitated hydrated dioxide was filtered off and carefully washed until the washings gave no indication of hydrogen dioxide. The moist precipitate was then washed into a beaker, one gram of potassium iodide added and 10 cm 3 of strong HC1. The precipitate dissolved quite readily in the cold and the iodine liberated was determined by standard sodium thiosulphate. The results appear in Table V. TABLE V. Bxp. CeO 2 taken. CeO 2 found. Error. grm. grm. grm. 1) 0.1142 0.1140 0.0002- 2) 0.1142 0.1147 0.0005+ i 0.1142 0.1142 0.1152 0.1159 0.0010+ 0.0017+ (5) 0.1142 0.1152 0.0010+ 0.1142 0.1156 0.0014+ Another series of these precipitates prepared in the same way was boiled with a definite amount of arsenious acid in acid solution, as previously described in the case of the ignited dioxide. The results which are recorded in Table VI VOLUMETRIC ESTIMATION OF CERIUM. 297 show, as in the case of the ignited dioxide, an insufficient reduction of the cerium by the arsenious acid. TABLE VI. Exp. CeO 2 taken. Ce0 8 found. Error. gnn. grin. grin. (1) 0.0881 0.0370 0.0011- 2) 0.0381 0.0361 0.0020- (3) 0.1142 0.1077 0.0064- (4) 0.1060 0.1002 0.0058- The Estimation of Cerium Oxalate ly Potassium Perman- ganate (with LEO A. LYNCH). Stolba* has stated that cerium oxalate may be estimated volumetrically after the same manner as calcium oxalate by treating the washed precipitate, suspended in warm water, to which a moderate amount of sulphuric acid has been added, by potassium permanganate. As the titration proceeds the precipitate disappears and the end reaction is sharp. He also finds that the permanganate does not oxidize the cerium from the lower to the higher condition. So far as we have been able to discover, no experimental evidence has been presented to prove the correctness of Stolba's statement, and the work to be described was undertaken to furnish such evidence. The solutions used were prepared and standardized as follows : The cerium solutions were made by dissolving 10 grams of pure cerium chloride in one liter of water, and standardized by precipitating measured and weighed portions, in a faintly acid solution, with ammonium oxalate, filtering, washing, igniting, and weighing as the dioxide (CeO 2 ). A solution of potassium permanganate was prepared and stan- dardized by titration against weighed amounts of ammonium oxalate. A solution of ammonium oxalate was made and its value determined by titrating measured amounts against * Sitzungsber. d. kgl. bohm. Gesellsck. d. Wissenschaften v. 4. Juli, 1879; Zeitschr. anal. Chem., xix, 194. 298 VOLUMETRIC ESTIMATION OF CERIUM. potassium permanganate. Definite portions of the cerium solution were drawn from a burette and after diluting with water from 100 to 200 cm 3 a definite amount of ammonium oxalate was added, care being taken to have an excess over the amount necessary, and the whole warmed to insure a more crystalline precipitation.* The precipitate was then filtered off on paper and carefully washed, the filtrate and washings being collected in a liter Erlenmeyer flask and set aside for future use. The precipitate was treated with about 10 cm 3 of hot (1 : 4) sulphuric acid, which dissolved it completely, if not at first, by running it through the filter a few times, and the solution and washings were collected hi another liter flask. The total volume of liquid was made up to about 500 cm 3 , warmed to about 70 C. to 80 C. and titrated with potassium permanganate to the appearance of the faint blush of color showing the complete oxidation of the oxalic acid. The filtrate from the cerium oxalate containing the excess of oxalic acid was diluted to 500 cm 3 , acidified with 10 cm 3 of dilute (1 : 4) sulphuric acid, one gram of manganous sulphate added to prevent the interfering action of the free hydrochloric acid upon the estimation of the oxalic acid,f and titrated with potassium permanganate after the same manner as the dissolved precipitate. A definite quantity of ammonium oxalate having been originally taken, it became possible, by subtracting from it the amount obtained, to derive the measure of the oxalate used in the precipitation of the cerium oxalate. By this procedure, it will be observed a check was made upon the results obtained by the titration of the precipitate. In experiments (1) to (6) the cerium oxalate was thrown down in neutral solution, in experiments (7) to (10) in acid solutions. The treatment of the filtrate in experiment (1) was made without the presence of the manganous sulphate. The results recorded in Table VII seem to uphold the statement of Stolba. * As shown by the table, the precipitation was sometimes in neutral, some- times in faintly acid solution. t Gooch and Peters, Am. Jour. Sci., vii, 461. This volume, p. 222. VOLUMETRIC ESTIMATION OF CERIUM. 299 TABLE VII. Bxp. Amount found. Amount found. Amount taken. Calculated as Error. Calculated as Error. Calculated CeCl 3 . Calculated as CeCl 8 . Calculated as as CeCl 3 . Treatment of CeCl 3 . Treatment of CeCl 8 . precipitate. filtrate. grm. grm. grlu. grm. grm. (1 [0.1091 0.1087 0.0004- 0.1023 0.0068-] (2 0.1091 0.1103 0.0012+ . g 3 0.1091 0.1087 0.0004- 0.1087 0.0004- 4 0.1364 0.1373 0.0009+ 0.1391 0.0027+ 1: 0.1364 0.2182 0.1367 0.2202 0.0003+ 0.0020+ 0.1367 0.2206 0.0003+ 0.0024+ (7) 0.1091 0.1087 0.0004- (8) 0.1519 o.is35 0.0016+ 0.1535 0.0016+ (9) 0.1364 0.1367 0.0003+ 0.1367 0.0003+ (10) 0.2182 0.2183 0.0001+ 0.2183 0.0000 XXXVII ON THE ESTIMATION OF THALLIUM AS THE CHROMATE. BY PHILIP E. BROWNING AND GEORGE P. HUTCHINS.* CROOKES has shown f that the chromate precipitated by the addition of potassium dichromate to an alkaline solution of a thallous salt has the constitution of a neutral salt and is very insoluble in water 100 parts of water at 100 C. dissolving about 0.2 parts and at 60 C. about 0.03 parts. He has also made use of this reaction J to effect a rough separation of thallium from cadmium. The object of this paper is to describe some work directed toward a study of the application of this reaction to the gravi- metric estimation of thallium and the best conditions under which to effect the precipitation. For the work a solution of thallous nitrate was made by dissolving 10 grms. in water and making up to a liter. The standard was determined by taking measured and weighed portions from a burette, precipitating with a slight excess of potassium iodide, agitat- ing to bring about a good separation of the thallous iodide, and allowing to stand until the supernatant liquid was clear. The iodide was then filtered off upon an asbestos felt contained in a perforated platinum crucible, the whole having been previously ignited and weighed, washed with a mixture of alcohol and water, dried over a low flame and weighed to a constant weight. The filtrate, which together with the washings seldom amounted to more than 50 cm 3 , was evaporated to dryness on a water bath, a few drops of water * From Am. Jour. Sci., viii, 460. t Chem. News, viii, 255. J Chem. News, vii, 145. ESTIMATION OF THALLIUM AS CHEOMATE. 301 added and thus the small amount of thallous iodide which had been dissolved recovered. This small insoluble residue, which seldom amounted to one milligram in weight, was filtered off, washed and weighed as previously described. Baubigny * has shown this method to give very satisfactory results, and the uniformity of our determinations certainly confirms his statements. For convenience in the calculations of results to be de- scribed later, a solution of potassium dichromate of definite strength was made. Portions of the thallium solution were drawn from a burette into test tubes of about 100 cm 3 capacity and weighed as a check on the burette reading. The solution was heated to about 70 C. to 80 C. and a few drops of ammonia or potassium carbonate solution added to distinct alkalinity. A definite amount of the potassium dichromate in solution was delivered from a burette, care being taken to have an excess, and the contents agitated to bring about a good separation of the precipitated chromate. After the precipitate had completely settled out and the solution had become cold the chromate was filtered upon asbestos, as described above, dried over a low flame and weighed to a constant weight. The filtrates from several determinations were evaporated to a small volume and in one or two cases a residue amounting to a few tenths of a milligram was obtained, but no appreciable quantity of dissolved chromate was thus recovered. It was found that when the precipita- tion was made in the cold the chromate did not flock well, but remained partly in a finely divided condition which would run through the felt and require repeated filtration. The addition of ammonium nitrate before precipitation prevented this largely, even in the cold, bat the best results were obtained by warming the solution before precipitation and using potassium carbonate rather than ammonium hydroxide. The results follow in Table I. An attempt was made to estimate the thallium volumetri- cally by determining the amount of chromate in the filtrate * Chem. News, Ixiv, 239. 302 ESTIMATION OF THALLIUM AS CHROMATE. TABLE I. Brp. TINOs taken. Calculated as 11,0. TliCrO 4 found. Calculated as Tl a O. Error. Calculated as T1 2 0. grm. grm. grm. (1) 0.0796 0.0791 0.0005- (2) 0.0792 0.0788 0.0004- (3) (4) 0.0792 0.1188 0.0786 0.1177 0.0006- 0.0011- (5) 0.1192 0.1186 0.0006- (6) 0.1185 0.1178 0.0007- (7) 0.1190 0.1185 0.0005- (8) 0.1189 0.1183 0.0006- (9) 0.1196 0.2000 0.0004+ (10) 0.1196 0.2005 0.0009+ (ID 0.1173 0.1173 0.0000 (12) 0.1171 0.1163 0.0008- from the thallous chromate, and by difference (the potassium dichromate originally added being known) the amount combined with the thallium in the precipitate. The method used to determine the standard of the dichromate solutions and also the chromate remaining in the filtrate was described by one of us in a previous paper from this laboratory.* According to this procedure the filtrate from the thallous chromate containing the excess of alkali chromate was acidified with sulphuric acid, a definite amount of a solution of arsenious oxide, previously standardized, was added and the whole was allowed to stand a few moments until the change from the yellow to the bluish green showed the complete reduction of the chromic acid. Potassium bicarbonate was added to distinct alkaline reaction and the arsenious oxide remaining was determined by titration with -standard iodine solution. The amount of the arsenious oxide oxidized is of course the measure of the chromate in the solution. The amount of chromate in the original solution used being known, by subtracting the amount thus determined in the filtrate the chromate in combination with the thallium may be readily found, and from it the thallium estimated. Filtrates from certain precipitates, of which the determinations are given in * Am. Jour. Sci., i, 35, 1896. Volume I, p. 344. ESTIMATION OF THALLIUM AS CHROMATE. 303 Table I, were treated in this way, and the results, indicated by corresponding numbers, follow in Table II. TABLE II. Bxp. TINO 3 taken. Calculated as Tl 8 CrO 4 found. Calculated as Error. Calculated as TlaO. Tl,0. TljO. grm. grm. grm. (5) 0.1192 0.1198 0.0006+ (8) 0.1189 0.1205 0.0016+ (9) 0.1196 0.1180 0.0016- (10) 0.1196 0.1192 0.0004- (11) 0.1173 0.1182 0.0009+ (12) 0.1171 0.1190 0.0019+ The method cannot be very accurate on account of the high molecular weight of thallium oxide as compared with that of the chromic acid determined, but the results check fairly well with the gravimetric method. xxxvm THE ETHERS OP ISONITROSOGUAIACOL IN THEIR RELATION TO THE SPACE ISOMERISM OF NITROGEN. BY JOHN L. BRIDGE AND WM. CONGEE MORGAN* WHEN the presence of isomerism in the quinoneoximes was first noted by one of us,f and the phenomenon exhibited by the two isomeric ethers then described was shown by Kehrmann J to be due to no structural differences, but to necessitate the assumption of a spatial arrangement about the nitrogen atom, according to the theory of Hantzsch and Werner, the plan was adopted of studying the various substituted quinoneoximes, by means of their acyl and alkyl ethers, with reference to this phenomenon. Accordingly we have investigated the toluquinoneoximes, both ortho and meta, producing them by the action of nitrous acid on the cresol as well as by hydroxylamine on toluquinone, and found that, whereas there is abundant evidence for the existence of stereoisomeric bodies in the metaoxime ethers, in the derivatives of the orthooxime, all such indication is wanting. The significance of this observation is furthermore increased by the fact that all oximes, in which isomerism has been reported, may be considered as derivatives of metasubstituted quinones, and it seemed not improbable, therefore, that these observations might be formulated into a general rule regarding the appearance of isomerism in the quinoneoximes. In the course of his investigation of the properties and reactions of isonitrosoguaiacol, among other derivatives, Pfob || * From Am. Chem. Jour., xxii, 485. t Ann. Chem. (Liebig), cclxxvii, 79. t Ann. Chem. (Liebig), cclxxix, 27. Am. Chem. Jour., xx, 761 ; xxii, 402. This volume, pp. 145, 283. II Monatsh. Chem., xyiii, 467. THE ETHERS OF ISONITROSOGUAIACOL 305 made the methyl and acetyl ethers, but did not announce the observation of any cases of isomerism. Isonitrosoguaiacol may be considered as the metamethoxyquinoneoxime, hence this metasubstituted quinoneoxime presented conditions differing widely from the other members of the same series above mentioned. Moreover, because of its close relationship to toluquinonemetaoxime, from which it differs only by the interposition of an oxygen atom between the ring and the methyl group, it seemed possible that isomeric modifications of the ethers might exist, which had been overlooked by the former investigator. When, furthermore, Rupe * made no mention of such appearance in his research on isonitrosoguaia- col, it seemed advisable to undertake anew the investigation of this body with the special purpose of discovering such isomerism, if possible, and to couple with it an investigation of the orthomethoxyquinoneoxune or isonitroso derivative of the monomethyl ether of resorcin, in order by this means to be able to parallel in these closely analogous bodies the experiments with the ortho- and metacresols. With this idea, the work of Pfob and Rupe was carefully repeated, so far as it pertained to the question in hand, but, aside from minor differences, our results served only to corroborate the testimony of these investigators. New derivatives, to be described later, were prepared in the hope that these bodies might show some variations leading to the discovery of isomeric modifications, but each appeared to be entirely homogeneous and no evidence for isomerism could be found. These results are of course only negative and do not disprove the existence of space isomers in the same bodies, yet the same methods, which gave very positive evidence of their presence hi other quinoneoxime ethers, were used to detect them in this instance. Aside from the difficulty of obtaining the pure monomethyl ether of resorcin in any considerable quantity, because of poor synthetical processes and inefficient methods of separation, inasmuch as preliminary experiments pointed to a multiplicity * Ber. Dtsch. chem. Ges., xxx, 2444. VOL. ii. 20 306 THE ETHERS OF ISONITROSOGUAIACOL. of products in the reaction with nitrous acid such as Kietaibl * found with the monoethyl ether, it was thought inadvisable in the light of the results obtained with the meta body, to con- tinue the work on the ethers of orthomethoxyquinoneoxime. Work along the general line will be continued, and the results of experimentation with mononitrosoresorcin will soon appear. EXPERIMENTAL PART. Isonitrosoguaiaeol and Salts. The isonitrosoguaiacol used in the investigation was pre- pared both by the method of Pfob, working with nascent nitrous acid in alcoholic solution, and also by the general method for the formation of the sodium salts of isonitroso bodies sug- gested by Walker, f To a concentrated alcoholic solution of sodium alcoholate, guaiacol is added in sufficient quantity to form the sodium salt by the resulting metathesis, then, to this solution of sodium guaiacol, slightly more than the theoretical quantity of amyl nitrite is added and, after thorough mixing, the liquid is allowed to stand over sulphuric acid for twenty- four hours, when the bright olive-green crystalline sodium salt of isonitrosoguaiacol separates. After washing thoroughly with ether, pulverizing and rewashing, the salt may be used directly for the preparation of derivatives, or, if further purifi- cation is desirable, it may be dissolved in water, acidified with hydrochloric acid, and the filtered and dried product dissolved in ether and shaken with animal charcoal, when, upon evapo- ration, the pure isonitrosoguaiacol crystallizes. Of the above methods of preparation the latter is much to be preferred, although Rupe mentions it unfavorably because of poor yields and impure products. He advocates the use of ethyl nitrite in a closed tube ; but on trial we were unable to obtain the quantitative yields which he reports and the seventy per cent yield which the amyl nitrite gives, makes this method quite equal, in efficiency as well as purity of reaction-product, to the other more tedious process. The silver salt, formed from the sodium salt by treating the aqueous solution with a slight excess of silver nitrate, comes * Monatsh. Chem., xix, 536. t Ber. Dtsch. chem. Ges., xvii, 399. THE ETHERS OF ISONITROSOGUAIACOL. 