U',M It iliJlilll!!:. mi mm ft,- EM' mm 1 mil'iS'i!;' , I If! I! RESEARCHES UPON THE ATOMIC WEIGHTS OF CADMIUM, MANGANESE, BROMINE, LEAD, ARSENIC, IODINE, SILVER CHROMIUM, AND PHOSPHORUS BY GREGORY PAUL BAXTER IN COLLABORATION WITH M. A. HINES, H. L. FREVERT, J. HUNT WILSON, F. B. COFFIN, G. S. TILLEY, EDWARD MUELLER, R. H. JESSE, Jr., AND GRINNELL JONES <"HHSTJMUT HiLi^ l^iASa. WASHINGTON, D. C. PUBLISHED BY THE CARNEGIE INSTITUTION OF WASHINGTON I9IO CARNEGIE INSTITUTION OF WASHINGTON PUBLICATION No. 135 CONTRIBUTIONS FROM THE CHEMICAL LABORATORY OF HARVARD COLLEGE QDy 63 'tK3 2132-34 THE UNIVERSITY PRESS, CAMBRIDGE, U.S.A. PREFACE. This collection of papers upon the atomic weights of certain common ele- ments embodies the results of researches of which the experimental work has been carried on in the Chemical Laboratory of Harvard College during the past six years. All of the papers have already been published separately both in American and in German periodicals, and references to the places of publi- cation are given at the beginning of each article. In reprinting the papers in the present form the only changes of importance which have been made are due to more exact knowledge of the fundamental atomic weights upon which the calculations depend. Many recent investiga- tions, especially that upon the analysis of lithium chloride and perchlorate by Richards and Willard,^ have shown that the atomic weight of silver, referred to oxygen 16.000, is certainly as low as 107.880, and possibly as low as 107.870. Since the International Committee upon Atomic Weights at the date of writ- ing have chosen the higher of these values, the calculations have been based upon the value 107.880 for silver, the atomic weights of chlorine and bromine being assumed to be 35.457^ and 79.916 ^ respectively. The effect of a change from 107.880 to 107.870 in the atomic weight of silver is, however, plainly indicated in each instance. In the case of cadmium the subject-matter of two papers has been rear- ranged in a manner differing considerably from that of the original publica- tion. In the case of iodine the subject-matter of two papers has been com- bined in one. In all other cases the presentation is essentially that of the original publication. Generous grants from the Carnegie Institution of Washington have been of the greatest assistance in the progress of this work, while grants from the Cyrus M. Warren Fund for Research in Harvard University have materially aided all the investigations. G. P. Baxter. * Publications of the Carnegie Institution, No. 125 (1910); Jour. Amer. Chem. Soc, 32, 4. * Puhlications of the Carnegie Institution, No. 28 (1905); Jour. Amer. Chem. Soc, 27, 459; Zeit. anorg. Chem., 47, 56. 8 Proc. Amer. Acad., 42, 201 (1906); Jour. Amer. Chem. Soc, 28, 1322; Zeit. anorg. Chem.. Soc, 389. (See page 49.) Digitized by the Internet Archive in 2010 with funding from Boston Library Consortium IVIember Libraries http://www.archive.org/details/researchesuponatOObaxt CONTENTS. Page Preface '. . iii I. A Revision of the Atomic Weight of Cadmium; The Analysis of Cadmium Chloride. By G. P. Baxter and M. A. Hines. Introduction 3 Purification of Materials 4 Preparation of Cadmium Chloride for the Preliminary Analyses 7 Method of Analysis lo Preliminary Series of Results 13 Action of Hydrochloric-Acid Gas upon Phosphorus Pentoxide 14 Preparation and Drying of Cadmium Chloride for the Final Analyses 15 Final Series of Results 16 II. A Revision of the Atomic Weight of Cadmium; The Analysis of Cadmium Bromide. By G. P. Baxter, M. A. Hines, and H. L, Frevert. Purification of Materials 21 Drying of Cadmium Bromide for Analysis 22 Method of Analysis 24 Results and Discussion 26 III. A Revision of the Atomic Weight of Manganese; The Analyses of Man- ganous Bromide and Chloride. By G. P. Baxter and M. A. Hines. Introduction . ." 33 Analysis of Manganous Bromide 34 Purification of Materials 34 Drying of Manganous Bromide 37 Method of Analysis 38 Density of Manganous Bromide 40 Results 42 Analysis of Manganous Chloride 44 Piurification of Materials 44 Drying of Manganous Chloride 44 Method of Analysis 45 Density of Manganous Chloride 46 Results and Discussion 46 IV. A Revision of the Atomic Weight of Bromine; The Synthesis of Silver Bromide and the Ratio of Silver Bromide to Silver Chloride. By G. P. Baxter. Introduction 51 Purification of Materials 54 Synthesis of Silver Bromide 57 Results _ 58 Conversion of Silver Bromide into Silver Chloride 59 Results and Discussion 60 V. A Revision of the Atomic Weight of Lead; The Analysis of Lead Chloride. By G. p. Baxter and J. H. Wilson. Introduction 65 Purification of Materials 67 Drying of Lead Chloride and Method of Analysis 68 Results and Discussion 69 v VI CONTENTS VI. A Revision of the Atomic Weight of Arsenic; The Analysis of Silver Arsenate. By G. P. B.\xter and F. B. Coffin. Introduction 73 Preparation of Trisilver Arsenate 74 Purification of Other Materials 76 JNIethods of Analysis 77 Insoluble Residue 81 Determination of Moisture in Dried Silver Arsenate 82 Specific Gravity of Silver Arsenate 84 Results and Discussion 85 VII. A Revision of the Atomic Weight of Iodine; The Synthesis of Silver Iodide and the Ratio of Silver Iodide to Silver Bromide and Silver Chloride. By G. P. Baxter. Introduction 91 Ratio of Silver to Silver Iodide 92 Purification of Materials 92 Method of Synthesis 94 Specific Gravity of Silver Iodide 96 Results 97 Ratio of Silver to Iodine 99 Results loi Ratio of Silver Iodide to Silver Chloride 102 Results 104 Ratio of Silver Iodide to Silver Bromide 105 Results 106 Ratio of Iodine to Silver and Silver Iodide 107 Discussion of Results no Ratio of Silver Bromide to Silver Chloride in Historical Discussion 112 Summary 114 VIII. A Revision of the Atomic Weights of Iodine and Silver; The Analysis OF Iodine Pentoxide. By G. P. Baxter and G. S. Tilley. Introduction 117 Purification of Materials for the First Series of Analyses 118 Conversion of Iodic Acid into Iodine Pentoxide 121 Determination of Iodine in Iodine Pentoxide 123 Determination of Moisture in Iodine Pentoxide 126 Specific Gravity of Iodine Pentoxide 129 Adsorption of Air by Iodine Pentoxide 130 Purification of Iodic Acid and Silver for the Second Series of Analyses 132 Method of Analyses 133 Discussion of Results 135 IX. A Revision of the Atomic Weight of Chromium; The Analysis of Silver Chromate. By G. P. Baxter, Ed. Mueller, and M. A. Hines. Introduction 139 Purification of Materials 141 Preparation of Silver Chromate 142 Drying of Silver Chromate 144 Determination of Silver in Silver Chromate 145 Determination of Moisture in Dried Silver Chromate 147 Specific Gravity of Silver Chromate 149 Discussion of Results 151 CONTENTS Vll X. A Revision of the Atomic Weight of Chromium; The Analysis of Silver Bichromate. By G. P. Baxter and R. H. Jesse, Jr. Introduction 155 Purification of Materials 156 Preparation of Silver Dichromate 157 Drj'ing of Silver Dichromate 158 Determination of Silver in Silver Dichromate , . 158 Determination of Moisture and Nitric Acid in Dried Silver Dichromate 159 Specific Gravity of Silver Dichromate 161 Discussion of Results 163 XI. The Revision of the Atomic Weight of Phosphorus; The Analysis of Silver Phosphate. By G. P. Baxter and Grinnell Tones. Introduction 167 Purification of Materials 169 Preparation of Trisilver Phosphate 171 Drying of Silver Phosphate 176 Determination of Silver in Silver Phosphate 177 Insoluble Residue 178 Determination of Moisture in Dried Silver Phosphate 181 Specific Gravity of Silver Phosphate 182 Adsorption of Air by Silver Phosphate 183 Ratio of Silver Bromide to Silver Phosphate 184 Discussion of Results 184 Summary 185 I. A REVISION OF THE ATOMIC WEIGHT OF CADMIUM. THE ANALYSIS OF CADMIUM CHLORIDE. By Gregory Paul Baxter and Murray Arnold Hikes. Journal of the American Chemical Society, 27, 222 (1905) ; 28, 770 {1906). Zeitschrift fur anorganische Chemie, 44, 158 (1905) ; 49. A^S (1906). Contributions from the Chemical Laboratory of Harvard College. A REVISION OF THE ATOMIC WEIGHT OF CADMIUM. THE ANALYSIS OF CADMIUM CHLORIDE. INTRODUCTION. From the following list of investigations upon the atomic weight of cad- mium it can be seen that this subject has attracted considerable attention, especially in recent years.^ Date. Investigator. Reference. Ratio determined. Result. i8i8 Stromeyer .... Schweigger's Jour., 22, Cd: CdO iii-S 366 Cd: CdS Cd: CI2 Cd:l2 II3-8 112.8 III. 7 i8S7 von Hauer .... J. pr. Chem., 72, 350 CdS04: CdS 111.9 i860 Lenssen J. pr. Chem., 281 CdCzOi: CdO X12.0 i860 Dumas Ann. Chem. Pharm., 113, CdCl2: 2Ag 112. 14 1881 Huntington . . . 27 Proc. Amer. Acad., 17, 28 CdBrz: 2Ag CdBr2: 2AgBr 112.18 112. 17 1890 Partridge .... Amer. Jour. Sci., (3) 40, CdCsO*: CdO 111.80 377 CdS04:CdS CdCaOi: CdS 111.73 111.67 1891 Morse and Jones Amer. Chem. Jour., 14, Cd: CdO 112.07 261 CdCjOi: CdO 112.02 1892 Lorimer and Smith Zeit. anorg. Chem., i, 364 CdO: Cd 112.04 189s Bucher Doctoral Dissertation, CdC204: CdO 111.88 Baltimore, Md. CdCzO^: CdS CdCU: 2AgCl CdBra: 2AgBr Cd: CdS04 Cd: CdO (porcelain) Cd: CdO (platinum) 112. 12 112.31 112.33 112.3s 112.08 111.89 1896 Hardin Jour. Amer. Chem. Soc, CdCl2:Cd 112.03 18, 1016 CdBra: Cd Cd:Ag 112.02 111.94 1898 Morse and Arbuckle Amer. Chem. Jour., 20, 536 Cd: CdO 112.38 The relative value of many of these determinations has already been several times discussed,^ and since it is invariably a difficult matter intelligently to * The greater portion of this list is to be found in Clarke's " Recalculation of the Atomic Weights," Smithsonian Misc. Coll., 1910. The results have been calculated with the use of the following atomic weights: = 16.00; C = i2.oo; 8=32.07; Cl = 35.46; Br = 79.92; Ag = io7.88; 1=126.92. ^ Clarke, Partridge, Morse and Jones: Loc. cit.; Richards: Amer. Chem. Jour., 20, 547 (1898). 3 4 RESEARCHES UPON ATOMIC WEIGHTS. criticize experimental work without an actual repetition of the experiments, for frequently some constant source of error is so securely hidden that it may be detected only by the most careful investigation, no attempt at criticism is made here. Nevertheless, it is worth while calling especial attention to the careful re- searches of Morse and Jones and Morse and Arbuckle upon the ratio Cd : CdO, which yielded the value 1 1 2.38, and that of Bucher upon the ratios CdCU : 2 AgCl and CdBr2 : '2AgBr which yielded the values 112.31 and 112.33. Our experi- ments indicate that the real value for the constant in question is sUghtly higher than any of these results. In a recent determination of the specific gravity of cadmium chloride/ anhydrous cadmium chloride was prepared by ignition of a double chloride of cadmium and ammonium in a current of hydrochloric-acid gas, in a state of so great purity that it was considered worth while to make use of the salt for a determination of the atomic weight of cadmiimi. PURIFICATION OF MATERIALS. CADMIUM CHLORIDE. The general method of purification of the cadmium material was that of fractionally precipitating cadmium sulphide. One kilogram of metalHc cad- mium was dissolved in aqua regia, and the solution, after being boiled to expel chlorine and oxides of nitrogen, was filtered, and diluted to about 4 liters. The solution contained traces of lead, copper, thallimn, nickel, iron, and zinc. When a current of hydrogen sulphide was passed through the solution, the first small fraction of cadmium sulphide which was precipitated was dark- colored, nearly black, owing to the presence of lead, copper, and thallium. This fraction was removed by filtration and rejected. A second larger fraction of the sulphide, although it contained no appreciable amount of the latter metals, also was discarded. The third fraction consisted of all that could be precipitated by saturating the solution with hydrogen sulphide. However, this did not contain more than one-quarter of the original material, for the solution was very strongly acid, owing to the large excess of acid used in dis- solving the metal and the accumulation of the acid formed during the precipi- tation. The solution was separated from the precipitate by decantation, and was then diluted to 16 liters. Upon saturating this solution with hydrogen sulphide a fourth fraction of cadmium sulphide was obtained, and a second dilution of the solution made possible the precipitation of nearly all the remainder of the cadmium in still a fifth fraction. The third, fourth, and fifth fractions of the sulphide were separately washed until free from chlorides. As the electrolytes were eliminated, the cadmium sulphide showed a tendency to pass into colloidal condition, which necessitated * Baxter and Hines: Amer. Chem. Jour., 31, 220 (1904). THE ANALYSIS OF CADMIUM CHLORIDE. 5 long standing for the precipitate to settle after each washing, although the flasks which contained the precipitates were kept warm by being placed upon a steam radiator. During the washing the original fine yellow precipitate was gradu- ally converted into an orange-red crystalline modification. When, as was usu- ally the case, both forms were present in the same flask, the red form quickly settled to the bottom with a sharp line of division from the yellow form. Nearly all the yellow form was changed into the red modification upon standing about three weeks. In order to free the sulphide from included and occluded impurities each frac- tion was dissolved and reprecipitated. The red form of the sulphide was ap- parently nearly insoluble in dilute sulphuric acid, for in one case the washed sulphide was boiled with the acid for 12 hours without any appreciable amount of solution. Finally, hydrochloric acid was used to dissolve the cadmium sulphide. The solution of each fraction was diluted to 8 liters and was sat- urated with hydrogen sulphide. Since only a portion of the cadmium was precipitated in this way, owing to the large excess of acid, the acid was par- tially neutralized with ammonia. This resulted in the precipitation of more cadmium sulphide, although the solution still contained considerable cadmium, for cadmium sulphide is soluble to a marked extent in an acid solution of ammonium chloride. The sulphide obtained from each of the original three fractions, both before and after the addition of ammonia, was combined and washed until free from chlorides. Each fraction was dissolved in redistilled nitric acid, then enough redistilled sulphuric acid to convert the nitrate into sulphate was added, and the solutions were evaporated and the residues heated until all volatile acids were expelled. Finally the sulphate was recrystallized three times from aqueous solution. The cadmium sulphate was converted into cadmium chloride by first obtain- ing metallic cadmimn electrolytically. A saturated solution of cadmium sul- phate was electrolyzed with about one ampere current per square decimeter in a platinum dish, which served as the cathode, until deposition ceased. After the deposit of metal had been thoroughly washed with hot water until free from sulphate, it was dissolved in hydrochloric acid which had been distilled with the use of a platinum condenser. In order to prepare the double chloride of cadmiiun and ammonium of the formula CdCl2NH4Cl, the calculated amount of ammonium chloride was added to the cadmium chloride and the solution evaporated to crystallization. This ammonium chloride was synthesized from hydrochloric acid and ammonia. The hydrochloric acid had been distilled in platinum, and the ammonia had been freed from amines and purified as follows : Ammonium chloride was boiled with concentrated nitric acid for about 20 hours, and then after crystallization was converted into ammonia by distillation with sodium hydroxide. The solu- tion of pure ammonia was distilled into the pure hydrochloric acid in a 6 RESEARCHES UPON ATOMIC WEIGHTS. platinum dish, and the solution of ammonium chloride was evaporated to crystallization. The cadmium ammonium chloride was crystallized in a platinum dish, eight times in the case of the first fraction, four times in the case of the second frac- tion. The third fraction of the sulphide was not converted into the double chloride, but was investigated in a later research upon cadmium bromide (see page 2i). The first fraction is designated as Sample I, the second as Sample II. A third specimen used in the analyses was a portion of that employed in the determination of the specific gravity of cadmium chloride.^ This is designated as Sample A. The method of purification of Sample A was almost exactly identical with that described above, except that the original material was pre- cipitated but not fractionated with hydrogen sulphide. SILVER. In the preparation of pure silver essentially the same method was employed as in other atomic weight investigations in this laboratory .^ In this case the various treatments consisted in thrice precipitating the silver from a dilute solution of silver nitrate in nitric acid by a large excess of hydrochloric acid, with intermediate reduction of the silver chloride in each case by means of invert sugar and sodium hydroxide. The sodium hydroxide for the third re- duction was freed from heavy metals by electrolysis. Both the silver chloride and the metallic silver were of course very thoroughly washed by decantation with pure water. The final product was fused on charcoal in the flame of a clean blowpipe. Next the buttons were converted into electrolytic crystals by slow deposition upon a pure silver cathode from a concentrated, nearly neutral solu- tion of silver nitrate, the anode being com-posed of the pure silver buttons. After thorough washing and drying the crystals were fused in a current of pure electrolytic hydrogen in a boat of pure lime.^ The lim.e boat was made by lining an unglazed porcelain boat with a mixture of freshly ignited lim.e and calcium nitrate, both having previously been carefully freed from iron and other heavy metals. The boat was thoroughly ignited before use.* During the fusion it was contained in a large Royal Berlin porcelain tube, the ends of which were closed with hollow brass stoppers made to fit tightly by means of narrow r'ngs of rubber, the stoppers being cooled by a current of cold water.^ Richards and Wells, in a recent investigation of the purity of silver purified by different * Loc. cit. » See especially Richards and Wells: Ptib. Car. Inst., No. 28, 16 (1905); Jour. Amer. Chem. Soc, 27, 472; Zeit. anorg. Chetn., 47, 70. ' Baxter: Proc. Amer. Acad., 39, 249 (1903); Zeit. anorg. Chem., 38, 237 (i904)- * Richards: Proc. Amer. Acad., 30, 379 (1894); Zeit. anorg. Chem., 8, 262 (1895); Rich- ards and Parker: Proc. Amer. Acad., 32, 63 (1896); Zeit. anorg. Chem., 13, 89 (1897). ' Richards and Parker: Loc. cit. THE ANALYSIS OF CADMIUM CHLORIDE. 7 methods, have found that silver prepared in the above fashion is at least as pure as any/ Since the buttons of silver obtained from the fusion in hydrogen were of very considerable size, they were cut into fragments of from i to 5 gm. by means of a clean steel chisel and anvil. A slight surface contamination with iron was removed by etching the fragments several times with dilute nitric acid, until the acid remained free from iron, and drying them at 200°. Two different samples, purified in the same way, were employed. One was prepared especially for this investigation and was used in analyses 4, 5, 6, and 14. The other was a portion of the material employed in an investigation upon the atomic weight of iodine by one of us ^ (analyses 7, 8, and 9). Still a third specimen of silver, used in analyses 13 and 15, was twice deposited electrolyt- ically before the final fusion in hydrogen. Water was purified by double distillation through tin condensers, first from alkaUne permanganate solution, finally with a trace of sulphuric acid. Con- nection between the flasks and the condensers was made by constricting the necks of the flasks to fit the ends of the condenser tubes, avoiding thus the use of rubber and cork.* Nitric acid was twice distilled with a platinum condenser, the first third being rejected in both distillations. The product of the first distillation contained only the merest trace of chlorine. PREPARATION OF THE CADMIUM CHLORIDE FOR THE PRELIMINARY ANALYSES. The method of analysis differed little from that used in the analysis of other halogen salts in atomic weight investigations in this laboratory. The cadmium chloride was freed from ammonium chloride by fusion. Then, after solution in water, the chlorine content was found either gravimetrically as silver chloride or by titration against weighed amounts of silver. The apparatus used for the expulsion of the ammonium chloride from the double salt was similar to that employed for a Uke purpose by Richards and Parker* in their analysis of magnesium chloride. Since this apparatus was used in two other researches described in this collection, a detailed description is given here. Hydrochloric-acid gas was generated by the action of concen- trated sulphuric acid upon concentrated hydrochloric acid in the flask A (fig. i), and, after bubbling through concentrated hydrochloric-acid solution in the wash bottle B, it was dried by passing through four towers about 30 cm. long and 4 cm. in diameter filled with glass beads saturated with concentrated sul- phuric acid, C, D, E, F. In the first series of experiments the hydrochloric-acid ' Loc. cit. ^ Baxter: Proc. Amer. Acad., 40, 419 (1904); Jour. Amer. Chem. Soc, 26, 1577; Zeit. anorg. Chem., 43, 14 (1905). (See page 92.) 3 Richards: Proc. Amer. Acad., ^o, 2,^0 {iZg^); Zeit. anorg. Chem. ,S, 261 (iSgs). * Proc. Amer. Acad., 32, 59 (1896): Zeit. anorg. Chem., 13, 85 (1897). 8 RESEARCHES UPON ATOMIC WEIGHTS. gas was further dried by passing through a tube containing resublimed phos- phorus pentoxide. This tube is not shown in the diagram since it was elim- inated in the final series of experiments. Nitrogen was prepared by Wanklyn's method of passing air through concen- trated ammonia solution in the bottle M and then over hot copper gauze in the hard glass tube N. The excess of ammonia was removed by dilute sulphuric =*=< Fig. I. — Apparatus for the fusion of chlorides in a current of hydrochloric-acid gas. acid in the bottles and P, and the nitrogen was purified and dried by means of beads saturated with silver-nitrate solution in the tower Q, solid potassium hydroxide in the tower R, concentrated sulphuric acid in the towers S, T, and U, and resublimed phosphorus pentoxide in the tube L. Air was purified and dried by reagents similar to those used in the purifica- tion of the nitrogen, in the towers G, H, I, J, K. The hydrochloric-acid apparatus was constructed wholly of glass with either ground or fused joints, while the nitrogen and air purifjdng trains had short rubber connections only at the beginning. Glass gridirons at suitable points gave sufficient flexibility to the apparatus. Ground joints were made tight by means of either concentrated sulphuric acid or with syrupy phosphoric acid. A portion of the cadmium ammonium chloride, contained in a weighed plat- inum boat in the hard-glass tube W, was heated gradually to fusion in a current of hydrochloric-acid gas and was kept fused until all the ammonimn chloride THE ANALYSIS OF CADMIUM CHLORIDE. 