COLUMBATLS. THLSIS PRESENTED TO THE. FACULTY OF THE. DEPARTMENT OF PHILOSOPHY OF THE. UNIVERSITY OF PENNSYLVANIA IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY BY MATTHEW HUME BEDFORD, A.B., CENTRAL UNIVERSITY, KENTUCKY. ESCHENBACH PRINTING COMPANY EASTON, PA. 1905 Digitized by the Internet Archive in 2017 with funding from University of Illinois Urbana-Champaign Alternates https://archive.org/details/columbatesOObedf tr it. w'. COLUMBATLS INTRODUCTION. This thesis represents but one chapter in a series of investiga¬ tions of columbium and tantalum which have been carried out in the past and are still being continued in this laboratory. It presents the results obtained in the preparation of sodium col- umbate by several methods and several other columbates derived therefrom. historical. In making a review of the literature on the columbates, it was noticed that each method of preparation gave salts to which different formulas were assigned. H. Rose 1 prepared a sodium columbate by the fusion of columbic oxide with sodium hydroxide in a silver crucible. He assigned to this salt the formula Na 2 0.Cb 2 0 5 .5H 2 0 and also obtained one with 7H 2 0. After Marignac had shown that Rose’s “Unterniob- saure,” Cb 2 0 3 , was really Cb 2 0 5 , Rammelsberg 2 recalculated Rose’s results and gave these salts the formula Na 2 O.Cb 2 0 5 .6H 2 0 and 9H 2 0. Rose said, “of all the salts of tantalic and columbic acids, this sodium salt is obtained in the best crystalline form, Xand it is also the most stable.” He also states that in the preparation of the neutral salt he obtained a basic salt, to which was assigned the composition 3Na 2 0.2Cb 2 0 5 .24H 2 0 3 . He gave the following as the sodium columbates that he found well defined: (1) 3Na 2 0.Cb 2 0 5 , obtained by long fusion of the oxide with sodium carbonate. (2) 3Na 2 0.2Cb 2 0 5 .24H 2 0. (3) Na 2 0.Cb 2 0 5 .6H 2 0 and 9H 2 0, the neutral salt. (4) 4Na 2 0.5Cb 2 0 5 , the only acid salt of ex:.Cc ratio, the other 1 Pogg. Ann., 113, 105. 2 Jour, fur Chem., 108, 77 (1869). 3 Rammelsberg : Jour, fur Chem., 108, 77 (1869). „15746 4 acid salts being mixtures (formulas from Rammelsberg’s recalcula- Hermann 1 prepared a sodium columbate to which he assigned the composition 5 Nap.4Cb 2 0,2iH 2 0. This salt was made by adding sodium hydroxide to a solution of potassium columbium oxyfluoride. In his paper he called attention to the great differ¬ ence between the found and calculated percentages in H. Rose s Marignac 3 obtained a crystalline potassium columbate, to which he assigned the formula, 4K2O.3CbA.4Hp +1zAq, but Rose and Hermann were unable to obtain a crystalline salt. T is was obtained by fusing columbic oxide with potassium carbonate. On recrystallization this salt approached the composition 8K 2 0.7Cb 2 0 5 .9H 2 0 + 3 2Aq. It had a tendency to form super- saturated solutions. He also described three other salts of the following composition: 3 K 2 0.2Cb 2 0 5 .6H 2 0 + 7 Aq. 3 K 2 0 .Na 2 0 . 3 Cb 2 0 5 . 9 H 2 0 . K 2 0 . 3 Cb A- 5 H 2 0 . He said “Columbic acid with sodium formed only powdery crystalline’salts, which suffered decomposition on washing with W Santessen 3 described a salt to which he assigned the composition K,0 2Cb 2 0 s .5iH 2 0. It was prepared by fusing columbic oxide with the calculated amount of potassium carbonate, and was in¬ soluble in water. When a large excess of potassium carbonate ™ L, h. obtained the «K **>■<%>,"**> “ * after extracting the melt with water. The salt HaP^bAd 2 was found to be soluble in water while the salt 2Nap.3CbA.9Hp, obtained by fusing the above salt with sodium hydroxide, was “ tw described four series of salts of the general composition RO Cb O 2R0.Cb 2 0 5 , 3 R 0 .Cb 2 0 5 and 4R0.Cb 2 0 5 . These salts were prepared by heating Cb 2 0 5 with a metallic chloride at a tem¬ perature somewhat below that at which the chloride volatilized ■ Jour, fur Chem., Ill, 373 (1871)- 2 Ibid., 97 * 449 (1866). 3 Bull. Soc. Chim., (2) 24, 52. ^ C. R., 81» 266, 1266. 