307 down as a brown gelatinous precipitate, which becomes crystal- line on gently warming, or may be obtained in crystalline form at once by heating the separate solutions to 50 C. before mix- ing. It is a very unstable salt, the dry product decomposing with a very gentle heating. Isonitrosoguaiacol Benzoyl Ether. The sodium salt formed as above was dissolved in as little water as possible and four or five tunes its volume of alcohol added. This solution was thoroughly shaken with a slight excess of benzoyl chloride, added drop by drop. The reaction is immediate and the ether soon begins to come out of the solution in almost pure condition. Recrystallized from alco- hol, it separates in straw-colored, branching crystals, which melt sharply at 188 C. when " dipped " for ten seconds. Heated gradually from normal temperatures, it begins to de- compose at 175 C. and liquefies at 185-188 C., the temper- ature depending on the rapidity with which heat is applied. Fractional crystallization from alcohol or other solvents did not essentially change the melting-point, nor were any differ- ent phenomena observed when the isonitrosoguaiacol was made by the acid reaction. This ether dissolved readily in chloro- form and glacial acetic acid, much less in benzene and ligroin, and is practically insoluble in ether and carbon disulphide. On analysis : 0.1100 gram of the substance, dried over H 2 S0 4 , gave 0.2632 gram C0 2 , and 0.0428 gram H 2 0. 0.1205 gram of the substance gave 5.45 cm 8 K at 15 C. and 772 mm. pressure. Calculated for d C 14 H U 4 N. Foun 0.0005- 0.0000 0.0000 0.00024- 0.0005 0.0004- * Loc. cit. VOLUMETRIC ESTIMATION OF COPPER. 355 TABLE m (continued). Exp. Cu taken as CuSO 4 . Element from which copper was separated. Oxalic acid. HN0 3 'So 8 )'' Volume at precip- itation. Cu found. Error. Sn taken as SnClj + HCl. I. grm. (50) (51) (51 a) (52) (53) (54) (55) grin. 0.1590 0.1590 0.1590 0.1590 0.1590 0.1590 0.1590 grin. 0.0468 0.0936 0.0936 0.0936 0.1873 0.2809 0.2809 grm. 2.0 2.0 2.0 2.0 2.0 2.0 3.0 cm 5.0 6.0 5.0 5.0 5.0 5.0 5.0 cm 8 65 60 60 60 65 70 75 grm. 0.1681 0.1603 0.1591 0.1694 0.1603 0.1914 0.1988 grm. 0.0009- 0.0013+ 0.0001+ 0.0004+ 0.0013+ 0.0324+ 0.0398+ Sn taken as SnCl 4 . K. (56) (57) (58) (59) 0.1590 0.1590 0.1590 0.1590 0.10 0.10 0.20 0.50 2.0 2.0 2.0 2.0 5.0 5.6 6.0 65 55 55 60 0.1581 0.1565 0.1577 0.1562 0.0009- 0.0026- 0.0013- 0.0028- CuO taken as CuSO 4 . FeO 3 taken as Fe(N0 3 ) 3 . L. CuO found. (60) (61) (62) (63) (64) (65) (66) 0.1990 0.1990 0.1990 0.1990 0.1990 0.1990 0.1990 0.136 0.272 0.364 0.544 0.272 0.544 0.218 2.0 2.0 2.0 2.0 2.0 2.0 2.0 6.0 6.0 5.0 5.0 2.6 2.0 60 60 60 65 60 60 65 0.1987 0.1983 0.1988 0.1971 0.1995 0.1998 0.1999 0.0003- 0.0007- 0.0002- 0.0019- 0.0005+ 0.0008+ 0.0009+ ZnO taken as ZnSO 4 . M. (67) (68) (69) (70) 0.1990 0.1990 0.1990 0.1990 0.028 0.057 0.057 0.085 2.0 2.0 2.0 2.0 5.0 6.0 6.0 6.0 60 65 65 70 0.2007 0.2008 0.2008 0.2036 0.0017+ 0.0018+ 0.0018+ 0.0045+ Separation from Arsenic, in Both Conditions of Oxidation. For the separation of arsenic, arsenious oxide dissolved in sodium carbonate, and di-hydrogen sodium arseniate were the forms of arsenic used. The results are accurate and are given in sections G and H of the table. In experiments (38)-(40) and (44)-(45) no nitric acid was added. While the presence of the nitric acid is not necessary for the separation of the copper from the arsenic, still the filtration in the absence of 356 VOLUMETRIC ESTIMATION OF COPPER. the nitric acid is so slow as to be objectionable. The presence of the nitric acid causes the precipitate to come down in a coarser condition, and in such condition it filters easily and is capable of being washed quickly. Separation from Tin, in Both Conditions of Oxidation. For the separation of copper from tin a preparation of stannous chloride (20 cm 3 giving 0.3746 grm. metallic tin by the battery) containing sufficient hydrochloric acid to prevent deposition of oxy-salts was used. The solution of stannic chloride contained 1.0 grm. metallic tin to every 10 cm 3 , and was used without hydrochloric acid. The results of the work are found in sections I and K of the table. The experiments go to show that while copper may be separated from small amounts of tin as stannous chloride yet there is a limit to the amount of tin which may be present. One-tenth of a gram of metallic tin is the largest amount that can be present, with 0.15 gm. copper oxide taken as the sulphate, without significant error. Practically the same statement can be made of the separation of copper from tin taken as stannic chloride. Experiment (57) shows a greater loss of copper when the nitric acid is omitted. Separation of Copper from Iron. A solution of ferric nitrate was used for the work on the separation of copper from iron. Low results were obtained when a solution of ferrous or ferric sulphate was used as the source of iron. The results of the experiments are recorded in section L of the table, and show that 0.20 grm. copper oxide as the sulphate may be separated from 0.2-0.3 grm. iron oxide taken as the nitrate. In experiment (64) a good result was obtained when no nitric acid was present, save that added in combination with the iron. A comparison of experiments (63) and (65) shows that it is best to avoid the use of large amounts of nitric acid when the larger amounts of ferric nitrate are present. VOLUMETRIC ESTIMATION OF COPPER. 357 For a practical application of the above separation of copper from iron a convenient amount of finely ground chalcopyrite (0.5 gnn.) was roasted 2-3 hours in a porcelain crucible until all sulphur was driven off, washed into a beaker, strong nitric acid, about 5 cm 3 , was added and, with the beaker covered, allowed to evaporate slowly on a hot plate, nearly to dryness. A little dilute nitric acid was added, the solution was filtered, the residue was washed with water containing dilute nitric acid, the filtrate, about 50 cm 3 in volume, was precipitated with 2.0 grm. oxalic acid, and the precipitate was estimated after standing 12-16 hours, as previously described. The washing with water acidified with nitric acid is important, because the finely ground ferric oxide remaining undissolved passes through the filter when washed with water alone, but gives no trouble if the water be acidic. The results of two estimations are here given. Chalcopyrite. Copper found by battery. Copper found by oxalate method. Difference. grm. % % % 0.5000 31.00 30.92 0.08- 1.0000 31.00 31.25 0.25+ Separation of Copper from Zinc. The separation of copper from zinc was not altogether successful owing to the tendency of the zinc oxalate to come down with the copper oxalate. Some experiments are given in section M of the table. The separations of copper from bismuth and antimony were unsuccessful. The work may be briefly summarized as follows : Copper exceeding in amount the equivalent of 0.0128 grm. of the oxide to 50 cm 3 of solution as the sulphate may be separated completely, even in the presence of a moderate amount of strong nitric acid, by the addition of sufficient amount of oxalic acid. Copper may be separated from cadmium, arsenic, iron, and small amounts of tin, when precipitated by oxalic acid in 358 VOLUMETRIC ESTIMATION OF COPPER. a volume of 50 cm 3 containing 5 cm 3 strong nitric acid. Inasmuch as the completeness of precipitation of the copper depends upon the presence of a certain minimum amount of the copper salt this method is not applicable when the amount of copper falls below 0.0128 grm. of the oxide to 50 cm 3 of solution. XLVI THE SULPHOCYANIDES OF COPPER AND SIL- VER IN GRAVIMETRIC ANALYSIS. BY K. G. VAN NAME * Cuprous Sulphocyanide. As early as 1854 attention was drawn by Rivotf to the possibility of estimating copper gravimetrically by weighing as cuprous sulphocyanide, and to the advantages which the process afforded in separating copper from other metals. Rivot's procedure consisted in dissolving the substance to be analyzed in hydrochloric acid, reducing copper with hypo- phosphorous or sulphurous acid, and precipitating with potas- sium sulphocyanide. The precipitate dried at a moderate temperature was weighed as cuprous sulphocyanide and then as a control converted by ignition with sulphur into cuprous sulphide and weighed in that condition. In his well known work upon quantitative analysis Fre- senius in one place J denies the practicability of the direct weighing of copper as cuprous sulphocyanide on account of the tendency of the latter to hold water even when heated to the temperature of incipient decomposition. As authority for this statement he cites Claus, who found 3 per cent of water in the precipitate after drying at 115, and Meitzendorff, who gave the percentage of water under the same conditions as 1.54. On a later page of the same volume, || however, Fresenius, after a trial of the process which gave 99.66 per cent of the * From Am. Jour. Sci., x, 451. t Compt. rend., xxxviii, 868. J Quant. Anal., 6. Aufl., i, 187. L. Gmelin, Handbuch, iv, 472. || Quant. Anal, 6. Aufl., i, 335. 360 SULPHOCYANIDES OF COPPER AND SILVER theory for copper, concludes that the method is practicable although apt to give low results, particularly in the presence of free acid. The process was again recommended in 1878 by Busse,* who had employed it for the estimation of copper, both alone and in the presence of iron, nickel, zinc, and arsenic, obtaining results very near the theory and plainly comparable with the figures obtained by afterwards igniting the cuprous sulphocyanide with sulphur in hydrogen. In spite of the evident advantages for certain purposes of Rivot's method over other modes of determining copper, it has never come into general use. The chief reason for this has apparently been the difficulty and inaccuracy attendant upon the weighing of the precipitate upon dried paper filters, a process which can hardly be depended upon unless managed with extreme care. In the experiments to be described this difficulty was avoided by performing the filtering and weighing upon asbes- tos in a perforated platinum crucible. The method of con- ducting a determination was as follows : A suitable quantity of a standard copper sulphate solution was run from a burette, diluted to a convenient volume, a few cubic centimeters of a saturated solution of ammonium bisulphite added, and the copper precipitated by an excess of ammonium sulphocyanide. The precipitate was allowed to settle, collected upon asbestos in a weighed crucible, washed with cold water, and dried at 110 until no further loss of weight took place. In Table I are given the results of a number of determi- nations made in this way. The copper sulphate solution was made up exactly decinormal and the standard confirmed electrolytically. As the ammonium sulphocyanide solution was slightly above decinormal, 13 cm 3 represent a small excess (about one' cubic centimeter) above the amount theoretically required to precipitate 25 cm 3 of the copper sulphate solution. The ammonium bisulphite, which had been recently prepared by saturating aqueous ammonia with sulphur dioxide, was * Zeitschr. anal. Chem., xvii, 63. IN GRAVIMETRIC ANALYSIS. 361 always used in sufficient quantity to give the liquid a strong and permanent odor of the latter. TABLE I. 25 cm 8 of ^ CuSO 4 solution, equivalent to 0.0795 grm. Cu, taken for each experiment. Exp. H 2 S0 4 . concentrated. HNH 4 SO S sat. sol. NH 4 SCN approz. ff Final volume. Time of standing. Cu found. Error. TIT cms cm 3 cm 3 cm 3 hrs. grui. grm. (1) None. 5 13 68 | 0.0795 0.0000 (2) None. 3 13 66 48 0.0793 0.0002- (3) None. 3 25 78 i 0.0796 0.00014 (4) None. 3 25 78 12 0.0796 0.0001+ (5) 1.5 10 13 85 12 0.0792 0.0003- (6) 1.6 8 13 105 48 0.0785 0.0010- (7) 1.5 3 25 85 4 0.0783 0.0012- (8) 1.5 5 25 85 21 0.0795 0.0000 (9) 5 5 25 85 3 0.0797 0.0002+ (10) 15 10 25 115 21 0.0793 0.0002- HCl concentrated. (11) 10 6 25 100 20 0.0795 0.0000 (12) 25 10 25 100 28 0.0784 0.0011- When there is no free acid present the time of standing before nitration and the amount of the excess of ammonium sulphocyanide are practically without effect, as experiments (1) to (4) of the table show. Experiments (5) to (10) were carried out in the presence of various amounts of free sulphuric acid up to 12 per cent of the total volume of liquid. The acid, at least within this limit, does not exert a sufficient solvent effect upon the cuprous sulphocyanide to interfere materially with the accuracy of the process, but it retards the precipitation, making it necessary to increase the time of standing before filtering in proportion to the amount of acid present. In several of these determina- tions the precipitation was visibly incomplete even after sev- eral hours' standing. This effect of the acid, however, hardly shows in the results of the table because the standing was always prolonged until the copper appeared to be all down before filtering. 362 SULPHOCYANIDES OF COPPER AND SILVER The low result of Experiment (7) was probably due chiefly to incomplete precipitation, although (9) shows that even with a much larger amount of acid precipitation may be complete within three hours. In general, however, it is safer to allow ample time (twelve hours or more) for the precipitation when there is much free acid present. Comparison of Experiments (5) and (6), for which only a bare excess of ammonium sulphocyanide was used, with (7) to (12) shows an apparent advantage in the larger excess in the presence of acid. Hydrochloric acid, judging from the results of (11) and (12), has no greater disturbing influence than sul- phuric acid, although in (12), where the concentrated acid constituted one-fourth of the entire volume, there was appar- ently a slight solvent action. The filtrate from this determi- nation when concentrated to about 25 cm 8 and treated with potassium ferrocyanide gave a strong test for copper, as did also the filtrate from (6). Several of the other nitrates were tested in the same way, but none showed more than an insig- nificant trace of copper. The nitrate of (7), however, was not tested. Table II contains the results of a series of experiments con- ducted as before, except that larger amounts of copper were employed. The copper sulphate solution was approximately f and standardized by the battery. The solution of ammo- nium sulphocyanide was the same previously used and a con- siderable excess was employed in every determination. More TABLE II. Bxp. Cu taken. H 2 S0 4 concen- trated. NH 4 SCN approx. ff TS' Final volume. Cu 2 S 2 (CN) 2 found, calcu- lated as Cu. Error. Cuin filtrate. | (3) (4) grm. 0.3175 0.3176 0.3175 0.3175 cms None. None. None. 10 cm 8 60 60 60 100 cm 8 500 500 500 500 grm. 0.3176 0.3177 0.3176 0.3176 grm. 0.0001+ 0.0002+ 0.0001+ 0.0000 None. None. None. None. (5) 0.3175 HC1 cone. 100 600 0.3165 0.0010- Distinct. 20 IN, GRA VIMETRIC ANAL YSIS. 363 than twice the amount theoretically required was used in every case where free acid was present, and at least twenty hours were allowed for the precipitation, which was made in cold, and as the table shows, rather dilute solutions. If the solu- tion is too concentrated the copper is apt to be thrown down in a finely divided condition, making it hard to filter. The time required to dry the cuprous sulphocyanide at 110 is in general from two to three hours. Heating much longer than this is not to be recommended, as a gradual increase in weight begins to take place, as is shown by the following example, which gives a series of weights of the same precipi- tate at different stages. Cu,S 8 (CN),. Calculated as Cu. gnu. gnu. After 2 hours at 110 . . . 0.6060 0.3167 4 " ... 0.6059 0.3167 19 " . . . 0.6067 0.3171 23 " ... 