SB 
 
 SOfl 
 
LIBRARY 
 
 UNIVERSITY OF CALIFORNIA. 
 
 RECEIVED BY EXCHANGE 
 
 Class 
 
The 
 
 Atomic Weight 
 
 of 
 
 VANADIUM 
 
 THESIS 
 
 PRESENTED TO THE FACULTY OF THE GRADUATE SCHOOL 
 
 OF THE UNIVERSITY OF PENNSYLVANIA IN PARTIAL 
 
 FULFILMENT OF THE REQUIREMENTS FOR THE 
 
 DEGREE OF DOCTOR OF PHILOSOPHY 
 
 BY 
 
 D. J. McADAM, JR. 
 WASHINGTON, PA, 
 
 1910 
 
 PHILADELPHIA 
 THE JOHN C. WINSTON CO 
 
 IQIO 
 
The 
 
 Atomic Weight 
 
 of 
 
 VAN ADI UM 
 
 THESIS 
 
 PRESENTED TO THE FACULTY OF THE GRADUATE SCHOOL 
 
 OF THE UNIVERSITY OF PENNSYLVANIA IN PARTIAL 
 
 FULFILMENT OF THE REQUIREMENTS FOR THE 
 
 DEGREE OF DOCTOR OF PHILOSOPHY 
 
 BY 
 
 D. J. McADAM, JR. 
 
 WASHINGTON, PA. 
 
 1910 
 
 PHILADELPHIA 
 
 THE JOHN C. WINSTON CO. 
 1910 
 

 ACKNOWLEDGMENT. 
 
 The writer wishes to express his thanks to Dr. Edgar 
 F. Smith for his kindness, encouragement and valuable 
 advice. This work was undertaken at Dr. Smith's sug- 
 gestion and carried on under his supervision. 
 
 The writer's sincerest thanks are also due to Prof. W. 
 B. Schober, of Lehigh University, for the use of the labo- 
 ratory and of valuable apparatus in the completion of this 
 work, also to Mr. F. B. Kingsbury and Mr. C. A. Pierle for 
 valuable assistance. 
 
INTRODUCTORY. 
 
 In this laboratory, use has frequently been made of dry 
 hydrochloric acid gas in making various separations, and 
 in several atomic weight determinations. 
 
 Smith and Hibbs (J. Am. Ch. S. 16, 578; Zeit. Anorg. 
 Ch. 7, 41) found that vanadium as well as a number of 
 other substances could be completely volatilized from their 
 alkali salts by heating in a current of dry hydrochloric acid 
 gas. It therefore seemed probable that this method might 
 be used in determining the atomic weight of vanadium. 
 
 A new determination of the atomic weight of this ele- 
 ment would be particularly desirable since the result* ob- 
 tained by previous investigators differ widely. 
 
 Historical. 
 
 Berzelius, in 1831 (Kongl. Vet. Acad, Handl. 1-65; 
 Ann. Phys. 22, 1-67), made four determinations of 
 the atomic weight of vanadium. In three of these, he 
 reduced V 2 O 5 to V 2 O 3 by heating in a stream of hydrogen ; 
 in the fourth, he reoxidized the V 2 O 3 to V 2 O 5 . As the 
 average of four fairly concordant results he obtained a 
 value which corresponds to 52.46, referred to oxygen as 16. 
 As Roscoe pointed out later, Berzelius' material undoubt- 
 edly contained phosphorus, a small trace of which inter- 
 feres greatly with the reduction of V 2 O 5 to V 2 O 3 . The 
 ammonium molybdate test for phosphoric acid was 
 unknown in Berzelius' time. 
 
 Roscoe, in 1867 (Phil. Trans. R. S. 158, 1-27), made 
 four determinations by the method of Berzelius, using 
 material carefully purified in several different ways. The 
 average of the four results was 51.371, but the maximum 
 and minimum differed by nearly one-half of one per cent. 
 
 3 
 
 228324 
 
He also obtained a series of nine values by determining 
 the amount of silver required to precipitate the chlorine 
 from vanadium oxychloride, VOC1 3 . He did not use the 
 proper end point for this reaction, but the final results 
 would not be affected very much by this error; loss of 
 chlorine on the treatment of the VOC1 3 with water was 
 probably a more serious source of error. 
 
 In still another series of eight experiments, Roscoe 
 weighed the silver chloride obtained from a known weight 
 of vanadium oxychloride. He does not mention any cor- 
 rection for the solubility of the silver chloride in the wash 
 water. If this was neglected, his value for the percentage 
 of the chlorine is too low, and consequently his value for the 
 atomic weight is too high. The variation in the percentage 
 of chlorine in these seventeen experiments is from 60.86 
 to 61.55, or about 1.13 per cent. This, of course, means a 
 variation of 1.13 per cent, in the calculated molecular 
 weight of VOC1 3 ; and, since the atomic weights of oxy- 
 gen and chlorine are assumed to be known, all this varia- 
 tion would apply to the vanadium alone, and would amount 
 to about 3.4 times 1.13 or 3.84 per cent. For example, the 
 maximum calculated value for the atomic weight would be 
 52.39, and the minimum, 50.44. The average of his results 
 by both methods is 51.24. 
 
 For over forty years, our knowledge of the atomic 
 weight of this element rested entirely on the work of Ros- 
 coe. Within the last year, however, two articles on the 
 atomic weight of vanadium by Wilhelm Prandtl and Benno 
 Bleyer have appeared. (Zeit. Anorg. Chem. 65, 152 and 
 67, 257.) These authors have used the same methods that 
 were used by Roscoe, but have used great care in avoiding 
 the sources of error in the work of the earlier author. 
 