5 Larsson , 1 in order to obtain columbates of a constant com¬ position, tried to crystallize the precipitated amorphous colum¬ bates by fusing them with the chloride of the corresponding metal or by crystallizing them from fused boron trioxide. He obtained the following salts. Those under I were obtained by heating the amorphous columbates with the metallic chlorides, and those under II by heating them with boron trioxide. The amorphous precipitates used in the fusions were obtained from potassium columbate, but he did not assign a formula to the latter. I 4Mg0.Cb 2 0 5 2 Ca 0 .Cb 2 0 5 Y 2 0 3 .Cb 2 0 5 II Mg0.Cb 2 0 5 Ca0.Cb 2 0 5 Y 2 0 3 .3Cb 2 0 5 CuO.Cb 2 O s Zn0.Cb 2 0 5 Cd0.Cb 2 0 5 3Mn0.5Cb 2 0 5 5Th0 2 .i6Cb 2 0 5 Zr 0 2 . 5 Cb 2 0 5 A number of other metals were tried but their ratios were as irregular as the last three salts under II. Melikoff and Pissarjewsky 2 describe the preparation of per- columbic acid, HCb0 4 + nH 2 0 and potassium percolumbate, K 4 Cb 2 O n + 3H 2 0. SOURCE OF CORUMBIUM. The columbic oxide used in this investigation had been carefully purified by Dr. R. D. Hall. It was prepared from the columbite of South Dakota. The oxide was free from tin, tungsten, and tantalum, but it contained a trace of titanium. EXPERIMENTAL PART. Preparation of the sodium salt by Hermann’s method. Fifty grams of potassium columbium oxyfluoride were dissolved in i litre of hot water to which was added a concentrated solution of 150 grams of sodium hydroxide. A white powdery precipitate was formed, and, after cooling, the supernatant liquid should be 1 Zeit. fiir anorg. Chem., 12, 188. 2 Ibid., 20, 340. 6 free from columbium. If this was not the case, more sodium hydrate was added. The precipitate was filtered out and washed with cold water until the filtrate became milky. This showed that the excess of sodium hydroxide had been washed out. It was then dried as far as possible by suction, and was recrystallized three times from distilled water. After the third crystallization it was allowed to dry in the air. This salt dissolved in water without leaving any residue and was very stable. Beautiful crystals could be obtained when the solution was allowed to cool slowly. It was found that it was best to cool rapidly and stir well as it formed supersaturated solutions very readily. method of analysis. At first it was thought that it would be necessary to get the water by ignition with tungstic oxide because Rose said that when sodium columbate was ignited, sodium hyroxide and columbic oxide remained. But it was found that the loss on ignition of the salt alone was the same as the loss on ignition of the mixture of the salt and tungstic oxide. The ignited sample was fused with 4-5 parts of potassium bisulphate and, after cooling, the melt was boiled up with water until completely disintegrated. The col¬ umbic oxide was filtered out and thoroughly washed, ignited, and weighed. The sodium was obtained by dissolving a sample of the i salt in 100 cc. of water and adding 5-10 drops of dilute sulphuric acid. After boiling a few minutes the precipitated columbium hydrate was filtered, washed with boiling water, the precipitate ignited, and weighed as columbic oxide. The filtrate was evaporated and the sodium weighed as sulphate after ignition. A trace of columbic oxide sometimes remained with the sodium sulphate which was removed by taking the sulphate up with water and filtering. After evaporation the sodium sulphate was again weighed. A. 0.3352 gram sample gave 0.0736 gram loss on ignition equal to 21.96 per cent. H 2 0 , and 0.2063 gram Cb 2 O s equal to 61.54 per cent. B. 0.2636 gram gave 0.0582 gram loss on ignition equal to 22.08 per cent. H 2 0 , and 0.1625 gram Cb 2 0 5 equal to 61.65 per cent. 7 C. 0.5058 gram gave 0.1922 gram Na 2 S 0 4 equal to 0.0839 gram Na 2 0 equal to 16.59 P er cent. D. 0.5254 gram gave 0.1993 gram Na 2 S 0 4 equal to 0.0870 gram Na 2 0 equal to 16.57 P er cent. Calculated Percentage found, percentage. /-*-■ 7Na 2 0. 434 16.58 16.59 16.57 6 Cb 2 0 6 . 1608 61.42 61.54 61.64 32H 2 0. 576 22.00 21.96 22.08 2618 100.00 100.09 100.29 H. Rose 1 assigned to his salt the formula Na 2 0 .Cb 2 0 5 . 