140 SMITH AND EXNER — ATOMIC WEIGHT OF TUNGSTEN. [April 7. 



values in a series of five trials ranged from 182.85 to 183.64. 

 Patient search was made for the reason, every step being tested 

 repeatedly, until eventually the conclusion was forced upon us that 

 carbon dioxide, in varying amounts, was disengaged through the 

 decomposition of the sodium carbonate in the final drying. Jac- 

 quelain {Jahresb., i860, p. 116; A. Ch. [4] 28, ^d, and A. Ch. 

 [3] 32, 205) showed that this salt loses from 0.03 to 0.05% in 

 weight at 400°, and other observers have shown that the loss con- 

 tinued with the length of the period of ignition and with the 

 temperature. Here, then, was a serious defect in the method which 

 would explain why the values found were low, and why they 

 differed so widely. The attempts to correct this weak point 

 proved futile, so that the method, having had a thorough trying-out, 

 was abandoned after months of arduous work. 



It was hoped that perhaps a normal silver tungstate might be 

 made, which after solution in potassium cyanide could be electro- 

 yzed and the value of tungsten obtained from a comparison with 

 the precipitated silver. Fifteen experiments were made. In one 

 series (the best) of five experiments the results varied from 

 184.00 to 184.39. It was found, after much search, that there 

 could be no certainty as to when a normal salt was really in hand. 

 Washing and drying, even when performed with the utmost care, 

 occasioned a change in the character of the salt. The method was 

 discarded. 



An effort was also made to obtain a cadmium salt of definite 

 composition. Much time was given to it, and experiments were 

 made in the electrolysis of bodies believed to be uniform in com- 

 position. The atomic values ranged from 181.90 to 185.71. 



Having subjected three new methods to vigorous tests in our 

 efforts to solve the problem along new lines, and having found them 

 utterly deficient, the hope still remained that possibly some of the 

 earlier methods might with pure material give satisfactory results. 

 The writers felt, without meaning to reflect in the slightest upon 

 earlier investigators, that their material possessed the merit of 

 superior purity ; and if that were really the case, older methods, 

 simple in principle and easy of execution, might well be expected 

 to give concordant values. Of the 175 experiments made by the 

 entire corps of previous investigators, there is but one short series, 

 namely, that of Pennington and Smith, in which there is that 

 degree of concordance which is desirable and necessary in fixing the 