9 had been expelled. After the salt had cooled, the hydrochloric acid was dis- placed by pure dry nitrogen and this in turn by dry air. Next, by means of a glass rod, the boat was pushed into the weighing-bottle contained in the soft- glass tube V, and the stopper was inserted without opening the apparatus or interrupting the current of dry air by rotating the tube V slightly so as to cause the stopper to roll from its position in the pocket of the tube into the main tube. By means of the rod the stopper was readily pushed into place. This botthng apparatus in its improved form was first used by Richards and Parker in their work upon magnesium chloride.^ The bottle was transferred to a desiccator and, after standing near the balance case for some time, it was weighed by substitution for a counterpoise similar in weight and volume as well as shape. Sublimation of the cadmium chloride always took place to some extent during the fusion, and the sublimed salt occasioned some difficulty since it flowed down the inside of the glass tube and, upon soHdification, firmly cemented the boat to the tube. Furthermore, the salt which adhered to the outside of the boat had thus been fused in contact with glass, and hence may have been impure. Both these difficulties were avoided by supporting the boat upon a carriage of heavy platimmi wire. While the salt was still warm and the current of hydrochloric- acid gas was still passing, the boat was pushed out of the carriage by means of a long glass rod. Neglect to observe the latter precaution usually resulted in the cementing of the boat to the carriage by the salt which had condensed upon the outside of the boat. It has already been shown that barium and calcium chlorides when they have been fused and allowed to solidify in an atmosphere of hydrochloric-acid gas, occlude none of the gas,^ for they give neutral solutions; hence it is reasonable to conclude that this is the case with cadmium chloride also. However, in order to test this point, in analysis 9 the hydrochloric acid was displaced by nitrogen while the salt was still warm, and in analysis 8 the salt was allowed to solidify only when the hydrochloric acid had been almost completely displaced by nitrogen. In one experiment where the hydrochloric acid had been completely displaced by nitrogen, the boat became covered with a gray coating which turned brown, and finally volatilized when the boat was ignited. This coating undoubtedly consisted of metallic cadmium, formed either by the dissociation of cadmium chloride vapor or by the action of the small amount of hydrogen con- tained in nitrogen produced by Wanklyn's method, owing to catalytic decom- position of the excess of ammonia by the hot copper. The close agreement of the results of analyses 8 and 9 with those obtained in the other analyses where the salt solidified and cooled in hydrochloric acid, shows conclusively that no ap- preciable amount of hydrochloric acid was occluded by the salt. * Loc. cit. ^ Richards: Proc. Amer. Acad.,2g, 59 (1893); Ze«^ anorg. Chem., 6, 93 (1894); Jour. Amer. Chem. Soc, 27, 376 (1902); Zeit. anorg. Chem., 31, 273. lO RESEARCHES UPON ATOMIC WEIGHTS. It was shown in our determination of the specific gravity of cadmium chloride, that the salt when prepared in this way contains no ammonium chloride. It is probable that the cadmium chloride contained no basic compound, since no insoluble salt was produced when the chloride was dissolved in water. The aqueous solution of the salt invariably contained a few tenths of a milligram of black insoluble matter which consisted chiefly of platinum. The presence of this platinum was undoubtedly due partially to slight attacking of the boat, owing perhaps to contamination of the hydrochloric acid with traces of air. The slight loss in weight of the boat which resulted in most of the analyses was not sufficient to accoimt for all the insoluble residue, which, therefore, must have had its source, in part, in the original material. Whether the platinum was dissolved from the platinum condenser during the distillation of the hydro- chloric acid, or from the platinum dish during the solution of the cadmiiun is uncertain. At all events, the temperature to which the salt was heated must have been sufficient to decompose all the platinic or platinous chlorides present, and since the insoluble residue was filtered out and weighed, and corrections applied to the weight of the salt both for the loss in weight of the boat and for the insoluble matter, no appreciable error could have been introduced by the platinum. THE METHOD OF ANALYSIS. After the salt had been weighed, the boat was transferred to a flask and the salt was dissolved in about 200 c.c. of the purest water. The weighing-bottle was rinsed and the rinsings were added to the solution. Next the solution was filtered into the precipitating flask through a tiny filter to collect the insoluble matter. Filter-paper and residue were then ignited in a weighed porcelain crucible. In the preliminary analyses the ratio of cadmium chloride to silver chloride was determined by adding to the solution of cadmium chloride, which had been diluted in the (Erlenmeyer) precipitating flask until not stronger than i per cent, a solution of a slight excess of silver nitrate of very nearly the same con- centration. The flask, which was provided with a ground-glass stopper, was shaken for some time, and was allowed to stand until the solution was clear. Then the precipitate of silver chloride was transferred to a Gooch crucible, after it had been washed by decantation six or eight times with about 150 c.c. of a 0.00 1 normal silver nitrate solution, and finally several times with pure water. Needless to say, the operations of precipitation and filtration were performed in a room hghted only with ruby light. The crucible with the precipitate was placed in an air-bath and heated for several hours at 130° to 140° C, and after it had cooled in a desiccator it was weighed. In order to determine how much moisture was retained by the precipitate in each case, it was transferred to a clean porcelain crucible and weighed, then the salt was fused by heating the small crucible, contained in a larger covered crucible, and again weighed. Two THE ANALYSIS OF CADMIUM CHLORIDE. II different specimens of silver chloride from analyses were separately dissolved in ammonia and reprecipitated with hydrochloric acid, and the filtrates, after evap- oration, were tested for cadmium. Negative results were obtained in both cases, showing both that silver chloride does not occlude cadmium salts to an appreciable extent and that the washing of the silver chloride had been thorough. The determination of the silver chloride dissolved in the wash-waters was the most difficult step in the analysis. At first the last few washings, those which had been carried out with pure water and which were the only ones which could have contained dissolved silver chloride, were evaporated to small bulk and an excess of silver nitrate was added. The precipitate of silver chloride, together with any asbestos which had been displaced from the Gooch crucible, was collected upon a small filter which was ignited and weighed. Owing to the com- bined effect of organic matter and light upon these solutions the precipitate was always too heavy. Hence this method was finally discarded. Four preliminary results obtained in this way varied between 112.34 and 11 2.40 for the atomic weight of cadmium. In order to avoid this error, in the next series the silver chloride dissolved in the wash- waters was determined by precipitating the chloride in 25 c.c. portions of the solution with an excess of silver nitrate, and comparing in a nephelometer the precipitate produced with that from solutions prepared from standard hydrochloric-acid solutions. At least two comparisons were made in each analysis. The nephelometer employed for the estimation of slight opalescences has al- ready been described in detail by Richards and Wells.^ All the precautions ne- cessary for the accurate use of this instrument were carefully observed. The two tubes to be compared were always of the same size. The source of light in the nephelometer was so adjusted that tubes containing exactly equal amounts of precipitate gave identical readings. It was found advantageous to insert a plate of ground glass between the source of light and the test-tubes. In making up the test solutions, great pains were taken that the concentration of electro- lytes in the two solutions and the conditions of precipitation should be as nearly as possible the same. Final readings were taken only after the ratio between the two tubes had become constant. The weight of pure silver required exactly to combine with the chlorine in cadmium chloride also was determined. From the weight of cadmium chloride very nearly the necessary quantity of pure silver was calculated. This silver was weighed out and dissolved, in a flask provided with a column of bulbs to prevent loss of silver by spattering (see fig. 2) in distilled nitric acid diluted with an equal volume of water. Ordinarily the silver was caused to dissolve so slowly that practically no gas was evolved. Careful experiments have shown, how- ^ Amer. Chem. Jour., 31, 235 (1904); 35, 510 (1906). 12 RESEARCHES UPON ATOMIC WEIGHTS. ever, that the column of bulbs is ample protection against loss by spattering, even when considerable effervescence takes place during the solution. After the silver was dissolved, the solution was diluted somewhat and heated until free from nitrous fumes. Then it was further diluted until not more con- centrated than I per cent, and was slowly added to the i per cent solution of cadmium chloride in the precipitating flask. After sev- eral minutes' shaking it was allowed to stand several days, with occasional shaking, until the solution was perfectly clear. Two 30 c.c. portions of the clear liquid were then pipetted into test tubes of similar size. To one portion was added i mg. of silver nitrate in the form of hundredth normal solution, to the other an equivalent amount of hydrochloric-acid solution, and the tubes were examined at frequent intervals in the nephelometer until the ratio of the opalescence shown by the two tubes became constant. Richards and Wells have shown that when equivalent amounts of silver and chloride have been used in the original precipitation, the nephelometer tubes show equal opalescence. If this was not the case in the first examination, the contents of the tubes were returned to the precipitating flask and either standard silver nitrate solution or standard hydrochloric-acid solution was added and the shaking and testing repeated until the amounts of chloride and silver in the solution were equivalent. FinaUy a considerable excess of silver nitrate was added to the analysis to precipitate dissolved silver chloride, and the silver chloride was determined gravimetrically as previously described. Correction was of course made for chloride introduced in the course of the nephelometric tests. A vacuum correction of -l- 0.000152 gm. was applied for every apparent gram of cadmium chloride, of -H 0.000071 gm. for every apparent gram of silver chlo- ride, and of —0.000031 gm. for every apparent gram of silver.^ All weighings were made by substitution, with tare vessels as nearly as pos- sible Hke those weighed. The gold-plated brass weights were twice carefully standardized to hundredths of a milligram. Fig. 2. — Flask for dissolving silver. ^ The specific gravity of cadmium chloride has been found to be 4.047. Baxter and Hines: Amer. Chem. Jour., 31, 220 (1904). Richards and StuU have determined the specific grav- ity of silver chloride to be 5.56, and Richards and Wells that of silver to be 10.49. Pub. Car, Inst., No. 28, II (1905); Jour. Amer. Chem. Sac, 27, 466; Zeit. anorg. Chem., 47, 64. The specific gravity of the weights is assumed to be 8.3. (See page 40.) The use of this low value for the specific gravity of the weights has led to slight changes in the vacuum cor- rections of cadmium and silver chlorides from the values used in the original publication of this paper. THE ANALYSIS OF CADMIUM CHLORIDE. 13 M M N ■^ '1" '1' w ci N H H H H H H M H s 1— 1 M c« p M 00 lO >0 •^ M O^ HI HI M 1" 1" ^ 't 't t cJ IN cj cs cJ cs W M M H M H M H M H H W On M M M OS CO H 00 liTO CO . CO H CO g,VO MOO 06 M CO 11 H H-T ^ 1 > ^ *c 'o 0 t^ . •^oo t>.^ g 00 10 q\oo q CO Cti to ■<* too J>- ^ g|i 0^ »0 CO M VO M . Tf t^ 0\ S CO t^ t^ VO r^oo 1° « On to Ov CO HI tOOO Tf CO 10 0> .00 ■* VO CO On S C^ NO 00 •* 0> CO 5 OnOO q 00 On !>• .NO ON 8 00 VO q\oq q co •^ M f» . N S «o^ 00 g>0 H 00 06 fJ CO m MO CO On N t^ M «s fa 6 6 6 6 6 6 .S ° ■ §11 i§§§ 000 11 l~. CN| t^ M W 5" IP? ^ K S3 Q Q ■ 5 5 S «» d d o' d d d OaJ .'§8^5' s 5, 6 6 6 VO CO On CO w CI On N NO t^NO M 00 to VO CO On H . M NO 00 "^ On CO g OnOO q 00 On t>- C» Tj- CO to to to CO m 0^ •«1- CO t~ . rfOO g CO t^ 1>- ^ 10 t>.oq 10 J>-od 1— 1 l-H 1— 1 ^ "o ■fto <^B a ^ lONO t^OO On "0 S ©"a iz; H M CO 14 RESEARCHES UPON ATOMIC WEIGHTS. The close agreement of the results in each series leaves little doubt of the identity of the different samples, although they represent material from different sources as well as different fractions of the same material. The slight discrep- ancy between the results by the two methods is undoubtedly due, in part, to the difl&culty in determining silver chloride with accuracy, owing, in the first place, to loss of chlorine by the silver chloride in the processes of manipulation and drying, and in the second place to the slight solubility of silver chloride even in dilute silver nitrate solutions. In the light of these possibilities, it is probable that the results of Series I are slightly too high. On the other hand, it is probable that the average of Series II is slightly too low, for the average of experiments 7, 8 and 9, in which the experience gained in the previous analyses was a very considerable aid, is 112.415, 0.006 of a unit higher than the average of the whole series. ACTION OF HYDROCHLORIC-ACID GAS UPON PHOSPHORUS PENTOXIDE. Some time after the completion of the foregoing series of experiments it was found that in fusing manganous chloride, in a current of hydrochloric-acid gas which had been dried by concentrated sulphuric acid and finally by means of phosphorus pentoxide, an insoluble residue of manganous phosphate was in- variably obtained when the salt was dissolved in water. The quantity of this residue varied with the amount of moisture contained by the salt when brought in contact with the hydrochloric-acid gas, being extremely slight if the salt was very nearly dry, but amounting to several milligrams if the salt still contained much of its crystal water. Although it seemed certain that the phosphorus had its origin in the phosphorus pentoxide, and was volatilized in the form of either phosphorus pentachloride or oxychloride through the action of the hydro- chloric acid upon the pentoxide, in order to obtain still more positive evidence that this was really the case, the experiment was tried of passing hydrochloric- acid gas which had been dried thoroughly by means of sulphuric acid, first over phosphorus pentoxide which had been freshly sublimed in a current of dry air, and then into water. The aqueous solution, upon evaporation and testing with ammoniimi molybdate, gave a considerable amount of the charac- teristic ammonium phosphomolybdate. This result confirms that of Bailey and Fowler,^ who have fovmd that both hydrochloric and hydrobromic acids react with phosphorus pentoxide at ordinary temperatures to form the oxy- chloride and bromide of phosphorus respectively. The manganous phosphate, then, must have been produced by the action of the volatilized chloride of phosphorus upon the moisture contained by the manganous chloride to form phosphoric acid, with subsequent displacement of hydrochloric acid from the salt by the phosphoric acid. * Chem. News, 58, 22 (i THE ANALYSIS OF CADMIUM CHLORIDE. 1 5 Although in our work with cadmium chloride, the double cadmium ammon- ium chloride was fused in a current of hydrochloric-acid gas which had been finally dried with phosphorus pentoxide, the salt, which contains no crystal water, was essentially free from moisture before coming in contact with the hydrochloric acid. Nevertheless it seemed desirable to repeat the experiments with cadmivun chloride in such a way that the danger mentioned above, could be completely avoided. This result was easily attained by drying the hydro- chloric-acid gas with concentrated sulphuric acid only. In order to ascertain whether concentrated sulphuric acid is appreciably at- tacked by hydrochloric-acid gas, a large quantity of this gas was conducted through the columns and then into water. The aqueous solution was then evaporated and tested for sulphate with barium chloride. Although a slight precipitate of baric sulphate was produced, the quantity was estimated, by comparison in a nephelometer with a standard solution of a sulphate, to be less than 0.05 mg. Evidently nothing is to be feared from this source. PREPARATION AND DRYING OF CADMIUM CHLORIDE FOR THE FINAL ANALYSES. The material for these experiments was prepared from a portion of fraction II of cadmimn sulphide, by first depositing the metal electrolytically from the sulphate. At first electrolysis was carried on in a solution of the sulphate in pure water, between two electrodes of platinum foil. A sponge of extremely small crystals was thus produced. These crystals contained occluded sulphate in considerable quantities, and no amount of washing with water was sufficient completely to leach out this occluded material. More satisfactory results were obtained by depositing the metal upon a platinum dish which had been covered with a very thin film of soft paraffine, so that the deposit could be readily sep- arated from the dish.^ The cadmium was first washed with water, then with ether, next with alcohol, and finally with water again. This treatment effect- ually cleansed the metal from paraffine. The metal was next dissolved in pure hydrochloric acid in a platinum dish. The chloride does not lend itself readily to crystallization from aqueous solu- tion on account of its great solubihty even at low temperatures, but by con- ducting hydrochloric-acid gas into the solution the much less soluble double salt with hydrochloric acid, CdCl22HCl7H20, was formed. The salt was thus crystallized 3 times with centrifugal drainage, to free it from the trace of sul- phates occluded by the metal during electrolysis. Finally, it was dried and freed from hydrochloric acid as far as possible in a vacuum desiccator contain- ing solid potassium hydroxide. Before fusion the salt was largely freed from water and hydrochloric acid by gentle heating. ^ Richards: Proc. Amer. Acad. ,2$, 200 (iSgo). i6 RESEARCHES UPON ATOMIC WEIGHTS. o H < a H U bO < .2 o B III -^11 Si. MM o &"s oS2 ■« °E jail's .£f— ^ >> fcD 00 o2 So 3 boMS m bo O M-S |2§ 5 O 1°^ 00 C> !>. Th Tj- ^ 00 M t^ g M3 Tj- O 6v d 00 5, H 00 lo 00 r^ i>. . « M f<^ s Tt r^ 0\ g,\o Tl- o\ d\ 6 00 + & o d ui t^ lo V5 ro Ov e J^ o o 5 M H O « M 00 >0 vc5 NO lo U o O MR O^ t. aOT3 .Mo3^ 5 MS 5) > |o| O t^ t^ M M M Tj- Tj- Tl- OOO fO . o^ o^ a\ g H Tj- CO ^-o o 5: \o 00 vd OO 00 t^ . O « 00 g 't o o . On M o^ ; OMO PO vd 00 >d "+ i^ o> O M o ooo too t^ m3 ooo 888 ^i ooo 5oi M-d 3 0\ l^ lO g O fO Ov >o o o ta lovo lo .a » 1— 1 1— 1 h-l -HHH t-l eg « .r > THE ANALYSIS OF CADMIUM CHLORIDE. 1 7 The average of these results is almost identical with that obtained in the first two series of analyses, 11 2.4 17, hence it is evident that no serious error was introduced in our earher work by the use of phosphorus pentoxide for drying hydrochloric-acid gas. The results of this investigation may be summarized as follows: 1. In the analysis of cadmium chloride, both gravimetrically by determina- tion of the chlorine as silver chloride and volumetrically by comparison with silver, the atomic weight of cadmium is foimd to be 11 2.418 referred to silver 107.880, or 112.408 if silver has the atomic weight 107.870. 2. Phosphorus pentoxide is found to be attacked by pure hydrochloric-acid gas, and hence is unsuited for drying this gas, thus confirming the results of Bailey and Fowler, 3 . It is shown that no appreciable error is introduced from this source, if a dry salt is fused in a current of hydrochloric acid which has been dried by phospho- rus pentoxide. II. A REVISION OF THE ATOMIC WEIGHT OF CADMIUM. THE ANALYSIS OF CADMIUM BROMIDE. By Gregory Paul Baxter, Murray Arnold Hines, and Harry Louis Frevert. Journal of the American Chemical Society, 28, 770 (1906). Zeitschrift fiir anorganische Chemie, 49, 415 (1906). Chemical News, 94, 224, 236, 248 (1906). Contributions from the Chemical Laboratory of Harvard College. A REVISION OF THE ATOMIC WEIGHT OF CADMIUM. THE ANALYSIS OF CADMIUM BROMIDE. Since the research described in the preceding paper indicates that the atomic weight of cadmium is nearly one tenth of a unit higher than the results of recent prior determinations by other investigators, in order to confirm or dis- prove the higher value the analysis of cadmium bromide was undertaken. PURIFICATION OF MATERIALS. CADMIUM BROMIDE. The cadmium material employed for the work consisted of fractions II and III of sulphide, which were obtained in the earlier investigation (page 4). Here also the sulphide was first converted into sulphate, and after crystalli- zation of the sulphate the cadmium was deposited electrolytically upon a platinum dish which had been coated with a thin film of paraflfine. The deposit was separated from the dish mechanically and after washing with water was freed from paraffine with redistilled ether and alcohol. By the method finally adopted for converting the cadmium into bromide, it was covered in a quartz dish with water slightly acidified with hydrobromic acid to prevent the formation of basic cadmium salts, and the purest bromine was added in small quantities until the metal was almost wholly dissolved. The solution was heated with the residual metallic cadmium upon a steam-bath until every trace of bromine had disappeared. Then it was filtered with a platinmn funnel into a platinimi dish, and was recrystalHzed three times, with centrif- ugal drainage in the platinum funnel after each crystalUzation.