0.6069 0.3172 This tendency to increase in weight is, however, usually less marked than in the above example, and in any case need not interfere materially with the accuracy of the process unless the drying is prolonged far beyond the necessary length of time. The method is easily handled and, as the results of Tables I and II show, is capable of considerable accuracy. From the nature of the process it is evident that it is much less likely to be interfered with by the presence of other metals than the other gravimetric methods for copper, and may therefore be directly applied with good results in many cases where the use of the electrolytic or the oxide method would involve a pre- vious separation. Silver Sulphocyanide. The sulphocyanide of silver, unlike that of copper, is solu- ble in an excess of ammonium or alkali sulphocyanides and this fact prevents the use of the latter to precipitate silver for gravimetric estimation. The reverse process, however, the precipitation of a soluble sulphocyanide by an excess of silver 364 SULPHOCYANIDES OF COPPER AND SILVER nitrate, as will be shown by the experiments to be described, furnishes a convenient means of standardizing sulphocyanide solutions and in general for estimating sulphocyanic acid. When freshly precipitated the sulphocyanide of silver resem- bles the chloride in appearance, but when allowed to stand a few hours becomes finely granular and is very easily filtered and washed. It may be safely dried to a constant weight upon an asbestos filter at 110 -120, but at a somewhat higher tem- perature is decomposed, leaving a residue of silver sulphide. The determinations which are tabulated below were made as follows. Portions of 25 cm 3 of an approximately decinormal solution of ammonium sulphocyanide were measured from a burette, diluted with 100 cm 3 of water and silver nitrate added in excess. The precipitate was collected upon asbestos in a platinum crucible, washed with cold water and dried to a con- stant weight at 115 the drying requiring usually between two and three hours. The filtering is facilitated by allowing a few hours for the precipitate to settle ; but this is by no means essential, as it is easy with a little care to obtain a clear filtrate even when the filtering is performed at once. The solution of ammonium sulphocyanide was prepared from a pure salt, especially tested and found free from choride. This point is of importance, as chlorine is a common impurity and its presence in any considerable quantity will vitiate the results. In order that the effect of varying the excess of silver might be investigated, an approximately decinormal solution of silver nitrate was titrated against the ammonium sulpho- cyanide and the ratio between the two solutions determined. This silver nitrate solution was used for the first five determi- nations of Table III. For the rest the quantity of silver nitrate was not measured but regulated by the eye alone, thus making the conditions the same as would be the case in practical use of the method. These results are as uniform as could be expected, considering the variations which would be produced by even very small GRAVIMETRIC ANALYSIS. 365 TABLE in. Final volume of liquid 150 cm 8 . 25 cm 8 of NH 4 SCN sol. equivalent to 25.15 cm 8 of AgN0 8 sol. Exp. NH 3 SCN. AgN0 8 . Excess of AgNO s . AgSCN found. cm 3 cm 3 cm 3 grm. (1) 25 25.3 0.15 0.4372 (2) 25 25.3 0.15 .0.4376 8 25 25 25.4 25.4 0.25 0.25 0.4373 0.4375 (5) 25 30.4 5.25 0.4382 (6) 25 Kough excess. 0.4366 (7) 25 Rough excess. 0.4381 (8) 25 Rough excess. 0.4873 (9) 25 Rough excess. 0.4372 (10 25 Rough excess. 0.4369 errors in measuring out 25 cm 3 of decinormal sulphocyanide solution. It is moreover clearly shown that there is no difference in the results whether a bare excess or a moderately large excess of the silver nitrate is used. The mean of the values in the last column is 0.4374, which is equivalent to 0.2006 grm. of ammonium sulphocyanide for every 25 cm 3 of the solution. The standard of the sulphocyanide solution was also determined volumetrically by Volhard's process. The mean of four titrations carried out with great care against a standard silver nitrate solution gave as the standard 0.2003 grm. of ammonium sulphocyanide for 25 cm 3 of solution. This differ- ence between the standards as determined by the two methods (one part in 670) is much less than the variations which frequently appear between successive determinations by Volhard's method, under like conditions as to strength of solutions and amounts used. It is about equal to the error that would be produced in a single volumetric determination by a mistake of one drop in measuring one of the solutions, or of one-half drop in the same direction on each. It is therefore evident that the standard of a sulphocyanide solution obtained in the above way may be applied directly to 366 SULPHOCYANIDES OF COPPER AND SILVER the estimation of unknown amounts of silver by Volhard's method without sensible error. To remove a possible doubt as to whether the silver sulphocyanide dried at 115 was entirely free from water, a number of electrolytic determinations of the silver contained in the previously weighed precipitates of Table III were made in the following way. The perforated platinum crucible containing the silver sulphocyanide and asbestos was hung in a loop of heavy platinum wire and served as the anode. For the cathode a deep platinum dish of about 200 cm 3 capacity was used. An ammoniacal solution of potassium cyanide was employed as the electrolyte and gave the best results when made up by dissolving 2 grm. of potassium cyanide in 15 cm 3 of strong ammonia and 15 cm 3 of water. The crucible which served as the anode was filled with this solution in full strength, and the remainder was put into the platinum dish and diluted to the required volume with water. In this medium the silver sulphocyanide is slowly dissolved and diffuses through the asbestos felt into the space between the electrodes where the silver is deposited in the usual way. This diffusion, is, however, aided but little if at all by the current, and there is a tendency for traces of the silver to remain behind in the crucible. The current density employed was about 0.0012 ampere per square centimeter of cathode surface and the time about twelve hours. After weighing the silver deposited, it was dissolved in nitric acid, precipitated by hydrochloric acid and weighed again as the chloride, giving a check upon the results. Seven of the ten determinations of Table III were thus treated, but owing to the imperfections of the process the results were all slightly low, the worst showing a deficiency of 0.0025 grm. of silver, an error of less than 0.9 per cent. The results of the two best of these determinations given below are, however, sufficient to prove the point hi question, namely that the silver sulphocyanide dried at 115 has the theoretical constitution and contains no water. The numbers are those under which the determinations appear in Table III. /.AT GRAVIMETRIC ANALYSIS. 367 Bxp. AgSCN taken. Calculated as Ag. Ag found by battery. Error. Weighed AgCl. Calculated asAg. Error. (4) (10) grm. 0.4375 0.4369 gnu. 0.2844 0.2840 grm. 0.2839 0.2838 grm. 0.0005- 0.0002- grm. 0.3765 0.3761 grm. 0.2834 0.2831 grm. 0.0010- 0.0009- It is clear therefore that the estimation of sulphocyanides by precipitation with silver nitrate and direct weighing of the precipitate is wholly permissible. The method is extremely simple and, as has been shown, the results are quite accurate. XLVII ON THE ESTIMATION OF CAESIUM AND EUBIDIUM AS THE ACID SULPHATES, AND OF POTASSIUM AND SODIUM AS THE PYROSULPHATES. BY PHILIP E. BROWNING. BUNSEN * is authority for the statement that the acid sulphate of rubidium does not lose sulphuric acid at a heat approaching redness. It is stated in the literature f that the acid sulphates of caesium and rubidium when subjected to a low red heat pass into the form of the pyrosulphates. R. Weber J found that by treating the dry sulphates of potassium, caesium, rubidium, and thallium with sulphuric anhydride in a closed tube and heating on a water bath two layers separated. In the lower layer he obtained crystalline bodies which proved to have the constitution R 2 O . 8SO 3 . On strong heating he obtained from these substances, bodies of the form R 2 . 2SO 3 and finally R 2 O . SO 3 . He also notes that in the case of the caesium salt the removal of the excess of the sulphuric anhydride was attended with greater difficulty. Baum states that the pyrosulphates of the alkalies may be obtained by heating the acid sulphates under atmospneric pressure at low redness, or under diminished pressure at a temperature between 260 C. and 320 C. In a recent paper || from this laboratory I have shown that thallium may be estimated as the acid sulphate by evaporating a thallous salt in solution with an excess of sulphuric acid and bringing the residue to a constant weight at a temperature of about 250 C. * Ann. Chem. (Liebig), cxix, 110. t Graham-Otto, Lehrbuch d. Chem., iii, 269, 278. | Ber. Dtsch. Chem. Ges., xvii, 2497. Ber. Dtsch. Chem. Ges. xx, 752. || Am. Jour. Sci., ix (1900), 137. This volume, p. 317. ESTIMATION OF CAESIUM AND RUBIDIUM. 369 The similarity which thallium bears in some of its combi- nations to the alkaline metals suggested the study of the sulphates of these elements under the same general conditions of procedure. My first experiments were made with a pure caesium salt as follows : A weighed amount of the nitrate was placed in a previously weighed platinum crucible and treated with an excess of sulphuric acid. The crucible was then placed upon a steam bath until the water and nitric acid were largely expelled and then removed to a radiator, consisting of a porcelain crucible fitted with a pipe-stem triangle so arranged that the bottom of the platinum crucible would be about midway between the top and bottom of the porcelain crucible. This improvised radiator was set in an iron ring and a thermometer so placed that the mercury bulb would be on a level with the bottom and close to the side of the platinum crucible. An ordinary Bunsen burner served as the source of heat and the temperature was kept so far as possible between 250 C. and 270 C. After the fuming attending the removal of the large excess of sulphuric acid ceased, the crucible and contents were removed to a desiccator, and, after being allowed to cool, weighed. This process of heating was continued for half-hour periods until the weights were constant. The results shown in Table I were obtained by this method of treatment. In experiments (1), (4), and (9) TABLE I. Ezp. CsN0 8 taken. CsHS0 4 calcu- lated. First constant weight. Second constant weight. Error CsHS0 4 . Cs 2 SO 4 calcu- lated. Cs 2 S0 4 found. Error Cs 2 S0 4 . grm. grm. grm. grm. grin. grm* grm. grm. (1) 0.1706 0.2013 0.2054 0.2020 0.0007+ (2) 0.1706 0.2013 0.2010 0.0003- t (3) 0.1032 0.1217 0.1201 0.0016- (4) 0.1032 0.1217 0.1252 0.1222 0.0005+ 0.0961 0.0948 0.0013- (5) 0.1218 0.1437 0.1458 ( 0.0021+ 0.1130 0.1118 0.0012- (6) 0.1214 0.1435 0.1430 0.0005- ( (7) 0.1214 0.1435 0.1422 , 0.0013- ( (8) 0.1150 0.1356 0.1330 . 0.0026- (9) 0.1050 0.1245 0.1272 0.1248 0.0003+ (10) 0.1056 0.1245 0.1252 0.0007+ VOL. ii. 24 370 ESTIMATION OF CESIUM AND RUBIDIUM. it will be noticed that the weights were constant somewhat above the condition of the acid sulphate, a fact which would go to show a tendency on the part of the caesium salt to hold an excess of sulphuric acid over the amount necessary to form the ordinary acid sulphate. The results show that by regulating the heat at a temperature between 250 C. and 270 C. caesium may be brought, with a fair degree of cer- tainty to the condition of the acid sulphate. As a check upon the results, the acid sulphate was, in a few cases, treated with a little ammonium hydroxide, the excess of this was removed upon a steam bath and the neutral sulphate obtained by ignition at a red heat to a constant weight. These determi- nations agreed fairly well with the theory. The same pro- cedure was followed with rubidium, a pure rubidium chloride having been chosen as the starting-point. The results are TABLE II. Exp. RbCl taken. RbHS0 4 calculated. RbHS0 4 found. Error. Rb 2 S0 4 calculated. Rb 2 S0 4 found. Error. grm. grm. grm. grm. grm. grm. grm. (1) 0.1252 0.1889 0.1878 0.0011- . . (2) 0.1212 0.1829 0.1840 0.0011+ 0.1460 0.1460 0.0000 (3) 0.1230 0.1856 0.1850 0.0006- (4) 0.1230 0.1856 0.1858 0.0002+ 0.1357 0.1350 0.0007- (5) 0.1610 0.2430 0.2416 0.0014- 0.1777 0.1772 0.0005- (6) 0.1360 0.2052 0.2032 0.0020- 0.1501 0.1490 0.0011- given in Table II. No tendency was observed on the part of this element to hold sulphuric acid in excess of the amount necessary for the formation of the acid sulphate. When the same method was applied to sodium and potassium salts, pure chlorides being used as the starting-point, a somewhat different result was obtained, in that the weight of the final product appeared to indicate the formation of the pyrosulphate. The results given in Tables III and IV, in which the sodium and potassium salts are calculated as pyrosulphates, are sufficiently satisfactory for purposes of quantitative estimation. As in the case of the caesium and rubidium salts, a number of determinations as the neutral ESTIMATION OF CAESIUM AND RUBIDIUM. 371 TABLE m. Exp. KC1 taken. K 2 S,0 7 calculated. K,8 2 7 found. Error. K,S0 4 calculated. Sffi Error. gnu. grm. grm. grin. grm. grm. grm. (1) 0.2172 0,3704 0.3698 0.0006- 2 0.1706 0.2909 0.2886 0.0023 0.1993 0.1972 0.0021- (3) 0.1192 0.2032 0.2022 0.0010- 0.1393 0.1381 0.0012- (4) 0.1074 0.1830 0.1823 0.0007- (5) 0.1096 0.1868 0.1860 0.0008- TABLE IV. Exp. NaCl taken. Na 2 8 2 7 calculated. Na 2 S 2 7 found. Error. Na.jS0 4 calculated. NajSO* found. Error. (1) (2) (3) (4) grin* 0.1042 0.1028 0.1093 0.1402 grm. 0.1978 0.1952 0.2075 0.2662 grm. 0.1972 0.1952 0.2065 0.2651 grm. 0.0006- 0.0000 0.0010- 0.0011- gnu. 0.1266 0.1328 0.1703 grm. 0.1254 0.1320 0.1696 grm. 0.0012- 0.0008- 0.0007- sulphate were made by ignition of the sodium and potassium pyrosulphates, with results which are recorded. In Table V, two determinations are recorded, in one of which the caesium and rubidium salts were treated together and in the other the sodium and potassium salts. TABLE V. Exp. RbCl + CsNOj taken. RbHS0 1 + CsHS0 4 calculated. RbHS0 4 + CsHS0 4 found. Error. Rb 2 S0 4 + Cs 2 SO 4 calculated. Rb 2 S0 4 + CB*S0 4 found. Error. (1) grm. (RbCl 0.1428 ) {CsN0 3 0.1264 f grm. 0.3646 grm. 0.3666 grm. 0.0020+ grm. 0.2749 grm. 0.2752 griii. 0.0003+ NaCl + KC1 taken. ^80,+ MLftfff calculated. Na 2 S,0 7 + K 2 S 2 O 7 found. Error. ^A calculated. Na28 g 4 ik K 2 SO 4 found. Error. (2) (NaCl 0.1233 ) |KCI 0.1340 ] 0.4627 0.4630 0.0003+ 0.3062 0.3040 0.0022- An application of this general method to a lithium salt gave no evidence of the existence of a stable acid sulphate or pyrosulphate. 372 ESTIMATION OF CAESIUM AND RUBIDIUM. The results may be summed up as follows : Csesium and rubidium salts of volatile acids when treated with sulphuric acid in excess and brought to a constant weight at a tem- perature between 250 C. and 270 C. form acid salts of the type RHSO 4 and the neutral salts of the type R 2 SO 4 on ignition. Some tendency of the caesium salt to hold more sulphuric acid than corresponds to the formation of the acid sulphate RHSO 4 was apparent at temperatures between 258 C. and 270 C., but upon raising the temperature above 300 C. the loss was excessive and showed a tendency on the part of the acid sulphate to pass, at this temperature, toward the con- dition of the pyrosulphate. Sodium and potassium salts, when heated under the con- ditions described, give pyrosulphates of the type R 2 S 2 O 7 which on ignition go into the neutral sulphate of the form R 2 SO 4 . Lithium gives neither salts of the type RHSO 4 nor R 2 S 2 O 7 under the conditions of these experiments. XLVIII THE ESTIMATION OF CALCIUM, STRONTIUM, AND BARIUM AS THE OXALATES. BY CHARLES A. PETERS. A FORMER article from this laboratory* describes the condi- tions under which oxalic acid may be titrated by potassium permanganate in the presence of hydrochloric acid, and states that the extra consumption of permanganate which ordinarily takes place when oxalic acid is titrated by permanganate in the presence of hydrochloric acid, may be prevented by the addition of a manganous salt. This fact led to the idea of effecting the solution of the alkaline earth oxalates in hydro- chloric acid and titrating the free oxalic acid with perman- ganate in the presence of a manganous salt, and so to the study of the conditions under which precipitates of strontium and barium oxalates could be obtained sufficiently insoluble for quantitative purposes, the conditions under which calcium oxalate is insoluble being already well known. The permanganate solution for this work was standardized against freshly recrystallized ammonium oxalate, and on oxalic acid, the standards agreeing. Calcium Oxalate. It is well known that calcium may be estimated by treating the precipitated oxalate with sulphuric acid and titrating by permanganate the oxalic acid set free.f In the work described in the present article, the precipitate of calcium oxalate has been dissolved in hydrochloric acid and the oxalic * Gooch and Peters, Am. Jour. Sci., vii, 461. This volume, p. 222. t Mohr, Titrirmethode, 6. Aufl., S. 227. 