 The vanadium oxychloride was carefully purified, and 
 kept out of contact with air until it had been weighed. 
 
Moreover, on treatment of the oxychloride with water, they 
 reduced the vanadium to the valence of four by means of 
 pure zinc, and thus avoided loss of chlorine. 
 
 In order to make allowance for the solubility of silver 
 chloride in water, they prepared twelve wash waters and 
 used them in regular order. These were prepared by wash- 
 ing the silver chloride precipitated in the first experiment 
 in each series, the twelve filtrates being preserved in the 
 same order in which they were obtained. 
 
 In the first article published by Prandtl and Bleyer 
 are given the results of two series of experiments by this 
 method. In series i, consisting of five determinations (the 
 first one being, of course, rejected), the average percentage 
 of chlorine obtained from the oxychloride was 61.3095 
 0.0158; the maximum was 61.352 and the minimum 61.284, 
 a variation of o.n per cent. This variation when applied 
 to the vanadium content would be multiplied by 3.4, making 
 0.37 per cent, variation in the calculated values for the 
 atomic weight. The average value for the atomic weight 
 in this series is given as 51.133 0.013. The authors, 
 however, have overlooked the fact that the probable error 
 in the percentage of chlorine must be multiplied by 3.4 to 
 give the probable error in the atomic weight of vanadium. 
 So it should be 0.044 instead of db 0.013. 
 
 In series II, consisting of six determinations, the aver- 
 age percentage of chlorine was 61.3696 0.0098, the 
 extremes being 61.405 and 61.348, a variation of 0.093 P er 
 cent. ; this would correspond to a variation of 0.316 per 
 cent, in the atomic weight. The average value for the 
 atomic weight in the series is 50.963, and the probable 
 error, as before, should be multiplied by 3.4, giving 0.027 
 instead of 0.008. As a mean of the two series they get 
 51.048 o.oio, but it should be 0.034. 
 
 In the second article, by the same authors, they have 
 a third series of four accepted values varying from 61.301 
 
6 
 
 to 61.333 for percentage of chlorine. As an average for 
 the thirteen accepted values in the three series, they obtain 
 51.061 for the atomic weight. 
 
 In the same article, these authors have recorded the 
 results of four determinations by the method of Berzelius. 
 In this series, they obtained results that are considerably 
 higher, giving an average of 51.374 0.033. But they 
 found that the V 2 O 3 takes up oxygen so rapidly that the 
 weight of the compound cannot be obtained with any cer- 
 tainty. It is, therefore, probable that the observed weights 
 of V 2 O 8 are too great, and consequently the calculated 
 values for the atomic weight of vanadium are too high. 
 
 The discrepancies in the results of their careful inves- 
 tigation have thus made more evident the need for a rede- 
 termination of the atomic weight of vanadium by a new 
 method. 
 
 Method and Apparatus. 
 
 The salt of vanadium used in these experiments was 
 sodium meta-vanadate (NaVO 8 ), the most stable of the 
 vanadates of sodium. Some attempts were made to use 
 other vanadates of sodium, but all except the meta-vana- 
 date were found to be unstable or indefinite. 
 
 The meta-vanadate was heated in hydrochloric acid 
 gas in order to drive out the vanadium and leave sodium 
 chloride. Suice hydrochloric acid has a tendency to reduce 
 vanadates, it seemed probable that a mixture of hydro- 
 chloric acid and chlorine would drive out the vanadium 
 more readily than the hydrochloric acid alone. Experi- 
 ments proved that the presence of a small proportion of 
 chlorine had a good effect. 
 
 The apparatus was constructed entirely of glass, with 
 connections of ground glass or sealed by sulphuric acid. 
 The arrangement of the gas generators and drying appara- 
 tus is shown in Fig. I. 
 
In the vessel A, the hydrochloric acid gas is generated 
 by allowing concentrated sulphuric acid to drop into strong 
 hydrochloric acid solution. In B, the chlorine is gener- 
 ated by allowing hydrochloric acid to drop into the flask 
 containing pure potassium permanganate moistened with 
 water. C, C, are drying vessels containing sulphuric acid. 
 
 FIG. i 
 
 D, D, D are connectors in which sulphuric acid is placed 
 so as to make the connection air tight. E is a drying tower 
 containing glass beads moistened with sulphuric acid. F, F 
 are stop-cocks so arranged that the gas from the generators 
 can be shut off and a current of pure air substituted. 
 
 Several modifications were made in the apparatus con- 
 taining the sodium vanadate, before the best form was 
 obtained. These are shown in Figs. 2 and 3. 
 
PREPARATION OF PURE MATERIAL. 
 
 Water. 
 
 The distilled water of the laboratory was redistilled 
 once from alkaline permanganate, rejecting the first fourth 
 of the distillate. It was then redistilled twice more, the last 
 time with a block tin condenser; in some of the later 
 preparations a quartz condenser was used. The water 
 seal invented by T. W. Richards was used in all distillations. 
 When used for crystallizations, the water was either allowed 
 to run from the condenser directly into a platinum dish, 
 or was collected in a quartz flask. 
 
 Sodium Meta-Vanadate. 
 
 The chief source was the iron vanadate that is used in 
 the preparation of ferro-vanadium. It contained not more 
 than a trace of phosphorus or molybdenum. 
 
 The iron vanadate was changed to sodium vanadate 
 by boiling with caustic soda or sodium carbonate, filtering 
 off the ferric hydroxide, and evaporating the filtrate to 
 crystallization. It was purified by several different methods. 
 
 Sample A. 
 