6 H 2 0 and this seems to have been the one generally accepted. It was therefore thought at first that the salt, obtained by the method of Hermann as above, contained an extra Na 2 0 which would be eliminated by precipitating an aqueous solution of the salt by alcohol. This was done and the following analyses of an air-dried sample show that the composition of the salt had not altered, except in its water content. The same method of analysis was used as in the case of the preceding salt. A. 0.2252 gram sample gave 0.1394 gram of Cb 2 0 6 equal to 61.90 per cent., and 0.0861 gram Na 2 S 0 4 equal to 0.0379 gram Na 2 0 equal to 16.43 per cent. 0.2062 gram sample lost 0.0448 gram on ignition equal to 21.73 P er cent. H 2 0 . B. 0.2236 gram sample gave 0.1392 gram Cb 2 O s equal to 62.24 per cent., and 0.0847 gram Na 2 S 0 4 equal to 0.0372 gram Na 2 0 equal to 16.62 per cent. 0.3499 gram sample lost on ignition 0.0761 gram equal to 21.46 H 2 0 . Calculated Percentage found. Percentage. .-*-, 7Na z O. 434 16.69 16.43 16.62 6 Cb 2 0 6 . 1608 61.85 61.90 62.24 3iH 2 0. 558 21.46 21.73 21.46 2600 100.00 100.06 100.32 BARIUM COLUMBATE, 7 Ba 0 . 6 Cb 2 0 5 .i 8 H 2 0 . The sodium salt prepared by Hermann’s method was dissolved in hot water and a dilute solution of barium chloride was added. 1 “Rammelsberg’s Recalculation,” Jour, fur Chem., 108, 77 (1869). 8 A heavy white precipitate was formed which settled very quickly. This precipitate was washed by decantation with boiling water until free from chlorides. It was then collected on a filter and dried on a water-bath for one and a half hours. It was ground for analysis. The color remained white. method of analysis. The sample was ignited in a platinum crucible and the loss was water. This ignited residue was fused with a flux consisting of equal parts of sodium and potassium carbonates and after allow¬ ing the melt to cool, it was taken up in a dilute solution of oxalic acid. After acidifying with hydrochloric acid everything went into solution on warming. When all of the barium oxalate had dissolved, the solution was brought to boiling and the barium precipitated as sulphate with sulphuric acid. The barium sul¬ phate was filtered out and treated as usual. The columbium was obtained in the filtrate from the barium sulphate by pre¬ cipitation with ammonium hydroxide. At times this precipitate was very difficult to filter. A. 0.2222 gram sample lost 0.0240 gram on ignition equal to 10.80 per cent., H 2 0 , and gave 0.1213 gram BaS 0 4 equal to 0.0796 gram BaO equal to 35.83 per cent., and Cb 2 0 5 in this by difference 53.37 per cent. B. 0.5281 gram sample gave 0.2902 gram BaS 0 4 equal to 0.1905 gram BaO equal to 36.08 per cent., and gave 0.2801 gram Cb 2 0 5 equal to 53.06 per cent. 0.5208 gram sample lost 0.0560 gram on ignition equal to 10.75 per cent., H 2 0 . This gives the following: Calculated Percentage found, percentage. ,-*-* Mean. 7®aO. 1071 35.66 35.83 36.08 35.96 6 Cb 2 0 6 . 1608 53.55 53.37 53.06 53.22 i 8 H 2 0 . 324 10.79 10.80 10.75 10.78 3003 100.00 100.00 9989 9996 SILVER COLUMBATE, 7 Ag 2 0 . 6 Cb 2 0 5 . 5 H 2 0 This salt was obtained by adding a solution of silver nitrate to a solution of the sodium columbate. It separated as a white pre¬ cipitate. It was washed with boiling water by decantation until i 9 free from silver and then collected on a filter. It was dried on a water-bath for an hour. It was ground to a powder for analysis. It had a slight yellow color. method of analysis. It was thought at first that the loss on ignition would be the water and the oxygen of the silver oxide. But it was found that the loss on ignition was not equivalent to the oxygen from the silver oxide. Nitric acid would not dissolve all of the silver from the ignited residue, so a sample was ignited in a current of dry air and the water was collected in a calcium chloride tube and weighed For the determination of the silver and columbium a sample was fused with potassium bisulphate. The melt was taken up with oiling water and the columbium determined as usual. The silver in the filtrate from the columbium hydrate was titrated with standard potassium sulphocyanide. 