^ The original solution contained only traces of sulphate, and, when tested with barium hy- droxide, the mother-Uquors of the third crystallization gave absolutely no test for sulphate, hence the crystals themselves must have been pure (Samples II and III). The crystals were dried over potassium hydroxide in a vacuum desiccator. BROMINE. Commercial bromine was freed from chlorine by two distillations from a con- centrated solution of a bromide, the bromide in the second distillation being almost free from chloride. The bromine was covered with water, and hydrogen sulphide, which had been thoroughly washed with water, was passed into the ^ Richards: Jour. Amer. Chem. Soc, 27, no (1905). 21 22 RESEARCHES UPON ATOMIC WEIGHTS. solution until reduction of the bromine was complete. The solution was boiled, after mechanical separation of the greater part of the free sulphur and bromide of sulphur, and was filtered. Iodine was eliminated by boiling the hydrobromic acid with several small portions of potassium permanganate and rejecting the bromine set free. By heating the remainder of the hydrobromic acid with an excess of permanganate, over half of the bromine was obtained in the free state. The process of reduction with hydrogen sulphide and oxidation with perman- ganate was then repeated with the resulting bromine, and the final product was redistilled shortly before use. SILVER. One sample of silver was purified especially for this research. The processes to which it was subjected consisted first of precipitation as chloride from nitric- acid solution and reduction with invert sugar and sodiimi hydroxide. After fusion with a blowpipe on a crucible of the purest lime the metallic buttons were freed from surface impurities by scrubbing with moist sand and etching with nitric acid. Next the buttons were dissolved in nitric acid and the solu- tion was reduced with ammonium formate.^ The precipitated silver was thoroughly washed and again fused on a Ume crucible. The final process of purification consisted in electrolyzing the silver as described on page 6. The electrolytic crystals were fused in a current of hydrogen on a lime boat, and the buttons, after cleansing with nitric acid and dr5dng at 200°, were cut into fragments of convenient size with a clean chisel and anvil. Then they were again treated with fresh portions of dilute nitric acid until free from iron, washed, dried, and finally heated to about 400° in a vacuum. This silver was employed in analyses 4 to 8. In the first three analyses a mixture of two specimens of silver was em- ployed, both of which had already been used in an investigation upon the atomic weight of iodine by one of us.^ One was prepared from silver nitrate which had been seven times recrystalhzed from nitric acid, five times recrys- talHzed from water, and finally precipitated with ammonium formate. The other was precipitated once as silver chloride, electrolyzed once, and finally reduced with ammonium formate. DRYING OF CADMIUM BROMIDE FOR ANALYSIS. The method of analysis was essentially that usually employed in this laboratory for the analysis of metalHc haUdes. Weighed portions of the bromide, after fusion in nitrogen and hydrobromic-acid gases, were first titrated against weighed portions of silver. Then the precipitated silver bromide was collected and weighed. ^ Richards: Pub. Car. Inst. No. 28, p. 19 (1905); Jour. Amer. Chem. Soc, 27, 475; Zeit- anorg. Chem., 47, 72, 2 Baxter: Proc. Amer. Acad.. 41, 79 (1905); Jour. Amer. Chem. Soc. ,27, 881; Zeit. anorg. Chem., 46, 42. (See page 108.) THE ANALYSIS OF CADMIUM BROMIDE. 23 The apparatus used for the fusion of the salt in nitrogen and hydrobromic- acid gases was employed in the preparation of ferrous bromide by one of us/ and is a modification of apparatus used for a similar purpose in the determina- tion of the atomic weight of cobalt,^ in this laboratory. Nitrogen was prepared by passing air through concentrated ammonia solution in the bottle F (fig. 3) and then over hot copper gauze in the hard-glass tube G. The excess of am- FiG. 3. — Apparatus for the fusion of bromides in a current of dry nitrogen and hydrobromic-acid gases. monia was removed by dilute sulphuric acid in the bottles H and I. The gas was then conducted into an apparatus constructed wholly of glass, with ground joints, which consisted of a tower, J, filled with beads moistened with silver ni- trate solution to remove sulphur compounds, two similar towers, K and L, con- taining dilute sulphuric acid to ehminate last traces of ammonia, and two towers, M and N, filled with sticks of fused potassiimi hydroxide to absorb moisture and carbon dioxide. The partially dried gas, after bubbling through bromine in a small flask, P, passed into a second flask, Q, containing concentrated hydrobromic-acid solution in which washed red phosphorus was suspended, to convert the bromine into hydrobromic acid. A U-tube, R, also containing red phosphorus and hydrobromic acid, removed traces of bromine which escaped reduction in the flask. Two additional U -tubes, S and T, containing beads moistened with concentrated hydrobromic acid only, served to eliminate phos- phorus compounds which were found, in the investigation upon ferrous bro- mide,^ to accompanji' the hydrobromic acid if the phosphorous acid in the reduc- tion flask was allowed to become very concentrated. Finally, the mixture of 1 Baxter: Proc. Amer. Acad., 39, 246 (1903); Zeit. anorg. Chem., 38, 233 (1904). 2 Richards and Baxter: Proc. ^raer.^cac?., 33, 117 (1897); Zeit. anorg.Chem., 16, 572 (i ' Loc. cit. 24 RESEARCHES UPON ATOMIC WEIGHTS. nitrogen and hydrobromic-acid gases was thoroughly dried, first by pure fused calcium bromide, in the tube U, and then by resublimed phosphorus pentoxide in the tube W. If desired the nitrogen could be passed directly through the phosphorus pent- oxide tube O into the botthng apparatus XY, in which the fusion took place. Air was purified and dried by passing over fused potassium hydroxide in the tower A, concentrated sulphuric acid in the towers B and C, and phosphorus pentoxide in the tubes D and E. The cadmium bromide, contained in a weighed platinum boat, was heated gently in a current of nitrogen until a small quantity of residual crystal water was expelled, then strongly in a current of nitrogen and hydrobromic acid imtil fused. After the salt had cooled, the hydrobromic acid was displaced by nitro- gen and this in turn by dry air. During the displacement of the hydrobromic acid by nitrogen and air a shght backward current of gas was maintained through the tube W and the trap V. The boat was then transferred to the weighing-bottle in which it was originally weighed, and the stopper was inserted without an instant's exposure of the salt to moisture, by means of the bottling apparatus which has been referred to on page 9. The weighing-bottle was then allowed to stand in a desiccator near the balance case for some time before it was weighed. METHOD OF ANALYSIS. Next the boat was transfered to a flask and the salt was dissolved in about 300 c.c. of the purest water. The weighing-bottle was rinsed and the rinsings were added to the solution. Then the solution was filtered into the glass-stop- pered precipitating flask through a tiny filter to collect a trace of insoluble matter, and the filter-paper and residue were ignited at a low temperature in a weighed porcelain crucible. This residue, which usually amounted to less than 0.1 mg. and was never as much as 0.2 mg., did not contain detectable quantities of cadmium, and probably consisted of silica and a trace of platinum re- moved from the boat during the fusion, for the boat, when reweighed, in most cases was found to have lost a few hundredths of a milligram. No change in weight could be found when the boat was first dried and weighed, then ignited and reweighed. The difference between the weight of the residue and the loss in weight of the boat was subtracted from the weight of the cadmium bromide. From the corrected weight of the cadmium bromide very nearly the requisite quantity of pure silver was calculated. This silver was weighed out and dis- solved, in nitric acid diluted with an equal volimie of water, in the flask de- scribed on page 12. After the silver was dissolved, the solution was diluted to twice its volume and was heated until free from nitrous fumes. Then it was still further diluted until not stronger than i per cent, and was slowly added, with constant stirring, to the i per cent solution of cadmium bromide in the precipi- THE ANALYSIS OF CADMIUM BROMIDE. 25 tating flask. In three analyses (4, 5 and 8), this procedure was varied by adding the bromide to the silver nitrate. After being shaken for some time, the solution was allowed to stand several days, with occasional shaking, until the supernatant liquid was clear. 30 c.c. portions of the solution were then tested with hundredth normal solutions of silver nitrate and sodium bromide in the nephelometer ^ for excess of bromide or silver, and, if necessary, either standard silver nitrate or sodium bromide solution was added, and the process of shaking and testing repeated, until the amounts of bromide and silver in the solution were equivalent. If the solution was perfectly clear when tested, and contained no considerable excess of bromide or silver, the test solutions were discarded, since they contained only negligible amounts of dissolved silver bromide; other- wise they were returned to the flask and a correction was applied for the silver bromide thus introduced. As soon as the exact end-point of the titration had been found, about 4 eg. of silver nitrate in excess were added, to precipitate dissolved silver bromide, and the flask was again shaken and allowed to stand until clear. The precipi- tate of silver bromide was collected upon a weighed Gooch crucible, after it had been v/ashed by decantation about eight times with pure water. Then it was heated in an electric air-bath, first for several hours at 140°, finally for an hour at 200°, and, after it had cooled in a desiccator, it was weighed. In order to determine how much moisture was retained by the precipitate, in each case it was transferred as completely as possible to a clean porcelain crucible and weighed; then the salt was fused by heating the small covered crucible, con- tained in a large crucible, and again weighed. After fusion the silver bromide was light yellow, with only a trace of darkening, showing that no appreciable reduction had taken place. The asbestos mechanically detached from the Gooch crucible, together with a minute quantity of silver bromide which oc- casionally escaped the crucible, was collected from the filtrate and wash-waters upon a small filter, the ash of which was treated with nitric and hydrobromic acids before weighing. Although the filtrates and first wash-waters were essen- tially free from dissolved silver bromide, the subsequent wash-waters usually contained a trace of this substance. The amount of dissolved salt was deter- mined with the nephelometer by comparison vnth. standard bromide solutions. Finally, the weight of silver bromide was corrected for the sodium bromide introduced. Although in our analyses of cadmium chloride no evidence could be obtained of appreciable occlusion of either cadmium or silver salts by silver chloride, especial precautions were taken to avoid any possibihty of such a difficulty in this research. In the first place both the cadmium bromide and the silver nitrate solutions were very dilute during precipitation, each one having a vol- ume of about I liter. In the second place the method of precipitation was 1 See page 11. 26 RESEARCHES UPON ATOMIC WEIGHTS. varied by sometimes adding the silver nitrate to the cadmium bromide (analyses I, 2, 3, 6, and 7), and sometimes adding the bromide to the silver nitrate (analyses 4, 5, and 8). And in the third place the solutions were allowed to stand varying periods before the titration was completed, so that occluded substances might have opportunity to be dissolved. Analysis i, in which the largest quan- tity of bromide was employed, over 11 gm., which is to be expected to give the most marked evidences of occlusion, was not tested for 5 days after precipita- tion, and the titration was completed 8 days later. In the other analyses the period between precipitation and the completion of the titration varied from 7 days in analysis 4 to 3 days in analysis 8. Furthermore, in some cases, after the end-point had been reached, the solutions were allowed to stand some days longer with occasional testing. No change in end-point with standing was ob- served. In spite of these differences in the method of procedure, the variations in the final results do not exceed the experimental error to be expected, except in the case of analyses 4 and 12. Evidently, occlusion of any sort must have been very slight if it existed at all. Analyses 4 and 12, performed with the same portion of bromide, differ so markedly from the others that, although no reason for the difference is known, they are rejected in computing the final average. The gold-plated brass weights were carefully standardized to hundredths of a milligram. Vacuum corrections of -}-o.oooo86 for cadmium bromide,^ of +0.000041 for silver bromide and of —0.000031 for silver were appUed. All weighings were made by substitution with counterpoises as nearly Uke the ob- jects to be weighed as possible. The analytical work was performed wholly by Dr. Hines. (See table on opposite page.) RESULTS AND DISCUSSION. The ratios of silver used to silver bromide obtained in the same analysis afford sufficient proof of the purity of the bromine and silver, as well as confirmatory evidence of the absence of appreciable occlusion by the silver bromide. Ag : AgBr. Analyses i and 9 57-4446 2 " 10 57-4466 3 " II 57-4438 4 " 12 57-4438 s " 13 57-4423 6 " 14 57-4440 7 " 15 57-4430 8 " 16 57-4431 Average, 57-4439 ^ The specific gravities of cadmium and silver bromides have recently been found to be 5.192 and 6.473 respectively. Baxter and Hines: Amer. Chem. Jour., 3i» 220 (1904). As in the case of cadmium chloride, slight changes in the vacuum corrections of cadmium and silver bromides have been made from the values employed in the original publication of this paper, owing to more exact knowledge of the density of the weights. THE ANALYSIS OF CADMIUM BROMIDE. 27 o"o a >A00 CO P» c» CO l^ 1 el's a mmOO^mhmh HI H M 2 tea Tj-Tj-^cO'tTf-^'^ ■* ■^ '1""1"'?'"1'"1'"1""1"t •* fJcjMMCSNcJcJ « C< ^11 ««cJ?JMe5wM M M M 0) . M N HH ^ COO CJ Tf. M CI 100 "lil "O "O |5« ^MOOOOOlMCOO g 00 M H t^oo r^ 10 .23 a «« . 0> ■* t^ t^vO « 0» tn ■g J M r^M t^O». J! -■p<)t^^>>o«0^'c^ £00 >^ M r^ ro vooo ! H 0\ Oi Ov t^OO 06 06 •3 ^ (3 tn ■«,- to .S 0) . K-H 0^^^lo^^c^ n rj- > l-lt^l-(00U-). q H 00 os OS "s" S) H O^t^lOt'^PJ N •<*• "§■ go ^ HVOvO t>»U5VoiOvri 11^ .i-tt>.HOO»OMOO\ H (u 2 K N c< -"^ t/^oo CO r^ J^ SjVO N 1000 CO "* Ov ■^00 r;- q H 00 0^ <:>■ bp (y p«^ 2 -g . cOOOMOvOmvo wvdo t>.vr)ioio>r> coOOvoOcscocQ ^ ^ H -4J ^H ca 2 0.000 0.000 0.000 0.000 0.000 0.000 0.000, 0.000 .vO C« «*-( <; CO Ov>0 (O O^VD bp Tj-00 00 CO t^ to M 10 c« p••^t^^.q q IN M 'S •2J 1^^ 10 d> On 6* t^od 06 00 66666666 fe «M 00000000 .9°^ 00000000 lol 0000 mOOOOOOO t^ .00000000 1^1 h3 * gqqqqoooq i§§§§§§§§ Qodddddo'o ^ g & 66666666 0000 MVOOOOOOO t~ t^ OwwOOOOw .-l^-S .00000000 «J| VO M N t^ XO^O 0\ fiOOOQOOOO gjOqoooooo M 0\ COOO >0 « w ►5 *"S %^l „• N N ■'^ 1000 CO r^oo g \0 c^ >O00 CO ■* 0\ ^ 2 66666666 Tj-oO »>■ w 00 On q^ HVOVO t>.«0»OV3lO H 10 M ooo CO -la- o\ a 'i-oo t;- q H 00 c^ q. |o| l-ll-thHhHh-IMI— IhH t— 11— (1— (l-HI-Hh-ll— II— 1 MVOVO tilOlOlOlO HI t-t 1— 1 HH 1— 1 :g s • HHI— (HHI— It— (1— Ih-IH- 1 1— 1 t— ( 1— 1 1— 1 1l M « CCt >OVO t^OO On M « CO 'l* >00 cS 28 RESEARCHES UPON ATOMIC WEIGHTS. The most probable value for this ratio has been shown by Baxter to be 574453-' In the preceding paper Bailey and Fowler's ^ statement that hydrochloric- acid gas is contaminated with volatile compounds of phosphorus, if it is dried with phosphorus pentoxide, is confirmed. Although Bailey and Fowler attrib- ute to hydrobromic acid an effect similar to that of hydrochloric acid, for several reasons it is certain that in the experiments upon cadmium bromide no appreciable amount of phosphorus was introduced into the salt by the action of the hydrobromic acid upon the phosphorus pentoxide. In the first place, the experiment of passing into water hydrobromJc-acid gas, formed as in our work by passing nitrogen through bromine and then through an emulsion of red phosphorus in concentrated hydrobromic-acid solution, and dried first by fused calcium bromide and then by phosphorus pentoxide, was performed in this laboratory some years ago in connection with the analysis of cobalt and nickel bromides. In this experiment no phosphorus could be discovered in the aqueous solution. In the second place, in two analyses of bromides which had been heated in hydrobromic-acid gas, the filtrates from the silver-bromide pre- cipitates were evaporated to small bulk and tested for phosphoric acid, with negative results, while the slight residues obtained by filtering the aqueous solu- tions of the original bromides also showed in one case the complete absence of phosphorus, and in the other the presence of only a minute trace of this sub- stance, although in the latter case the salt had been sublimed in a current of hydrobromic acid and therefore contained maximum amounts of phosphorus.^ This result was to be expected from a consideration of the fact that the hydro- bromic-acid gas used in these experiments was diluted with at least twice its volume of nitrogen. In the light of this evidence it seems safe to assume that in the numerous analyses of bromides which have been carried out in this labo- ratory in recent years, no error was introduced by the use of phosphorus pent- oxide as drying agent for the hydrobromic-acid gas. Nevertheless, with more concentrated hydrobromic acid, doubtless it would be unwise to use this drying agent. It is interesting to compare the results of the analyses of the different frac- tions of material in this research and in the preceding one. (See table on op- posite page.) The close agreement of the results from fraction II by different methods and of the results from all three fractions leaves no doubt of the identity of the dif- ferent specimens of material. No matter how the results are averaged, the same conclusion is reached as in the previous paper, i. e., that the atomic weight of cadmivun hes very near the * Proc. Amer. Acad., 42, 210 (1906); Jour. Amer. Chem. Soc, 28, 1332; Zeit. anorg. Chem., 50, 398. (Seepage 51.) 2 Chem. News, 58, 22 (1888). 3 Baxter: Proc. Amer. Acad., 39, 248 (1903); Zeit. anorg. Chem., 38, 236. THE ANALYSIS OP CADMIUM BROMIDE. 29 value 112.417 (Ag = 107.880). If the atomic weight of silver is 107.870, then the atomic weight of cadmium is 112.407. Fraction. I . . . I . . . II, Series i II, Series i II, Series 2 II, Series 2 II . . . II . . . Ill ... Ill ... Ratio. CdClz CdCla CdCla CdCl2 CdCl2 : CdCl2 : CdBr2 CdBra : CdBra CdBra 2Ag 2AgCl 2Ag 2AgCl 2Ag 2AgCl 2Ag 2AgBr 2Ag 2AgBr Average Atomic weight. II2.415) I12.419 ) 112.4031 112.429 ) I12.418 1 II2.418 ' I12.418 I 112.414 ' 112.4231 112.413/ I12.417 Average. 112.417 112. 416 112.418 112.416 112.418 112.417 Attention should be called to the agreement with ours of the results of Morse and Arbuckle's synthesis of cadmium oxide, which )delded the value 112.38, and of Bucher's painstaking work upon the halogen compounds of cadmium. Bucher's values from cadmium chloride vary between 112. 21 and 112.41, with an average of 112.32, but if the first 7 of his 21 experiments are rejected, his average becomes 112.35, and 6 of his results are as high as 112.38. His analyses of the bromide vary between 112.25 and 112.41, with an average of 112.34. The results of this investigation are then as follows: (i) The value for the atomic weight of cadmium previously found by analy- sis of cadmium chloride, 112.42 (Ag = 107.880), is supported by the analysis of cadmimn bromide. If silver is 107.870, cadmium becomes 1 12.41. (2) It is pointed out that phosphorus pentoxide is not perceptibly attacked by hydrobromic-acid gas which is diluted with twice its volume of nitrogen. III. A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. THE ANALYSES OF MANGANOUS BROMIDE AND CHLORIDE. By Gregory Paul Baxter and Murray Arnold Hines. Journal of the American Chemical Society, 28, 1560 (1906). Zeitschrift fvir anorganische Chemie, 51, 202 (1906). Chemical News, 95, 102, iii, 123 (1907). Contributions from the Chemical Laboratory of Harvard College. A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. THE ANALYSES OF MANGANOUS BROMIDE AND CHLORIDE. INTRODUCTION. The following table, adapted from Clarke's "A Recalculation of the Atomic Weights" ^ gives a brief resimie of previous work upon the atomic weight of manganese which has other than historical interest. Date. Investigator. Reference. Ratio determined. Result. 1830 Berzelius . . . Ann. Physik. Chem., 18, 74 MnCU: 2AgCl S5-IO 1831 Turner .... Trans. Roy. Soc, Edinb. 11, 143 MnCla: aAgCl 54-90 1857 von Hauer . . J. pr. Chem., 72, 360 MnS04: MnS 54-91 1859 Schneider . . . Ann. Physik. Chem., 107, 605 Mn: 2CO2 54-03 Rawack .... Ibid. MngO^iHaO 54.08 i860 Dumas .... Ann. Chem. Pharm., 113, 25 MnCU: 2Ag 54-96 1883 Dewar and Scott Proc. Roy. Soc, 35. 