374 ESTIMATION OF CALCIUM, STRONTIUM, acid titrated by permanganate in the presence of a manganous salt. The process was as follows : The boiling hot solution of calcium chloride was precipitated with ammonium oxalate, allowed to stand 12 hours, and the supernatant liquid de- canted on asbestos. The precipitate was washed two or three tunes by decantation with 50-100 cm 3 of cold water and brought on the felt. The crucible containing the precipitate was returned to the beaker, 100-200 cm 3 of water were added, together with 5-10 cm 3 of strong hydrochloric acid and 0.5-1.0 grin, of manganous chloride, and the oxalic acid titrated at a temperature of 35 -45. The results, given in Table I, are obviously excellent. TABLE I. CaO taken as CaClj. Ammonium oxalate. Volume at precipitation. CaO found. Error. grm. grm. cm 3 gnu. grm. 0.0656 0.3 100 0.0657 0.0001+ 0.0656 0.3 100 0.0656 0.0000 0.0656 0.3 150 0.0658 0.0002+ 0.0656 0.3 100 0.0655 0.0001- 0.0985 0.5 175 0.0981 0.0004- 0.1313 0.6 150 0.1315 0.0002+ 0.1313 0.6 200 0.1315 0.0002+ Extended washing with hot water, however, is to be avoided after the precipitant, ammonium oxalate, has been removed. In one experiment, for example, in which the precipitate, on the felt, was washed fourteen times with portions of about 50 cm 3 each of hot water, each portion bleached from 2-6 drops of approximately ^ permanganate, making a total loss of 0.0034 grm. of calcium oxide. Strontium Oxalate. Souchay and Lenssen * state that strontium oxalate is solu- ble in 12,000 parts of water. This fact would seem sufficient to warrant the study of the quantitative separation of stron- tium as the oxalate. In the work which follows strontium Ann. Chem. (Liebig), cii, 35. AND BARIUM AS THE OXALATE S. 375 oxalate has been precipitated both in alcoholic solution and in water solution, and for convenience these two conditions of precipitation will be discussed separately. All the strontium salts, of established purity, were stan- dardized by precipitation with sulphuric acid in a solution containing at least one-half its volume of alcohol, and with some solutions confirmatory standards were also obtained by evaporation with sulphuric acid. Precipitation in Alcoholic Solution. To determine the com- pleteness of the precipitation in alcoholic solution strontium nitrate was precipitated by ammonium oxalate in a solution containing one-third of its volume of alcohol, the mixture was allowed to stand over night, the liquid was filtered off on asbestos, and the precipitate was treated in the capped filter- ing crucible with sulphuric acid, ignited, and weighed as the sulphate. The results are given in Table II. It is plain from TABLE IL SrO taken as Sr(N03) 2 . Ammonium oxalate. Volume at precipitation. Volume of alcohol. SrO found as SrS0 4 . Difference. grm. grm. cm* griii. griii. 0.2434 0.8 180 0.2440 0,0006+ 0.2434 0.8 180 0.2437 0.0003+ 0.0022 0.2 100 0.0022 0.0000 0.0013 0.2 100 0.0014 0.0001+ 0.0004 0.04 100 0.0004 0.0000 the results in this table that the precipitation of even small amounts of the strontium salt from a solution containing one- third of its volume of alcohol is practically complete. To determine the minimum amount of alcohol necessary for the complete precipitation of the strontium oxalate, experiments were made using varying proportions of 85 per cent alcohol with different amounts of ammonium oxalate, and the filtrates from such experiments were tested for strontium by the addi- tion of more alcohol. The results given in Table III show that when a moderate excess of ammonium oxalate is present, a volume of 85 per cent alcohol, amounting to one-fifth of 376 ESTIMATION OF CALCIUM, STRONTIUM, TABLE III. SrO present as Sr(N0 3 ) 3 . Ammonium oxalate. Volume of liquid. Proportion of 85 per cent alcohol. SrO found in filtrates, weighed as SrSO 4 . grin. grm. cm 3 grm. 0.1 0.4 100 1 0.0000 0.1 0.4 100 JL 0.0000 0.1 0.4 100 1 0.0004 0.1 0.2 100 1 0.0000 0.1 0.2 100 JL 0.0009 0.1 0.2 100 JL 0.0020 0.1 0.1 100 * 0.0002 the whole, is sufficient to complete the precipitation of the strontium as the oxalate. The conditions under which strontium oxalate is insoluble having been determined, the process for the volumetric esti- mation of strontium was carried out as follows: The hot solution of a strontium salt was precipitated with ammonium oxalate, 85 per cent alcohol, amounting to from one-fifth to one- third the total volume, was added, the mixture was allowed to stand over night, and the clear liquid was decanted on an asbestos filter. The precipitate was washed with a mixture of equal parts of 85 per cent alcohol and water, transferred to the filter, dried in the filtering crucible over a flame to free it from alcohol, returned to the beaker previously dried, treated with sulphuric acid, or with 5-10 cm 3 of hydrochloric acid (in the latter case 0.5-1.0 grm. of a manganous salt being added) and the liberated oxalic acid was titrated by perman- ganate. The results obtained by this method are accurate and are given in Table IV. In the last experiment in which a comparatively large amount of strontium salt was present and the dilution low, there is a slight tendency towards a minus error, due probably to the occlusion of some oxalic acid by the strontium sulphate formed. This phenomenon would favor titration at greater dilution when sulphuric acid is used to liberate the oxalic acid from large amounts of strontium oxalate. Precipitation in Water Solution. In order to determine the AND BARIUM AS THE OXALATES. 377 TABLE IV. VOLUME DURING TITBATION 150-250 CM S . SrO taken as Ammonium Volume at precipita- tion. Propor- tion of 85 per cent alcohol. Acid pres- ent during titratiou. SrO found. Error. grm. 0.0974 grm. 0.4 cm 3 100 HC1 grm. 0.0973 grm. 0.0001- 0.0974 0.4 100 HC1 0.0983 0.0009+ 0.0974 0.4 100 HC1 0.0975 0.0001+ 0.0974 0.8 100 HC1 0.0981 0.0007+ 0.1948 0.4 200 HC1 0.1943 0.0005- 0.1948 0.8 200 HC1 0.1942 0.0006- 0.0974 0.4 100 H 2 S0 4 0.0970 0.0004- 0.0974 0.4 100 H 2 S0 4 0.0977 0.0003+ 0.0974 0.4 100 ] H. 2 S0 4 0.0976 0.0002+ 0.1948 0.6 150 H 2 S0 4 0.1938 0.0010- degree of precipitation of strontium salts in water solution, strontium oxide, taken as the nitrate, was precipitated by ammonium oxalate, the mixture was allowed to stand over night, filtered on asbestos, the precipitate was washed with water containing one-half its volume of 85 per cent alcohol, treated hi the capped crucible with a few drops of sulphuric acid, ignited, and weighed as the sulphate. The result gave 0.0973 grm. of strontium oxide instead of 0.0974 grm. taken. The precipitation, therefore, of strontium oxalate, in water solution with a sufficient excess of ammonium oxalate present, is practically complete. To determine the amount of ammonium oxalate necessary for the precipitation of strontium salts in water solution, experiments were made in which strontium oxalate was pre- cipitated in the presence of varying amounts of ammonium oxalate, allowed to stand over night, the clear liquid was decanted on asbestos, and the precipitate was washed twice with 10-20 cm 3 of cold water. The results obtained by the estimation of the oxalic acid by permanganate show that an amount of ammonium oxalate several times larger than that required for the theoretical formation of strontium oxalate is necessary for the separation of the strontium oxalate. The experiments are recorded in Table V. 378 ESTIMATION OF CALCIUM, STRONTIUM, TABLE V. SrO, taken as Sr(NOa),. Ammonium oxalate. Volume at precipi- tation. Acid present during titratiou. SrO found. Error. gnn. grm. cms grm. grm. 0.0487 0.0487 0.064 0.0768 100 100 H 2 S0 4 H 2 SO 4 0.0441 0.0465 0.0046- 0.0022- 0.0487 0.16 100 H 2 SO 4 0.0488 0.0001+ 0.0974 0.128 100 H 2 S0 4 0.0939 0.0025- 0.0974 0.16 100 H 2 S0 4 0.0959 0.0015- 0.0974 0.32 100 H 2 S0 4 0.0976 0.0001+ The solvent action of a large amount of water on a pre- cipitate of strontium oxalate was tested by washing a pre- cipitate equivalent to 0.0974 grm. of the oxide with 150 cm 3 of cold water. The precipitate, when weighed as the sulphate, showed a loss of 0.0033 grm., as the oxide, which amount was subsequently recovered from the filtrate by the addition of ammonium oxalate and alcohol. Plainly excessive washing with water is to be avoided. In the estimation, therefore, of strontium precipitated as the oxalate in water solution, the amount of water used in washing was limited. It was found that 40-50 cm 3 of water judiciously applied was sufficient to wash out the ammonium salt without producing appreciable solvent effect upon the strontium oxalate. The process of treatment was similar to that used in the precipitations from alcoholic solution, excepting that no alcohol was added to the solution, that the washing was effected with a limited amount of water, and that, there being no alcohol present to effect the titration, the precipitate was not dried before treatment with permanganate. The results are given hi Table VI. In the results recorded in section A of Table VI, the stron- tium oxalate was treated with sulphuric acid and titrated at 80, the volume being 200-300 cm 3 ; while hi the experiments given in section B, the precipitate was treated with hydro- chloric acid and titrated at 35-45, at a volume of 100-200 cm 3 , after the addition of 0.5-1.0 grm. of manganous chloride. The results show that 0.1 grm. of strontium salt, calculated as the AND BARIUM AS THE OXALATES. 379 oxide, may be estimated as the oxalate with a fair degree of accuracy when precipitated in 100-250 cm 3 of water by a TABLE VI. SrO, taken as Sr(NOs),. Ammonium oxalate. Volume at precipi- tation. Acid present during titration. SrO found. Error. A. grm. 0.0974 0.0974 0.0974 0.0974 0.0974 0.0974 0.0974 0.0974 0.0974 grm. 0.5 0.5 0.5 0.5 0.8 0.8 1.0 2.0 2.0 Cfflg 100 100 100 100 100 100 100 100 100 H2SO 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 SO 4 grm. 0.0966 0.0985 0.0977 0.0963 0.0981 0.0966 0.0965 0.0963 0.0970 grm. 0.0008- 0.0011+ 0.0003+ 0.0011- 0.0007+ 0.0008- 0.0009- 0.0011- 0.0004- SrO, taken as B. 8rCl 2 . 0.0778 0.0778 0.0778 0.0778 0.5 0.5 0.5 0.5 100 100 100 100 H 2 S0 4 H2S0 4 H 2 SO 4 H 2 S0 4 0.0792 0.0767 0.0776 0.0776 0.0014+ 0.0011- 0.0002- 0.0002- SrO, taken as Sr(N0 8 ) 2 . 0.0974 0.0974 0.0974 0.0974 0.0974 0.0974 0.0974 0.0974 0.8 2.0 0.8 0.8 0.8 0.8 0.8 0.8 250 250 100 100 100 100 100 100 H 2 S0 4 H 2 S0 4 HCI HCI HCI HCI HCI HCI 0.0973 0.0975 0.0971 0.0980 0.0975 0.0980 0.0973 0.0978 0.0001- 0.0001+ 0.0003- 0.0006+ 0.0001+ 0.0006+ 0.0001- 0.0004+ C. 0.2425 0.2436 0.2436 0.2436 0.2436 0.2436 0.384 0.384 0.64 0.8 2.0 2.0 125 125 125 125 125 125 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 0.2376 0.2402 0.2411 0.2367 0.2376 0.2402 0.0049- 0.0034- 0.0025- 0.0069- 0.0060- 0.0034- D. 0.2436 0.2436 0.2436 0.2436 0.8 0.8 2.0 2.0 250 250 250 250 H 2 SO 4 H 2 S0 4 H 2 S0 4 H 2 S0 4 0.2443 0.2446 0.2440 0.2431 0.0007+ 0.0010+ 0.0004+ 0.0005- 380 ESTIMATION OF CALCIUM, STRONTIUM, TABLE VI (continued). SrO, taken as Sr(N0 8 ),. Ammonium oxalate. Volume at precipi- tation. Acid present during titratiou. SrO found. Error. E. grui. grm. cm 8 grm. grm. 0.2436 0.8 500 H 2 S0 4 0.2396 0.0040- 0.2436 2.0 500 H 2 S0 4 0.2403 0.0033- 0.2436 2.0 500 H 2 S0 4 0.2413 0.0023- 0.2436 4.0 600 H 2 S0 4 0.2410 0.0026- 0.2436 8.0 500 H 2 S0 4 0.2407 0.0029- 0.4872 2.0 500 H 2 SO 4 0.4837 0.0035- 0.4872 4.0 500 H 2 S0 4 0.4855 0.0017- 0.5430 5.0 500 HnS0 4 0.5422 0.0008- 0.4579 10.0 500 H 2 SO 4 0.4554 0.0025- 0.7307 4.0 500 HC1 0.7262 0.0045- sufficient excess of ammonium oxalate. In the experiments recorded in section C, in which the amount of strontium salt in 125 cm 3 of water is increased, a negative error is intro- duced, which is not diminished by the presence of a large amount of ammonium oxalate, but when the dilution is in- creased to 250 cm 3 , as is the case in experiments given in section D, so that the conditions correspond more nearly to those recorded in sections A and B, the errors fall to a minimum. In the experiments recorded in section E, in which the dilution is increased to 500 cm 3 , an error is in- troduced which is not prevented by the presence of a large excess of ammonium oxalate and which is independent of the amounts of strontium salt used. Eight of the water filtrates and wash waters obtained in the experiments recorded in Table VI were tested for traces of strontium by the addition of alcohol, and in all cases a small amount of strontium was found, amounting, in the average, to 0.0010 grm. in 100 cm 3 of water. Barium Oxalate. Barium oxalate according to Souchay and Lenssen * is soluble in 2590 parts of cold water, and according to Berg- * Ann. Chem. (Liebig), xc, 102. AND BARIUM AS THE OXALATES. 381 man * is scarcely at all soluble in alcohol. The attempt was made to estimate barium by precipitation with ammonium oxalate in a mixture containing alcohol. It was found that in filtrates from oxalate precipitations in which 0.1-0.2 grm. of barium oxide, taken as the nitrate, had been precipitated in volumes of 100 cm 8 , containing 30 cm 3 of absolute alcohol, and allowed to stand over night, treatment with sulphuric acid gave barium sulphate amounting in the average to no more than 0.0001 grm. of barium oxide. The insolubility of barium oxalate under these conditions, therefore, is practically complete. The process for the estimation of barium was as follows : Ammonium oxalate was added to a solution of a barium salt, TABLE VII. BaO taken as Ba(N0 3 ) 2 . Ammonium oxalate. Volume at precipita- tion. Acid present during titration. BaO found. Error. A. grm. 0.1165 0.1165 0.1165 0.1165 0.1165 0.1165 0.1165 0.2330 0.2330 0.2330 grin. 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.4 0.4 0.4 cm 3 100 100 100 100 100 100 100 100 100 100 HCI HCI HCI HCI HCI HCI HCI HCI HCI HCI gnu. 0.1177 0.1170 0.1164 0.1151 0.1165 0.1176 0.1164 0.2319 0.2335 0.2342 grm. 0.0012+ 0.0005+ 0.0001- 0.0014- 0.0000 0.0011+ 0.0001- 0.0011- 0.0005+ 0.0012+ BaO taken as BaCl 2 . 0.4 0.4 0.4 0.4 0.4 100 100 100 100 100 HCI HCI HCi HCI HCI 0.0952 0.0939 0.0941 0.1893 0.1892 0.0010+ 0.0003- 0.0001- 0.0009+ 0.0008+ 0.0942 0.0942 0.0942 0.1884 0.1884 B. 0.0942 0.1884 0.0942 0.2 0.4 0.2 200 200 500 H 2 S0 4 H 2 S0 4 H 2 S0 4 0.0858 0.1732 0.0857 0.0084- 0.0152- 0.0085- * Bergman's Essays, i, 320. 382 ESTIMATION OF CALCIUM, STRONTIUM, containing 30 per cent of its volume of absolute alcohol, the mixture was allowed to stand over night, filtered on asbestos, the precipitate was washed by decantation with 100-200 cm 3 of water containing 30 per cent of absolute alcohol, and dried over a flame to insure the removal of the alcohol. The cruci- ble containing the precipitate was returned to the beaker also previously dried over a flame, 100-200 cm 3 of water, 5-10 cm 3 of strong hydrochloric acid, and 0.5-1.0 grm. of manganous chloride were added, and the solution was titrated at 35 -45 with permanganate. The results of the experiments, given in Table VII, A, show that barium, either as the nitrate or chloride, may be estimated in the manner described with a fair degree of accuracy. In the experiments given in section B of Table VII, the precipitate of barium oxalate was treated with sulphuric acid after the addition of the stated amount of water. The results show a large loss of oxalic acid probably due to the occlusion of some of the oxalic acid by the barium sulphate. This fact must prevent the use of sulphuric acid in an analytical process which depends upon the liberation of oxalic acid from barium oxalate. Gravimetric Estimation of the Oxalates of Strontium and Barium. It is well known that calcium may be weighed as the carbonate after a careful ignition of the oxalate, and it would seem probable that strontium might also be weighed as the carbonate. Precipitates of strontium oxalate, on asbestos, were ignited in a capped crucible from 2-8 minutes in the flame of a Bun sen burner and weighed as the carbonate, and in a single case the carbonate thus produced was converted by treatment with sulphuric acid to the sulphate and weighed as such. The results are given in Table VIII, and while they all show a very slight loss, which amounts in experiment (3) to one milligram, when one-fourth of a gram of strontium oxide taken as the nitrate was used, still the results are fairly accurate. AND BARIUM AS THE OXALATES. TABLE VUL 383 Ezp. BrO taken as Sr(NO s ),. SrO calculated from SrC0 3 found. SrO calculated from SrSO 4 found. grm. grm. grm. (1) (2) 0.1120 0.1120 0.1113 0.1116 .... (3) 0.2435 0.2425 0.2437 Precipitates of barium oxalate were also ignited from 5-10 minutes and weighed as the carbonate. The results are given in Table IX, and are fairly accurate. TABLE IX. Exp. BaO taken as Ba(N0 3 ) 2 . BaO calculated from BaCO 8 found. Difference. (1) (2) (3) griii. 