4 . 0.3002 gram sample gave 0.1470 gram Cb 2 O s equal to 4807 per cent. 4 w B. 0.3051 gram sample gave 0.1487 gram Cb 2 0 6 equal to 48 75 per cent. ^ ' /0 '• 0,2682 gram S f m P ,e S ave ° I3I2 gram Cb 2 0 6 equal to 4893 per cent, and o. 1306 gram of Ag 2 0 equal to 48.70 per cent 1 0 2818 gram sam P le gave 0.0071 gram of water equal to 2 52 per cent., and o. 1382 gram Cb 2 0 5 equal to 49.05 per cent., and 0.1374 gram Ag 2 0 equal to 48.74 per cent. Calculated Percentage found. 7 Ag- O ^ percentage. Mean. . 2t V 9 5h 2 o. 5 .. 16,8 4,4 ° y 2 - 7 i 2.52 v> lit . ZZ 22 100.00 100.17 ammelsberg ascribed to H. Rose’s silver columbate the follow- ig composition: Calculated Percentage found . 376 cb A . pe ?;r (H ; Eose,) 12O . I92 4558 47-31 Ac l.. • 18 _ __ _ r -77 1.98 1 T ».. 1018 100.00 no Jour, fur Chem., 108, 77 (1869). 99 74 IO It is described as having a slight yellow color on precipitation and as turning black on drying at 100°. zinc columbate, 7Zn0.6Cb 2 0 5 .25H 2 0. This salt was prepared by adding a solution of zinc sulphate to a solution of the sodium columbate. A white precipitateseparaJed Tattled auicklv It was washed by decantation with boiling “£» sulphate and « »» «*• « “ ‘ This precipitate wa, dried on a water-bath and ground to a A fnr nrifllvsis It remained white after drying. Tt might be mentioned here that the water content may not al¬ ways remain the same when the salt is dried as indicated above method of analysis. The sample was ignited in a platinum crucible, the loss repre senting the water. This ignited sample was fused with potassiun sen ting _ determined as usual. Th< renting the water. This ignnea bcunpi^ * bisulphate and the columbic oxide determined as usua . zinc was obtained by difference. A. 0.3018 gram sample lost 0.0520 gram on ignition equal 17.21 per cent. H 2 0 . *• <0 poo ^018 gram sample iubt &- - per cent. H 2 0 and gave 0.1842 gram Cb 2 0 6 equal to 61.0 P er cent - . rb __or cent -7t,0 . Calculated percentage. 21.69 APh O . . . . . 1608 61.18 2 cHoO . 17-13 2628 100.00 Percentage found. Mean. 21.86 60.87 17-27 99-99 In order to show that any other formula, that wouldapproa this composition, will not give the “ rreCt «“ given.' culated percentages for the 1 :1 and 8 . 7 ratios are g Percentage. Percentage. ZnO. *9-3* Cb 2 O t . 6 3-59 4 H 2 0. *7°9 Percentage. _ _ 21.23 7Cb 2 O s . 61 16 30II/.). 17.61 8 ZnO 100.00 100.00 Since the salts which have been described show conclusively that they were not of the RO.Cb 2 0 5 and probably not of the 8 R 0 . 7 Cb 2 0 type, it was thought that sodium columbate should be prepared by the method of Rose. PREPARATION OF THE SAET BY ROSE’S METHOD. Ten grams of the ignited columbic oxide were fused with 40 grams of sodium hydroxide in a silver crucible. The excess of sodium hydroxide was washed out and the sodium columbate was filtered by suction. It was then twice recrystallized from water and dried in the air. Very beautiful crystals of this salt were obtained and they were completely soluble in water. They were slightly dark in color, due to a small quantity of silver oxide which could not be removed by filtration from their solution. method of analysis. The same method of analysis was employed as that described for the sodium salt prepared by Hermann’s method. A. 0.3540 gram sample lost 0.0777 gram on ignition equal to 21.95 per cent. H 2 0 , and gave 0.2175 gram Cb 2 O s equal to 61.45 per cent. B. 0.3362 gram sample lost 0.0740 gram on ignition equal to 22.01 per cent., H 2 0 and gave 0.2066 gram Cb 2 0 6 equal to 61.46 per cent. C. 0.4698 sample gave 0.2883 gram Cb 2 0 5 equal to 61.38 per cent., and gave 0.1792 gram Na 2 S 0 4 equal to 0.0783 gram Na 2 0 equal to 16.65 P er cent. 7Na 2 0. 5 Cb 2 0 6 i 32 H 2 0 . Percentage calculated. --Percentage found.-, A. B. C. 434 1608 576 16.58 61.42 22.00 61.45 61.46 61.38 21.95 22.01 . 