44 AgMn04: AgMnO AgMn04: KBr 55-01 5S-04 i88s Marignac . . . Arch. sci. phys. nat., [3] 10, 21 MnO:MnS04 55-01 1890 Weeren .... Dissertation, Halle MnO: MnSOi MnS:MnS04 55-00 55.00 The close agreement of the greater part of these determinations is striking, the experiments of Schnieder and Rawack being the only ones which indicate a value for manganese very different from 55.0. The variations of their results from the others is not surprising, however, since manganoso-manganic oxide and manganous oxalate, with which they worked, are undoubtedly difi&cult to obtain in a pure condition. The remaining determinations all fall within limits two tenths of a unit apart, and all but two agree within thirteen hundredths of a unit. For this investigation the substances chosen for examination were manganous bromide and chloride, since the analysis of halogen compounds may be ef- fected with great accuracy. Furthermore, these compounds have not been in- vestigated by any of the more recent experimenters except Dewar and Scott,^ who performed one analysis each of the chloride and bromide and obtained the values 54.89 and 54.95 respectively. » Smith. Misc. Coll., The Constants of Nature, Part V., 1910. The results have been calculated with the use of the following atomic weights: = 16.00; C = 12.00; S = 32.07; CL= 35-46; K = 39.10: Ag = 107.88. ' Loc. cit. 33 34 RESEARCHES UPON ATOMIC WEIGHTS. THE ANALYSIS OF MANGANOUS BROMIDE. PURIFICATION OF MATERIALS. All of the water used in either the purification or the analyses was twice dis- tilled, once from a dilute alkaline solution of potassium permanganate and then from a very dilute sulphuric-acid solution. Block-tin condensers were used in both distillations and the apparatus contained no rubber or cork connections. The water was collected as a rule in Jena glass flasks, although for special pur- poses either platinum or quartz receivers were substituted. Acids and ammonia also were distilled shortly before use, either platinum or quartz condensers and receivers being employed when necessary. Solid reagents were recrystallized, usually with centrifugal drainage. Special pains were taken in all the work to prevent the introduction of alka- lies or silica into the purest materials, by avoiding as far as possible the use of glass vessels. MANGANOUS BROMIDE. Four different specimens of manganous bromide were employed, which were obtained from different sources and were purified in different ways. In the case of Samples A and B, purification of the manganese from other heavy metals was accompUshed by recrystallization of Merck's "chemically pure" potassium permanganate. Sample A was crystallized three times only, while Sample B was thus treated ten times, the last two crops of crystals being thoroughly freed from the mother-liquors by centrifugal drainage. In order to free the manganese from potassium and convert it into the bro- mide, the following processes were employed with Sample A: First the perman- ganate was dissolved in water and was reduced by passing sulphur dioxide into the solution. This sulphur dioxide was made by heating copper turnings with concentrated sulphuric acid and was purified from copper compoimds mechani- cally carried along by passing through three gas washing-bottles, each con- taining a solution of sulphurous acid, and one column of beads moistened with a similar solution. From the solution of potassium and manganous sulphates the manganese was precipitated by the addition of an alkaline solution of am- monium carbonate. The manganous carbonate was washed with water imtil the washings were free from sulphates, then it was dissolved in nitric acid, which had been redistilled until free from chlorine, and the manganous nitrate was recrystallized six times from a solution strongly acid with nitric acid, four times in a glass vessel, twice in platinum. Usually it was necessary to start crystallization by inoculation, and cooling with ice was found advisable for the sake of economy in material. From a dilute solution of the purified nitrate in a platinum vessel, the manganese was again precipitated as carbonate, by means of ammonium carbonate which had been freshly made by passing pure carbon dioxide into distilled ammonia in a plati- A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. 3$ num flask. The resulting manganous carbonate, after thorough washing with water containing a small amount of ammonia to prevent colloidal solution of the carbonate, was readily converted into bromide by solution in hydrobromic acid. Since it was probable that the carbonate contained occluded nitrate, and since a portion of the material had been oxidized to the manganic state during the washing, it was obvious that bromine would be set free during the solution in hydrobromic acid. The use of a platimun vessel for this purpose was there- fore precluded. In order to avoid the introduction of silica, fused quartz dishes were employed, instead of glass vessels. The former have been shown to be practically insoluble in acid solutions.^ The free bromine was expelled from the solution of manganous bromide by prolonged heating on a steam-bath in a quartz dish. Finally it was crystallized six times, thrice in quartz, and, after filtration with a platinum funnel, thrice in platinum with centrifugal drainage after each crystallization. The crystals were dried as far as possible over stick potash in a vacuum desiccator. From the mother-liquors, by means of six similar crystallizations. Sample A2 was obtained. In the conversion of Sample B from permanganate to bromide minor changes were introduced. The ammonium carbonate was prepared in a pure state by distilling a solution of commercial ammonium carbonate in a platinum still. Instead of expelling free bromine from the solution of manganous bromide by prolonged heating upon the steam-bath, the solution was evaporated as far as possible upon the steam-bath and the residue was heated to 200° in an electric oven. The bromide was dissolved in water, and after filtration of the solution, was crystallized three times in a platinum dish. The third crop of crystals is designated as Sample B. Sample C was prepared from a commercial specimen of pyrolusite. This was first dissolved in hydrochloric acid and the solution was boiled to expel chlorine. Hydrogen sulphide was passed into the diluted solution of manganous chloride to saturation, and the precipitate of sulphur and sulphides was removed by fil- tration. After the excess of hydrogen sulphide had been expelled by boiling, the solution was fractionally precipitated with sodium hydroxide until the precipi- tate was free from iron. Finally, the manganese was precipitated with ammo- nium carbonate and the precipitate was washed and dissolved in nitric acid. The nitrate was recrystallized and converted into bromide exactly as in the case of Sample A. The source of Sample D was Merck's "chemically pure" manganous sul- phate. A solution of 500 gm. of this salt was first saturated with hydrogen sul- phide, and the precipitate, which consisted chiefly of manganous sulphide, was removed by filtration. After the addition of a small amount of ammonia, hy- drogen sulphide was again passed into the solution to saturation, and the pre- » Mylius and Meusser: Zeit. anorg. Chem., 44, 221 (1905). (See also page 36 of this paper.) 36 RESEARCHES UPON ATOMIC WEIGHTS. cipitate discarded. In a similar way third and fourth fractions of sulphide were removed. Next the solution was thrice fractionated with small portions of potassium hydroxide, the precipitate being rejected in each case. Then the manganese was twice precipitated as carbonate by means of ammonium car- bonate and the manganous carbonate was converted into bromide exactly as in the case of Sample B. The first crop of thrice recrystallized bromide is desig- nated Sample Di, a second similar crop obtained from the mother-liquors is Sample D2. HYDROBROMIC ACID. Commercial bromine was freed from chlorine by twice converting the bro- mine into hydrobromic acid by means of thoroughly washed hydrogen sulphide and water, and heating the hydrobromic acid, after distillation, with recrystal- lized potassium permanganate. The bromine was thus twice distilled from a bromide, the bromide in the second distillation being almost free from chloride. Iodine was eliminated by boiUng the hydrobromic acid in each case with a small quantity of permanganate and rejecting the bromine set free. A portion of the final product, when converted into ammonium bromide by means of ammonia, and added to a solution of 3.46875 gm. (in vacuimi) of pure silver, yielded 6.03855 gm. (in vacuum) of fused silver bromide, whence the ratio of silver to silver bromide is 57.443, while 57.445 is the value to be expected.^ By treating this bromine, covered with water, with washed hydrogen sulphide, hydrobromic acid was again produced. The solution was boiled, after mechani- cal separation of the greater part of the free sulphur and bromide of sulphur, and was then filtered. In order to remove the sulphuric acid produced during the action of the bromine upon the hydrogen sulphide, the hydrobromic acid was first distilled. Then it was diluted, and a small quantity of recrystallized barium hydroxide was added to precipitate last traces of sulphuric acid. The slight precipitate of bariimi sulphate was collected upon a filter, and the acid was three times distilled with rejection of the first and last portions, with a glass retort and condenser. Finally the acid was once distilled with the use of a quartz condenser. The product of the final distillation was collected in quartz vessels and was used immediately for dissolving the manganous carbonate. That this acid was free from solid impurities, such as alkalies and silica, was shown by evaporating 30 c.c. in a weighed platimmi crucible. No weighable residue remained after the crucible had been heated to very dull redness. NITRIC ACID. This acid was twice distilled, all but the last third of the distillate being rejected in each distillation. This acid gave no test for chloride in a nephel- ometer. * Baxter: Proc. Amer. Acad., ^2, 210 (1906); Jour. Amer. Chem. Soc, 28, 1332; Zeit. anorg. Chem., 50, 398. (See page 59.) A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. 37 SILVER. Five different specimens of silver were employed, a portion of each one of which had already been used in an atomic weight research, and had been shown to be of the highest grade of purity. Two of these specimens. Samples H and J, were used in an investigation upon the atomic weight of iodine by one of us.^ Sample H was prepared from silver nitrate which had been seven times re- crystallized from nitric acid, five times recrystaUized from water, and finally precipitated with ammonium formate. Sample J was precipitated once as silver chloride, electrolyzed once, and finally precipitated with ammonium formate. Sample K was used in our first investigation upon the atomic weight of cad- mium.2 This sample was thrice precipitated as silver chloride and once elec- trolyzed. Sample L was precipitated once as chloride, once as metal by am- monium formate and was once electrolyzed. This sample has been used in the analysis of cadmium bromide.^ Sample M was prepared for an investigation upon the atomic weight of bromine, and had been twice electrolyzed after a preliminary purification."* Samples H, J, and L also were used in the latter re- search, and were found to give values identical with those obtained with Sample M. All five samples were finally fused in a current of pure hydrogen in a lime boat. The fused lumps were cleaned with dilute nitric acid, cut into fragments either with a clean steel chisel and anvil, or with a jeweler's saw, treated with dilute nitric acid until free from iron, washed, dried, and finally heated to about 300° in a vacuum. DRYING OF MANGANOUS BROMIDE. The method of analysis was essentially that usually employed in this laboratory for the analysis of metallic halides, and has been described in the preceding papers on the cadmium halides. Weighed portions of the bromide, after fusion in hydrobromic acid, were first titrated against weighed portions of pure silver. Then the precipitated silver salt was collected and weighed. The apparatus used for the fusion of the manganous bromide in a current of nitrogen and hydrobromic-acid gases has already been described in detail on page 23. The manganous bromide, contained in a weighed platinum boat, was heated gently in a current of nitrogen, until the greater part of the crystal water was expelled, then strongly in a current of nitrogen and hydrobromic acid until fused. After the salt had cooled, the hydrobromic acid was displaced by nitro- gen and this in turn by pure dry air, the purif)dng apparatus being constructed in such a way that by means of stop-cocks any one gas or mixture of gases could be employed, to the exclusion of the others. The boat was then transferred to ^ See page 108. ^ See page 6. ' Seepage 22. * See page 56. 38 RESEARCHES UPON ATOMIC WEIGHTS. the weighing-bottle in which it was originally weighed, and the stopper was in- serted without an instant's exposure of the salt to moisture, by means of the bottling apparatus described on page 9. The weighing-bottle was then al- lowed to stand in a desiccator near the balance case for some time before it was weighed. METHOD OF ANALYSIS. Next the boat was transferred to a flask and the salt was dissolved in about 300 c.c. of the purest water. The weighing-bottle was rinsed and the rinsings were added to the solution. Then the solution was filtered into the glass stop- pered precipitating flask through a tiny filter to collect a trace of insoluble matter, and the filter-paper and residue were ignited in a weighed porcelain crucible. From the weight of manganous bromide very nearly the requisite quantity of pure silver could be calculated. This silver was weighed out and after solu- tion in nitric acid as described on page 12, and dilution until not stronger than I per cent, the solution was slowly added with constant stirring to the i per cent solution of manganous bromide in the precipitating flask. After having been shaken for some time, the solution was allowed to stand several days, one week in the case of analyses 14 and 15, with occasional shaking, imtil the super- natant liquid was clear. 30 c.c. portions of the solution were then tested with hundredth-normal solutions of silver nitrate and sodiimi bromide in the nephel- ometer for excess of bromide or silver, and, if necessary, either standard silver nitrate or sodium bromide solution was added, and the process of shaking and testing repeated, until the amounts of bromide and silver in the solution were equivalent. If the solution was perfectly clear when tested, and contained no considerable excess of bromide or silver, the test solutions were discarded, since they contained only negligible amounts of silver bromide, otherwise they were returned to the flask and a correction was applied for the silver bromide thus introduced. As soon as the exact end-point of the titration had been found, about 4 eg. of silver nitrate in excess were added to precipitate dissolved silver bromide and the solution was again shaken and allowed to stand until clear. The pre- cipitate of silver bromide was collected upon a weighed Gooch crucible after it had been washed with water by decantation about ten times. Next it was heated for several hours at 140°, then for 2 hours at about 230° in an electric air-bath, and, after it had cooled in a desiccator, it was weighed. In order to determine how much moisture was retained by the precipitate, in each case it was trans- ferred as completely as possible to a clean porcelain crucible and weighed; then the salt was fused by heating the small crucible, contained in a large covered crucible, and again weighed. The fused salt was light yellow as a rule, showing that no appreciable reduction had taken place. The asbestos mechanically de- tached from the Gooch crucible together with a minute quantity of silver bro- A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. 39 mide which occasionally escaped the crucible, was collected from the filtrate and wash-waters upon a small filter, the ash of which was treated with nitric and hydrobromic acids before weighing. Although the filtrates and first wash- waters were essentially free from dissolved silver bromide, the subsequent wash- waters usually contained a trace of silver bromide. The amount of dissolved salt was determined with the nephelometer by comparison with standard, bro- mide solutions. Several difliculties were met in carrying out the analyses. In the first place it proved difficult to wash the platinum boat absolutely clean. When rinsed with cold water only, and dried at ioo°, the weight was in many cases a few hun- dredths of a milligram greater than before fusion of the bromide. Ignition to redness of the boat thus treated then produced a slight loss in weight. Rinsing with hot water reduced the gain in weight of the boat after drying but did not wholly prevent a slight loss on ignition. The cause of the variation was not discovered, hence it seems safer in the calculations to use the weight of the boat after drying. The total variation is so slight, however, that it scarcely affects the final result. Two other difficulties arose from the fact that when a solution of a manganous salt, even as dilute as the filtrates from analyses, is filtered through filter-paper, in spite of long-continued washing a small amount of manganese is tenaciously retained by the paper. This was discovered from the fact that the asbestos residues always contained manganese. In analyses 29 to 31 it was found possible to eliminate the manganese completely by washing the filter finally with 5 per cent hydrobromic acid. In two cases the residues were analyzed for manganese and were found to contain 0.00023 3-nd 0.00057 S^- o^ ^1^3 O4 respectively. The average of these two quantities is, however, larger than the total residue in some cases, hence this value can not be used to correct the previous analyses. In order to determine accurately the proper correction for this error, a solution containing manganous nitrate in the porportion in which it was usually con- tained in the filtrate of an analysis was passed through filter-papers and the fil- ters were then washed as thoroughly as possible with water. The ash of these papers invariably contained manganese, the weights of manganic oxide in sev- eral experiments being found to be 0.00018, o.oooii, 0.00006, 0.00018 and 0.00005 S^- with an average of 0.00012 gm. This quantity was subtracted from the weight of asbestos shreds in all cases except analyses 29 to 31, where the paper was washed with hydrobromic acid. The residue obtained by the filtration of the manganous bromide proved to contain manganese and to be free from detectable amounts of platinum and silica. Probably this insoluble residue consisted chiefly of oxides of mangan- ese, although prolonged fusion in hydrobromic acid failed to reduce materially the proportion of insoluble matter. The discovery of adsorption in the case of manganous nitrate led to the suspicion that at least a portion of this residue 40 RESEARCHES UPON ATOMIC WEIGHTS. was due to adsorption of manganese compounds by the filter-paper. In order to test this point a solution of manganous bromide, containing about 5 gm. in 200 c.c, after one filtration was again filtered through a second filter about 3 cm. in diameter. Filters of this size were used in filtering the manganous bromide solution in the analyses. This filter was then washed with water as thor- oughly as in an analysis, and was ignited. The weight of manganic oxide ob- tained was 0.00008 gm. Two repetitions of the experiment pelded 0.00008 and o.oooio gm. respectively. If, as is probable, the manganese is adsorbed, not as bromide, but as some basic compound, possibly as manganic hydroxide, the bro- mine would have remained partially, if not wholly, in the solution. In that case a suitable correction could be applied by subtracting from the weight of the residue the average of the quantities of manganese adsorbed in the above exper- iments. An attempt to prevent the difficulty by adding dilute sulphuric acid to the solution of manganous bromide did not diminish the extent of the adsorp- tion. Hence a negative correction of 0.00009 S^- is applied to the residue in all cases. It has already been shown that chlorides and bromides which have been fused and allowed to solidify in an acid atmosphere occlude none of the gas, for they give neutral solutions.^ THE DENSITY OF MANGANOUS BROMIDE. In order to find accurately the vacuum correction for manganous bromide, it was necessary to determine the density of this salt. The experiments were carried out exactly as described in our determinations of the specific gravities of cadmium halides,^ with the following results: Density of MnBr2. Density of Toluol 25°/ 4° = 0.86156. Weight of MnBr2 in vacuum. Weight of toluol displaced in vacuum. Density of MnBrj. 2S°/4° gm. 3.0098 3.0342 gm. 0.5914 0.5963 4.38s 4-384 Aver. 4.385 The average density of several of the brass weights was found by displace- ment of water to be 8.3. 1 Richards: Proc. Amer. Acad., 29, 59 (1893); Zeit. anorg. Chem., 6, 93 (1894); Jour. Amer. Chem. Soc, 24, 376 (1902); Zeit. anorg. Chem., 31, 273; Richards and Baxter: Proc. Amer. Acad., 34, 367 (1899); Zeit. anorg. Chem., 21, 269; Baxter and Hines: Jour. Amer. Chem. Soc, 27, 227 (1905); Zeit. anorg. Chem., 44, 163. ' Amer. Chem. Jour., 31, 220 (1904). A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. 41 All weights were reduced to the vacuum standard by applying the following corrections for each apparent gram of substance. Specific gravity. Vacuum correction. Weights MnBra AgBr Ag Toluol 8.3 4.385 6.473 10.50 0.862 +0.000129 4-0.00004I —0.000031 +0.00126 The balance was a new Troemner, No. 10, and was easily sensitive to 0.02 mg. with a load of less than 50 gm. The weights, which were of brass, gold- plated, were occasionally carefully standardized to hundredths of a milligram. The corrections did not vary with time, however. All weighings were made by substitution, with tare vessels as nearly like those being weighed as possible. (See tables on pp. 42 and 43.) The close agreement of the averages of the two series is conclusive evidence that no serious error, such as occlusion by the silver bromide, affected the method of analysis. This is strikingly shown by the ratio between the silver used and the silver bromide obtained in the same analysis. Ag : AgBr Analyses i and 18 57.4438 2 " 19 57-4459 3 " 20 57.4426 4 " 21 57.4425 5 " 23 57-4439 8 " 24 57-4439 9 " 25 57-4457 10 " 26 . 57-4432 11 " 27 57-4423 12 " 28 57.4421 13 " 29 57-4454 14 " 30 57.4387 15 "31 57-4456 Average, 57-4435 Average, rejecting analyses 14 and 30 . . . 