 A mixture of vanadic acid and acid vanadates of 
 sodium was boiled with a slight excess of C. P. sodium 
 carbonate. It was then recrystallized five times in glass 
 from ordinary distilled water, and twice in platinum from 
 twice distilled water. The crystallization was carried on 
 in a room free from acid fumes, so no special precautions 
 were taken to keep the material out of contact with the 
 air of the room. The sample was dried and preserved in 
 porcelain dishes in a desiccator, over sulphuric acid that 
 had been boiled with ammonium sulphate. 
 
9 
 
 Sample B. 
 
 Some meta-vanadate containing phosphates was 
 recrystallized a few times, rejecting the first crop of crys- 
 tals each time. The phosphate was thus almost completely 
 removed. The partially purified meta-vanadate was now 
 recrystallized six times in glass and once in platinum from 
 twice distilled water, being drained by suction each time. 
 
 It was next crystallized six times more in platinum by 
 dissolving in hot water, cooling, and distilling into the dish 
 some twice distilled alcohol. The substance separated out 
 in small crystals and settled rapidly, leaving a large volume 
 of liquid, which was poured off. It was then recrystallized 
 twice from water by allowing the crystals to form slowly 
 without stirring ; for each crystallization several weeks were 
 required, and considerable liquid was left, which was poured 
 off. The substance was dried and kept for about two years 
 in a desiccator over caustic soda. 
 
 Sample C. 
 
 Some sodium vanadate, which had been crystallized 
 several times, was recrystallized twice more in porcelain, 
 from ordinary distilled water. It was then recrystallized 
 in platinum three times from ordinary distilled water and 
 three times from thrice distilled water. It was kept in a 
 desiccator over caustic soda during crystallization. 
 
 Centrifugal drainage was used in the last five crys- 
 tallizations. The centrifugal machine had an arrangement 
 by which the material could be placed in a perforated plati- 
 num cone. Not enough time was allowed to each crystal- 
 lization, consequently the crystals formed in a pasty mass, 
 which did not drain well. This error was avoided in later 
 preparations. 
 
 The crystals were now dissolved in hot water and 
 allowed to stand in a desiccator over caustic soda ; crystals 
 formed slowly and the remaining liquid was poured off. 
 
10 
 
 The solution was yellow at first, but as crystallization pro- 
 ceeded it became colorless. 
 
 The fused material was quite hygroscopic, indicating 
 that it still contained an excess of alkali. 
 
 Sample D. 
 
 The mother liquor from sample C was put in a desicca- 
 tor over caustic soda and allowed to evaporate to dry- 
 ness. A little over two grams were obtained, and this was 
 used later as a test of the neutrality of sample C. 
 
 It showed some excess of alkali. 
 
 Sample E. 
 
 Since sample C still contained excess of alkali, a por- 
 tion of it was recrytallized twice more with centrifugal 
 drainage, and was washed each time with a little distilled 
 water. The crystallization was allowed to take place 
 slowly, so that the substance was not pasty as in sample C ; 
 consequently it drained readily, giving a light yellow fil- 
 trate. Contact with the air of the room was prevented as 
 much as possible by keeping it in a desiccator over caustic 
 soda during crystallization. 
 
 The material had now been put through eleven crys- 
 tallizations altogether, and was apparently free from excess 
 of alkali. 
 
 While the fused material of sample C was so hygro- 
 scopic that it could hardly be weighed to o.oooi gm. the 
 weight of sample E could be obtained without difficulty to 
 o.ooooi gm. 
 
 Sample F. 
 
 Crude sodium vanadate was acidified with sulphuric 
 acid, and the precipitated vanadic acid washed repeatedly 
 with distilled water, but it still gave a test for sulphuric 
 acid. 
 
II 
 
 It was then treated with a large quantity of water, 
 and reduced to vandyl oxalate by heating with oxalic acid. 
 A slight excess of ammonium oxalate was added and the 
 double ammonium oxalate crystallized out on evaporation 
 and cooling. After three crystallizations, with drainage by 
 suction, no trace of sulphate could be detected, and the 
 material was probably free from phosphoric and molybdic 
 acids. 
 
 By addition of excess of ammonia, a brown precipi- 
 tate of ammonium vanadate was obtained which was fil- 
 tered and washed several times. It was then heated to get 
 rid of ammonia, and fused with sodium carbonate. Since 
 the water solution was still greenish, the oxidation was com- 
 pleted by addition of a little sodium peroxide. The analy- 
 ses of the sodium carbonate and sodium peroxide, accord- 
 ing to J. T. Baker, showed the following : 
 
 Na 2 CO s Na.0, 
 
 Fe 0.0005% Fe 0.003 % 
 
 A1 2 O 0.0002% A1 2 O o.ooi % 
 
 CaO 0.008 % Cl .' . . 0.007 % 
 
 SiO 2 o.ooi % SO o.ooi % 
 
 Cl 0.085 % Na 2 O 3 83.1 % 
 
 SOs o.ooi % COa 0.32 % 
 
 H,O 0.95 % 
 
 The sodium vanadate was now crystallized eight times 
 in porcelain and three times in platinum, from ordinary dis- 
 tilled water, suction being used after each crystallization. 
 It was now practically neutral to phenol-phthalein. 
 
 It was next crystallized four times more from thrice 
 distilled water, each time with centrifugal drainage and 
 washing. The last three times the water came in contact 
 with nothing but quartz and platinum. Special precautions 
 were taken to avoid contact with carbon dioxide of the air 
 during the last few crystallizations. 
 
12 
 
 The effect of carbon dioxide of the air could be only 
 slight, but experiments showed that it affects the equi- 
 librium to a certain extent. The gas prepared by the action 
 of dilute sulphuric acid on sodium bicarbonate, washed in a 
 sodium bicarbonate solution, was passed through a vessel 
 containing glass beads moistened with water, and introduced 
 into a nearly colorless solution of meta-vanadate, which it 
 soon turned to a deep orange red. The equation might be 
 represented as follows: 
 
 4NaVO 3 + H 2 O + 2CO 2 = Na 2 V 4 O u + 2NaHCO,. 
 