0.2912 0.2912 0.2912 grm. 0.2909 0.2901 0.2901 grm. 0.0003- 0.0011- 0.0011- The results of this work may be summarized as follows: In the estimation of calcium by titration of the oxalate with permanganate, accurate results may be obtained when hydro- chloric acid (with a manganous salt) is used as the solvent. Strontium salts may be precipitated by ammonium oxalate with practical completeness in a solution containing one-fifth of its volume of 85 per cent alcohol, and with approximate completeness from water solutions at a dilution not exceeding 250 cm 3 . Furthermore strontium oxalate may be titrated by permanganate with accuracy when either sulphuric acid or hydrochloric acid (with a manganous salt) is used to liberate the oxalic acid. Barium may be precipitated with practical completeness by ammonium oxalate in a solution containing 30 per cent of alcohol, and the barium oxalate thus obtained may be dissolved in hydrochloric acid and titrated by per- manganate after the addition of a manganous salt. Strontium and barium oxalates may be converted to carbonates by ignition, and weighed as such. XLIX THE ACTION OF SODIUM THIOSULPHATE ON SOLUTIONS OF METALLIC SALTS AT HIGH TEMPERATURES AND PRESSURES. BY JOHN T. NORTON, JR. THE use of sodium thiosulphate as a substitute for hydro- gen sulphide in effecting precipitations and its application in the case of arsenic, antimony, copper, and platinum was suggested by Himly * before the middle of the present cen- tury. Thirteen years later Vohl f and Slater, independently $ drew attention to this use of sodium thiosulphate and ex- tended the investigation to salts of tin, mercury, silver, gold, lead, bismuth, and cadmium. Slater in addition studied the action of sodium thiosulphate upon chromic acid, molybdates, ferrous and ferric ferrocyanides, ferric sulphocyanides and potassium permanganate. Following out these lines, the precipitation of copper, together with arsenic antimony, by treating with sodium thiosulphate the hot solution contain- ing sulphuric acid, and the separation of these elements from tin, zinc, iron, nickel, cobalt, and manganese has been advo- cated by Westmoreland ; and quite recently Faktor || has studied the action of sodium thiosulphate upon neutral salts of several of the elements mentioned, as well as the modi- fying influence of ammonium chloride and other salts upon the course of the reaction. Subsequently to the work of Himly, Vohl, and Slater, Chancel ** developed his well known method for the precipi- * Ann. Chem. (Liebig), xliii, 150. t Ann. Chem. (Liebig), xcvi, 237. t Chemical Gazette, 1855, p. 369. Jour. Soc Chem. Ind., v, 61. || Chem. Centralblatt, 1900, ii, 20, 67, 239, 594. * Comp. rend., xlvi, 987. ACTION OF SODIUM THIOSULPHATE, ETC. 385 tation of aluminum as the hydroxide and its separation from salts of iron by boiling with sodium thiosulphate the nearly neutral solution, containing the salt of aluminum and iron, at suitable dilution ; and upon an extension of the principle of Chancel's separation of aluminum from iron Stromeyer * founded his well known processes for the separation of titan- ium and zirconium from iron. The latter process appears to be fairly trustworthy ; but of Chancel's method, although it has met with wide acceptance, it was shown by Wolcott Gibbs, very soon after its announcement,! that it fails to bring about complete separation of alumina within a reasonable period of boiling, and this result has been confirmed by Zimmer- man, J who has shown that the boiling must be continued fifteen hours in order to complete the precipitation of the alumina. It was shown by Dr. Gibbs that when the treatment of salts of aluminum by thiosulphate was carried on in sealed tubes under pressure at 120 C., the precipitation of alumina was complete, and further that the precipitation of sulphides of nickel, cobalt, and iron, though partial under ordinary atmos- pheric pressure, was made complete by heating in sealed tubes to 120-140 C. In repeating the experiments of Dr. Gibbs qualitatively and extending them, I have made use of the well known Pfungst tube to secure the necessary pressure. In each experiment a test-tube containing the mixture of an excess of sodium thiosulphate with the salt whose action was studied was placed within the Pfungst tube containing some water, the cover of the latter was set in place and firmly bolted upon a washer of lead, and the whole was submitted to tempera- tures varying from 140 to 200 C. for an hour by immersing in a bath of paramne. After cooling, the test-tube was taken out, the precipitate was filtered off, and the filtrate tested by appropriate reagents to determine the completeness of pre- cipitation. The following table records the details of these experiments. * Ann. Chem. (Liebig), cxiii, 127. t Zeit. anal. Chem., iii, 389. t Inaug. Diss. Berlin, 1887. VOL. u. 25 386 ACTION OF SODIUM THIOSULPHATE ON TABLE I. ACTION OP NA 2 S 2 O 3 ON SALTS UNDER PRESSURE. Salts used. Precipitates. Degree of precipitation. SULPHIDES. NiSO 4 . CoSO 4 . FeCl 3 . ZnSO 4 . Pb0 2 (C 2 H 8 0) a . Hg(N0 3 V AgN0 8 . CuSO 4 . CdS0 4 . KSbC 4 H 4 (X. Bi(N0 8 ) 8 . NiS + S. CoS + S. FeS + S. ZnS + S. PbS + S. HgS + S. Ag 2 S + S. CuS,Cu 2 S, + S. CdS + S. Sb 2 S 8 + S. Bi 2 S 8 + S. Complete. Complete. Complete. Complete. Complete. Complete. Complete. Complete. Complete. Complete. Complete. HYDROXIDES. (NH 4 )A1(S0 4 ) 2 .12H 2 0. K 2 Cr 2 O 7 . K 2 ZrF 6 . KoTiFa. Th(N0 8 ) 4 . A10 8 H 3 + S. Cr0 3 H 3 + S. Zr0 4 H 4 + S. Ti0 4 H 4 + S. Th0 4 H 4 + S. Complete. Complete. Complete. Complete. Complete. ELEMENTS. SeO 2 . Te0 2 . Se + S. Te + S. Complete. Complete. SULPHIDES. MnS0 4 . AuCl 8 . (NH 4 ) 2 Mo0 4 . MnS + S. Au 2 S + S. MoS 8 (?) + S Red liquid. Partial. Partial. Partial. HYDROXIDES. BeCLj. Be0 2 H 2 + S. Partial. UNDETERMINED. (NH 4 ) 2 U 2 7 . K 2 PtCle. CeCl 8 . CaCl 2 . SrCL. Bad-. MgS0 4 . NH 4 VO 8 . H 2 KAsO 4 . Black. Gray, reddish brown liquid. White, yellow liquid. White, yellow liquid. White, yellow liquid. White, yellow liquid. , Brown liquid. Partial. Partial. Partial. Partial. Partial. Partial. None. None. None. SOLUTIONS OF METALLIC SALTS. 387 A perusal of this table brings to light several interesting facts. It appears that salts of nickel, cobalt, iron, zinc, lead, mercury, silver, copper, cadmium, antimony and bismuth are completely precipitated as sulphides by sodium thiosulphate under the prevailing conditions of temperature and pressure. In the case of manganese precipitation is only partial, and arsenic does not seem to be precipitated from an arsenate with- out the addition of acid. Tin, curiously enough, is not thrown down as the sulphide from a stannous salt, but gives a dirty white precipitate of uncertain composition. Salts of aluminum, chromium, titanium, zirconium and thorium are completely precipitated as the hydroxides ; but in the case of beryllium, which one might expect to act similarly, the precipitation as the hydroxide is incomplete. Salts of selenium and tellurium are reduced, and the elements are precipitated. The precipi- tates obtained with barium, strontium, and calcium were white in a bright yellow liquid, but no study was made of the constitution of either precipitate or liquid. In the case of magnesium there was no precipitate. Salts of molybdenum, vanadium and uranium gave dark colored liquids. Thallium yielded a white spongy mass which on compression was re- duced to a very small bulk without disintegrating. Salts of gold and platinum gave slight dark precipitates, presumably sulphides, surrounded by dark colored liquids. The apparatus used in these experiments and described above is easily handled and answers sufficiently well for qualitative purposes. But, obviously, the introduction into precipitates of foreign matter caused by the action of water on the glass of the test-tube and porcelain lining of the Pfungst tube, precludes the possibility of an exact quanti- tative study of the reactions involved. For the subsequent experiments, therefore, conducted upon the same general lines, a digester with an interior cylindrical cavity of about 12 cm. in depth by 5 cm. in diameter, and provided with a pressure gauge was employed. As a container for the solu- tions to be tested, use was made of a platinum cylinder, 4 cm. in diameter and 10 cm. deep, provided with a loose cover. 388 ACTION OF SODIUM THIOSULPHATE ON With this apparatus the following quantitative experiments which deal with those elements which are precipitated as hydroxides namely, aluminum, beryllium, chromium, zir- conium, and titanium were made. In each case a weighed quantity of the salt taken for the experiment was dissolved in 50 cm 3 of water in the platinum vessel and to this a known amount of sodium thiosulphate was added. The vessel was placed in the digester, and the latter was heated by a Bunsen burner in the customary way until the required pressure was shown on the gauge. The apparatus was then cooled and the platinum vessel removed from the digester. The precipitate was filtered off on ashless paper, ignited, and weighed. Experiments with a Salt of Aluminum. In a series of experiments made according to the method of Chancel, the results of which are shown in Table II, the solution in water of a weighed portion of pure ammonium alum was treated with an excess of sodium thiosulphate and boiled vigorously for periods varying from ten minutes to half an hour. TABLE II. Ezp. Amount of Alum taken Amount of A1 2 3 found. Error. as A1,O 3 . grm. grm. grm. grm. (1) 0.0537 Large excess. 0.0471 0.0066- (2) 0.0537 Large excess. 0.0397 0.0140- is 0.1083 Large excess. 0.0931 0.0152- (4) 0.1137 5 grin. 0.0979 0.0158- (5) 0.1139 2 grm. 0.1002 0.0137- The results substantiate the observations of Gibbs * and of Zimmerman f and show clearly that the boiling of solutions of the aluminum salt and sodium thiosulphate for a reasonable time does not effect the complete precipitation of aluminum as the hydroxide. Table III shows the result of submitting solutions of * Loc. cit. t Loc. cit. SOLUTIONS OF METALLIC SALTS. 389 ammonium alum treated with varying quantities of sodium thiosulphate to a pressure of 20 atmospheres in the digester. It usually required about 40 minutes to raise the pressure to the limit set; but this limit once reached, the digester was allowed to cool slowly. The duration of an experiment was about two hours. TABLE III. Alum taken as A1 2 3 . Amount of Na,S 2 3 used. A1 3 3 found. Error. grm. grm. grm. grm. 0.0565 5 0.0633 0.0068+ 0.1132 10 0.1154 0.0022-f 0.1153 5 0.1186 0.0033-f- 0.1128 3 0.1129 0.0001-f- 0.1126 3 0.1142 0.0016-f- 0.1128 2 0.1120 0.0008- 0.1136 2 0.1121 0.0015 0.1128 2.5 0.1136 0.0008-f- 0.1124 2.5 0.1127 0.0003+ 0.1134 2.25 0.1133 0.0001- This table shows that sodium thiosulphate precipitates aluminum completely as the hydroxide when pressure is employed. The high results seen in some of the experiments appear to be due to the difficulty of removing by ignition the large amounts of sulphur found in the action, as well as to the salts mechanically included in the precipitate. The amounts of sulphur and contaminating salts present depend upon the amount of thiosulphate taken ; therefore this should be as small as possible, 2-3 grm. being sufficient to precipitate all the alumina in a gram of alum. When the amount of thiosulphate is reasonably restricted the weights of alumina obtained accord fairly well with the theory. Experiments with a Salt of Chromium. Up to the tune of the completion of this work nothing appears to have been done upon the precipitation of chromium as the hydroxide by means of sodium thiosulphate. Slater * and Rose f make mention of the action of sodium thiosulphate * Loc. cit. t Traite de Chimie Analytique, vol. i, p. 479. 390 ACTION OF SODIUM THIOSULPHATE ON upon chromic acid, bichromates, and neutral chromates, but give no quantitative data. Recently, however, F. Faktor * has studied the action of sodium thiosulphate on chromium compounds. This investigator has found that if aqueous solutions of potassium bichromate and sodium thiosulphate are boiled together a brown precipitate of hydrated Cr 2 O 8 , CrO 8 separates out and the liquid turns yellow owing to the formation of normal chromate. A solution of potassium chromate is unaffected by boiling with thiosulphate but in presence of ammonium or of magnesium chloride the chromium is separated rapidly and completely in the same form as with the bichromate, and after continued boiling with an excess of thiosulphate all the chromium present is precipitated. Faktor also found that a solution of chromic chloride is completely decomposed by continued boiling with thiosulphate, chromic hydroxide and sulphur being precipitated. In the experiments shown in Table IV a weighed quantity of pure potassium bichromate was dissolved in water, a known amount of sodium thiosulphate added, and the whole submitted to a pressure of 20 atmospheres in the digester. After cooling, the precipitate was filtered off on an ashless paper, ignited and weighed as Cr 2 O 3 . TABLE IV. Exp. K,Cr 2 O 7 taken as Cr 2 O 3 . Amount of Na 2 S,0 6 . Cr 2 O s found. Error. grm. grm. grm. grm. (1) 0.1330 3 0.1341 0.0011+ (2) 0.1330 2.5 0.1326 0.0004- (3) 0.1322 2.5 0.1318 0.0004- (4) 0.1303 2 0.1303 0.0000 (5) 0.1301 2 0.1310 0.0009+ 6) 0.1320 2 0.1322 0.0002+ The results of these experiments are very satisfactory, and show that under pressure sodium thiosulphate precipitates chromium rapidly and completely as the hydroxide. It is advisable to use as small a quantity of thiosulphate as possible Zeitschr. anal. Chem., 1900, xxxix, 345. SOLUTIONS OF METALLIC SALTS. 391 in order to prevent the presence of much free sulphur in the precipitate. Experiments with a Salt of Beryllium. In experiments dealing with beryllium the salt used was the chloride, a certain amount of which was dissolved in water diluted to a liter and the amount of beryllium present deter- mined by precipitating with ammonia and weighing as the oxide. Measured quantities of this solution were drawn from a burette as required. "When a solution of a salt of beryllium and sodium thiosulphate are merely boiled together nearly all the beryllium remains in solution. It was expected that the use of pressure Would 'throw out all the beryllium, but, curiously enough, when solutions of beryllium chloride and sodium thiosulphate were submitted in the digester to pressures ranging from 10 to 80 atmospheres only a partial precipitation of the hydroxide took place. Experiments with Salts of Zirconium. To prepare a standard solution of the salt of zirconium it was found to be most convenient to heat the double fluoride of potassium and zirconium with sulphuric acid, evaporate to dryness in platinum, dissolve the zirconium sulphate remaining in water and enough sulphuric acid to prevent the precipitation of the basic salt, and dilute to standard volume. Measured portions of the solution were taken from a burette as required for the experiments. The presence, however, of so large an amount of sulphuric acid as was necessary to keep the zirconium salt in solution tends to decompose sodium thio- sulphate so rapidly that it was found necessary to nearly neutralize the solution with ammonium carbonate before adding the sodium thiosulphate. The solution of zirconium sulphate was standardized by precipitating with ammonia and weighing as the oxide. In experiment (1) of Table V, the solutions of zirconium sulphate and sodium thiosulphate were boiled together for a few minutes and then the precipitate filtered off, ignited, and 392 ACTION OF SODIUM THIOSULPHATE ON weighed as the oxide. In experiments (2)-(5) inclusive similar solutions were submitted to a pressure of 20 atmospheres in the digester. TABLE V. Exp. Zr0 2 taken. Na 2 S 2 3 taken. ZrO. found. Error. grin. grru. grm. grm. (1) 0.0658 3 0.0651 0.0007- 2 0.0658 3 0.0676 0.0016+ (3) 0.0666 2 0.0670 0.0004+ (4) 0.0641 2 0.0648 0.0007+ (5) 0.0641 2 0.0645 0.0004+ These results clearly show that sodium' thiosulphate precipi- tates zirconium completely as the hydroxide either with or without the aid of pressure. Experiments with a Salt of Titanium. The solution of the salt of titanium was obtained by treating the double fluoride of potassium and titanium with sulphuric acid, evaporating to dryness, and dissolving the residue in sulphuric acid and water. The solution was standardized by precipitating the titanium hydroxide with ammonia and then adding an excess of acetic acid as recommended by Gooch.* This method of procedure avoids the tendency to excessive weight observed when the titanium hydroxide is precipitated by ammonia in presence of salts of the alkalies. In the following table is shown the effect of treating a solution of titanium sulphate with sodium thiosulphate. TABLE VI. Exp. Ti0 2 taken. Na 2 S 2 taken. TiO 2 found. Error. grm. grm. grm. gnn. (1) 0.0240 2 0.0237 0.0003- (2) 0.0240 2 0.0240 0.0000 (3) 0.0240 2 0.0240 0.0000 * Am. Chem. Jour., vii, 285. SOLUTIONS OF METALLIC SALTS. 