57.4439 The most probable value for this ratio has recently been shown to be 57.4453-^ Although the foregoing figures furnish strong e\ddence that the atomic weight of manganese lies very close to 54.93, it seemed advisable to attempt to confiirm this result by the analysis of other manganese compounds. The suc- cess which accompanied this investigation of the bromide led to the selection of manganous chloride for the next series of experiments. A comparison of the different specimens of material and a discussion of all the results is to be found at the end of this paper. * Baxter: Proc. Amer. Acad., ^2, 210 (1906); Jour. Amer. Chem. Soc, 28, 1332; Zeit. anorg. Chem. 50, 398. (See page 59.) 42 RESEARCHES UPON ATOMIC WEIGHTS. ^ m & >w d .2°S t>. « «/5 Ov 00 « PI p« ■* 00 to vo « CO •^ S5§ rO PO CO CO CO CO ■«»• CO rl- CO p) CO CO P* s o> q> » Ov 9- q. o> qv qv Ov Ov qv qv qv qv O M §) •* ■/1 vo V> lO lO to to to to to to to to to to 13 <« 0) o to 00 't CO O H Ov 00 p> O vo O Ov H f in rO «o M o • £; M< 00 qv NO 00 VO PJ vq qv to vq t^ to t^ O 4> VO ■* •^ to V} vd 4 CO to CO •* CO ■4 CO to Hi's ^ CO tn t» Ov Tj- CO Ov •* VO 00 p» O PI rO 8 Ov tH 00 HI o CO to »^ Ov VO t^ I:Sm ^- l^ O Ov Ov Ov 00 CO PI M o CO vO Ov 00 CO OV 00 Ov Ov VO Ov o t^ r^ Ov o g§ 00 •4 oq 4 vq >o to vd t-4 to CO to qv CO 4 vq CO l^ ■* to CO vq to O ^ t^ >o Q O O o o O Q „ o o 00 ^ to O M CO s? M o to g § o o o § § § § 8 o O o o o o o o o § § o 8 o 8 o Q O d d d d d d d d d d d d d d 1 1 1 1 1 + + + 1 + 1 + vc5 H ON l>. "o a M « T)- CO o ^4 Ov 00 H H o vo PI CO vo II t» Tt lO CO w to o t- M PI ^ o PI CO M ^ *^ CO 00 vo O OO '^l- t~ o PI i~ vo CO to S o fO o to M H 00 Tj- Ov Ov < l-< 5 >o 00 Ov vq 00 VO P» vq M qv to vq »>. to t^ . •* to o Ov ■* t MB u •^ .9 rO CO C) CO CO •* CO 00 Ov i^ S CO o p) Ov o ov 00 CO tH o •* vo M CO CO Ov Ov Ov Ov VO P( OV o t^ t^ Co >9 00 CO vq t>. to to tH Ov ■ M ►-» l-» M M HI w H- 1 + =g M < < u u < U m <5 m m « u p n P 02 S .a M « CO ^ to vo t» 00 Ov o H CI CO rf to ^ = g " *^ " H M " (S A REVISION OF THE ATOMIC WEIGHT OF MANGANESE. 43 ^ Atomic weight of man- ganese. c^ VO c^ ■* xn Ov CO « t^ H VO t^ •* VO CO PI CO CO fO ro p) M T^ CO PI CO PI W P< PI o PI O PI On Ov Ov On qv Ov Ov Ov Ov OV Ov ov Ov Ov qv Ov CO CO 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 4 Ov qv lO lO IT) lO lO in m in m in VO VO VO to VO VO 4 4 to to -So . H »0 O VO c« 00 CO VO H Ov 00 H O to Ov O VO "* O O ^ 8 S^ O t^ VO VO t^ Tt- M VO Tj- g VO CO CO C) vO CS P) Ov ov lO t^ Ov VO 00 VO CO H CO VO M T^ ov PI VO >o Ov Tt- VO VO 53 t>. 00 '^ Tj- m 00 Tt H C^ o 00 00 PI co PI O dv dv w 00 00 dv M H d >d dv vd r>. vd 00 vd dv o" CO 1— t "S^^ 'O M 00 »0 l>. Ov VO ■* o> M- VO 00 M p> o p) .'" 1 . M CO CO O Ov H (N P) O CO M lo r>. Ov VO J>. 'co n- t^ o o o VO ov 00 CO P» M o CO VO o> >, "1^ CO CO M 00 CO cs Ov Ov VO PI Ov O J>. t>. Ov "rt 1— 1 >o vq >o 00 00 vq in t>. in H qv rf vq l>. to vq ai O o ir, lO vo' 4 4 in vd in CO VO CO 4 CO 4 CO to bO .S (J "o Weight of AgBr from wash- waters. t^ in VO 00 in Tf M O O o o M o o o o § § § § g ^ CQ d d d d d d d S3 "WS m > < -2°o o> CO 00 M Ov VO M H CO CO PI Ov CO f>. VO O r^ t>. CO CO Tj- VO VD M P) H VO w VO ■>d- CO S-"^ 23 * o O O O o 8 § o O O O o VO III O O O O O O 8 8 8 8 8 O O O O g O O 8 H d d d d d d d d d d d d d d d d II °§ M O Ov Tf pj ov 00 CO ■<*■ VO p« to Tt o 00 o . ro CO cq ri oq O H H W o H M tH PI (H p) M 03.2 o o o O o o o o < pq m in 5 o o o o o Q o o o o o o o 3«2 o O o o o o o o o o o o o pq a d 6 6 6 d d d d d d d d d d d d • 5m i 'ij- irj VO Ov Ov w P) CO « w Ov t>. M 00 M o . w Ov '^ 0\ w r^ CO VO VO n- CO P) CO t^ ^ CO i,vo 01 O M VO O^ N H P) Ov ov lO t^ 00 to 00 VO CO M CO rt- H "* ov P) VO VO Ov Tt to VO HH ^=1 *^ 00 Tt rj- in 00 'i- H C^ q 00 00 P^ CO PI qv 1— 1 dv dv M CO 00 dv M d vd dv vd t4 vd 00 vd dv en o 00 00 H M H O 00 rl- Tj- P) o t^ o in O 00 o Ov P> o H t>. o O O O o O H O w o M O H o o H . o O O Q O O 5 o o o o o o o o o o o o II 1 o o 8 8 o o o o o o o o Q o o o o o o o o o o o o o 60 < «a"S d 6 6 6 d d d d d d d d d d d d + + i + 1 + ! + 1 1 i + 1 1 -a<" U O qj t^ M Q Ov ^■"S « CO CO H 00 CO c^ Ov Ov VO PI Ov O r^ t^ ov .s ^ vq wo 00 00 vq in t^ »o M qv "1" '■°. t^ to vq lO »o vd 4 4 in vd VO CO to CO 4 CO 4 CO to affl is o < < < < U o o < < pq pq pa o Q Q Q VO t^ 00 o> o M W CO Tf lO VO r^ 00 Ov o M 03 H H t-( M ?< W " M H r^ C) On O NO HI OnnO r^ t^NC) >o>o «o t>-Nd O On W CO COOO On r~- o>oo •^00 ■*oo §• Onoo 00 00 O On O c< M O NO »o r^ cj CsNO ^^•^'^'^On Tt CO CO CO CO •* CO 6 O O CO OO NO CO lo t^ H w 00 CJ On p NO w Onno r>- t^vo loO <0 t^NO § q o M 00 lO lO O O w O o o o o COOO o o o o "do I 1 I 4-+ I I S o Ql o d o o d d »0 On M On •^00 On 00 Onoo lO M Tj-00 S> Onoo 00 00 h On O M OCOlOTfT}-ON "t CO to CO CO "i- CO I-:) I-] H^ 1^ M h4 ;^ pqpqpqWWpqW M N CO ^ lONO tN. CJ < 0) o ^ " MO ^ .g o^ 5 ® t^oO t~» CO CO CO c^ cs CO CO On On O^ On On On On tJ- -4 Tt •* -4 Tt ■* lo >o vo >o vj >o >o H OO w P» On M O •«tND CO CO C) t^ •"^ ^Nooo r^ONM cow ft Tj-cococ^NO O CO C»VOO»01HOOP< 0> d 00 t^od t^ d 00 O On H CO COOO On t^ OnOO •*oo ■^oo ^ OnOO 00 00 O On O ftMC^ONO>Ot^C« ConO ^co»0'^'4-qN Tj- CO CO CO CO '^ CO 00 On t^OO M M e» On H O P» •^ On ■<* 8CS W H M M M O O Q Q O O I O O O O 6 O 5 d d d 6 6 6 6 o o o o o d d d d d M NO 00 t^ r^ o ■* CO H OnnO Th h no lONO »0 !>■ On M C?> rt- CO CO N lO O C» «0 O >0 w 00 M On d 00 t^OO ti d 00 M H O M 00 to »0 e»ooooooo I I 1++ I I »0 On M On ^00 On 00 OnOO Vt M Tj-00 ^ OnOO 00 00 m On O gc»fjONO»ot>.c< C»vq IOC0»O'^t1-0n ■^ CO CO CO CO •^ CO pqpqmwWPQW 00 On O H « CO •* Sf c IV. A REVISION OF THE ATOMIC WEIGHT OF BROMINE. THE SYNTHESIS OF SILVER BROMIDE AND THE RATIO OF SILVER BROMIDE TO SILVER CHLORIDE. By Gregory Paul Baxter. Proceedings of the American Academy of Arts and Sciences, 42, 201 (1906). Journal of the American Chemical Society, 28, 1322 (1906). Zeitschrift fUr anorganische Chemie, 50, 389 (1906). Chemical News, 94, 260, 261 (1906). Experimentelle Untersuchungen iiber Atomgewichte, page 768. T. W. Richards. 1909. Contributions from the Chemical Laboratory of Harvard College. A REVISION OF THE ATOMIC WEIGHT OF BROMINE. THE SYNTHESIS OF SILVER BROMIDE AND THE RATIO OF SILVER BROMIDE TO SILVER CHLORIDE. INTRODUCTION. In numerous investigations in this laboratory upon the atomic weights of certain metals, in which metallic bromides were first titrated against the purest silver, and then the precipitated silver bromide was collected and weighed, the relation between the silver used in the titrations and the silver bromide ob- tained has yielded data from which the atomic weight of bromine may be calcu- lated. Furthermore, in all these investigations, as a check upon the purity of the silver and bromine employed, silver bromide was synthesized directly from weighed quantities of silver and an excess of ammonium bromide or hydrobro- mic acid. Many of these results have already been collected and discussed by Richards,^ nevertheless they are cited in the following table together with a few more recent determinations. (See table on page 52.) From the first of these ratios the atomic weight of bromine, referred to silver 107.880, is found to be 79.919, and from the second 79.918. Very recently, in experiments in which silver iodide was heated first in a cur- rent of air and bromine until the iodine was completely displaced, and then in a current of chlorine to displace the bromine, the ratio of silver bromide to silver chloride was determined in six cases. From the results of these experi- ments the atomic weight of bromine was calculated to be 79.916,^ if the atomic weight of chlorine is assumed to be 35.457. These values for bromine are in close agreement with those of Stas.' In his experiments weighed quantities of pure silver and bromine were first titrated against each other, and then the precipitate of silver bromide was collected and weighed. Of the four results by the first method, one should be rejected ac- cording to his own statements, since the bromine was not thoroughly dried. The remaining three, 79.922, 79.924, and 79.923, give as an average 79.923. From the weight of silver bromide foiu: values were obtained, 79.913, 79.9XSi 79.918, and 79.920, with an average of 79.917. * Proc. Amer. Phil. Soc, 43, 119 (1904). * Baxter: Proc. Amer. Acad., 41, 82 (1905); Jour. Amer. Chem. Soc, 27, 884; Zeit. anor g. Chem., 46, 45. (See page III.) * (Euvres Completes, i, 603. 51 52 RESEARCHES UPON ATOMIC WEIGHTS. Marignac ^ also determined the ratio of silver to silver bromide, with some- what lower results, — 79.922, 79.904, and 79.915; average, 79.913. Scott,^ in his analyses of ammonium bromide, obtained six values for the same ratio, varying between 79.899 and 79.911, with an average of 79.906. One of his results is here rejected, since the silver used in this experiment was known to be impure. iNDrRECT Determinations. Ratio No. Bromide analyzed. No. of experiments. Analyst. Reference. Ag AgBr I BaBrz Last seven Richards Proc. Amer. Acad., 28, 28 57-444 2 SrBr2 Seven Richards Ibid., 30, 389 57-444 3 ZnBrz One Richards Ibid., 31, 178 57-445 4 NiBr2 Seven Cushman Ibid., 33, III 57-444 5 CoBr2 Last five Baxter Ibid., 33, 127 57-446 6 UBr4 Three Merigold Ibid., 37, 393 57-447 7 CsBr Three Archibald Ibid., 38, 466 57-444 8 FeBr2 Two Baxter Ibid., 39, 252 57-443 9 CdBra Eight Hines Jour. Amer. Chem. Soc, 28, 783 57-444 10 MnBr2 Thirteen Hines Ibid., 28, 1578 57-444 Average, weighted according to the number of determinations 57-4443 Direct Det erminations. II HBr Two Richards Proc. Amer. Acad., 28, 17, 18 57-445 12 NH4Br One Richards Ibid., 30, 380 57-446 13 HBr Two Richards Ibid., 31, 165 57-444 14 NHiBr One Cushman Ibid., 33, 106 57-445 I."; NH4Br One Baxter Ibid., 33, 122 57-444 16 NH^Br Two Baxter Ibid., 34, 353 57-447 17 NHiBr Three Baxter Ibid., 39, 250 57-444 18 NHiBr One Hines Jour. Amer. Chem. Soc, 28, 1565 57-443 Average, weighted accor ding to the nun iber of determinations 57-4447 Dumas ^ by heating silver bromide in chlorine found the values 80.02, 79.87, and 79.94. In computing the atomic weight of bromine from these data, great weight is always given to Stas's determinations, the value 79,918 being usually assumed as the most probable one for the constant in question. Certainly, as pointed out by Richards,^ the true value must lie between 79.91 and 79.92. Clarke calculates the value 79.912 as the weighted average of the different investiga- tions previous to Scott's.^ ^ (Euvres Completes, 1, 81. * Jour. Chem. Soc. Trans., 79, 147 (1901). » Ann. Chem. Pharm., 113, 20 (i860). * Proc. Amer. Phil. Soc, 43, 119 (1904). ^ A Recalculation of Atomic Weights, Smith. Misc. Coll., 1897. A REVISION OF THE ATOMIC WEIGHT OF BROMINE. 53 Considerable uncertainty exists as to the purity of the materials employed in much of the foregoing work. Richards and Wells ^ have already exhaustively investigated the various methods of preparing pure silver, and have found that while it is a comparatively simple matter to free this substance from metallic impurities, the absence of gaseous impurities is by no means so easy to secure. Oxygen may be eliminated best by fusion in an atmosphere of pure hydrogen gas,2 or by prolonged fusion in a vacuum, while a lime boat was found to be the most suitable support for the silver during fusion. In most of the experiments cited on page 52, one of the final steps in the purification of the silver was fusion of electrolytic crystals on lime, in many cases in a vacuum, but without especial care to prolong the fusion. Silver pre- pared in this way was found by Richards and Wells to contain traces of oxy- gen, derived from silver nitrate occluded by the electrolytic crystals. In cases 8, 9, 10, 17, and 18, however, the silver was fused in hydrogen. Richards and Wells showed also that Stas's silver contained at least o.oi per cent of impurity, since it yielded o.oi per cent less silver chloride than their purest silver.^ Scott's silver in three cases was merely heated, not fused, in hydrogen, and in two of the others was fused before a blowpipe on calcic phosphate. In one experiment only the metal was fused on lime. No details are given as to the purification of the silver used by Marignac. Bromine also may be freed from impurities only with some difficulty. Ex- perience in this laboratory has shown that chlorine may be eliminated most conveniently by distilling or precipitating the bromine from solution in a bro- mide. One such distillation is sufficient to remove chlorine completely only when the substance is initially comparatively pure. If, however, the process is repeated by converting a portion of the partially purified product into a bro- mide, and dissolving the remainder of the bromine in this comparatively pure bromide, the chlorine is eliminated so completely that further repetition of this process has no apparent effect.^ The removal of iodine may be easily effected by converting the bromine into hydrobromic acid or a soluble bromide, and boiling the solution with a small quantity of free bromine. Here again it is well to repeat the process several times, since the reaction betwteen free bro- mine and the iodine ion, like that between free chlorine and the bromine ion, is undoubtedly incomplete. The greater part of the results given on page 52 were obtained with bro- mine which had been purified with due observance of these precautions. Of the other investigators, Stas seems to have been the only one to use sufficient * Pub. Car. Inst. No. 28, 16; Jour. Amer. Chem. Soc, 27, 472; Zeit. anorg. Ghent., 47, 70. 2 Baxter: Proc. Amer. Acad., 39, 249 (1903); Zeit. anorg. Chem., 38, 232 (1904). ^ Loc. cit., page 62. * Attention has already been called to these points by Richards and Wells: Proc. Amer. Acad., 41, 440 (1906); Zeit. physikal. Chem., 56, 354. 54 RESEARCHES UPON ATOMIC WEIGHTS. pains to secure purity of the bromine. Stas removed iodine by shaking potas- sium bromide several times with free bromine and carbon disulphide, and in the course of the prolonged purification distilled the bromine twice from solu- tion in a bromide. Marignac's purification consisted solely in crystallization of barium bromate and Scott's in distillation of hydrobromic acid. Of the methods employed in these early determinations, that involving the analysis of metallic halides is least suited for the purpose, on accotmt of the danger of occlusion of metallic salts by the precipitated silver bromide. That such an error actually exists to a slight extent is shown by the fact that the average of the "indirect" determinations is slightly larger than the average of the "direct" determinations. Obviously, if silver bromide is precipitated by means of either ammonium bromide or hydrobromic acid, occluded ammonium salts or free acids can be easily expelled by fusion of the bromide. This pre- caution was observed in most of the determinations recorded on page 52, and is absolutely essential for the complete elimination of water from the salt. Stas and Marignac both fused the silver bromide in their syntheses, but this opera- tion was omitted by Scott, who dried the bromide at 180°. Scott's statement that the loss on fusion of silver bromide which had been dried at 180° was due to the presence of asbestos is contradicted by the experiments recorded later in this paper, in which the loss on fusion amounted to about o.oi per cent in the case of silver bromide which had been dried in a similar fashion and which was almost entirely free from asbestos. From this brief discussion of the more important errors which may have in- fluenced previous determinations of the atomic weight of bromine, it is evident that some uncertainty still exists as to the true value of this constant. In the hope of throwing new light upon the subject, experiments were carried out by two of the methods outlined above, with especial precautions to insure purity of materials and to eliminate known possible errors in the experimental methods. Both the methods chosen — S)nithesis of silver bromide from a weighed amount of silver, and conversion of silver bromide into silver chloride — have already been recently tested in this laboratory ,1 and have been found to be at least as satisfactory as any. PURIFICATION OF MATERIALS. BROMINE. In purifying bromine for this research, the principles set forth on page 53 of this paper were appHed; but in some cases the purifying processes were re- peated after the product was apparently pure, in order to make certain that further treatment had no effect. ^ Baxter: Proc. Amer. Acad., 40, 419; 41, 73; Jour. Amer. Chem. Soc, 26, 1577; 27, 876; Zeit. anorg. Chem., ^Z, T-A] 46,36; Richards and Wells: Pj<&. Car. /«rf., No. 28; /owr.^wer. Chem. Soc, 27, 459; Zeit. anorg. Chem., 47, 56. A REVISION OF THE ATOMIC WEIGHT OF BROMINE. 55 Sample I was first completely dissolved in calcic bromide which had been made from about one-third of the original material by means of lime and am- monia, and was then distilled from the solution. The product was covered with several times its volume of water, and was converted into hydrobromic acid by means of pure hydrogen sulphide which had been generated from fer- rous sulphide with dilute sulphuric acid, and which had been thoroughly washed with water. After filtration from the precipitated sulphur and bromide of sul- phur, the acid was boiled for some time, with occasional addition of small quan- tities of recrystallized potassium permanganate to eliminate the iodine. Finally the residual hydrobromic acid was heated with an equivalent amount of re- crystalHzed permanganate, and the bromine was condensed in a flask cooled with ice. Sample II was first converted into hydrobromic acid by means of red phospho- rus and water, and the hydrobromic acid was then distilled, after having been boiled with an excess of bromine. An equivalent amoimt of permanganate was added, and the bromine liberated was separated from the solution by distilla- tion. About one-fourth of the product was next transformed into calcic bro- mide by means of ammonia and lime which was free from chloride, and the re- maining three-fourths of the bromide were dissolved in the calcic bromide and distilled. Still a third distillation from a bromide was carried out by reducing the product of the second distillation with hydrogen sulphide and subsequently oxidizing the hydrobromic acid with the purest recrystallized potassium per- manganate, after boiling the acid with several small portions of permanganate to eliminate last traces of iodine. Sample III was obtained by preparing calcic bromide from a portion of Sample II and distilling the remainder of Sample II from solution in this bromide. In the case of Sample IV the processes of reduction to hydrobromic acid with hydrogen sulphide and oxidation of the hydrobromic acid with pure permanganate were four times repeated. After each reduction the hydro- bromic acid was boiled with free bromine to remove iodine. Sample V was three times reduced with hydrogen sulphide and oxidized with permanganate. One-fourth the product was converted into calcic bromide and the remainder was dissolved in this calcic bromide and distilled. Thus Sample I was twice distilled from a bromide; Sample II was treated three times in the same way; and Samples III, IV, and V four times. Shortly before use each sample was distilled and converted into ammonium bromide by slow addition to an excess of redistilled ammonium hydroxide. The solution was then boiled to expel the excess of ammonia. SILVER. Several different samples of silver were employed, many of which have already been used in atomic weight researches in this laboratory, and have S6 RESEARCHES UPON ATOMIC WEIGHTS. shown evidence of great purity. For details concerning the purification the papers referred to should be consulted. Sample A was employed in a determination of the atomic weight of iodine.' This specimen had been twice precipitated as chloride and once electrolyzed. Sample B was used in experiments upon the atomic weight of iodine ^ and of manganese.' It was precipitated once as chloride, electrolyzed once, and finally precipitated as metal with ammonium formate. Sample C also was employed in a determination of the atomic weight of man- ganese, and was purified by recrystallizing silver nitrate, 7 times from nitric acid and 5 times from aqueous solution. Finally the silver nitrate was reduced by means of ammonium formate. Sample D was prepared for the determination of the atomic weights of cad- mium ^ and manganese, by precipitation as chloride, precipitation with am- monium formate, and electrolysis. Sample E was first purified in part by precipitation as chloride, in part by precipitation with ammonium formate. The combined material was then subjected to two electrolyses. In all cases the electrolytic crystals were fused in a boat of the purest lime, contained in a porcelain tube, in a current of electrolytic hydrogen. After the buttons had been cleansed with dilute nitric acid and dried at 200°, they were cut into fragments of from 4 to 8 gm. either by means of a clean chisel and anvil or with a fine jeweller's saw. The latter method was employed in the case of samples D and E, because it proved easier completely to free the silver from surface contamination with iron by etching the fragments with nitric acid, than when a chisel was used. The cleansing process with nitric acid was repeated until the solution thus obtained, after precipitation with hydrochloric acid and evaporation, proved free from iron. That every trace of iron could be removed by this treatment was proved by testing for iron the evaporated filtrates from several of the analyses subsequently recorded in this paper. Negative results were obtained in all cases. After thorough washing with water and drying at 100°, the pieces of metal were heated to about 400° in a vacuum, and were preserved over solid potas- sium hydroxide in a desiccator. 1 Baxter: Proc. Amer. Acad., 40, 420 (1904); Jour. Amer. Chem. Soc, 26, 1578; Zeit. anorg. Chem., 43, 15. (See page 92.) 2 Baxter: Ibid., 41, 79 (1905); Joiir. Amer. Chem. Soc, 27, 881; Zeit. anorg. Chem., 46, 42. (See page 108.) ' Baxter and Hines: Jour. Amer. Chem. Soc, 28, 1560 (1906); Zeit. anorg. Chem., 51, 207. (See page 37.) * Baxter and Hines: Jour. Amer. Chem. Soc, 28, 772 (1906); Zeit. anorg. Chem. 4g, 417. (See page 6.) A REVISION OF THE ATOMIC WEIGHT OF BROMINE. 