 On boiling this orange colored solution, or putting it 
 in a desiccator over caustic soda, the reverse reaction takes 
 place, CO 2 is given off and the solution becomes colorless. 
 If the meta-vanadate contains excess of alkali or sodium 
 carbonate, only a faint yellow color appears on passing in 
 CO 2 , since the increased mass of bicarbonate favors the 
 reverse reaction. 
 
 Meta-vanadate solutions are usually colorless when hot, 
 and yellow when cold, but an experiment tried on some of 
 the pure material seems to indicate that the yellow color is 
 due to traces of acid vanadate caused by CO 2 of the air. 
 A solution of the meta-vanadate, which was light yellow 
 in color, was heated to boiling in a glass flask. On cooling 
 in contact with air, a yellow color appeared ; but, if the flask 
 was stoppered while the liquid was boiling, no color 
 appeared on cooling. 
 
 These experiments prove that CO 2 of the air can have 
 no effect on the solutions when they are hot; therefore, its 
 effect would be negligible in preparing sodium meta-vana- 
 date, if the material is kept in a desiccator over caustic 
 soda. 
 
 The substance which had thus been put through fif- 
 teen crystallizations was cream colored. It was neutral in 
 
13 
 
 hot or cold solution to methyl orange, litimus or phenol- 
 phthalein. The fused substance was colorless and trans- 
 parent. 
 
 Samples G, H and I have not been finished, but are to 
 be used in later work. 
 
 Sample G. 
 
 Iron vanadate was boiled with caustic soda, filtered 
 and acidified with sulphuric acid. The preciptated vanadic 
 acid was washed twice with water and then heated with 
 sulphuric and oxalic acids ; by this process the vanadium 
 was reduced from the valence five to the valence four, and 
 with the sulphuric acid formed vanadyl sulphate, V 2 O 4 3SO 3 . 
 On evaporation, small blue deliquescent crystals were 
 formed, which were filtered from the syrupy liquid with 
 suction. It was crytallized five times in this way. 
 
 After the thi'rd crystallization, it was tested for phos- 
 phoric acid by adding ammonium molybdate and nitric acid, 
 dissolving the reddish brown precipitate in ammonia and 
 adding magnesia mixture, and alternating with ammonium 
 molybdate and magnesia mixture until a yellow precipitate 
 of phospho-molybdate was obtained. The amount was 
 quite small, and a comparison test showed it to be less than 
 o.i per cent. 
 
 After the fifth crystallization of the vanadyl sulphate, 
 it was dissolved in a large quantity of distilled water, and 
 an excess of sodium carbonate of especial purity, accord- 
 ing to J. T. Baker, was added. The brownish precipitate 
 was washed fourteen times with water (in which its solu- 
 bility is considerable), but it still showed traces of sul- 
 phates. 
 
 In order to convert it into sodium vanadate, it was now 
 treated with sodium peroxide having the following anaysis, 
 according to J. T. Baker : 
 
 Fe.. 0.005% A1 2 O 3 . .0.001% Cl.. 0.002% SO 3 . .none, 
 insoluble matter trace, phosphates ^ none. 
 
14 
 
 To combine with the excess of alkali, CO 2 was passed 
 into the liquid. The CO 2 was generated by action of sul- 
 phuric acid on C. P. sodium bicarbonate, and was washed 
 by passing through three bottles of distilled water. The 
 solution was now evaporated to crystallization and it was 
 necessary to leave it at this point. 
 
 Sample H. 
 
 The preparation of this sample was begun before any 
 of the others; but, since it was not possible to work at it 
 continuously, the process has extended over several years. 
 
 Some C. P. ammonium vanadate from Kahlbaum was 
 treated with ammonium sulpho-cyanide acidified with hydro- 
 chloric acid and shaken with ether. A slight pink color 
 appeared in the ether, and became deeper when the solution 
 was reduced by zinc, thus indicating traces of iron and 
 molybdenum. 
 
 In order to remove molybdenum, the ammonium vana- 
 date was ignited to drive off ammonia and ammonium 
 chloride, and was dissolved in strong HC1; the excess of 
 the latter was removed by evaporation and the residue dis- 
 solved in water and filtered. The liquid was now saturated 
 with H 2 S and heated in a 2^2 liter pressure flask at 85C. 
 for several hours. A brown precipitate appeared, which 
 was principally V 2 S,, and should contain all the molyb- 
 denum. 
 
 The next process was to remove traces of arsenic. The 
 blue filtrate from the molybdenum sulphide was evaporated 
 and oxidized with nitric acid. It was then saturated with 
 ammonia, diluted to four or five liters, and saturated with 
 H 2 S ; then, from the ammonium sulpho-vanadate thus 
 formed, the vanadium sulphide was precipitated by HC1. 
 The sulphide was washed several times by decantation, 
 dried, and gently heated in a hard glass tube in a current 
 of dry HC1 gas. This should remove the arsenic, according 
 to Smith and Hibbs. 
 
15 
 
 The remaining V 2 S 5 now gave no test for molybdenum 
 in several grams of material, although the test was deli- 
 cate enough to show a color with less than o.oooi gm. of 
 MoO 8 . 
 