393 Experiment (1) was conducted by merely boiling a solution of the reagents named above, filtering off the precipitate and weighing as the oxide. In experiments (2) and (3) the solution of titanium sulphate and sodium thiosulphate was submitted to a pressure of 20 atmospheres in the digester. These results show that titanium is completely precipitated by sodium thiosulphate either with or without the aid of pressure. To recapitulate : I have shown that sodium thiosulphate will completely precipitate aluminum, chromium, zirconium and titanium as the hydroxides with the aid of high tempera- ture and pressure. Beryllium is only partially precipitated under similar conditions. Mere boiling for a reasonable time will not precipitate aluminum and chromium, but it is suffi- cient in the case of zirconium and titanium. SYSTEMATIC INDEX. LABORATORY APPLIANCES AND PREPARATIONS. Laboratory Apparatus (Gooch), I, 141. Generation of Chlorine (Gooch and Kreider), 1, 260. Preparation of Perchloric Acid (Kreider), 1, 282. Labora- tory Apparatus (Kreider), I, 306. INORGANIC CHEMISTRY. Interaction of Potassium Permanganate and Sulphuric Acid (Gooch and Dan- ner), I, 145. Reducing agents on lodic Acid (Roberts) 1,250. Existence of Selenium Monoxide (Peirce), I, 385. Condition of Oxidation of Man- ganese precipitated by the Chlorate Process (Gooch and Austin), II, 85. Action of Carbon Dioxide on Soluble Borates (Jones), II, 100. Action of Acetylene on Oxides of Copper (Gooch and Baldwin), II, 276. Action of Sodium Thiosulphate on Solutions of Metallic Salts at High Temperatures and Pressures (Norton), II, 384. ORGANIC CHEMISTRY. Blue Iodide of Starch (Roberts) I, 236. Action of Urea and Sulphocarbanilide on Acid Anhydrides (Dunlap), I, 355. Action of Urea and Primary Amines on Maleic Anhydrides (Dunlap and Phelps), 11,42. Ethers of Toluquinone- oxime and Space Isomerism of Nitrogen (Bridge and Morgan), II, 145. Space Isomerism of Toluquinoneoxime Ethers (Morgan), II, 283. Ethers of Isonitrosoguaiacol and Space Isomerism of Nitrogen (Bridge and Morgan), 11,304. MINERALOGICAL CHEMISTRY. Rhodochrosite from Franklin Furnace (Browning), I, 57. So-called Perofskite from Magnet Cove (Mar), I, 60. ANALYTICAL CHEMISTRY. QUALITATIVE ANALYSIS. Detection of Iodine, Bromine and Chlorine (Gooch and Brooks), I, 47. Detec- tion of Strontium and Calcium (Browning), 1, 121. Detection of Arsenic with Antimony and Tin (Gooch and Hodge), I, 231. Detection of Per- chlorates (Gooch and Kreider), I, 246. Reduction of Arsenic Acid (Gooch and Phelps), I, 265. Separation and Identification of Potassium and Sodium (Kreider and Breckenridge), I, 401. Detection of Sulphides, Sulphates, Sulphites, Thiosulphates (Browning and Howe), II, 134. Separation of Nickel from Cobalt (Browning and Hart well), II, 344. 396 SYSTEMATIC INDEX. QUANTITATIVE ANALYSIS. Colorimetric Methods. Detection and Approximative Estimation of Minute Amounts of Arsenic in Copper (Gooch and Moseley), I, 272. Electrolytic Methods. Determination of Halogens in mixed Silver Salts (Gooch and Fairbanks), I, 290. Spectroscopic Methods. Determination of Potassium (Gooch and Hart), I, 92. Determination of Ru- bidium (Gooch and Phinney), I, 157. Gravimetric Methods. Determination of Chlorine in Alkaline Chlorides and Iodides (Gooch and Mar), I, 18. Determination of Bromine in Alkaline Bromides and Iodides (Gooch and Ensign), I, 37. Estimation of Barium as Sulphate (Mar), I, 63. Sepa- ration of Strontium from Calcium (Browning), I, 107. Separation of Barium from Calcium (Browning), I, 116. Determination of Barium in presence of Calcium and Magnesium (Mar), I, 125. Separation of Barium from Stron- tium (Browning), I, 168. Influence of Nitric Acid and Aqua Regia on the Precipitation of Barium as Sulphate (Browning), I, 181. Treatment of Barium Sulphate (Phiuney), I, 187. Separation of Copper from Cadmium (Browning), I, 226. Determination of Potassium (Kreider), I, 282. De- termination of Carbon Dioxide (Gooch and Phelps), I, 302. Determination of Selenium (Peirce),!, 365. Estimation of Cadmium as Oxide (Browning and Jones), 1, 409. Separation of Aluminum from Iron (Gooch and Havens), II, 20. Separation of Aluminum and Beryllium (Havens), II, 47. Esti- mation of Manganese as Sulphate and as Oxides (Gooch and Austin), II, 77. Estimation of Manganese separated as Carbonate (Austin), II, 96. Separations of Aluminum by Hydrochloric Acid (Havens), II, 106. De- termination of Manganese as Pyrophosphate (Gooch and Austin), II, 121. Separation of Nickel and Cobalt by Hydrochloric Acid (Havens), II, 141. Estimation of Boric Acid (Gooch and Jones), II, 172. Ammonium Magne- sium Phosphate of Analysis (Gooch and Austin), II, 190. Volatilization of Iron Chlorides and Separation of Oxides of Iron and Aluminum (Gooch and Havens), II, 215. Double Ammonium Phosphates of Beryllium, Zinc, Cadmium (Austin), II, 252. Separation of Iron from Chromium, Zirconium, Beryllium, by Gaseous Hydrochloric Acid (Havens and Way), II, 266. Ammonium Magnesium Arseniate of Analysis (Austin), II, 209. Estimation of Thallium as Acid and Neutral Sulphates (Browning), II, 317. Separation and Determination of Mercury as Oxalate (Peters), II, 325. Sulphocyanides of Copper and Silver (Van Name), II, 359. Estima- tion of Caesium and Rubidium as the Acid Sulphates and of Sodium and Potassium as the Pyrosulphates (Browning), II, 368. Volumetric Methods. Standard Solutions. Tartar Emetic (Gruener), I, 216. Potassium Permanganate (Roberts), I, 269. lodometric Processes. Determination of Iodine in Haloid Salts (Gooch and Brown- ing), 1, 1. Reduction of Arsenic Acid (Gooch and Browning), I, 30. De- SYSTEMATIC INDEX. 397 termination of Antimony and its Condition of Oxidation (Gooch and Gruener), 1, 73. Estimation of Chlorates (Gooch and Smith), I, 82. Sepa- ration of Antimony from Arsenic (Gooch and Banner), I. 36. Determina- tion of Nitrates (Gooch and Gruener), I, 132. Determination of Iodine in Haloid Salts by action of Arsenic Acid (Gooch and Browning), I, 173. Determination of Nitrates (Gruener), I, 193. Estimation of Chlorates and Nitrates, and of Nitrites and Nitrates (Roberts), 1, 219. Estimation of Tellu- ric Acid (Gooch and Rowland), I, 277. Reduction of Acids of Selenium by Hydriodic Acid (Gooch and Reynolds), I, 310. Determination of Per- chlorates (Kreider), I, 316. Reduction of Selenic Acid (Gooch and Evans), I, 331. Reduction of Selenic Acid (Gooch and Scoville), I, 335. Deter- mination of Selenious and Selenic Acids (Gooch and Peirce), I, 338. Inter- action of Chromic and Arsenious Acids (Browning), I, 344. Separation of Selenium from Tellurium (Gooch and Peirce), 1, 348. Determination of Car- bon Dioxide (Phelps), I, 369. Estimation of Molybdic Acid (Gooch and Fair- banks), I, 375. Determination of Phosphorus in Iron (Fairbanks), I, 391. Reduction of Vanadic Acid (Browning), I, 397. Estimation of Vanadium (Browning and Goodman), II, 4. Determination of Oxygen in Air and Aqueous Solution (Kreider), II, 11. Estimation of Molybdenum (Gooch), II, 27. Application of lodic Acid to Analysis of Iodides (Gooch and Walker), II, 33. Titration of Sodium Thiosnlphate with lodic Acid (Walker), II, 52. Determination of Molybdenum (Gooch and Norton), II, 111. Analysis of Alkalies and Acids (Walker and Gillespie), II, 162. Influence of Hydrochloric Acid in Titrations by Thiosulphate, and Estimation of Se- lenious Acid (Norton), II, 206. Estimation of Iron in the Ferric State (Norton), II, 230. Determination of Tellurous Acid in presence of Haloid Salts (Gooch and Peters), II, 238. Estimation of Boric Acid (Jones), II, 244. Determination of Gold (Gooch and Morley), II, 269. Estimation of Cerium (Browning), II, 289. Estimation of Thallium (Browning and Hutchins), II, 300. Titration of Mercury by Sodium Thiosulphate (Norton), II, 328. Estimation of Arsenic Acid (Gooch and Morris), II, 336. Alkalimetric Processes. Estimation of Boric Acid (Jones), II, 182. Oxidimetric Processes. Determination of Selenious Acid (Gooch and demons), I, 297. Titration of Oxalic Acid in presence of Hydrochloric Acid (Gooch and Peters), II, 222. Determination of Tellurous Acid in presence of Haloid Salts (Gooch and Peters), II, 238. Separation and Determination of Mercury as Mercurous Oxalate (Peters), II, 320. Estimation of Copper as Oxalate, with Separations (Peters), II, 347. Estimation of Calcium, Strontium, and Barium as the Oxalates (Peters), II, 373. Precipitation Processes. Determination of Tellurium by Precipitation as the Iodide (Gooch and Morgan), II, 1. Gasometric Processes. Reduction of Nitric Acid by Ferrous Salts (Roberts), I, 203. Estimation of Chlorates and Nitrates, and of Nitrites and Nitrates (Roberts), I, 219. INDEX OF AUTHORS. AUSTIN, MARTHA. Estimation of Manganese as Sulphate and Oxide (with Gooch, F. A.) H, 77 Condition of Oxidation of Manganese precipitated by Chlorate Process (with Gooch, F. A.) II, 85 Estimation of Manganese Separated as Carbonate II, 96 Determination of Manganese as Pyrophosphate (with Gooch, F. A.) II, 121 Constitution of Ammonium Magnesium Phosphate of Analysis (with Gooch, F. A.) II, 190 Double Ammonium Phosphates of Beryllium, Zinc, Cadmium, in Analysis II, 252 Constitution of Ammonium Magnesium Arseniate of Analysis . . II, 309 BALDWIN, DEFOREST. Action of Acetylene on Oxides of Copper (with Gooch, F. A.) II, 276 BRECKENRIDGE, J. E. Separation and Identification of Potassium and Sodium (with Kreider, D. Albert) I, 401 BRIDGE, JOHN L. Ethers of Toluquinoneoxime, and Space Isomerism of Nitrogen (with Morgan, Wm. Conger) II, 145 Ethers of Isonitrosoguaiacol and Space Isomerism of Nitrogen (with Morgan, Wm. Conger) II, 304 BROOKS, F. T. Detection of Iodine, Bromine, and Chlorine (with Gooch, F. A.) I, 47 BROWNING, PHILIP E. Determination of Iodine in Haloid Salts (with Gooch, F. A.) I, 1 Reduction of Arsenic Acid in Analysis (with Gooch, F. A.) ... I, 30 Analysis of Rhodochrosite from Franklin Furnace I, 57 Quantitative Separation of Strontium from Calcium by Amyl Alcohol on Nitrates I, 107 Quantitative Separation of Barium from Calcium by Amyl Alcohol on Nitrates I, 116 Separation and Detection of Strontium and Calcium by Amyl Alcohol on Nitrates . . 1,121 Quantitative Separation of Barium from Strontium by Amyl Alcohol on Bromides I, 168 Determination of Iodine in Haloid Salts by Action of Arsenic Acid (with Gooch, F. A.) I, 173 Influence of Nitric Acid and Aqua Regia on Precipitation of Barium as Sulphate I, 181 Separation of Copper from Cadmium by Iodide Method .... I, 226 Interaction of Chromic and Arsenious Acids I, 344 Reduction of Vanadic Acid by Hydrobromic and Hydriodic Acids, and Estimation by Iodine I, 397 400 INDEX OF AUTHORS. VOL. PAGE Estimation of Cadmium as Oxide (with Jones, Louis C.) . . . . I, 409 Application of Organic Acids to Estimation of Vanadium (with Goodman, Richard J.) II, 4 Detection of Sulphides, Sulphates, Sulphites, and Thiosulphates (with Howe, Ernest) II, 134 Volumetric Estimation of Cerium (with Hanford, G. A. ; Hall, F. J. ; Cutter, Wm. D. ; Lynch, Leo A.) II, 289 Estimation of Thallium as Chromate (with Hutchius, George P.) . II, 300 Estimation of Thallium as Acid and Neutral Sulphates 11,317 Qualitative Separation of Nickel from Cobalt by Ammonia on the Ferricyanides (with Hartwell, John B.) II, 344 On the Estimation of Caesium and Rubidium as the Acid Sul- phates, and of Potassium and Sodium as the Pyrosulphates . . II, 368 CLEMONS, C. F. Determination of Selenious Acid by Potassium Perman- ganate (with Gooch, F. A.) I, 297 CUTTER, WM. D. Volumetric Estimation of Cerium (with Browning, Philip E.) 11,294 DANNER, E. VV. Separation of Antimony from Arsenic by Hydrochloric and Hydriodic Acids (with Gooch, F. A.) I, 86 Interaction of Potassium Permanganate and Sulphuric Acid (with Gooch, F. A.) I, 145 DONLAP, FREDERICK L. Action of Urea and Sulphocarbanilide on Acid Anhydrides 1, 355 Action of Urea and Primary Amines on Maleic Anhydride (with Phelps, Isaac K.) 11,42 ENSIGN, J. R. Determination of Bromine in Alkaline Bromides and Iodides (with Gooch, F. A.) I, 37 EVANS, P. S., JR. Reduction of Selenic Acid by Hydrochloric Acid (with Gooch, F. A.) 1,331 FAIRBANKS, CHARLOTTE. Estimation of Halogens in Silver Salts (with Gooch, F. A.) . . 1,290 lodometric Estimation of Molybdic Acid (with Gooch, F. A.) . . I, 375 lodometric Determination of Phosphorus in Iron I, 391 GILLESPIE, DAVID H. M. Iodine in Analysis of Acids and Alkalies (with Walker, Claude F.) II, 162 GOOCH, F. A. Determination of Iodine in Haloid Salts (with Browning, P.E.) I, 1 Determination of Chlorine in Alkaline Chlorides and Iodides (with Mar, F. W.) I, 18 Reduction of Arsenic Acid in Analysis (with Browning, P. E.) . . I, 30 Determination of Bromine in Alkaline Bromides and Iodides (with Ensign, J. R.) I, 37 Detection of Iodine, Bromine, and Chlorine (with Brooks, F. T.) . I, 47 Determination of Antimony and its Condition of Oxidation (with Gruener, H. W.) I, 73 Estimation of Chlorates (with Smith, C. G.) I, 82 Separation of Antimony from Arsenic by Hydrochloric and Hydri- odic Acids (with Danner, E. W.) I, 86 Detection and Determination of Potassium Spectroscopically (with Hart, T. S.) I, 92 lodometric Determination of Nitrates (with Gruener, H. W.) . . I, 132 Laboratory Apparatus I, 141 INDEX OF AUTHORS. 401 VOL. PAGE Interaction of Potassium Permanganate and Sulphuric Acid (with Banner, E. W.) I, 145 Quantitative Determination of Rubidium by the Spectroscope (with Phinney, J. I.) I, 157 Determination of Iodine in Haloid Salts by Action of Arsenic Acid ( with Browning, P. E.) 1,173 Detection and Separation of Arsenic with Antimony and Tin (with Hodge, B.) 1,231 Detection of Alkaline Perchlorates with Chlorides, Chlorates, and Nitrates (with Kreider, D. Albert) I, 246 Generation of Chlorine (with Kreider, D. Albert) I, 260 Reduction of Arsenic Acid by Hydrochloric Acid and Potassium Bromide (with Phelps, I. K.) I, 265 Detection and Estimation of Minute Amounts of Arsenic in Copper (with Moseley, H. P.) I, 272 lodometric Estimation of Telluric Acid (with Rowland, J.) ... I, 277 Estimation of Halogens in Silver Salts (with Fairbanks, Charlotte) I, 290 Determination of Selenious Acid by Potassium Permanganate (with demons, C. F.) I, 297 Precipitation and Gravimetric Determination of Carbon Dioxide (with Phelps, I. K.) I, 302 Reduction of Acids of Selenium by Hydriodic Acid (with Reynolds, W.G.) 1,310 Reduction of Selenic Acid by Hydrochloric Acid (with Evans, P. S., Jr.) 1,331 Reduction of Selenic Acid by Potassium Bromide in Acid Solution (witb Scoville, W. S.) I, 335 lodometric Determination of Selenious and Selenic Acids (with Peirce, A. W.) I, 338 Separation of Selenium from Tellurium by difference in Volatility of Bromides (with Peirce, A. W.) I, 348 lodometric Estimation of Molybdic Acid (with Fairbanks, Char- lotte) 1,375 Determination of Tellurium by precipitation as Iodide (with Mor- gan, W. C.) II, 1 Separation of Aluminum from Iron (with Havens, F. S.) . . . . II, 20 Estimation of Molybdenum lodometrically II, 27 Application of lodic Acid to Analysis of Iodides (with Walker, C. F.) II, 33 Estimation of Manganese as Sulphate and Oxide (with Martha Austin) II, 77 Condition of Oxidation of Manganese precipitated by Chlorate Process (with Austin, Martha) II, 85 lodometric Determination of Molybdenum (with Norton, John T., Jr.) 11,111 Determination of Manganese as Pyrophosphate (with Austin, Martha) II, 121 Estimation of Boric Acid (with Jones, Louis Cleveland) . . . . II, 172 Constitution of Ammonium Magnesium Phosphate of Analysis (with Austin, Martha) II, 190 Volatilization of Iron Chlorides, and Separation of Oxides of Iron and Aluminum (with Havens, Franke Stuart) II, 215 VOL. n. 26 402 INDEX OF AUTHORS. VOL. PAGE Titration of Oxalic Acid by Potassium Permanganate in presence of Hydrochloric Acid (with Peters, C. A.) II, 222 Determination of Tellurous Acid in presence of Haloid Salts (with Peters, C. A.) II, 238 lodometric Determination of Gold (with Morley, Frederick H.) . II, 269 Action of Acetylene on Oxides of Copper (with Baldwin, De Forest) II, 276 lodometric Estimation of Arsenic Acid (with Morris, Julia C.) . . II, 236 GOODMAN, RICHARD J. Application of Organic Acids to Estimation of Vanadium (with Browning, Philip E.) II, 4 GRUBNER, H. W. Determination of Antimony and its condition of Oxi- dation (with Gooch, F. A.) I, 73 lodometric Determination of Nitrates (with Gooch, F. A.) ... I, 132 lodometric Determination of Nitrates I, 193 Stability of Standard Solutions of Tartar Emetic I, 216 HALL, F. J. Volumetric Estimation of Cerium (with Browning, Philip E.) II, 290 HANFORD, G. A. Volumetric Estimation of Cerium (with Browning, Philip E.) 11,290 HART, T. S. Detection and Determination of Potassium Spectroscopi- cally (with Gooch, F. A.) I, 92 HARTWELL, JOHN B. Qualitative Separation of Nickel from Cobalt by Ammonia on the Ferricyanides (with Browning, Philip E.) . . II, 344 HAVENS, FRANKE STUART. Separation of Aluminum from Iron (with Gooch, F. A.) II, 20 Separation of Aluminum and Beryllium by Hydrochloric Acid . . II, 47 Further Separations of Aluminum by Hydrochloric Acid . . . . II, 106 Separation of Nickel and Cobalt by Hydrochloric Acid II, 141 Volatilization of Iron Chlorides, and Separation of Oxides of Iron and Aluminum (with Gooch, F. A.) II, 215 Separation of Iron from Chromium, Zirconium, and Beryllium by Action of Gaseous Hydrochloric Acid on the Oxides (with Way, Arthur Fitch) II, 266 HODGE, B. Detection and Separation of Arsenic with Antimony and Tin (with Gooch, F. A.) I, 231 HOWE, ERNEST. Detection of Sulphides, Sulphates, Sulphites, and Thiosulphates (with Browning, Philip E.) II, 134 HOWLAND, J. lodometric Estimation of Telluric Acid (with Gooch, F. A.) I, 277 HUTCHINS, GEORGE P. Estimation of Thallium as Chromate (with Browning, Philip E.) II, 300 JONES, Louis CLEVELAND. Estimation of Cadmium as Oxide (with Browning, Philip E.) I, 409 Action of Carbon Dioxide on Soluble Borates II, 100 Estimation of Boric Acid (with Gooch, F. A.) II, 172 Volumetric Estimation of Boric Acid 11,182 lodometric Estimation of Boric Acid . II, 244 KREIDER, D. ALBERT. Detection of Alkaline Perchlorates with Chlo- rides, Chlorates, and Nitrates (with Gooch, F. A.) I, 246 Generation of Chlorine (with Gooch, F. A.) I, 260 Preparation of Perchloric Acid and Determination of Potassium . I, 282 Laboratory Apparatus I, 306 Quantitative Determination of Perchlorates I, 316 INDEX OF AUTHORS. 