57 SYNTHESIS OF SILVER BROMIDE. The ratio of silver to silver bromide was determined as follows: Weighed quantities of silver were dissolved in the purest redistilled nitric acid diluted with an equal volume of water, in the dissolving flask described on page 12, Next, the acid solution of the silver was diluted with an equal volume of water, and was heated until free from nitrous acid and oxides of nitrogen. After still further dilution, the solution was added slowly with constant agitation to a dilute solution of an excess of ammonium bromide in a glass-stoppered precipi- tating flask, and the whole was violently shaken for some time to promote coag- ulation. By adding the silver solution to the bromide, occlusion of silver nitrate was almost wholly precluded. In some experiments the solutions were as dilute as twentieth normal, in others as concentrated as fourth normal. The final re-' suits seem to be independent of the concentration of the solutions. At the end of about 24 hours the flask with its contents was again shaken, and then it was allowed to stand until the supernatant liquid was perfectly clear. The pre- cipitate of silver bromide was collected upon a weighed Gooch crucible, after thorough washing by decantation with water, and was dried in an electric oven, first for several hours at 130°, finally for about 14 hours at 180°. Then it was cooled and weighed. The operations of precipitation and filtration were performed in a large cup- board lighted with red light, and if the flask was taken out of this cupboard it was enveloped in several thicknesses of black cloth. Even after the prolonged drying, traces of moisture were retained by the salt, and could be expelled only by fusion. This was done by transferring the bulk of the silver bromide, freed as completely as possible from asbestos, to a small porcelain crucible which was weighed with its cover. The silver bromide was then fused by heating the small crucible, contained in a large crucible to prevent direct contact with the flame of the burner. A temperature much above the fusing point of silver bromide was avoided so that volatilization of the salt could not take place. This treatment must have eliminated occluded ammon- ium salts as well as water. Finally, in order to convert any occluded silver ni- trate, metallic silver, or silver sub-bromide into silver bromide, the salt was again fused in a current of dry air containing bromine vapor. This treatment seldom produced any measurable effect either upon the weight or the appearance of the salt, which was perfectly transparent and of a light yellow color, even after the first fusion in air. A few shreds of asbestos displaced from the crucible, together with an occa- sional trace of silver bromide which escaped the crucible, were collected upon a tiny filter paper, which was then ignited in a porcelain crucible. Before weighing, the ash was either treated with a drop of nitric and hydrobromic acids and again heated, or else was heated for some minutes in a current of air and bromine. The filtrate and washings were evaporated to small bulk. The precipitating flask and all other glass vessels used in the analysis were rinsed with ammonia 58 RESEARCHES UPON ATOMIC WEIGHTS. u o « s.yg \< k. O O o > a Sq^ o 5 -S O M M M Ov o c> O^ vO fl CO On !>. lo lO lOl O 3 ■* Ti- Tj- "* '^ ^ •<1- t>. t^ On 00 M M M M On On On On On On On ON t^ t-« O. t^ On M lO 00 Tj- >0 Tj- ■* ■*■<}• Tl- •* ■«1- •* ^ Tt- f. t^ t^ J^ lO VO lO VO O NO M C< iONO M M W C» M M On On On On On On On On On On On On t>. t^ t>. t^ r^ r^ •* -"l- •* •* Tt PO CO e< On VD On CO CO tJ CO CO On O 6 M cooo O Co M t^ cooo 00 00 6 On M- t^ t>. CO On P) CO OnOO NO . r^ t^ lO c» ••t •* lo to to >o ■* Tt T^ rh •. l^ r^ lO lO lO lO t^NO t-( M On On O CN| lO VO P< NO C< On cq li^ COOO On Tf to On On P) lO COOO I^ CO W M o o o o o o o o o o o 6 6 6 6 6 t^ CN< On M t^ O VO cs OnnO >* VO 1 I 60 M < < ■3 O ID C.9 2 m 5 a r^ CO O •* ■* O 000 000 000 odd On Tl- O NO IN t^O CO t^ « CO VO r-* VO VO TtOO VO « cs p) l-l M M P) H M P» 1 U C) (^ § 8 8 88 g Q Q 8 P« t>. On O NO 00 On O CO PO t>. O M COOO O P« t>. COOO VO M 00 NO CO CO P) OnOO VO P< NO H 00 PO 00 00 O On M-OO 00 w On VO H M C» l^ On CONO 't P« t^ PJ !>. Tj- On t>. CO O r^ VO O >o Tf PI t^ M M 00 ti d CO w M d pooo r^ 00 p< NO POOO 't On •* VO On On PJ VO cooo tt,>B PO VOOO P» VO PI M On . 00 t^OO On SH M NO PI g, t^ q OnnO ^ 10 VO 10 M 00 IH M CO M NO M r^ CO r^ O w q 00 00 VO ■* On VO VO VO O NO t^ w PI NO 00 On CI M rt t^ M NO t^NO M O 00 CO M 00 On ^ CO P< rj- lo tivd NO NO 00 H 00 t^NO CO r^ PI NO 00 w 00 M q •<*• OnOO d >>> >>>>>> MhHM So^ <:nNq tJ- VO NO O t^ o o M CO On t^ VO On 06 00 I (— t o MH a a 2 Xi a a A REVISION OF THE ATOMIC WEIGHT OF BROMINE. 59 and the rinsings were added to the evaporated filtrate and wash-waters. The whole was then tested in a nephelometer for silver and the quantity found was estimated by comparison with standard silver solutions. In most cases the correction thus obtained was less than o.i mg. The asbestos which formed the felt in the Gooch crucible, after having been shredded, was digested for some hours with aqua regia and was then thor- oughly washed with water. Before the empty crucible was weighed, the felt was ignited with a Bunsen burner. Crucibles thus treated and then, after being moistened with water, again heated to iSo**, did not change in weight. In the table on page 58 are cited all the analyses which were completed with- out accident. Vacuum corrections of —0.000031 for every apparent gram of silver and of 4-0.000041 for every apparent gram of silver bromide are applied.^ The platinum-plated brass weights were standardized from time to time and were found to retain their original values within a very few hundredths of a milligram in all cases. CONVERSION OF SILVER BROMIDE INTO SILVER CHLORIDE. The ratio of silver bromide to silver chloride was determined much as de- scribed in previous papers upon the atomic weight of iodine.^ Pure silver bromide was prepared by precipitation of silver nitrate with an excess of ammonium bromide. The silver employed was purified either by precipitation as chloride and reduction with invert sugar, or by electrolysis, or by precipita- tion with ammonium formate. The metal was then fused before a blowpipe upon a crucible of the purest lime, and the buttons were thoroughly cleansed with nitric acid. No further purification was considered necessary since the weight of the metal was of no consequence. After the silver bromide had been washed by decantation with water, in some cases it was collected in a Gooch crucible in which a disk of filter paper was employed instead of asbestos, and after drying at 100° it was carefully sepa- rated from the filter paper. In other cases the precipitate was transferred to a platinum dish, and was drained with a platinum reverse filter ^ with a disk of filter paper. In still others a platinum Gooch crucible with small holes was found to belsufl&ciently effective as a filtering medium without the use of either asbestos or filter paper. Before being weighed the silver bromide was fused in a current of air saturated with bromine in a weighed quartz crucible. The air was purified by passing successively over beads moistened with silver nitrate solution, over sodium carbonate, and finally over concentrated sulphuric acid which had been heated to its boiling-point with a small quantity of recrystallized potassium dichromate to eliminate volatile and oxidizable impurities, The air was then passed * See page 41. ^ Baxter: Proc. Amer. Acad., 40, 432 (1904); 41, 75 (1905); Jour. Amer. diem. Soc, 26, 1590; 37, 876; Zeit. anorg. Chem.,43, 27; 46, 36. (See page 102.) ' Cooke: Proc. Amer. Acad., 12, 121 (1876). 6o RESEARCHES UPON ATOMIC WEIGHTS. through dry bromine in a small bulb. This apparatus was constructed entirely of glass with ground joints. The tube which conducted the gases into the cru- cible passed through a Rose crucible-cover of glazed porcelain in all experiments except analyses 28 to 31, in which a quartz cover was employed. The quartz crucibles were always contained in large porcelain crucibles while being heated. They remained almost absolutely constant in weight during the experiments. The bromine was in each case a portion of the sample from which the silver bromide had been made. Next the bromide was heated barely to fusion in a slow current of chlorine, generated by the action of hydrochloric acid upon manganese dioxide, and dried by means of concentrated sulphuric acid. The apparatus for this pur- pose also was constructed wholly of glass. When the bromine was apparently completely displaced, the silver chloride was heated in the air for a few min- utes to expel dissolved chlorine, and then was cooled and weighed. A repetition of the heating in chlorine seldom affected the weight of the salt more than a few hundredths of a milligram, although occasionally a third heating was necessary to effect this result. That no loss of silver chloride by volatiHzation took place is certain for two reasons. In the first place the cover of the crucible and the delivery- tube for the bromine, when rinsed with ammonia and the solution treated with a sUght excess of hydrochloric acid, gave no visible opalescence in the nephelometer. In the second place the weight of the chloride became constant without diffi- culty. It has already been shown that silver chloride which has been fused in chlorine, if subsequently heated in air, retains no excess of chlorine.^ The following vacuum corrections were applied: silver bromide, -f- 0.000041 ; silver chloride + 0.000071.^ (See table on page 61.) RESULTS AND DISCUSSION. Aside from the close agreement of all the results of Series I, the fact is to be emphasized that of the last seven analyses, which were consecutive, only two differ from the average of the series, 79.916, by as much as o.ooi unit. Fur- thermore, there is no evidence of any dissimilarity in the different preparations of bromine. Material which has received only two distillations from a bromide gives values no lower than bromine which has been thus treated four times. The various specimens of silver also show no difference in purity. In the case of Series II, the extreme variation of the results is only 0.004 unit, and only one of the 13 experiments yielded a value which differs from the aver- age by more than o.ooi unit. Finally, the difference between the averages of Series I and II is only 0.0007 unit. It is extremely unlikely that constant errors could have affected both ^ Baxter: Proc.^wer. ^ca^/., 40, 432 (1904); Jour. Amer. Chem. Soc. ,26, isg^; Zeit.anorg. Chem., 43, 29. (See page 103.) ^ See page 41. A REVISION OF THE ATOMIC WEIGHT OF BROMINE. 6i series equally, so that this striking agreement is strong proof that both series are free from such errors. The Atomic Weight of Bromine. Series II. AgBr:AgCl. Ag = 107.880 CI = 35-457- No. of analysis. Sample of bromine. Weight of silver bromide in vacuum. Weight of silver chloride in vacuum. ^■"»!ff Atoniic weight of bromine. 19 20 21 22 23 24 25 26 27 28 29 30 31 II II II IV IV V V V V I I III III Total gm. 8.03979 8.57738 13.15698 12.71403 13.96784 13.0S168 12.52604 II.I1984 8.82272 II.93192 12.53547 17.15021 10.31852 153.94242 gm. 6.13642 6.54677 10.04221 Average 9.70413 10.66116 i\verage 9.98469 9-56059 8.48733 6.73402 Average 9.10721 9.56767 Average 13.09009 7.87572 Average 131.0176 131.0170 131.0168 79.916 79-915 79-915 131.0171 79.915 131.0167 131.0162 79.915 79.914 131.0164 79-915 131.0174 131.017s 131.0170 131.0172 79.916 79.916 79-915 79.916 131.0173 79.916 131.0162 131.0190 131.0176 79.914 79.918 79.916 131.0167 131.0168 79-915 79-915 131.0168 79-915 117.49801 131.0170 79-915 Average of all 13 Averacre of .Spripfs sxperiments 131.0171 79-915 79.916 I and II It has already been pointed out that the average of Stas's syntheses, 79.917, probably represents with considerable accuracy the atomic weight of bromine, and that certainly his determinations are more accurate than those of later experimenters. His syntheses are few in number, however, and differ among themselves by several thousandths of a unit, so that they do not define within this amount the constant in question. Their average, however, confirms the value obtained in this paper. From all the experiments here described the number 79.916 seems to be the most probable value for the atomic weight of bromine, if silver has the atomic weight 107.880. If silver is taken at 107.870, bromine becomes 79.909. In conclusion, attention may be called to the fact that a diminution in the atomic weight of bromine referred to silver raises slightly all atomic weights resulting from the analysis of metallic bromides by precipitation with silver, I am deeply indebted to the Cyrus M. Warren Fund for Research in Harvard University for assistance in pursuing this investigation. V. A REVISION OF THE ATOMIC WEIGHT OF LEAD. 777^ ANALYSIS OF LEAD CHLORIDE. By Gregory Paul Baxter and John Hunt Wilson. Proceedings of the American Academy of Arts and Sciences, 43, 365 (1907). Journal of the American Chemical Society, 30, 187 (1908). Zeitschrift fiir anorganische Chemie, 57, 174 (1908). Chemical News, 98, 64, 78 (1908). Contributions from the Chemical Laboratory of Harvard College. A REVISION OF THE ATOMIC WEIGHT OF LEAD. THE ANALYSIS OF LEAD CHLORIDE. INTRODUCTION. Although lead is one of the most common elements, its atomic weight has te- ceived comparatively little attention, the value at present accepted being based almost wholly upon the work of Stas.^ Of the earlier determinations of this constant those of Dobereiner ^ and Longchamps ^ can hardly be considered as possessing other than historic interest. The first results which can lay claim to accuracy are those of Berzelius,^ who obtained values ranging from 206.7 to 207.3 by reduction of litharge in a current of hydrogen. Berzelius also S3Tithe- sized the sulphate from metallic lead with the result 207.0.^ Shortly after, Turner ® criticized the first method employed by Berzelius and attributed the irregularity of his results to the action of lead oxide on the silicious matter of the tube at the temperature employed in the reduction. By the conversion of both the metal and the oxide into sulphate Turner in a painstaking research deduced the values 207.0 and 207.6 respectively, and by converting the nitrate into sulphate, 204.2. Marignac ^ converted metallic lead into the chloride by heating in a stream of chlorine and obtained the result 207.42. Both Marig- nac ^ and Dumas ^ analyzed lead chloride. Marignac, who dried the salt at 200°, by titration against silver found the atomic weight of lead to be 206.81, and from the ratio of lead chloride to silver chloride, 206.85. Dumas subse- quently showed that lead chloride, even when dried at 250°, retains moisture and is somewhat basic, and in one analysis, in which corrections are applied for these errors, found a somewhat higher value, 207.07, as was to be expected. Chloride analyses by early investigators are, however, to be universally dis- trusted, owing to neglect of the very considerable solubility of silver chloride, thus producing too low results. Stas's work upon the syntheses of lead nitrate and sulphate from the metal is undoubtedly the most accurate contribution upon the subject,^" although a 1 Earlier work on the atomic weight of lead has been carefully summarized by Clarke. Smithsonian Miscellaneous Collections, Constants of Nature, "A Recalculation of the Atomic Weights," 1910. The earlier results have been recalculated on the basis of the following atomic weights: = 16.00; N = 14.01; S = 32.07; CI = 35.46; Ag = 107.88. * Schweig. Jour., 17, 241 (1816). ' Ann. Chim. Phys., 34, 105 (1827). * Pogg. Ann., 19, 314 (183c). ^ Lehrbuch, sth ed., 3, 1187 (1845). ^ Phil. Trans., 527 (1833). ' Lieh. Ann., 59, 289 (1846). ^ Jour. Prakt. Chem., 74, ziS (1858). ^ Lieb. Ann., 113, 35 (i860). ^^ (Euvres Completes, i, 383. 6s 66 RESEARCHES UPON ATOMIC WEIGHTS. careful consideration of his work discloses minor defects, many of which he recognizes himself. The metallic lead used in the syntheses was finally fused under potassium cyanide. Whether or not this treatment introduced impuri- ties into the metal is uncertain. Stas himself suspected the presence of alkali metals. Since the nitrate could not be dried above 150° without decom- position, it undoubtedly contained moisture, and Stas calls attention to this point. The sulphate was made by treatment of lead nitrate, resulting from the nitrate syntheses, with sulphuric acid. The sulphate was dried finally at dull redness, and was probably free, or nearly free, from moisture, although it may have contained traces of lead oxide resulting from occluded nitrate, as well as sulphuric acid. Most of these probable errors tend to lower the observed atomic weight, so that Stas's value from the series of nitrate syntheses, 206.81, and that from the sulphate series, 206.92, are to be regarded as minimum values. The reader of Stas's own account of his work upon lead can not fail to be impressed with the fact that he was somewhat dissatisfied with the outcome of his research. Mention should also be made of the work of Anderson and Svanberg ^ on the conversion of lead nitrate into oxide, although the method was primarily employed in an endeavor to fix the atomic weight of nitrogen. Their results yield the value 207.37. The discrepancies between the results of these various experiments only serve to emphasize the need of a redetermination of the value in question, and it was with this object in view that the work embodied in this paper was undertaken. The search for a suitable method for determining the atomic weight of lead failed to reveal any more promising line of attack than those already employed for the purpose. With an element of so high an atomic weight as lead, in any method involving the change of one of its compounds into another, errors which may be insignificant with elements of small atomic weight are magnified in the calculations to undesirable proportions. Furthermore, during the following investigation, reduction of the chloride and oxide in hydrogen was investigated far enough to show that complete reduction of either compound was extremely difficult, if not impossible, without loss of material from the containing vessel by sublimation, aside from the fact that all available material for containing ves- sels is acted upon by either the fused salt or the reduced metal. The elimination of moisture from lead nitrate or lead sulphate without decomposition of the salts seemed likely to prove a stimibling-block in the use of these substances. Finally, in spite of the slight solubility of lead chloride, the determination of the chlorine in this salt by precipitation with silver nitrate was chosen as pre- senting fewest difficulties. In the first place, the determination of a halogen can be effected with great accuracy. In the second place, the elimination of mois- ture from lead chloride is an easy matter, since the salt may be fused in a plat- inum vessel in a current of hydrochloric-acid gas without attacking the platinum * Ann. Chim. Phys. (3), 9, 254 (1843). A REVISION OF THE ATOMIC WEIGHT OF LEAD. 67 in the least and without the production of basic salts. In the third place, silver chloride, which has been precipitated from a dilute solution of lead chloride by means of silver nitrate, was shown experimentally not to contain an amount of occluded lead salt large enough to be detected. PURIFICATION OF MATERIALS, Water, hydrochloric acid, and nitric acid were carefully purified by distilla- tion as described in the preceding papers. In the preparation of pure silver also the usual methods were employed, one precipitation as chloride, one as metal by ammonium formate, and one by electrolysis being followed by the final fusion in hydrogen on a boat of pure lime. LEAD CHLORIDE. Three samples of lead chloride from two entirely different sources were em- ployed. Sample A was prepared from metallic lead. Commercial lead was dis- solved in dilute nitric acid, and the solution, after filtration, was precipitated with a slight excess of sulphuric acid. The lead sulphate was thoroughly washed, suspended in water, and hydrogen sulphide was passed in until the sulphate was almost completely converted into sulphide. Next the sulphide was washed with water, dissolved in hot dilute nitric acid, and the solution was freed from sulphur and unchanged sulphate by filtration. The lead nitrate thus obtained was crystallized twice, dissolved in water, and precipitated in glass vessels with a slight excess of hydrochloric acid. The chloride was washed several times with cold water and then crystallized from hot water eight times, the last five crys- tallizations being carried out wholly in platinum, with centrifugal drainage after each crystallization. In crystallizing the lead chloride the whole sample was not dissolved at one time, but the same mother-liquor was used for dissolv- ing several portions of the original salt. Needless to say, the chloride was not exposed to contact with the products of combustion of illuminating gas, lest lead sulphate be formed. Sample B was prepared from commercial lead nitrate. This salt was dissolved and crystallized from dilute nitric acid once in glass and six times in platinum vessels, with centrifugal drainage. Hydrochloric acid was then distilled into a large quartz dish, and the solution of the nitrate was slowly added with constant stirring with a quartz rod. The chloride was freed from aqua regia as far as possible by washing with cold water, and was once crystallized from aqueous solution in quartz dishes to remove last traces of aqua regia. Finally the salt was crystallized three times in platinum. It could reasonably be expected that both of these samples were of a high degree of purity; nevertheless, upon heating the salt in an atmosphere of hy- drochloric acid, the salt itself turned somewhat dark, and upon solution of the fused salt in water a slight dark residue remained. Although in a few preliminary 68 RESEARCHES UPON ATOMIC WEIGHTS. experiments attempts were made to determine this residue oy miration anu ignition, it was subsequently found that even a small filter paper adsorbs ap- preciable amounts of lead compounds from a solution of the chloride, which can not be removed by washing with water. From 0.03 mg. to 0.13 mg. of residue were obtained in several blank experiments, by ignition of filters through which 0.5 per cent solutions of lead chloride had been passed, with subsequent very thorough washing. In order to avoid the uncertainty of this correc- tion, further attempts were made to obtain a sample of the salt which would give a perfectly clear solution in water after fusion, and thus render filtra- tion unnecessary. With this end in view a considerable quantity of