 The vanadium sulphide was next oxidized with redis- 
 tilled nitric acid and heated to form V 2 O 5 . In order to vola- 
 tilize the vanadium and leave behind various impurities, 
 the V 2 O 5 was then heated in a hard glass tube in dry HC1 
 gas. (Smith and Hibbs, Jr. Am. Ch. S. 17, 682-686.) 
 The volatile portion was collected in distilled water, evapo- 
 rated, oxidized with redistilled nitric acid, and heated to 
 form V 2 O 5 . 
 
 The purified substance was now dissolved in redis- 
 tilled ammonia in, platinum vessels and thle excess /of 
 ammonia boiled off. Acetic acid (which gave no precipi- 
 tate with H 2 S, NH 4 OH, (NH 4 ) 2 C 2 O 4 or (NH 4 ) 2 S) was 
 added to form the tetra-vanadate, which was precipitated 
 by redistilled alcohol, and the supernatant liquid poured off. 
 
 After standing in this way for about two years, it was 
 recrystallized from redistilled water and the liquid removed 
 by centrifugal drainage. It was then boiled with a slight, 
 excess of especially pure sodium carbonate until no more 
 ammonia fumes were given off, and evaporated to crystalli- 
 zation. 
 
 Sample I. 
 
 An acid vanadate of magnesium was prepared by boil- 
 ing vanadic acid with excess of magnesia, filtering, con- 
 centrating the filtrate and adding acetic acid. An orange 
 colored crystalline precipitate was formed, which could be 
 recrystallized by using a very large quantity of water ; only 
 a few grams would be obtained from several liters of water. 
 With each crystallization, some of the salt was decomposed 
 into an insoluble acid vanadate, nevertheless, about eighty 
 grams passed through the sixth crystallization. 
 
i6 
 
 Owing to the large amount of water used, this ought 
 to be free from impurities. Even after the third crystalli- 
 zation, it gave no test for molybdenum or phosphoric acid, 
 although the original material contained traces. It is 
 intended to convert this magnesium salt into sodium vana- 
 date and use it later. 
 
 Hydrochloric and Sulphuric Acids. 
 
 For all but the last two experiments, ordinary C. P. 
 acids were used. Since the method of generating the hydro- 
 chloric acid gas was a purification in itself, it was thought 
 that the ordinary acids would be sufficiently pure. 
 
 If the hydrochloric acid contained any hydrobromic 
 acid, the bromine would be set free and driven out by the 
 action of the chlorine. No arsenic would remain with the 
 sodium chloride in the final weighing, since it is easily 
 volatilized by heating in contact with hydrochloric acid gas. 
 Any sulphur dioxide in the sulphuric acid would be oxi- 
 dized by chlorine, and remain in the drying vessels. 
 
 Nevertheless, in the last two experiments, special 
 hydrochloric and sulphuric acids were used, containing no 
 arsenic, nitric acid, or sulphurous acid, and only a trace 
 of iron, according to the analysis of J. T. Baker. 
 
 Balance and Weights. 
 
 A Staudinger balance sensitive to o.ooooi gm., was 
 used in these experiments. It was kept in the basement, in a 
 room that was used for no other work. The balance rested 
 on a brick and stone support that was sunk directly into 
 the ground. 
 
 The weights were kindly loaned by Professor Frank- 
 lin, of the Physics Department of Lehigh University. The 
 larger weights were of brass, gold plated, and the smaller 
 weights were of platinum. They were carefully standard- 
 ized and allowance was made for the buoyancy of the air; 
 
17 
 
 the standardization was repeated several times in the course 
 of the experiments. 
 
 All weighings were by substitution, and were cor- 
 rected for the buoyancy of the air. A counterpoise was 
 always used, of the same size and material as the vessel to 
 be weighed, and the vessel was always left near the bal- 
 ance for several hours before weighing. 
 
 Dehydration of the Sodium Meta-Vanadate. 
 
 A portion was heated for several hours at 385 C. in a 
 Caries furnace in a current of dry air, and weighed. After 
 being fused and again weighed, a further loss of 0.05 per 
 cent was observed. 
 
 The substance melts at a dull red heat and on cooling 
 forms white radiating crystals, slightly hygroscopic. If 
 heated for a time at bright redness, it appears to lose weight 
 continuously, although very slowly. By fusion at dull red- 
 ness, however, it is easy to get a constant weight. 
 
 Specific Gravity of Anyhydrous Sodium Vanadate. 
 
 The specific gravity of the fused salt was determined 
 by putting a known weight of the substance into a pyknom- 
 eter, filling the latter with alcohol, and weighing it. 
 
 The weight of alcohol required to fill the pyknometer 
 was 14.593, the weight of sodium vanadate used was 5-4375, 
 the weight of the sodium vanadate and alcohol required to 
 fill the vessel was 18.4335. Therefore the weight of alcohol 
 displaced by the vanadate was 14.593 (18.4335 5.4375) = 
 i. 600. Since the specific gravity of the alcohol was 0.82, 
 the specific gravity of the sodium vanadate is 
 
 5^37x0.82 = 2.79. 
 i. 600 
 
 Corrections for Buoyancy of the Air. 
 The correction for the weight of one gram of sodium 
 vanadate is equal to the weight of i c. c. of air multiplied 
 
i8 
 
 by the difference between the volume of one gram of vana- 
 date and I gram of weights. 
 
 The temperature varied from 20 C. to 30 C. in the 
 progress of the experiments, but the pressure did not vary 
 much from 750 mm. ; the average correction was, therefore, 
 calculated for 25 C. and 750 mm. The weight of i c. c. 
 of air under those conditions is 0.00117, the volume of I 
 gram of vanadate is 0.358, and of I gram of brass, 0.119; 
 therefore the correction for I gram of vanadate is 0.00117 
 X (o-358 0.119) = + 0.00028. For i gram of sodium 
 chloride, using the accepted value 2.14 for the specific 
 gravity, the correction is 0.00117 (0.468 '0.119) = 
 +0.00041. 
 