2618 100.00 Mean. 16.65 61.43 21.98 IOO.06 H. Rose 1 gave the following data for the salt prepared in this nanner. 1 Recalculated by Rammelsberg, Jour, fur Chem., 108, 77 (1869). 12 Theory. Found. Na 2 0. . 62 14.16 15-68 Cb 2 0 5 . . 268 61.20 60.82 6H 2 0. . 108 24.64 438 100.00 The sodium salt was also prepared by fusing 8 grams of columbic oxide with 25 grams of sodium carbonate in a platinum crucible. The melt was taken up in water and the excess of sodium car¬ bonate washed out and the sodium columbate filtered by suction. This was recrystallized from water. There was such a small quan¬ tity of the salt that it could not be crystallized well because of its tendency to form a supersaturated solution. However, on adding a small quantity of a solution of sodium carbonate, it came out slowly as a fine powder. This was washed with a little cold water, and allowed to dry in the air. This salt was completely soluble in water. method of analysis. The method of analysis used for this salt was the same as that described for the sodium salt prepared by the other two methods. Percentage Percentage found, calculated. Mean. 7Na 2 0‘. 434 16.13 15.84 6Cb 2 O s . 1608 59-78 60.20 36H 2 0. 648 24.09 24.16 2690 100.00 100.20 CONCLUSIONS. Sodium columbate was prepared by three different methods. The results obtained showed that in each case the salts had the same proportion of sodium oxide to columbic oxide. The one pre¬ pared by fusing columbic oxide with sodium carbonate gave a salt with four molecules of water more than the other two. But this may be explained by its being precipitated with sodium carbonate. Sodium oxide could not be dissolved out with alcohol to form a salt of the ratio Na 2 O.Cb 2 0 6 . The 7 :6 ratio held throughout the barium, silver, and zinc salts. The barium and zinc salts did not give the percentages required by an 8 : 7 ratio, although the re- i i3 quirements of the two ratios were fairly well fulfilled by the results obtained for most of the other salts. Rammelsberg allowed too large an error in recalculating Rose’s work, allowing as much as 3.1 per cent, for columbic oxide, and 3.41 per cent, for mercury in his mercurous columbate. Some of Rose’s results come very close for a 7 : 6 ratio. SEPARATION OF COLUMBIC AND TUNGSTIC ACIDS. This study arose from the great difficulty that was experienced in trying to separate columbic and tungstic oxides in a salt ob¬ tained by dissolving moist tungstic oxide in a boiling solution of sodium columbate. The thought, at that time, was to put col¬ umbic oxide in with tungstic oxide to form a columbo-tungstate analogous to the phospho-tungstates and vanado-tungstates. A salt was obtained that crystallized in the same form as the sodium columbate and contained about 22.50 per cent, of tungstic oxide and 43.3 per cent, of columbic oxide. This salt could be recrystal¬ lized without undergoing any decomposition, as the following will show: I. Sample from first crystallization gave 15.64 Na 2 0 and 84.35 Cb 2 0 5 +W 0 3 . II. Sample from second crystallization gave 15.74 Na 2 0 and 84.24 Cb 2 0 5 +W 0 3 . The weight of the water was subtracted from the weight of the sample before the calculations were made, but on trying to pre¬ pare this salt again it was found that a salt of a slightly different composition was obtained. It seems that with further study a columbo-tungstate can be made. In the separation of the columbic and tungstic acids in this salt concordant results were not obtained by the usual method. The first method tried was the fusion of the sample with a mixture of sodium carbonate and sulphur. It will be sufficient to state here that the results varied within wide limits, even on mixtures of weighed quantities of columbic and tungstic oxides. Digestion of the ignited sample with a strong solution of sodium hydroxide was next tried, but did not give satisfactory results. A short review of the methods that have been used to get columbium and tantalum free from tin and tungsten will be given here. Berzelius 1 digested the oxides with yellow ammonium sulphide to remove the tin and tungsten. Wohler 2 said