403 VOL. PAGE Separation and Identification of Potassium and Sodium (with Breckenridge, J. E.) . . I, 401 Determination of Oxygen in Air and Aqueous Solution .... II, 1 1 LYNCH, LEO A. Volumetric Estimation of Cerium (with Browning, Philip E.) II, 297 MAR, F. W. Determination of Chlorine in Alkaline Chlorides and Iodides (with Gooch, F. A.) I, 18 So-called Perofskite from Magnet Cove I, 60 Estimation of Barium as the Sulphate I, 63 Determination of Barium in presence of Calcium and Magne- sium I, 125 MORGAN, WM. CONGER. Determination of Tellurium hy precipitation as Iodide (with Gooch, F. A.) II, 1 Ethers of Toluquinoneoxime and Space Isomerism of Nitrogen (with Bridge, John L.) II, 145 Space Isomerisms of Toluquinoneoxime Ethers II, 283 Ethers of Isonitrosoguaiacol and Space Isomerism of Nitrogen (with Bridge, John L.) II, 304 MORLEY, FREDERICK H. lodometric Determination of Gold (with Gooch, F. A.) H, 269 MORRIS, JULIA C. lodometric Estimation of Arsenic Acid (with Gooch, F. A.) II, 336 MOSELEY, H. P. Detection and Estimation of Minute Amounts of Arsenic in Copper (with Gooch, F. A.) I, 272 NORTON, JOHN T., Jr. lodometric Determination of Molybdenum (with Gooch, F. A.) II, 111 Hydrochloric Acid in Titrations by Sodium Thiosulphate, and Esti- mation of Selenious Acid II, 206 Estimation of Iron in Ferric Condition by Sodium Thiosulphate and Iodine II, 230 Titration of Mercury by Sodium Thiosulphate II, 328 The Action of Sodium Thiosulphate on Solutions of Metallic Salts at High Temperatures and Pressures II, 384 PEIRCE, A. W. lodometric Determination of Selenious and Selenic Acids (with Gooch, F. A.) I, 338 Separation of Selenium from Tellurium by difference in Volatility of Bromides (with Gooch, F. A.) I, 348 Gravimetric Determination of Selenium I, 365 Existence of Selenium Monoxide I, 385 PETERS, CHARLES A. Titration of Oxalic Acid by Potassium Permangan- ate in presence of Hydrochloric Acid (with Gooch, F. A.) ... II, 222 Determination of Tellurous Acid in presence of Haloid Salts (with Gooch, F. A.) 11,238 Determination of Mercury as Mercurous Oxalate H, 320 Volumetric Estimation of Copper with Separation from Cadmium, Arsenic, Tin, Iron, and Zinc II, 347 The Estimation of Calcium, Strontium, and Barium as the Oxalates. II, 373 PHELPS, ISAAC K. Reduction of Arsenic Acid by Hydrochloric Acid Potassium Bromide (with Gooch, F. A.) I, 265 Precipitation and Gravimetric Determination of Carbon Dioxide (with Gooch, F. A.) I, 302 lodometric Determination of Carbon Dioxide I, 369 404 INDEX OF AUTHORS. VOL. PAGE Action of Urea and Primary Amines on Maleic Anhydride (with Duulap, Frederick L.) II, 42 Combustion of Organic Substances in the Wet Way II, 62 PHINNEY, J. L Quantitative Determination of Rubidium by the Spec- troscope (with Gooch, F. A.) I, 157 Treatment of Barium Sulphate in Analysis I, 187 REYNOLDS, W. G. Reduction of Acids of Selenium by Hydriodic Acid ( with Gooch, F. A.) 1,310 ROBERTS, CHARLOTTE F. Reduction of Nitric Acid by Ferrous Salts I, 203 Estimation of Chlorates and Nitrates, and of Nitrites and Nitrates . I, 219 Blue Iodide of Starch I, 236 Action of Reducing Agents on lodic Acid I, 250 Standardization of Potassium Permanganate in Iron Analysis . . I, 269 SCOVILLE, W. S. Reduction of Selenic Acid by Potassium Bromide in Acid Solution (with Gooch, F. A.) I, 335 SMITH, C. G. Estimation of Chlorates (with Gooch, F. A.) .... I, 82 VAN NAME, R. G. The Sulphocyanides of Copper and Silver in Gravi- metric Analysis II, 359 WALKER, CLAUDE. Application of lodic Acid to Analysis of Iodides (with Gooch, F. A.) II, 33 Titration of Sodium Thiosulphate with lodic Acid II, 52 Iodine in Analysis of Acids and Alkalies (with Gillespie, David H. M.) II, 162 WAY, ARTHUR FITCH. Separation of Iron from Chromium, Zir- conium, and Beryllium, by action of Gaseous Hydrochloric Acid on the Oxides (with Havens, Franke Stuart) II, 266 INDEX OF SUBJECTS. VOL. PAGE Acetylene, action of, on oxides of copper (Gooch and Baldwin) . . . 11,276 Acids, application of iodine in analysis of (Walker and Gillespie) . . II, 162 of selenium, reduction of, by hydriodic acid (Gooch and Reynolds) . I, 310 Acid anhydrides, action of urea and sulphocarbanilide upon (Dunlap) . I, 355 Alkalies, application of iodine to analysis of (Walker and Gillespie) . II, 162 Alkaline bromides, determination of bromine in (Gooch and Ensign) . I, 37 Alkaline chlorides, determination of chlorine in alkaline iodides mixed with (Gooch and Mar) I, 18 Alkaline iodides, determination of chlorine in (Gooch and Mar) . . I, 18 determination of bromine in (Gooch and Ensign) ...... I, 37 Alkaline perchlorates, detection of, associated with chlorides, chlorates, and nitrates (Gooch and Kreider) 1,246 Aluminum salts, action of sodium thiosulphate upon, at high tempera- tures and pressures (Norton) II, 388 Aluminum, separation of, by hydrochloric acid, from iron (Gooch and Havens) II, 20 separation of, by hydrochloric acid, from beryllium (Havens) . . II, 47 separation of, by hydrochloric acid, from bismuth, copper, and . . . mercury (Havens) II, 109 separation of, by hydrochloric acid, from zinc (Havens) .... II, 107 Aluminum oxide, separation of oxides of iron from (Gooch and Havens) II, 215 Ammonium magnesium arseniate in analysis, constitution of (Austin) . II, 309 Ammonium beryllium phosphate in analysis (Austin) II, 253 Ammonium cadmium phosphate in analysis (Austin) II, 262 Ammonium magnesium phosphate in analysis (Gooch and Austin) . . 11,190 Ammonium zinc phosphate in analysis (Austin) II, 257 Amyl alcohol, use of, in detecting strontium and calcium (Browning) . I, 121 use of, in separating strontium and calcium (Browning) use of, in separating barium and calcium (Browning) use of, in separating barium and strontium (Browning) .... Antimony, detection of arsenic associated with (Gooch and Hodge) . . determination of, and its condition of oxidation (Gooch and Gruener) separation of, from arsenic, by hydrochloric and hydriodic acids (Gooch and Banner) Antimonious chloride, decomposition of nitrates by (Gruener) .... Antimouic acid, salts of, reduced by potassium iodide and sulphuric acid, and estimated iodometrically (Gooch and Gruener) Apparatus burette clip (Gooch) 121 116 ^168 231 73 86 199 73 141 ,264 ,308 chlorine generator (Gooch and Kreider) force pump (Kreider) hot filter (Kreider) I, 306 mercury washer (Gooch) 1,143 steam evaporator (Gooch) I, 142 support (Gooch) I, 142 406 INDEX OF SUBJECTS. VOL. PAGE valve (Kreider) 1, 307 used in analysis of iodides by iodic acid (Gooch and Walker) ... II, 37 used in combustion of organic substances in the wet way (Phelps . II, 68 used in estimation of carbon dioxide gravimetrically (Gooch and Phelps) 1,302 used in estimation of iodine in haloid salts (Gooch and Browning) . I, 12 used in estimation of molybdenum (Gooch and Norton) . . . . 11,114 used in estimation of molybdic acid (Gooch and Fairbanks) . I, 378, 382 used in estimation of molybdic acid (Fairbanks) I, 394 used in estimation of nitrates (Gooch and Gruener) I, 137 used in estimation of oxygen in air and aqueous solution (Kreider) II, 17 used in estimation of selenium iodometrically, by volatilization of the bromide (Gooch and Peirce) I, 350 used in reduction of arsenic acid (Gooch and Browning) .... I, 33 used in reduction of nitric acid by ferrous salts (Roberts) . . . 1,208 Arsenic, detection of, associated with antimony and tin (Gooch and Hodge) 1,231 detection and approximative estimation of, in copper (Gooch and Moseley) 1, 272 separation of antimony from, by hydrochloric and hydriodic acids (Gooch and Danner) I, 86 separation of copper as oxalate from (Peters) II, 347 Arsenious acid, action of, upon cerium dioxide (Browning and Cutter) II, 294 interaction of, with chromic acid (Browning) I, 344 Arsenic acid, determination of, by reduction with potassium iodide and sulphuric acid, and titration by iodine in alkaline solution (Gooch and Browning) I, 30 iodometric estimation of (Gooch and Morris) . II, 336 reduction of, in analysis (Gooch and Browning) II, 30 reduction of, by action of hydrochloric acid and potassium bro- mide (Gooch and Phelps) . 1, 265 use of, to liberate iodine in quantitative estimation of iodides (Gooch and Browning) I, 1 use of, in determination of iodine in haloid salts (Gooch and Brown- ing) I, 1 Aqua regia, influence of, on the precipitation of barium as the sul- phate (Browning) I, 181 Barium, determination of, in presence of calcium and magnesium (Mar) I, 125 estimation of, as oxalate (Peters) II, 373 points in estimation of, as sulphate (Mar) .... ... I, 63 precipitation of, as sulphate, in presence of nitric acid and aqua regia (Browning) I, 181 quantitative separation of, from calcium by amyl alcohol on the ni- trates (Browning) I, 116 quantitative separation of, from strontium by amyl alcohol on the bromides (Browning) I, 168 Barium chlorides, precipitation and separation of, from calcium and mag- nesium, by hydrochloric acid and ether (Mar) I, 125 Barium sulphate, influence of hydrochloric acid upon precipitation of (Mar) I, 63 purification of, by crystallizing from sulphuric acid (Mar) .... I, 71 treatment of, in analysis (Phinney) I, 187 INDEX OF SUBJECTS. 407 VOL. PAGE Beryllium, separation of, from aluminum, by action of hydrochloric acid (Havens) II, 47 separation of iron from, by action of hydrochloric acid (Havens and Way) 11,266 Beryllium ammonium phosphate in analysis (Austin) II, 253 Beryllium salt, action of sodium thiosulphate upon, at high tempera- tures and pressures (Norton) n, 391 Bismuth, separation of aluminum from (Havens) II, 109 Blue iodide of starch (Eoberts) I, 236 Borates (soluble), action of carbon dioxide on (Jones) . II, 100 Boric acid, estimation of (Gooch and Jones) II, 172 iodometric method for estimation of (Jones) II } 244 use of calcium oxide as a retainer for (Gooch and Jones) .... 11,175 use of sodium tungstate as a retainer for (Gooch and Jones) . . . II, 178 volumetric estimation of (Jones) II, 182 Bromine, detection of, in presence of chlorine and iodine (Gooch and Brooks) I, 47 , determination of, in alkaline bromides and iodides (Gooch and Ensign) I, 37 volatilization of, from aqueous solutions of bromide and chloride by action of sulphuric acid and nitrous acid (Gooch and Ensign) . I, 43 Cadmium, estimation of, as oxide (Browning and Jones) I, 409 Separation of copper from, by the iodide method (Browning) . . I, 226 Separation of copper from, as oxalate (Peters) II, 354 Cadmium ammonium phosphate, in analysis (Austin) II, 262 Caesium, estimation of, as the acid sulphate (Browning) II, 368 Calcium, determination of barium in presence of (Mar) I, 125 estimation of, as oxalate (Peters) II, 373 quantitative separation of barium from, by action of amyl alcohol on the nitrates (Browning) ' I, 1 1 6 separation of, from strontium, and detection of, by action of amyl alcohol on the nitrates (Browning) 1,121 Calcium oxide, use of, as a retainer for boric acid (Gooch and Jones) . II, 175 Carbon dioxide, action of on soluble borates (Jones) II, 100 iodometric method for determination of (Phelps) 1,369 precipitation and gravimetric determination of (Gooch and Phelps) I, 302 Cerium, modified Bunsen method for determination of (Browning, Han- ford, and Hall) II, 290 volumetric estimation of (Browning) II, 289 Cerium dioxide, action of arsenious acid upon (Browning and Cutter) . II, 294 Cerium oxalate, estimation of, by potassium permanganate (Browning and Lynch) 11,297 Chlorates, detection of -perchlorates associated with (Gooch and Kreider) I, 246 estimation of (Gooch and Smith) I, 82 Chlorates and nitrates, estimation of, in one operation (Roberts) ... 1,219 Chlorate process, condition of oxidation of manganese precipitated in (Gooch and Austin) II, 85 Chlorides, detection of perchlorates associated with (Gooch and Kreider) I, 246 Chlorine, detection of, in presence of bromides and iodides (Gooch and Brooks) I, 47 direct determination of, in alkaline chlorides and iodides (Gooch and Mar) I, 18 408 INDEX OF SUBJECTS. generation of, by hydrochloric acid and potassium chlorate (Gooch and Kreider) I, 260 Chromium salt, action of sodium thiosulphate upon, at high tempera- tures and pressures (Norton) II, 389 Chromium, separation of iron from, by gaseous hydrochloric acid (Ha- vens and Way) II, 266 Chromic acid, interaction of, with arsenious acid (Browning) .... I, 344 use of, in combustion of organic substances in the wet way (Phelps) II, 67 Cobalt, separation of, from nickel (Havens) II, 141 separation of nickel from, by action of ammonium hydroxide on the ferricyanides (Browning and Hart well) II, 344 Combustion of organic substances in the wet way (Phelps) II, 62 Copper, detection and approximate estimation of minute amounts of arsenic in (Gooch and Moseley) I, 272 estimation of, as oxalate, with separation from cadmium, arsenic, tin, iron, and zinc (Peters) II, 347 preparation of, free from arsenic (Gooch and Moseley) I, 275 separation of aluminum from (Havens) II, 109 separation of, from cadmium by the iodide method (Browning) . . I, 226 Copper oxides, action of acetylene on (Gooch and Baldwin) 11,276 Copper sulphocyanide in gravimetric analysis (Van Name) 11,359 Dibrommaleinamide, preparation of, from urea and dibrommaleic anhy- dride (Dunlap) I, 358 Dibrommale'inuric acid, preparation of, from urea and dibrommaleic an- hydride (Dunlap) 1, 358 Dibromtoluquinonemetaoxime beuzoyl ether (Bridge and Morgan) . . II, 157 Dibromtoluquinoneorthooxime benzoyl ether (Bridge and Morgan) . . II, 161 Dibromtoluquinoneorthooxime methyl ether (Bridge and Morgan) . . II, 159 Dichlormaleinimide preparation of, from urea and dichlormale'ic anhy- dride (Dunlap) 1, 357 Dichlormale'inuric acid, preparation of, by action of urea on dichlorma- leic anhydride (Dunlap) I, 356 Double ammonium phosphates of beryllium, zinc, and cadmium in ana- lysis (Austin) II, 252 Electrolytic iron, use of, in standardizing permanganate solutions (Rob- erts) 1,269 Ethers of toluquinoneoxime, and their bearing on the space isomerism of nitrogen (Bridge and Morgan) II, 145 Ferric alum, use of, with nitric acid, to liberate iodine from haloid salts (Gooch and Mar) I, 23 Ferrous salts, use of, in reduction of nitric acid (Roberts) I, 203 Gold, iodometric determination of (Gooch and Morley) II, 269 Halogens, estimation of, in mixed silver salts (Gooch and Fairbanks) . I, 290 Haloid salts, determination of iodine in (Gooch and Browning) ... I, 2 determination of tellurous acid in presence of (Gooch and Peters) . II, 238 Hydriodic acid, action of, with hydrochloric acid in separation of anti- mony from arsenic (Gooch and Danner) I, 86 use of, in reduction of acids of selenium (Gooch and Reynolds) . . I, 310 use of, in reduction of vanadic acid ( Browning) I, 397 Hydrobromic acid, use of, in reduction of vanadic acid (Browning) . . I, 397 Hydrochloric acid, action of, with hydriodic acid, in separation of anti- mony from arsenic (Gooch and Danner) I, 86 INDEX OF SUBJECTS. 