 Preliminary Experiments on the Action of Hydrochloric 
 Acid Gas on Sodium Vanadate 
 
 Some sodium vanadate of sample A was placed in a 
 porcelain boat, in a glass tube, through which a mixture of 
 hydrochloric acid gas and chlorine was passed. The tube 
 was heated to 385 C. in a Caries furnace, but after sev- 
 eral hours the material still showed by its color the pres- 
 ence of vanadium. It was then heated in a combustion 
 furnace for an hour and a half, but still showed a yellow 
 tinge. Nevertheless, it was weighed and the amount of 
 vanadium estimated by the aniline test described later. It 
 was found by this test that 0.0013 gm. of V 2 O 6 remained. 
 The corrected weights were: 
 
 Wt. of Na V O, 0.8843 
 
 " " Na Cl 0.4241 
 
 Assuming the atomic weights of sodium and chlorine 
 to be 23.00 and 35.46, respectively, this would give 50.90 
 for the atomic weight of vanadium. 
 
19 
 
 To see if it would be possible to remove all the van- 
 adium by means of HC1 gas and chlorine, a still smaller 
 weight of sodium vanadate was used; yet after several 
 hours' heating, the residue still showed a yellow color and 
 the aniline test showed 0.003 gm. V 2 O 5 . The corrected 
 weights were: 
 
 Na VOs 0.4538 
 
 Na Cl 0.2175 
 
 This would give 50.97 for the atomic weight. 
 
 No importance is attached to these values, for the 
 quantities used were too small. 
 
 The aniline test was worked out by Witz and Osmond 
 (Bull. Soc. Chim. (2) 45,309), and is of value in estimating 
 very small quantities of vanadium. It depends on the fact 
 that vanadium salts are easily oxidized and reduced, and 
 therefore can act as oxygen carriers. 
 
 If a solution of potassium chlorate is added to a solu- 
 tion of aniline hydro-chloride, heated to boiling and allowed 
 to stand at room temperature, no coloration appears. If, 
 however, a trace of vanadium is present, a color appears, 
 varying from light yellow to black, according to the amount 
 of vanadium present; more than a trace of vanadium will 
 cause a precipitate of aniline black. The sensitiveness is 
 increased by the presence of a small quantity of dilute HC1, 
 but a larger quantity will cause a precipitation of aniline 
 black, even when no vanadium is present 
 
 In making this test, a standard vanadium solution was 
 used, containing about o.oooi gm. V 2 O 5 to I c. c. ; this 
 was used for color comparison, as in the Nessler test for 
 ammonia, and in neutral solution was about as delicate. 
 
 Since the vanadium could not all be removed by treat- 
 ment in a porcelain boat, some trials were made with a Jena 
 glass bulb of the form shown in Fig. 2, a form used by T. 
 W. Richards in some of his work. 
 
20 
 
 G is the bulb containing the vanadium. H, H are 
 extra bulbs to catch any material that may spatter out of G. 
 I, I are flasks containing water, through which the acid 
 fumes and volatile vanadium compounds are drawn by 
 suction. At J the apparatus is connected to the generator 
 at the point marked J in Fig I. A counterpoise of the same 
 size and shape as the bulb G was used, and the difference 
 in weight was determined by substitution. 
 
 The sodium vanadate was fused in a platinum capsule, 
 then powdered and transferred to the bulb, which was held 
 
 FIG. 2 
 
 slightly inclined. After weighing the bulb and vanadate, 
 it was heated in a current of hydrochloric acid gas and 
 chlorine. The heat was furnished by Bunsen burners placed 
 the proper distance above and below the bulb ; a small flame 
 was also placed under the exit tube and thus any crawling 
 of material prevented. 
 
 When as much as possible of the vanadium had been 
 removed, a little water was added to the substance in the 
 bulb and it was again dried in the stream of gas, being 
 rotated at the last, so as to spread the material over the 
 inner surface of the vessel. In this process the sodium 
 chloride first dissolved in the water, leaving the vanadium ; 
 
21 
 
 then, as the HC1 was passed in, the vanadium dissolved 
 and the sodium chloride crystallized out. As the liquid 
 evaporated the vanadium compound was left at the surface, 
 where it could be attacked by the gases. The process was 
 thus a combination of crystallization and distillation. 
 
 After three or four of these treatments, the vanadium 
 could all be driven out of more than five grams of vanadate, 
 leaving a perfectly white residue, which gave no indication 
 of vanadium by the aniline test. 
 
 After displacing the hydrochloric acid gas by air, the 
 sodium chloride was heated nearly to fusion, and this 'was 
 repeated until the weight was constant. 
 
 It was found, however, that there are two disadvan- 
 tages to this form of apparatus. The fused sodium vana- 
 date, which is slightly hygroscopic, takes up a little mois- 
 ture while it is being pulverized and transferred to the bulb. 
 It does not give up all of this moisture even after standing 
 over night in a desiccator; experiment showed that seven 
 grams of vanadate retained 0.0025 gram of moisture or 
 0.035 per cent. But the most serious disadvantage was 
 that the sodium chloride decrepitated, when heated, and a 
 small quantity was thrown out of the vessel into the extra 
 bulbs H. 
 
 Nevertheless, three experiments were made with this 
 form of apparatus, using material of sample A. In each 
 case 0.035 P er cent - was subtracted from the observed 
 weight of sodium vanadate to make allowance for moisture, 
 and any sodium chloride thrown out by decrepitation was 
 weighed and its weight added to that of the main quantity. 
 The results are given in the following table: 
 
 Weight, 
 NaV0 3 . 
 