that it was sufficient to wash the oxides with ammonium sulphide on the filter. He also stated that it would be safer to fuse the oxides with three times their weight of an alkaline carbonate and sulphur, and then wash with ammonium sulphide. H. Rose 3 fused the oxides with sodium carbonate and sulphur and washed with ammonium sulphide. Blomstrand 4 used the method of Berzelius instead of Rose’s. He said that fusion with sodium carbonate and sulphur did not give good results. If heated too high some columbate was formed and would go into solution along with the sulpho-tungstate and sulpho-stannate. And if it was not heated high enough the action would not be complete. Hall 5 found that digesting even four or five times with am¬ monium sulphide did not remove all of the tin and tungsten. He then had recourse to Rose’s method. After a second fusion with sodium carbonate and sulphur he could not get a test for tin or tungsten. This proved the best course to pursue in separating columbium from tungsten and tin. But for a quantitive separation it does not give concordant re¬ sults. The reasons advanced by Blomstrand seem to give the best explanation. Therefore, another method for the quantitative separation of columbium and tungsten was sought. One of the thoughts that suggested itself was precipitation of the columbic oxide with a magnesia mixture in an alkaline solution. This is the usual method of separating phosphoric and tungstic acids. It was soon found that the magnesium columbate could not be weighed as such. The method was tried on weighed quantities of columbic and tungstic oxides. 1 Pogg. Ann., 4, 6. 2 “Mineral Analyse,” p. 140. 3 Handbuch der Anal. Cliem., 2, 336. 4 Jour, fur Chem., 99, 40. 5 Jour. Am. Chem. Soc., 26, 1235. 15 procedure. Weighed quantities of the two oxides were fused with potassium carbonate (because the potassium salts are more soluble than the sodium). The melt was taken up in 150 cc. of water. To the potassium carbonate solution an excess of a magnesia mixture was added. The precipitate formed was allowed to stand several hours and then filtered. It was washed on the filter five or six times with the precipitant, dried, and ignited. After all of the carbon had burned off, the residue was fused with potassium bisulphate, boiled up with water, filtered, washed, ignited and the columbic oxide weighed. The tungstic oxide was obtained by difference. No. of Cb 2 0 5 W0 3 Cb 2 0 5 W0 3 sample. taken. taken. found. by difference. 1 . 0.2086 0.0986 0.2078 O.0994 2 . 0.3218 0.1500 0.3208 0.1510 3 . 0.5004 0.1000 0.4951 0.1053 4 . 0.2026 O.IOOO 0.2032 0.0994 5 . 0.1503 0.1200 0.1506 0.1197 When this method was applied to a solution containing un¬ known quantities of columbic and tungstic acids, a solution of mercurous nitrate was added, and then a slight excess of nitric acid. Freshly precipitated mercuric oxide was then added and the solution was boiled five minutes. As soon as the precipitate had settled, it was filtered out and washed with boiling water. It was dried, ignited, and the columbic and tungstic oxides weighed together. The two oxides were then separated by the method described above. Very fair results were obtained. They varied 0.3-0.5 per cent, from the mean. CONCLUSIONS. From a study of the results obtained it can be seen that this method gives very good results. Better results were obtained from this method than from a sodium carbonate and sulphur fusion. It can be carried out in a short time and is very easy to manipulate. It was not tried on a mixture of a trace of tungstic oxide and a large quantity of columbic oxide but it would seem that very good results ought to be obtained. This method was not applied to a separation of tungstic and tantalic acids, but, from the results with columbium, there should be no difficulty in applying it to a separation of these two acids. acknowledgment. This investigation was undertaken at the suggestion of Prof. Edgar F. Smith, and I take this opportunity of sincerely thanking him for his encouragement and ready advice throughout the work.