409 VOL. PAGE influence of, upon the precipitation of barium sulphate (Mar) ... II, 63 influence of, in titrations by sodium thiosulphate, with special refer- ence to the estimation of selenious acid (Norton) II, 206 use of, in reduction of seleuic acid {Gooch and Evans) I, 331 use of, in separation of aluminum from iron (Gooch and Havens) . II, 20 use of, in separation of aluminum from zinc, copper, mercury, bis- muth (Havens) II, 106 use of, with ether, to precipitate barium chloride in presence of salts of magnesium and calcium (Mar) 1,125 use of, with potassium bromide, in reducing and volatilizing arsenic acid (Gooch and Phelps) I, 265 use of, with potassium bromide, in separating arsenic from copper (Gooch and Moseley) I, 272 use of, with potassium chlorate, to generate chlorine (Gooch and Kreider) 1, 260 use of, with potassium iodide in volatilizing arsenic (Gooch and Hodge) 1, 231 titration of oxalic acid by potassium permanganate in presence of (Gooch and Peters) 11,222 volatility of, in aqueous solutions containing sulphuric acid and so- dium chloride (Gooch and Mar) I, 19 Hydrochloric acid (gaseous) use of, in separation of iron from chromium, zirconium, and beryllium (Havens and Way) II, 266 lodic acid, action of iodine on, in presence of hydrochloric acid (Roberts) I, 257 action of reducing agents on, in presence of hydrochloric acid (Rob- erts) I, 252 application to the analysis of iodides (Gooch and Walker) .... II, 33 use of, in absorption of nitric oxide (Roberts) 1,250 use of, in titration of sodium thiosulphate (Walker) II, 52 Iodides, application of iodic acid to the analysis of (Gooch and Walker) . II, 33 Iodide method, use of, in separating copper from cadmium (Browning) . I, 226 Iodine, action of, on iodic acid in presence of hydrochloric acid (Roberts) I, 257 application of, in analysis of alkalies and acids (Walker and Gilles- pie) 11,162 detection of, in presence of chlorine and bromine (Gooch and Brooks) I, 47 .determination of, in haloid salts, by action of arsenic acid (Gooch and Browning) I, 1 liberation of, from haloid salts, by arsenic acid (Gooch and Brown- ing) I, 1 liberation of, from haloid salts, by ferric alum with nitric acid (Gooch and Mar) I, 23 liberation of, from haloid salts, by nitrous acid (Gooch and Mar) . I, 27 use of, in estimating iron reduced from the ferric state by sodium thiosulphate (Norton) II, 230 lodometric determination of gold (Gooch and Morley) II, 269 of molybdenum (Gooch and Norton) 11,111 of nitrates (Gooch and Gruener) I, 132 of nitrates (Gruener) 1,193 of selenious and selenic acids (Gooch and Peirce) I, 338 lodometric estimation of alkalies and acids (Walker and Gillespie) . . II, 162 of antimonic acid (Gooch and Gruener) I, 73 410 INDEX OF SUBJECTS. VOL. PAGE of antimony separated from arsenic (Gooch and Danner) .... I, 86 of arsenic acid (Gooch and Browning) I, 30 of arsenic acid (Gooch and Morris) II, 336 of boric acid (Jones) n, 244 of carbon dioxide (Phelps) I ? 359 of chlorates (Gooch and Smith) I, 82 of chromic acid (Browning) I ? 344 of cerium (Browning, Hanford, and Hall) II, 290 of gold (Gooch and Morley) II, 269 of iodides (Gooch and Walker) II, 33 of iodine in haloid salts (Gooch and Browning) I, i of iron (Norton) II, 230 of mercury (Norton) II, 328 of molybdenum (Gooch) II, 27 of molybdenum (Gooch and Norton) 11,111 of molybdic acid (Gooch and Fairbanks) I, 375 of nitrates ( Gooch and Gruener) 1,132 of nitrates (Gruener) I, 193 of oxygen, in air and in aqueous solution (Kreider) II, 11 of oxygen, in perchlorates (Kreider) I, 316 of phosphorus in iron (Fairbanks) I, 391 of selenious acid (Gooch and Reynolds) I, 310 of selenious acid (Gooch and Peirce) 1,338 of selenic acid (Gooch and Reynolds) 1,314 of selenic acid (Gooch and Peirce) I, 338 of selenium associated with tellurium (Gooch and Peirce) .... I, 348 of tellurous acid (Gooch and Peters) II, 238 of vanadic acid (Browning) 1,397 of vauadic acid (Browning and Goodman) II, 4 lodometric method for the determination of carbon dioxide (Phelps) . I, 369 for the determination of phosphorus in iron (Fairbanks) .... I, 391 for the estimation of boric acid (Jones) II, 244 Iron, estimation of, in the ferric state by reduction with sodium thiosul- phate and titration with iodine (Norton) II, 230 iodometric method for determination of phosphorus in (Fairbanks) I, 391 method for the separation of aluminum from (Gooch and Havens) . II, 20 separation of, from chromium, zirconium, and beryllium, by gase- ous hydrochloric acid ( Havens and Way) II, 266 separation of copper oxalate from (Peters) II, 347 Iron analysis, standardization of potassium permanganate in (Roberts) . I, 269 Iron oxides, separation of, from aluminum oxide (Gooch and Havens) . II, 215 Isonitrosoguaiacol, and salts of (Bridge and Morgan) II, 306 Isonitrosoguaiacol benzoyl ether (Bridge and Morgan) II, 307 Isonitrosoguaiacol benzoyl ether dibromide (Bridge and Morgan) . . II, 307 Isonitrosoguaiacol, ethers of, in their relation to the space isomerism of nitrogen (Bridge and Morgan) II, 304 Isomerism (space) of nitrogen, in ethers of isonitrosoguaiacol (Bridge and Morgan) II, 304 in ethers of toluquinoneoxime (Bridge and Morgan) 11,145 Isomerism (space) of the toluquinoneoxime ethers (Morgan) .... II, 283 Laboratory apparatus (Gooch) I, 141 Laboratory apparatus (Kreider) 1,306 INDEX OF SUBJECTS. 411 VOL. PAGE Magnesium, determination of, by precipitation as ammonium magne- sium phosphate (Gooch and Austin) 11,190 determination of barium in presence of (Mar) I, 125 Maleic anhydride, action of primary amines upon (Dnnlap and Phelps) II, 44 action of urea and primary amines upon (Dunlap and Phelps) . . II, 42 Maleiiric acid, preparation of, from urea and maleic anhydride (Dunlap andPhelps) II, 42 Manganese, determination of, as the pyrophosphate (Gooch and Austin) II, 121 condition of oxidation of, precipitated by the chlorate process (Gooch and Austin) II, 85 estimation of, as the sulphate and as the oxides (Gooch and Austin) II, 77 estimation of, separated as the carbonate (Austin) II, 96 Manganous chloride, use of, in hydrochloric acid, in detection of oxidiz- ing agents (Gooch and Gruener) I, 134 use of, in estimating nitrates (Gooch and Gruener) I, 132 Mercury, gravimetric estimation of, as the oxalate (Peters) II, 325 separation of aluminum from ( Havens) 11,109 titration of, by sodium thiosulphate (Norton) II, 328 Mercurous oxalate, separation and determination of (Peters) .... II, 320 Metallic salts, action of sodium thiosulphate upon, in solution at high temperatures and pressures (Norton) II, 384 Molybdenum, estimation of iodometrically (Gooch) II, 27 iodometric determination of (Gooch and Norton) 11,111 Molybdic acid, iodometric estimation of (Gooch and Fairbanks) ... I, 375 Monobromisonitrosoguaiacol benzoyl ether (Morgan) II, 308 Monobromtoluquinouemetaoxime benzoyl ether ( Bridge and Morgan) . II, 1 58 Monobromtoluquinoneorthooxime benzoyl ether (Morgan) II, 286 Naphthylmaleamic acid j8, preparation of, from 0-naphthylamine and maleic anhydride (Dunlap and Phelps) II, 45 Nickel, separation of, from cobalt (Havens) II, 141 separation from cobalt by action of ammonium hydroxide on the ferricyanides (Browning and Hart well) II, 344 Nitrates, action of phosphoric acid and potassium iodide upon (Gruener) I, 193 decomposition of, by antimonious chloride (Gruener) I, 199 detection of perchlorates associated with (Gooch and Kreider) . . I, 246 iodometric determination of (Gooch and Gruener) I, 132 odometric determination of (Gruener) I, 193 Nitrates and chlorates, estimation of, in one operation (Roberts) ... I, 219 Nitrates and nitrites, estimation of, in one operation (Roberts) ... I, 222 Nitric acid, influence of, in precipitation of barium as the sulphate (Browning) I, 181 reduction of, by ferrous salts (Roberts) I, 203 Nitric oxide, absorption of, by iodic acid (Roberts) I, 250 Nitrogen, space isomerism of, and bearing of ethers of toluquinoneoxime (Bridge and Morgan), (Morgan) II, 145, 283 space isomerism of, in ethers of isonitrosoguaiacol (Bridge and Morgan), (Morgan) II, 304 Nitrous acid, use of, in liberating iodine (Gooch and Mar) I, 27 use of, in liberating iodine (Gooch and Ensign) I, 43 Oxalic acid, titration of, by potassium permanganate in presence of hy- drochloric acid (Gooch and Peters) II, 222 Oxygen, amount of, required to oxidize an organic substance (Phelps) . II, 71 412 INDEX OF SUBJECTS. VOL. PAGE determination, In air and in aqueous solution (Kreider) II, 11 Organic acids, application of, in estimation of vanadium (Browning and Goodman) II, 4 Organic substance, amount of oxygen required for oxidation of (Phelps) II, 71 combustion of, by chromic acid in the wet way (Phelps) .... II, 67 combustion of, by potassium permanganate in the wet way (Phelps) II, 62 Perchlorates, quantitative determination of (Kreider) 1,316 Perchloric acid, application of, to the determination of potassium (Kreider) I, 286 preparation of (Kreider) I, 282 Permanganate solutions, standardization of, by electrolytic iron (Roberts) I, 269 Perofskite (so-called), analysis of, from Magnet Cove, Ark. (Mar) . . I, 60 Phosphorus, iodometric method for the determination of, in iron (Fairbanks) 1,391 Phosphoric acid, determination of, by precipitation as ammonium mag- nesium phosphate (Gooch and Austin) II, 204 Phosphoric acid, use of, with potassium iodide, in determining nitrates (Gruener) I, 193 Phthalanil, preparation of, from phthalic anhydride and sulphocar- banilide (Dunlap) I, 361 Phthalauilic acid, preparation of, from phthalic anhydride and sulpho- carbanilide (Duiifap) 1, 361 Phthalimide, preparation of, by action of urea on phthalic anhydride (Dunlap) 1,355 Primary Amines, action of, on maleic anhydride (Dunlap and Phelps) . II, 44 Potassium, detection and determination of, spectroscopically ( Gooch and Hart) I, 92 determination of, by perchloric acid (Kreider) I, 282 estimation of, as the pyrosulphate (Browning) II, 368 separation of, from sodium (Kreider and Breckenridge) .... 1,401 Potassium Spectrum, brightening of, by sodium chloride (Gooch and Hart) I, 101 Potassium bromide, use of, in reduction of selenic acid (Gooch and Scoville) 1,335 use of, in reduction of arsenic acid (Gooch and Phelps) .... I, 265 use of, with hydrochloric acid, in separation of arsenic from copper (Gooch and Moseley) I, 272 Potassium perchlorate, decomposition of, by anhydrous zinc chloride (Gooch and Kreider) I, 247 Potassium permanganate, estimation of cerium oxalate by (Browning and Lynch) II, 297 action of sulphuric acid upon (Gooch and Danner) I, 145 standardization of, in iron analysis (Roberts) I, 269 titration of oxalic acid by, in presence of hydrochloric acid (Gooch and Peters) II, 222 use of, in combustion of organic substances in the wet way (Phelps) II, 62 use of, in estimation of copper, with separation from cadmium, arsenic, tin, iron, and zinc (Peters) II, 347 use of, in the volumetric estimation of mercury as the oxalate (Peters) 11,320 use of, in the estimation of selenious acid (Gooch and demons) . . I, 297 use of, in the estimation of tellurous acid (Gooch and Danner) . . I, 154 INDEX OF SUBJECTS. 413 VOL. PAGE use of, in the estimation of tellurous acid (Gooch and Peters) . . II, 238 Reducing agents, action of, on iodic acid (Roberts) I, 250 Rhodochrosite, analysis of, from Franklin Furnace, N. J. (Browning) . I, 57 Rubidium, estimation of, as the acid sulphate (Browning) II, 370 quantitative spectroscopic determination of (Gooch and Phinney) . I, 157 Selenic acid, iodometric determination of (Gooch and Peirce) .... I, 341 reduction of, by hydrochloric acid (Gooch and Evans) I, 331 reduction of, by potassium bromide in acid solution (Gooch and Scoville) l t 335 Selenious acid, determination of, by potassium permanganate (Gooch and demons) I ? 297 influence of hydrochloric acid in thiosulphate titrations of (Norton) II, 206 iodometric determination of (Gooch and Peirce) 1,338 Selenium, gravimetric determination of (Peirce) I, 365 method for separation of, from tellurium (Gooch 'and Peirce) ... I, 348 reduction of acids of, by hydriodic acid (Gooch and Reynolds) . . I, 310 Selenium monoxide, on the existence of (Peirce) 1,385 Silver salts, electrolytic reduction of, in estimation of halogens (Gooch and Fairbanks) I, 290 Silver sulphocyanide in gravimetric analysis (Van Name) II, 359 Sodium, estimation of, as the pyrosulphate (Browning) 11,371 separation of, from potassium (Kreider and Breckenridge) ... I, 401 Sodium chloride, brightening of potassium spectrum by (Gooch and Hart) I, 101 Sodium thiosulphate, action of, on solutions of metallic salts at high temperatures and pressures (Norton) II, 384 influence of hydrochloric acid in titrations by, with special reference to the estimation of selenious acid (Norton) II, 206 reduction of iron in ferric state by (Norton) II, 230 use of, in titration of mercury (Norton) II, 328 titration of, by iodic acid (Walker) II, 52 Sodium tungstate, use of, as a retainer for boric acid (Gooch and Jones) II, 178 Space isomerism of nitrogen, bearing of ethers of toluquiuoneoxime on (Bridge and Morgan) (Morgan) 11,145,283 Spectroscopic determination of potassium (quantitative) (Gooch and Hart) I, 92 Spectroscopic determination of rubidium (quantitative) (Gooch and Phinney) I, 157 Standard solutions of tartar emetic, stability of (Gruener) I, 216 Standardization of potassium permanganate in iron analysis (Roberts) . I, 269 Starch, blue iodide of (Roberts) I, 236 Starch blue, conditions governing formation and decomposition of (Roberts) . 1, 236 Strontium, estimation of, as oxalate (Peters) 11,374 quantitative separation of, from calcium, by action of amyl alcohol on the nitrates (Browning) I, 121 separation of barium from, by action of amyl alcohol on the brom- ides (Browning) 1,168 separation of, from calcium, by action of amyl alcohol on the nitrate (Browning) I, 121 Succinanil, preparation of, from succinic anhydride and sulphocarban- ilide (Dunlap) 1,363 Succinic anhydride, action of sulphocarbanilide upon (Dunlap) ... I, 363 414 INDEX OF SUBJECTS. VOl. PAGB Succinimide, preparation of, from urea and succinic anhydride (Dunlap) I, 359 Sulphates, detection of, in presence of sulphides, sulphites, and thiosul- phates ( Browning and Howe) 11,134 Sulphides, detection of, in presence of sulphates, sulphites, and thiosul- phates (Browning and Howe) II, 134 Sulphites, detection of, in presence of sulphides, sulphates, and thiosul- phates (Browning and Howe) II, 134 Sulphocarbanilide, action of, on certain acid anhydrides (Dunlap) . . I, 355 action of, on phthalic anhydride (Dunlap) I, 355 action of, on succinic anhydride (Dunlap) 1,359 Sulphocyanides of copper and silver in gravimetric analysis (Van Name) II, 359 Sulphuric acid, action of potassium permanganate upon (Gooch and Dan- ner) I, 145 Tartar emetic, stability of standard solutions of (Gruener) I, 216 Tellurium, determination of, by precipitation on the iodide (Gooch and Morgan) n, 1 method for the separation of selenium from (Gooch and Peirce) . I, 348 Telluric acid, iodometric method for the estimation of (Gooch and How- land) 1, 277 Tellurous acid, determination of, by potassium permanganate (Gooch and Danner) I, 154 determination of, in presence of haloid salts (Gooch and Peters) . . II, 238 Thallium, estimation of, as acid and neutral sulphates (Browning) . . II, 317 estimation of, as chromate (Browning and Hutchins) II, 300 Thiosulphates, detection of, in presence of sulphides, sulphites, and sul- phates (Browning and Howe) II, 134 Tin, detection of arsenic associated with (Gooch and Hodge) .... I, 231 separation of copper as oxalate from (Peters) H, 347 Titanium salt, action of sodium thiosulphate upon, at high temperatures and pressures (Norton) II, 392 Toluquinoneoxime, ethers of, and their bearing on the space isomerism of nitrogen (Bridge and Morgan) II, 145 Toluquinoneoxime ethers, space isomerism of (Morgan) II, 283 Toluquinonemetaoxime acetyl ether (Bridge and Morgan) 11,153 Toluquinonemetaoxime benzoyl ether (Bridge and Morgan) II, 154 Toluquinonemetaoxime methyl ether (Bridge and Morgan) II, 152 Toluquinonemetaoxime, sodium salt of (Morgan) II, 285 Toluquinoneorthooxime acetyl ether (Bridge and Morgan) II, 160 Toluquinoneorthooxime benzoyl ether (Bridge and Morgan) . . . . II, 160 Toluquinoneorthooxime benzoyl ether dichloride (Morgan) II, 287 Toluquinoneorthooxime methyl ether (Bridge and Morgan) . . . . II, 159 Tolylmaleamic acid (o), preparation of, from maleic anhydride and o-toluidine (Dunlap and Phelps) II, 145 Tolylmaleamic acid (p), preparation of, from maleic anhydride and p-toluidine (Dunlap and Phelps) II, 44 Urea, action of, on certain acid anhydrides (Dunlap) I, 355 action of, on maleic anhydride (Dunlap and Phelps) II, 42 Vanadic acid, reduction of by hydriodic and hydrobromic acids (Brown- ing) 1,397 estimation of, iodometrically (Browning) 1,397 Vanadium, application of certain organic acids to estimation of (Brown- ing and Goodman) II, 4 INDEX OF SUBJECTS. 415 VOL. PAGE Volatilization of the iron chlorides in analysis (Gooch and Havens) . . 11,215 Volumetric estimation of mercury (Peters) 11,320 Zinc, separation of aluminum from (Havens) II, 107 separation of copper as oxalate from (Peters) II, 357 Zinc ammonium phosphate in analysis (Austin) II, 257 Zinc chloride (anhydrous), use of, in detecting perchlorates (Gooch and Kreider) 1, 247 Zirconium, separation of iron from, by gaseous hydrochloric acid (Havens and Way) 11,266 Zirconium salt, action of sodium thiosulphate upon, at high temper- atures and pressures (Norton) II, 391 UNIVERSITY OF CALIFORNIA LIBRARY THIS BOOK IS DUE ON THE LAST DATE STAMPED BELOW JAN 31 1916 30m-l,'15 04198