 Weight, 
 NaCl. 
 
 Molec. Wt. 
 NaVO 3 . 
 
 Atomic Wt., 
 Vanadium. 
 
 (I) 4-8850 
 (2) 5-5793 
 (3) 5.6975 
 
 2-3393 
 2.6691 
 2.7244 
 
 I22.O8 
 I22.2O 
 122.23 
 
 51.08 
 51.20 
 51.23 
 
 Average 51.17 
 
22 
 
 Not much importance, however, is attached to this 
 value on account of the uncertainty of the corrections. So, 
 in order to avoid these sources of error, an apparatus of the 
 form shown in Fig. 3 was devised, and it finally proved sat- 
 isfactory. 
 
 K is the flask containing the sodium vanadate. The 
 small bulbs on the inner and outer vessels prevent loss by 
 decrepitation; and, by means of a platinum wire fastened 
 around it at L, the flask may be suspended vertically for 
 
 FIG. 3. 
 
 weighing. M is a quartz dish to support the flask while 
 being heated. N is a hard glass tube used as a connector. 
 
 In order to get the weight of the sodium vanadate, it 
 was fused in a platinum capsule and rotated just before 
 solidifying, so as to spread the substance in a thin layer. 
 The total weight of capsule, vanadate and a platinum rod 
 was then taken. By means of the platinum rod, the vana- 
 date was then broken up and as much as possible trans- 
 ferred to the flask, which was held vertically. After again 
 heating the capsule and rod, the loss of weight was obtained, 
 
23 
 
 which was the weight of the sodium vanadate transferred 
 to the flask. 
 
 The heating with hydrochloric acid gas and chlorine 
 was the same as in the experiments with the glass bulb. 
 The quartz dish; M, was heated gently by a Bunsen burner, 
 and two other burners were arranged above the vessel, so 
 as to radiate sufficient heat downward. After about four 
 treatments with the gas, followed each time by adding a 
 little water and again drying in the gas, pure sodium 
 chloride was left. The white color of this substance proved 
 to be a very good indication of its freedom from vanadium. 
 The time required to get all the vanadium out was about 
 thirty hours of actual heating, and seemed to be inde- 
 pendent of the quantity of vanadate used. 
 
 An attempt was made to use a Jena glass flask, but it 
 cracked at a critical moment, so a quartz flask with straight 
 neck and Jena gjlass inner tube was next tried. Only one 
 experiment was completed, because the inner tube fused 
 in contact with the quartz, when the sodium chloride was 
 finally heated to fusion. The results of this experiment are 
 recorded under (i) in the Table on page (26). The 
 material used was sample A. 
 
 It was necessary to send to Germany and have two 
 quartz flasks made of the form shown in Fig. 3, with both 
 inner and outer tubes of quartz. For this reason the work 
 had to be left for a year. 
 
 An advantage of the use of the quartz apparatus is 
 that the sodium chloride could be heated to complete fusion 
 in the vessel and all moisture thus removed. It was found 
 that the fusion caused practically no loss by volatilization, 
 and a quite definite weight for the salt could be obtained. 
 The losses on heating were about as follows : The flask and 
 salt were weighed after fusion of the salt on the edges, then 
 after complete fusion, there would be a loss of 0.0012 gm. 
 
24 
 
 After again fusing it for several minutes to the highest 
 temperature of the blast lamp, there would be a further 
 loss of o.oooi, or sometimes 0.0002 gm. The last weight 
 was taken as correct. The bulb on the inner tube nearly 
 closes the outlet to the flask, when it is held vertically, and 
 this evidently prevents loss by volatilization. 
 
 Another quartz apparatus of the same size and shape 
 was used as a counterpoise, the difference in weight by sub- 
 stitution of the flask and counterpoise being taken. On the 
 opposite pan of the balance a quartz dish of the same weight 
 was used as a tare. It was necessary to leave the apparatus 
 in the balance case for several hours before the final weigh- 
 ing was taken. On being transferred from desiccator to 
 balance case, the flask seemed to lose about aooi gm. in 
 weight in about an hour; after that the weight would be 
 practically constant. The loss was undoubtedly due to the 
 fact that the flasks were filled first with the dry air of the 
 desiccator and then with the moister air of the balance case. 
 
 After getting the final weight of the flask and fused 
 sodium chloride, the flask was rinsed out with distilled 
 water, heated with a blast lamp and again weighed. The 
 difference in weight is recorded in the Table as "loss on 
 rinsing/' 
 
 In experiment II, there was a slight increase in weight 
 of the flask, and at the same time a slight greenish yellow 
 coloration appeared in the neck of the flask. The yellow 
 color then remained through all the experiments, and the 
 weight of the flask was practically constant, except in 
 experiment IV, where there was a considerable gain in 
 weight of the flask, and an increase in the yellowish color 
 on the neck. The deeper color persisted until the inner sur- 
 face had been rubbed with a cloth moistened with various 
 reagents, such as acids, ammonia and caustic potash ; this 
 treatment caused the color to become lighter and brought 
 
25 
 
 the weight back to normal. On heating the neck of the 
 flask with the blast lamp, the yellow color would travel 
 around to the opposite side from the flame, but could not 
 be driven away. Our explanation is that some vanadium 
 vapors were taken up by the quartz in the neck of the appa- 
 ratus, where the heat was much less than in the body of the 
 flask. By taking the loss in weight on rinsing, this would 
 not cause any error in the results. 
 
