114 BRIDGMAN. 



Bismuth. An attempt was made to obtain measurements on this 

 metal by the radial flow method, but without success. It did not 

 prove possible to get sufficiently good thermal contact with the fine 

 copper tubing, and I have already mentioned that the attempt to use 

 silver tubing failed because of the extremely rapid alloying action 

 between silver and bismuth. Measurements were finally made on 

 three difl^erent specimens by the longitudinal flow method. 



A great deal of time was spent in the endeavor to obtain pure 

 material. In my previous work on the effect of pressure on electrical 

 resistance I had purifled the metal by electrolysis. I now endeavored 

 to repeat this, but without success; I could not make the bismuth 

 form a coherent deposit. The procedure of the previous work was 

 exactly repeated as far as I could tell. Previously the hydrosilico- 

 fluoric acid had been obtained from a German source; this was no 

 longer available and acid from the J. T. Baker Chemical Co. was used 

 instead. The acid was of high purity as indicated b^^ the analysis on 

 the label, but there is a possibility that some impurity not covered by 

 the analysis might have been responsible. I then obtained some 

 bismuth from the U. S. S. Metals Refining Co. I have to thank them 

 for supplying me with six pounds of the metal, in two lots. Their 

 product is prepared electrolytically, in distinction from the ordinary 

 commercial product, and is guaranteed by them to have less than 

 0.1% impurity. Ordinary commercial bismuth has about 3% im- 

 purity. My test for purity has been the temperature coefficient of 

 electrical resistance. This electrolytic bismuth showed a very low 

 coefficient, only 0.0022 between 0° and 30°. Ordinary commercial 

 bismuth is higher. Professor F. A. Saunders was kind enough to 

 make a spectroscopic analysis; he found a very strong silver line, 

 which seemed to indicate a rather considerable impurity. I con- 

 sulted the U. S. S. Metals Refining Co. again, and they were so kind 

 as to send me a second sample, which they had submitted to chemical 

 analysis, and found to contain only silver in detectible quantity, and 

 this was less than 0.06%. But the temperature coefficient of this 

 new sample was again very low. I attempted a purification by slow 

 crystallization from the melt, with the result of bringing the coefficient 

 up to only 0.0025. Ordinary commercial material, purified in the 

 same manfier, showed a coefficient of 0.0034. Professor Saunders 

 was again kind enough to make a spectroscopic analysis and found 

 again the strong silver line, which seemed to him could only arise from 

 a rather large amount of impurity. He found traces of Cu and Pb, 

 and no traces of Sn, Cd, Zn, Li, As, or Sb. Professor G. P. Baxter 