Sample A was fused in a large platinum boat in a current of hydrochloric acid. The fused salt was powdered in an agate mortar, dissolved in water in a platinvma vessel, and the solution was freed from the residue by filtration through a tiny filter in a platinum funnel into a platinum dish, where it was allowed to crystallize. This sample was then twice recrystallized with centrifugal drainage. Not- withstanding the drastic treatment to which it had been subjected, when a portion of this material was fused in hydrochloric acid, the same darkening as before was observed, and the same residue was obtained. The suspicion that the difiiculty was due to dissolving of the filter paper by the solution of the salt ^ led to a second more successful attempt by crystallization from hydro- chloric acid solution in platinum vessels. In this way it was found possible to prepare salt which showed no tendency to darken upon heating, and which, after fusion, left absolutely no residue upon solution in water. Portions of Samples A and B were thus recrystallized three times more. Since these two specimens of material gave identical results, for two final experiments, portions from each of these samples were mixed and then subjected to three additional crystalHzations. This last sample was designated Sample C. DRYING OF LEAD CHLORIDE AND METHOD OF ANALYSIS. The method of analysis did not differ materially from that used in the analy- sis of cadmium and manganese chlorides (pages 7 and 45). The lead chloride, contained in a weighed platinum boat, was first fused in a current of hydro- chloric-acid gas, and the boat was transferred to its weighing-bottle and weighed. On account of the small solubility of lead chloride it was a somewhat troublesome matter to dissolve the fused material. This was done in most of the analyses by prolonged contact with water nearly at the boiling point in a Jena glass flask. In the last two analyses, in order to show that no error was introduced through the solubility of the glass, the solution was prepared in a large platinum retort and was transferred to the precipitating flask only when cold. ^ Mr, P. B. Goode in this laboratory has recently found a similar difficulty with the chlo- rides of the alkaline earths. A REVISION OF THE ATOMIC WEIGHT OF LEAD. 69 When the lead chloride was dissolved, a dilute solution of a very neariy equiv- alent amount of pure silver was added, and, after standing, the amounts of chloride and silver were carefully adjusted until exactly equivalent by means of the nephelometer. When the exact end-point had been found, about 0.2 gm. of silver nitrate in excess was added to precipitate dissolved silver chloride, and the precipitate was collected and determined upon a Gooch crucible. Dis- solved silver chloride in the filtrate and wash-waters was estimated by com- parison with standard solutions in the nephelometer. Asbestos displaced from the Gooch crucible was collected, and the moisture retained by the dried pre- cipitate was found by loss of weight during fusion. In order to find out whether lead or silver nitrates were appreciably adsorbed by the filter paper, a solution containing lead nitrate, silver nitrate, and nitric acid of the concentration of these filtrates, was passed through several small filter papers, which were then very carefully washed. In four cases, after in- cineration of the papers, there was found, — o.ooooi, -{- 0.00002, + 0.00003, -f- 0.00001 gm. of residue, exclusive of ash. This correction is so small that it is neglected in the calculations. In all the analyses the platinum boat behaved admirably, the loss in weight never amounting to more than a few hundredths of a milligram. The balance used was a short-arm Troemner, easily sensitive to 0.02 mg. The gold-plated brass weights were carefully standardized to hundredths of a milligram. All the weighings were made by substitution with tare vessels as nearly like those to be weighed as possible. Vacuum corrections. The values of the density of lead chloride as given by various observers range from 5.78 to 5.805,^ the mean of the more accurate de- terminations being 5.80. This gives rise to a vacuum correction of -j- 0.000062 for each apparent gram of lead chloride, the density of the weights being as- sumed to be 8.3 . The other vacuum corrections applied were for silver chloride, -f 0.000071, and for silver, —0.000031. All analyses which were carried to a successful completion are recorded in the tables on page 70. RESULTS AND DISCUSSION. The close agreement of the averages of the two series is strong evidence that no constant error, such as occlusion, affects the results. In all, 19.55663 gm. of silver produced 25.98401 gm. of silver chloride, whence the ratio of silver chloride to silver is 132.865, a value in close agreement with the result 132.867 obtained by Richards and Wells. Furthermore, the different samples, A, B, and C, all give essentially identical results. It appears, then, that if the atomic weight of silver is taken as 107.880, the atomic weight of lead is 207.09, nearly 0.2 unit higher than the value now in * Landolt-Bomstein-Meyerhoffer, Tabellen. 70 RESEARCHES UPON ATOMIC WEIGHTS. use. If the atomic weight of silver is 107.87, lead becomes 207.08, a number still much higher than that depending upon Stas's syntheses, as is to be ex- pected. The Atomic Weight of Lead. Series I. PbClj: aAg. Ag = 107.880 CI = 35-457 No. of analysis. Sample of PbCl,. Weight of PbClz in vacuum. gm. 4.67691 3-67705 4.14110 4-56988 5.12287 3-85844 4.67244 3-10317 4.29613 Weight of Ag in vacuum. gm. 3.63061 2-85375 3-21388 3-54672 3-97596 2.99456 3.62628 2.40837 3-33427 Weight of Ag added or subtracted. gm. — 0.00074 0.00000 +0.00020 0.00000 — 0.00028 0.00000 0.00000 0.00000 —0.00020 Corrected weight of Ag. gm. 3.62987 2-85375 3.21408 3-54672 3-97568 2.99456 3.62628 2.40837 3-33407 Atomic weight of Pb. 207.083 207.093 207.077 207.089 207.105 207.090 207.093 207.092 207.106 Average 207.092 Series II. PbCl2: 2AgCl. No. Sample of fof analysis . PbClj. 10 A II A 12 B 13 B 14 C IS C Weight of PbCl, in vacuum. gm. 4.67691 4.14110 5.12287 3-85844 3-10317 4.29613 Weight of AgCl in vacuum. gm. 4.82148 4.26848 5.28116 3-97759 3-19751 4.42730 Loss on fusion. gm. O.OOIOO 0.00020 0.00054 0.00035 0.00045 0.00020 Weight of asbestos. gm. 0.00021 0.00008 0.00013 0.00033 0.00014 0.00004 Wt. AgCl from wash- waters. gm. 0.00204 0.00180 0.00197 0.00192 0.00189 0.00268 Corrected weight of AgCl. gm. 4-82273 4.27016 5.28272 3-97949 3.19909 4.42982 Atomic weight of Pb. 207.092 207.096 207.085 207.040 207.165 207.108 Average 207.097 Average, rejecting the least satisfactory analyses, 13 and 14 207.095 Average of Series I and II 207.094 Vl. A REVISION OF THE ATOMIC WEIGHT OF ARSENIC. THE ANALYSIS OF SILVER ARSENATE. By Gregory Paul Baxter and Fletcher Barker Coffin. Proceedings of the American Academy of Arts and Sciences, 44, 179 (1909). Journal of the American Chemical Society, 31, 297 (1909). Zeitschrift fiir anorganische Chemie, 62, 50 (1909). Contributions from the Chemical Laboratory of Harvard College. A REVISION OF THE ATOMIC WEIGHT OF ARSENIC THE ANALYSIS OF SILVER ARSENATE. INTRODUCTION. Below is a summary of the previous work upon the atomic weight of arsenic/ the results obtained by the several investigators having been recalculated with the use of the most recent ^ atomic weights; Date. Investigator. Reference. Ratio determined. Result. i8i6 Thomson Schweigger's Jour., 17, 421 2 As: AsjOb 76.3s 1818 Berzelius Pogg. Ann., 8, i 2AS2O3: 3S02 75-03 1845 Pelouze Compt. Rend., 20, 1047 AsClaisAg 74.93 I8SS Kessler Pogg. Ann., 95, 204 3AS2O3: 2K2Cr207 3AS2O3: 2KCIO8 74-95 75-23 1859 Dumas Ann. Chim. Phys. (3), 55, 174 AsCl3:3Ag 74-87 185Q Wallace Phil. Mag. (4), 18, 279 AsBr3:3Ag 74.20 1861 Kessler Pogg. Ann., 113, 140 3AS2O3: 2K2Cr207 75 -oi 1896 Hibbs Doctoral Thesis, Univ. of Penn. Na4As207:4NaCl 74-88 1902 Ebaugh Jour. Amer. Chem. Soc ., 24, 489 Ag3As04:3AgCl Ag3As04:3Ag Pb3(As04)2:3PbCl2 Pb3(As04)2:3PbBr2 75.02 74.92 75-06 74-88 A glance at this rather discordant series of results shows the necessity for a redetermination of the atomic weight of arsenic. Even in the more recent in- vestigations of Hibbs and Ebaugh there exists an extreme variation of nearly 0.2 imit in the averages of the five series. In this research silver arsenate was chosen for analysis, first, because the compound is unchanged by moderate heating, and hence may be dried at a temperature high enough to expel all but a very small amount of moisture. In the second place, silver compounds may be analyzed with great ease as well as accuracy by precipitation of the silver as silver halogen compounds. Fur- thermore, preliminary experiments confirmed the statement by Ebaugh that * Clarke: A recalculation of the atomic weights, Smith. Misc. Coll., Constants of Nature, Part V (1910). For an excellent critical discussion by Brauner of previous work, see Abegg's Eandhuch der anorganischen Chemie, 3, (2), 491 (1907). * O = 16.00; Na = 22.98; S = 32.07; CI = 35.46; K = 39.10; Cr = 52.01; Br =■ 79.92; Ag = 107.88; Pb = 207.09. 73 74 RESEARCHES UPON ATOMIC WEIGHTS. it is possible completely to convert the arsenate into chloride by heating in a current of hydrochloric-acid gas. Such a process has the advantage that no transfer of material is involved. THE PREPARATION OF TRISILVER ARSENATE. All the samples of silver arsenate were prepared by adding to a j&f teenth nor- mal solution of silver nitrate a solution of similar concentration of an equiva- lent amount of an arsenate of sodium or ammonium, the differences between the different samples consisting chiefly in the nature of the soluble arsenate em- ployed. Precipitation was carried out in a room lighted only with ruby light. After the silver arsenate had been washed by decantation many times with pure water, it was dried in a preliminary way by centrifugal settling in platinum crucibles, and then by being heated in an electric oven at about 130° C. The salt was powdered in an agate mortar before the final heating in a quartz tube or platinum boat, as explained later. It was shown by tests with diphenyl- amine that the arsenate could be washed free from nitrates. Although one of the hydrogens of arsenic acid resembles the hydrogen of strong acids in its dissociating tendency, the other two hydrogens are those of weak acids.^ Hence perceptible hydrolysis takes place in solutions of salts of this acid, even when the base is strong, that of the tertiary salts being of course greatest in extent. It is not an easy matter to predict the effect of this hydrolysis upon the composition of a precipitate of silver arsenate; for while the Phase Rule allows the existence of only one solid in equilibrium with the arsenate solution except at certain fixed concentrations, the possibility of the occlusion of either basic or acid arsenates by the silver arsenate still exists. Experiments only are able to throw light on this point. Accordingly arsenate solutions of different conditions of acidity and alkalinity were used in the precipitations, and the compositions of the different precipitates were compared. Sample A. Commercial C. P. disodium arsenate was recrystallized four times, all but the first crystallization being conducted in platinum vessels. The mother-liquor from the fourth crystallization, after the removal of the arsenic by hydrogen sulphide, gave no test for phosphate. The calculated amount of redistilled ammonia to make disodium ammonium arsenate was added to a solution of the purified salt before the precipitation of the silver arsenate. Dur- ing this precipitation the mother-liquor remains essentially neutral. Sample B. This sample was made from disodium arsenate which had been recrystallized five times in platinum vessels. Silver arsenate was precipitated * Washburn calculates from Walden's conductivity measurements the constant for the first hydrogen of arsenic acid to be 4.8X 10-'. Jour. Amer. Chem. Soc, 30, 35 (1908). The con- stants for the second and third hydrogens are probably lower than those of phosphoric acid, 2.0 X 10-^ and 3.6 X 10-". Abbott and Bray: Ibid., 31, 755 (1909). A REVISION OF THE ATOMIC WEIGHT OF ARSENIC. 75 with a solution of this salt without the addition of ammonia. Here the mother- liquor becomes more and more acid as precipitation proceeds. Sample C. Commercial C. P. arsenic trioxide was recrystallized three times from dilute hydrochloric-acid solution, and, after being rinsed with water and centrifugally drained, it was converted into arsenic acid by means of nitric and hydrochloric acids in a porcelain dish. The hydrochloric and nitric acids were expelled by evaporation nearly to dryness, and the residue was twice evaporated to dryness with nitric acid in a platinum dish. After the residue had been dissolved in water, the solution was allowed to stand for some time in order to allow pyro- and meta-arsenic acids to be converted as completely as possible into ortho-arsenic acid. Then sodium carbonate, which had been twice crystallized in platinum, was added to the solution in amount sufficient to form disodium arsenate, and the product was crystallized four times in platinum vessels. The precipitation of silver arsenate by adding a solution of this salt to a solution of silver nitrate resembles the preparation of Sample B. Sample D. A portion of the arsenic acid made for the preparation of Sample C was converted into ammonium dihydrogen arsenate by adding the calculated amount of redistilled ammonia, and the salt was recrystallized five times in platinum, A sufficient quantity of ammonia to form tri- ammonium arsenate was added to a solution of this salt before the precipi- tation of the silver arsenate. One specimen of silver arsenate made in this way was discarded, since its composition was very irregular. Sample E. To a portion of the arsenic acid used for Sample C recrystallized sodium carbonate was added in amount sufficient to form disodium arsenate. After the solution had been evaporated to dryness, the salt was recrystallized four times in platinum. Enough ammonia to form disodium ammonium arse- nate was added to a solution of this salt before the precipitation of the silver arsenate. This material resembles Sample A. Sample F. A portion of the disodium arsenate prepared for Sample B was converted into trisodium arsenate by means of recrystallized sodium carbonate, and the trisodium arsenate was recrystallized six times in platinum vessels. Sample G. Arsenic trioxide was twice resublimed in a current of pure dry air and then once crystallized from dilute hydrochloric-acid solution. Next the arsenious acid was oxidized to arsenic acid exactly as described under Sample C. Finally the arsenic acid was converted into trisodium arsenate by means of pure sodium carbonate, and the salt was crystallized four times in platinum. Samples F and G are evidently very similar. In all the foregoing crystallizations the crystals were thoroughly drained in a centrifugal machine employing large platinum Gooch crucibles as baskets,^ and each crop of crystals was once rinsed with a small quantity of pure water and subsequently drained in the centrifugal machine. ' Baxter: Jour. Amer. Chem. Soc, 30, 286 (1908). 76 RESEARCHES UPON ATOMIC WEIGHTS. PURIFICATION OF OTHER MATERIALS. SILVER NITRATE. The silver nitrate used in the preparation of the different samples of silver arsenate was recrystallized several times in platinum vessels, with centrifugal drainage, until the mother-liquor gave no opalescence upon dilution when tested in the nephelometer. HYDROBROMIC ACID. One quarter pound of commercial bromine was converted into potassium bromide by addition to recrystallized potassium oxalate. In the concentrated solution of this bromide, in a distilling flask cooled with ice, 3 pounds of bromine were dissolved in several separate portions, each portion being distilled from the solution into a flask cooled with ice before the addition of the next succeeding portion. A portion of the purified bromine was then converted into potassium bromide with pure potassium oxalate as before, and the re- mainder of the bromine was distilled in small portions from solution in this pure potassium bromide. The product obtained was thus twice distilled from a bromide, the bromide in the second distillation being essentially free from chlorine. This treatment has already been proved sufficient to free bromine from chlorine.^ Hydrobromic acid was synthesized from the pure bromine by bubbling hy- drogen gas (made by the action of water on "hydrone") through the bromine warmed to 40° — 44°, and passing the mixed gases over hot platinized asbestos in a glass tube. The apparatus was constructed wholly of glass. The hydrogen was cleansed by being passed through two wash bottles containing dilute sul- phuric acid, and through a tower filled with beads also moistened with dilute sulphuric acid. The hydrobromic-acid gas was absorbed in pure water contained in a cooled flask. In order to remove iodine the solution of hydrobromic acid was diluted with water and twice boiled with a small quantity of free bromine. Then a small quantity of recrystallized potassium permanganate was added to the hydrobromic-acid solution, and the bromine set free was expelled by boiling. Finally, the acid was distilled with the use of a quartz condenser, the first third being rejected. It was preserved in a bottle of Nonsol glass pro- vided with a ground-glass stopper. The purification of the hydrobromic acid was carried on in conjimction with Dr. Grinnell Jones, who was engaged in a parallel research upon the atomic weight of phosphorus. Using this acid, he found that 10.48627 gm. of silver bromide were obtained from 6.02386 gm. of the purest silver. This ratio of silver bromide to silver, 100.0000 to 57.4452, is in close agreement with the most probable value, 100.0000 to 57.4453.^ ^ Baxter: Proc.Amer. Acad. ,42, 201 {igo6); Jour. Amer.Chem.Soc, 28,1322; Zeit.anorg. Chem., 50, 389. (See page 59.) * Baxter: Loc. ciU A REVISION OF THE ATOMIC WEIGHT OF ARSENIC. 77 A solution of hydrochloric acid was purified by disrillation after dilution. Hydrochloric-acid gas was generated by dropping C. P. concentrated sul- phuric acid into C. P. concentrated hydrochloric acid. The acids were shown to be essentially free from arsenic. All the water used in the research was purified by distilling the ordinary dis- tilled water of the laboratory, once with alkaline permanganate and then once alone, in both cases with the use of block-tin condensers which required no cork or rubber connections to the distilHng flasks. Quartz or platinum vessels were always employed in place of glass, whenever glass was unsuitable. METHODS OF ANALYSIS. The first method of analysis employed was that of converting the silver arse- nate into silver chloride by heating in a current of hydrochloric-acid gas. Since this process does not involve transfer of material it should be capable of giving results of great accuracy. Glass and porcelain are unsuitable for containing the arsenate during this process on account of the certainty of their being at- tacked. The first attempts at using quartz for the purpose resulted in slight etching of the surface of the tube where it came in contact with the salt. Ex- perience showed, however, that with careful manipulation the attacking of the quartz could be wholly prevented. The vessel used to contain the arsenate was a quartz tube nearly 2 cm. in diameter but joined to small tubes at each end. These tubes were exactly like those employed by Richards and Jones in the conversion of silver sulphate into silver chloride.^ After the tube had been weighed by substitution for a counterpoise similar in shape and size, a suitable quantity of silver arsenate was introduced, and the tube and contents were heated in a current of pure dry air for between 7 and 8 hours at 250° C. Al- though this treatment is not sufficient to expel last traces of moisture, it was hoped that by uniform treatment of the arsenate in all the analyses the propor- tion of water retained by the salt could be reduced to a constant percentage, which could be determined in separate experiments. The complete drying of the salt by fusion was not permissible because of de- composition of the arsenate at temperatures in the neighborhood of its fusing point. During the drying of the arsenate the quartz tube was surrounded with a cylinder of thin platinum foil and was contained in a hard-glass tube connected with an apparatus for furnishing a current of pure dry air. The hard-glass tube was heated by means of two aluminum blocks 15 cm. by 13 cm. by 5 cm., one placed above the other, the upper surface of the lower block and the lower sur- face of the upper being suitably grooved to contain the tube. The blocks were bored to contain a thermometer, the bulb of which was located near the middle of the tube. This oven (fig. 4) could be readily maintained at constant temper- ^ Pub. Car. Inst., No. 69, 69(1907); Jour.Amsr.Chem.Soc.,2g,Ss3; Zeit. anorg. Chem., 55, 80. 78 RESEARCHES UPON ATOMIC Fig. 4. — Solid aluminum drying oven. ature within a very few degrees by means of a Bunsen flame. We are indebted to Dr. Arthur Stabler of the University of Berlin for the suggestion of this method of heating. In order to purify and dry the air it was passed through a tower filled with beads moistened with dilute silver-nitrate solution, through a tower filled with small lumps of solid potassium hydroxide, then through 3 towers filled with beads moistened with concentrated sulphuric acid, and finally through a tube filled with resubUmed phos- phorus pentoxide. The apparatus was con- structed wholly of glass, with ground joints, and was simi- lar to that shown in fig. i, page 8. After being heated, the quartz tube was transferred to a desiccator and was allowed to come to the temperature of the balance case before being weighed. The quartz tube was then placed upon hard- glass supports, in a horizontal position, one end being slipped into a larger tube, through which could be passed a current of either dry hydrochloric- acid gas or dry air. The other end of the quartz tube slipped into one of the arms of a large U-tube filled with glass pearls, which served to con- dense any silver-chloride vapor, which might escape from the quartz tube. The other arm of the U-tube was connected with the flue of a hood, the suction thus caused being sufficient to prevent the escape of gaseous arsenic compounds from the apparatus. The quartz tube was protected from dust by a covering of sheet mica. ^ The usual method of procedure was as follows: The quartz tube containing the silver arsenate being in place, a current of hydrochloric-acid gas was passed through the tube, and the tube was slowly revolved with pincers tipped with platinum wire in order that the salt might be thoroughly exposed to the action of the acid. Neglect to do this at the commencement of the reaction always resulted in the caking of the salt in the tube, thereby rendering the action of the acid less rapid. The hydrochloric acid was dried by passing through three towers containing beads moistened with concentrated sulphuric acid. The ap- paratus for generating and purifying the acid was constructed wholly of glass, and was similar to that shown in fig. i, page 8. In the earlier experiments the salt was gently heated from the commence- ment of the reaction. To all outward appearance it was entirely converted into silver chloride in a few hours. Upon fusion, however, it presented a very cloudy appearance, owing to the presence of arsenic compounds, which could A REVISION OF THE ATOMIC WEIGHT OF ARSENIC. 