 The solution obtained by rinsing out the vessel showed 
 a very slight turbidity which looked like silica. In one 
 case, a few shining scales could be noticed, which evidently 
 came from the quartz flask. This was undoubtedly due to 
 the contraction of the fused sodium chloride on cooling; 
 although it was spread as thin as possible, it always broke 
 apart with considerable crackling. The solutions obtained 
 in the last four experiments were filtered through a small 
 filter whose ash weighed 0.00003 gm. After ignition in a 
 platinum crucible, the weight of the silica was obtained. 
 
 Just how to apply this correction was somewhat of a 
 puzzle. If the silica came from the quartz vessel, its loss 
 was evidently balanced by other materials absorbed ; for the 
 weight of the vessel did not decrease, and yet in one case 
 the corrosion of its walls was observed after rinsing. It 
 was finally decided to assume that the silica came from the 
 vessel, and subtract its weight from the "loss on rinsing." 
 If part of it came from the sodium vanadate, our method 
 of correction would not be far wrong, since the silica would 
 affect the observed weight of sodium chloride over twice as 
 much as that of the vanadate. 
 
 In practically all cases the filtrate from the silica was 
 tested for vanadium by the aniline test, and any trace thus 
 indicated was subtracted from the loss on rinsing. 
 
26 
 
 The results of five experiments with quartz appa- 
 ratus are given in the Table in the order in which they were 
 obtained. 
 
 Date 
 
 1909 
 
 A 
 Colorless 
 
 4-8550 
 
 O.OOI4 
 4-8564 
 re 0.8690 
 
 3.1876 
 
 1910 
 
 B 
 
 Nearly 
 colorless 
 
 5.6388 
 
 0.0016 
 5.6404 
 2.1286 
 2.1282 
 
 0.5746 
 
 2.7028 
 0.0006 
 
 None. 
 2.7022 
 
 O.OOII 
 
 2.7033 
 121.976 
 50.976 
 
 1910 
 B 
 
 Nearly 
 colorless 
 
 4.4251 
 
 0.0012 
 4-4263 
 2.1282 
 2.1283 
 
 0.0062 
 
 2. 122 1 
 0.00094 
 
 0.00006 
 2.I2II 
 
 O.OOO9 
 2.1220 
 121.946 
 50.946 
 
 1910 
 E 
 Nearly 
 colorless 
 
 57789 
 
 0.0016 
 57805 
 2.1283 
 2.1257 
 
 0.6450 
 
 2.7707 
 0.0008 
 
 0.00005 
 2.7699 
 
 O.OOII 
 
 2.7710 
 121.952 
 50.952 
 
 1910 
 F 
 
 Colorless 
 9.4875 
 
 O.O027 
 94902 
 2.1283 
 2.1282 
 
 2.4187 
 
 4.5469 
 O.OO09 
 
 O.OOOO5 
 4.5460 
 
 O.OOlS 
 4.5478 
 121.097 
 50.097 
 
 Sample 
 
 Color of fused 
 vanadate 
 
 Wt. of NaVOs 
 in air 
 
 Correction for 
 buoyancy of 
 air 
 
 Wt. of NaVO, 
 in vacuum . . . 
 Excess wt. " 
 of counter- befo 
 poise over 
 flask . aftei 
 
 Excess wt. of 
 flask and Na Cl 
 over counter- 
 poise 
 
 Loss of Wt. on 
 rinsing 
 
 Wt. of silica . . . 
 
 
 Wt. of vana- 
 dium left 
 Wt. of NaCl 
 in air 
 Correction for 
 buoyancy of 
 air .... 
 
 None. 
 2.3267 
 
 O.OOIO 
 
 2.3277 
 121.966 
 50.066 
 
 Wt. of NaCl 
 in vacuum . . . 
 Mol. Wt. of 
 Na VO-, 
 
 Atomic Wt. of 
 vanadium . . . 
 
27 
 
 The average is 50.967 0.006. The atomic weights 
 of sodium and chlorine were assumed to be 23.00 and 35.46, 
 respectively. 
 
 The sample used in V had been the most carefully pre- 
 pared of any in the series, and the amount of material used 
 was so large that the percentage error in the weighing of the 
 sodium chloride would be considerably diminished. So, 
 perhaps, this should be given more weight in making up 
 the average. 
 
 The experiments had to be interrupted at this point, 
 but it is intended to make another series at some future 
 time, using from seven to ten grams of vanadate to each 
 experiment. 
 
CONCLUSION. 
 
 This series of fairly concordant results, obtained with 
 material prepared in several different ways, proves that 
 sodium meta-vanadate is a definite and stable compound 
 and that it can be prepared free from excess of alkali and 
 from acid vanadates. 
 
 The errors which would make our calculated value for 
 the atomic weight too high are : Presence of acid vanadate, 
 silica or moisture in the fused sodium vanadate, and loss 
 of sodium chloride by decrepitation or volatilization. 
 
 The errors which would make the calculated value too 
 low are : Presence of excess of alkali in the fused vanadate, 
 and presence of moisture in the sodium chloride. 
 
 It is hard to see how our calculated value can be too 
 low, since correction has been applied for silica, and for any 
 vanadium left in the sodium chloride, and the moisture must 
 have been completely removed. 
 
 This work in connection with that of Prandtl and 
 Bleyer therefore indicates that the value for the atomic 
 weight of vanadium is about 51.0 instead of 51.2. 
 
r.M\ KRSITY OF CALIFORNIA LIBRARY, 
 BERKELEY 
 
 THIS BOOK IS DUE ON THE LAST DATE 
 STAMPED BELOW 
 
 Books not returned on time are subject to a fine of 
 .">(>< per volume after the third day overdue, increasing 
 to $1.00 per volume after the sixth day. Books not in 
 demand may !><> removed if application is made before 
 expiration of loan period. 
 
 4 
 
 27Ma/53Yl 
 
 3 1953 Q 
 
 20m-l,'2'J