79 not be completely removed even by keeping the silver chloride fused in the cur- rent of hydrochloric acid for as long as 8 hours. This is the cause of the larger quantities of arsenic found in the chloride obtained in the earlier analyses. Furthermore, the longer period of heating at a temperature above the fusing point of silver chloride accoimts for the larger amounts of volatilized silver chloride found in these experiments. As experience was gained, it was found best to expose the salt first in the cold for about 8 hours to the action of the hydrochloric-acid gas, next to heat the salt gently below its fusing point for from 10 to 15 hours, and finally to keep it barely fused for from 5 to 10 hours longer. When the reaction was apparently at an end, the current of hydrochloric-acid gas was stopped, and dry air was passed through the tube for about 15 minutes in order to eUminate hydro- chloric acid. The silver chloride was then allowed to solidify in a uniform thin layer around the inside of the quartz tube by slowly revolving the tube during solidification. The platinum wire used in weighing the tube was slipped on, the tube was transferred to its desiccator, and after standing several hours beside the balance it was weighed. In order to make sure that the reaction was complete the silver chloride was again fused, and exposed to the action of hydrochloric acid for several hours longer. As a rule, no change in weight was observed. In all cases constant weight was obtained upon heating in the same way for a third time. After making certain that only a small quantity of arsenic, if any, remained in the silver chloride, the contents of the quartz tube were dissolved in ammonia, and the silver chloride was reprecipitated by boiling the solution to expel the ammonia and adding a small quantity of sulphuric acid. The solution, after evaporation, was added to a Berzelius-Marsh apparatus containing arsenic-free zinc and sulphuric acid, and a mirror of arsenic was deposited in a hard-glass capillary tube in the usual way. The hydrogen was dried by calcium chloride before passing into the hard-glass tube, and the generating flask was cooled with water to prevent the evolution of hydrogen sulphide. The arsenic mirror formed was compared with a photograph of standard arsenic mirrors,^ the original mirrors showing that comparison with the photo- graph was equally satisfactory. The correction was applied by assuming that the arsenic was present in the silver chloride as arsenic trichloride, although if present as silver arsenate the correction would be much smaller. In any case the correction for residual arsenic is so small as to be almost without effect upon the final result. Ebaugh used essentially the same method of heating the arsenate in hydro- chloric acid, although the periods were shorter, so that it is probable that the small quantities of arsenate used (scarcely over one gram in any analysis) did not retain weighable amounts of arsenic. * Sanger: Proc. Amer. Acad., 26, 24 (1891). 8o RESEARCHES UPON ATOMIC WEIGHTS. The U-tube used for collecting volatilized silver chloride was washed out thoroughly with dilute ammonia, and the solution was made up to definite vol- ume after nearly all the ammonia had been expelled by boiling. The silver content of the solution was then compared in the nephelometer with that of standard solutions of silver, care being taken that the tubes were treated in as nearly as possible the same way. The second method of analysis consisted in heating the silver arsenate in a platinum boat and, after weighing, dissolving the arsenate in nitric acid and precipitating the silver as chloride or bromide. The platinum boat was heated in a hard-glass tube forming part of a bottling apparatus,^ and was thus trans- ferred without exposure to moist air to the weighing-bottle in which it was always weighed. The boat and bottle were weighed by substitution by com- parison with a counterpoise similar both in shape and volume. After the silver arsenate had been weighed, the boat with its contents was transferred to a flask, and the salt was dissolved in warm 5N nitric acid. The weighing-bottle was rinsed with acid, and the rinsings were added to the main solution; then the solution was carefully transferred to a large glass-stoppered precipitating flask and diluted to a volume of about i liter. From the weight of silver arsenate the amount of either hydrochloric or hydrobromic acid necessary to precipitate the silver was calculated. A slight excess of one acid or the other was then diluted to about 600 c.c, and the solution was slowly poured into the solution of silver arsenate in the precipitat- ing flask. After a few moments' shaking the precipitate was allowed to stand for several days, with occasional agitation. The precipitated silver chloride or silver bromide was next collected upon a weighed Gooch crucible, after it had been washed by decantation about ten times with dilute hydrochloric acid in the case of silver chloride, with water in the case of silver bromide. After several hours' heating in an electric air bath at 150° C, and about 2 hours' heating at 200° C, the precipitate was cooled in a desiccator and weighed. In order to determine the moisture retained by the precipitate it was trans- ferred as completely as possible to a small porcelain crucible and weighed. Then the salt was fused by heating the small covered crucible contained in a large crucible and was again weighed. The asbestos mechanically detached from the Gooch crucible, together with a small quantity of silver chloride or silver bromide which escaped the crucible, was collected upon a small filter through which the filtrate and wash-waters were passed, and the filter paper was ignited in a small weighed porcelain cru- cible. Before being weighed the ash was treated with a drop of nitric and a drop of either hydrochloric or hydrobromic acid and again heated. ^ See page 9. A REVISION OF THE ATOMIC WEIGHT OF ARSENIC. 8l The filtrate and wash- waters were evaporated to small bulk. The precipitat- ing flask was rinsed with ammonia, and the rinsing was added to the evapo- rated filtrate and wash-water. Then the solution was diluted to definite volimae, and the silver content was determined by comparison with standard silver solutions in the nephelometer. The operations of precipitating and collecting the silver halides were all car- ried out in a room lighted only with ruby light. INSOLUBLE RESIDUE. All the specimens of silver arsenate, after being heated at 250° C, when dis- solved in dilute nitric acid, were found to contain a small amount of insoluble residue, which would dissolve only in rather concentrated nitric acid. Although the proportion of this residue was apparently increased by exposure to light, specimens of the arsenate which had been prepared wholly in the dark room were not free from it. No process of purification to which the soluble arsenate used in the preparation of the silver arsenate was subjected seemed to have the slightest effect upon the proportion of insoluble matter. A similar phenomenon was met by Dr. Grinnell Jones in the preparation of silver phosphate (page 178). Although the amount of this residue in one gram of silver arsenate which had been treated as in the analyses for silver was not over 0.00005 gm., it was important to determine its silver content. This was done in three cases in which the proportion of residue had been purposely increased as much as pos- sible by exposure to hght. The arsenate was dissolved in dilute nitric acid, and the residue was collected upon a weighed platinum Gooch crucible, the detached asbestos shreds being carefully determined by filtration upon a filter paper. The weight of residue was found by reweighing the crucible. After the residue had been dissolved in concentrated nitric acid and the solution had been diluted to definite volume, the silver content of the solution was ascer- tained by comparison with standard silver solutions in a nephelometer. Weight of silver arsenate. Weight of insoluble residue. Weight of silver. Per cent of silver. gm. 4.26 10.00 9.28 gm. 0.00198 0.00228 0.00657 gm. 0.00143 0.00162 0.00500 72.3 71.I 76.1 Average 7^.2 Theoretical per cent of silver in silver arsenate . . . 70.0 The first of the above determinations was made with a sample of silver arse- nate which had been exposed to bright light inside a desiccator for a month. 82 RESEARCHES UPON ATOMIC WEIGHTS. During this time the quartz tube containing the salt showed no perceptmie change in weight. The third determination also was made with a sample of salt which had been exposed to bright light for 3 weeks in a dry state. In the second determination the salt was exposed to light under water for one week. Two facts show that the presence of the small proportion of the residue in the arsenate could have had no important efifect upon the results. In the first place, the formation of the insoluble matter under the influence of light is not attended by change in weight. In the second place, the silver content of the residue is very near that of silver arsenate. Nevertheless, care was taken to protect the arsenate as far as possible from exposure to light. DETERMINATION OF MOISTURE IN DRIED SILVER ARSENATE. T. W. Richards ^ and others have already drawn attention to the fact that it is not possible, without fusion, to dry completely a substance formed in aqueous solution, owing to the mechanical retention of liquid in pockets within the solid. In the case of silver arsenate, although it is possible to fuse the salt, the temperature necessary is so high that decomposition of the salt takes place to some extent. Hence the loss in weight on fusion can not be used as a true measure of the water content of the salt. Since decomposition of the salt could produce only easily condensible substances and oxygen, the difficulty was over- come in the present instance by fusing weighed quantities of the salt in a current of pure dry air and collecting the water vapor in a weighed phosphorus pentox- ide tube. Of course great pains were taken to treat the salt used in the water determinations in exactly the same way as that used in the analyses for silver. The procedure was as follows: A sample of salt very nearly as pure as that used in the silver analyses was weighed out in a copper boat which had been previously cleaned and ignited in the blast lamp to remove organic matter. The boat was placed in a hard-glass tube and was heated for between 7 and 8 hoiu-s at 250° C, in a current of dry air. In these experiments, before passing through the drjdng towers the air had first been passed over hot copper oxide in order to oxidize any organic matter it might contain. Furthermore, the con- centrated sulphuric acid in the drying towers had been heated with a small quantity of potassium dichromate. One end of the hard-glass tube was con- nected to the apparatus for suppl3dng pure air, by means of a well-fitting ground joint upon which no lubricant was used. The other end was sealed to a small hard-glass tube which was surrounded with a damp cloth during the fusion of the salt in order to facilitate condensation of any silver or arsenic compounds vaporized from the salt. As a matter of fact, very little sublimation actually took place. In order to fuse silver arsenate within the hard-glass tube it was necessary to use the hottest flame of the blast lamp, the tube being covered with a semi- * Zeit. physik. Chetn., 46, 194 (1903). REVISION OF THE ATOMIC WEIGHT OF ARSENIC. 83 cylinder of sheet iron. Furthermore, since at this temperature even the hard glass became very soft, it was found necessary to wrap the tube with asbestos and closely wound iron wire for several inches at the piont where fusion took place. This also served to distribute the heat more evenly and to prevent the tube from cracking during the experiment. Just before the salt was fused a carefully weighed U-tube containing resub- limed phosphorus pentoxide was attached to the end of the tube, and beyond this was joined another similar tube which served as a protection against any moisture which might diffuse back into the weighed tube from the outside air. These phosphorus pentoxide tubes were provided with one way ground glass stopcocks lubricated with Ramsay desiccator grease. The salt was heated for 25 minutes with the hottest flame of the blast lamp, being then completely fused, and was further heated for 35 minutes at a con- siderably lower temperature in order to make certain that all moisture was car- ried into the absorption tube by the current of air. Finally the phosphorus pentoxide tube was reweighed. The pentoxide tube was weighed by substitution with the use of a counter- poise of the same size and weight filled with glass beads. Before being weighed both tubes were carefully wiped with a damp cloth and were allowed to stand near the balance case for one hour. One stopcock in each tube was opened im- mediately before the tube was hung upon the balance, in order to insure equi- librium between the internal and external air pressure. The stopcock of the counterpoise was left open during the weighing. Owing to the considerable length of time required for the tubes to come to equilibrium on the balance, it was considered unsafe to leave the stopcock of the pentoxide tube open during the weighing. As a check on the first weight of the pentoxide tube one stopcock was opened and closed and its weight determined a second time. Ordinarily no change in weight was observed. Weight of silver arsenate. Weight of water. Weight of water per gram of salt. gm. 11.09 13-59 17-23 12.57 gm. 0.00083 0.00073 0.00085 0.00057 gm. 0.000075 0.000054 0.000049 0.000045 Average 0.000056 Since it seemed possible that the hard-glass tube itself, when heated nearly to fusion, might give off traces of water vapor, two blank determinations were made by heating the empty hard-glass tube in exactly the same way as in the water determinations. These determinations showed a gain in weight of the pentoxide tube of 0.00022 and 0.00037 gm. respectively, the average being 84 RESEARCHES UPON ATOMIC WEIGHTS. 0.00030 gm. This correction was confirmed in another experiment in which the hard-glass tube was kept at the highest temperature obtainable with the blast lamp for an hour. The observed gain in weight of the absorption tube was 0.00048 gm. A negative correction of 0.00030 gm. was applied in each water determination. In order to allow for moisture the weight of the arsenate was therefore always corrected by subtracting 0.000056 gm. per gram of salt. Ebaugh took no notice of the water contained in silver arsenate which had been dried at only 170°. SPECIFIC GRAVITY OF SILVER ARSENATE. In order that the apparent weight of the silver arsenate might be corrected to a vacuum standard, the specific gravity of the arsenate was found by deter- mining the weight of toluol displaced by a known quantity of salt. The toluol was first dried by means of stick soda and was then distilled, with rejection of the first portion of the distillate. Its specific gravity at 25" referred to water at 4° was found to be 0.8620. Pains were taken to remove air from the arsenate when covered with toluol by placing the pycnometer in an exhausted desiccator. Weight of silver arsenate in vacuum. Weight of displaced toluol in vacuum. Specific gravity of silver arsenate 25°/4°. gm. S.1690 5.6729 gm. 0.6688 0.7350 gm. 6.662 6.652 Average 6.657 The following vacuum corrections were applied: Specific gravity. Vacuum correction. Weights Toluol Silver arsenate .... Silver chloride Silver bromide .... 8.3 0.862 6.657 5.56 6.473 + 0.00126 + 0.000036 + 0.000071 + 0.000041 All weighings were made upon a nearly new short-armed Troemner balance, easily sensitive to 0.02 mg. with a load of 50 gm. The gold-plated Sartorius weights were several times carefully standardized to hundredths of a milligram by the method described by Richards,^ and were used for no other work. * Jour. Amer. Ckem., Soc, 22, 144 (1900). A REVISION OF THE ATOMIC WEIGHT OF ARSENIC. 85 RESULTS AND DISCUSSION. Series I. sAgCl: Ag3As04. No. of analysis. Sample AgsAsOj. Corrected weight of AggAsOi in vacuum. Weight of AgCl in vacuum. Residual A8CI3. Volatilized AgCl. Corrected weight of AgCl in vacuum. Ratio 3 AgCl: AgsAtOi. I 2 3 4 5 A A A B C gm. 3-17276 2.65042 3.51128 5.83614 5-72252 gm. 2.94912 2.46364 3-26395 5.42499 S-31947 gm. 0.00004 0.0CX)04 O.CX>OOI 0.00001 0.00001 gm. 0.00014 0.00007 0.00002 0.00005 0.00001 gm. 2.94922 2.46367 3.26396 5-42503 S-31947 0.929544 0-929539 0.929564 0.929558 0.929568 Avera Z6 ... 20 c c c Series II. 3 AgCl: AgsAsO^. No. of anal- ysis. Sample of AgsAsO^ Corrected weight of AgsAsOi Weight of AgCl Weight of asbestos. Dissolved AgCl from filtrate. Loss on fusion. Corrected weight of AgCl Ratio 3 AgCl: AgjAsOa. gm. 4-59149 3.38270 gm. 4.26815 3.14401 gm. 0.CXK)08 0.00037 gm. 0.0I2 0.00013 gm. o.o ^^ used. Proc. Roy. Soc, 53, 147. i84 RESEARCHES UPON ATOMIC WEIGHTS. RATIO OF SILVER BROMIDE TO SILVER PHOSPHATE. The following table contains all of the analyses not vitiated by a known im- purity in the sample or by an accident during the analysis. One feature of this table requires further explanation. In analysis 5 the silver was determined by precipitation as chloride instead of bromide. For every gram of silver phos- phate there was obtained 1.02727 gm. of silver chloride. Since Baxter found AgBr:AgCl= 1,31017:1.00000, this analysis indicates that one gram of sample N is equivalent to 1.02704 X 1.31017 = 1.34560 gm. of silver bromide. This result is placed in the table for comparison with the other analyses and is used in the computation of the mean. Series I. 3 AgBr: AgsPO^. No. of Sample Weight of Weight of Weight Dissolved AgBr. Loss Corrected Ratio analy- of AgsPO* AgBr in of on weight of 3AgBr sis. AgsPOi m vacuum. vacuum. asbestos. fusion. of AgBr. AgsPOj. gm. gm. gm gm. gm. gm. I 6.20166 8.34427 0.00036 0.00034 0.00007 8.34490 1-34558 2 6.35722 8.55386 0.00041 0.00003 0. 0001 1 8.55419 I-345S9 3 N 5.80244 7.80792 0.00029 0.00005 0.00007 7.80819 1-34567 4 N 5-05845 6.80658 (AgCl) 0.00019 0.00020 0.00012 6.80685 / AgCl \ \ 3-43544 / 1-34564 5 N 3-34498 3-43514 0.00029 0.00009 0.00008 1.34560 6 P 7-15386 9.62648 0.00046 0.00013 0.00013 9.62694 1-34570 7 P 7.20085 9.68929 0.00023 0.00005 O.OOOIO 9.68947 1.34560 8 R 6.20182 8.34466 0.00041 0.00027 0.00012 8.34522 1-34561 9 R 5.20683 7-00543 0.00029 0.00040 0.00007 7.00605 1-34555 Ave rag e . . . . 1.34562 Per cent of Ag in A K.PO. . . . 77.300 DISCUSSION OF RESULTS. A careful study of these results shows that the composition of silver phosphate is very nearly, if not quite, independent of the changes in the acidity of the solu- tions from which it is precipitated. Samples O and R were prepared under slightly more acid conditions than Samples N and P. The average amount of silver bromide obtained from one gram of Samples and R is 1.34558 (77.297 per cent of silver), whereas the average from Samples N and P is 1.34564 (77.301 percent of silver). This difference, if real and significant, is probably due to a very slight occlusion of disilver hydrogen phosphate. It does not seem probable that any basic salt was present in Samples N and P, because silver shows little tendency to form basic salts and the conditions of precipitation were not favorable for the formation of basic salts. The difference between composition of the samples is so slight, both in abso- lute amount and by comparison with the differences between different analyses of the same sample, that in the present state of our knowledge it does not ^ Loc. cit. A REVISION OF THE ATOMIC WEIGHT OF PHOSPHORUS. 185 seem justifiable to reject the analyses of Samples R and 0. This conclusion is supported by the fact that the water determinations failed to show a difference between these samples. The results, however, indicate that the average ratio 1.34562 (77.300 per cent of silver) may be very slightly too low, owing to the presence of disilver hydrogen phosphate. The ratio 1.34562, assuming the atomic weight of silver to be 107.880, and assimiing that silver bromide con- tains 57.4453 per cent of silver, leads to an atomic weight of 31.043 for phos- phorus, whereas the ratio 1.34564 derived from Samples N and P gives the value 31.037. The rounded-off value, 31.04, may be considered to be essentially free from error from this source. SUMMARY. 1. A careful study has been made of the conditions necessary for the prep- aration of pure trisilver phosphate. 2. It is found that silver phosphate can be almost completely dried without fusion by heating in a current of dry air. 3. The density of silver phosphate is foimd to be 6.37. 4. It is found that silver phosphate does not adsorb a significant amou,nt of air. 5. Nine analyses, made with four different s'amples, show that one gram of silver phosphate yields 1.34562 gm. of silver bromide, whence the per cent of silver in silver phosphate is 77.300. Therefore, If Ag = 107.88 P = 31.04 If Ag= 107.87 P = 31-03 Date Due nnj n bi 984 1 IJAV 1 i MAY 1 ^ i999 ' I f) BOSTON COLLEGE 3 9031 01575182 9 Boston College Library Chestnut Hill 67, Mass. Books may be kept for two weeks unless a shorter time is specified. Two cents a day is charged for each 2-week book kept overtime; 25 cents a day for each over- night book. If you cannot find what you want, inquire at the delivery desk for assisstance. W 9-48