320 THE ATOMIC WEIGHTS. 



not invalidate the Pd : KCl ratio. In a second series the palladiochlo- 

 ride was dried in vacuo at 100°, giving the following data: 



K.PdCh. Pd. 2KCI. Ratio. 



1.3635 .4422 .6186 7 ••484 



3.0628 .9944 1-3929 71-391 



1.4845 .4816 .6782 71 -01 1 



1.7995 .5838 -8206 71.143 



Mean, 71.257, d= -0736 



These experiments seem to be less concordant than the preceding set. 

 It must be noted, however, that the authors reject the KCl determina- 

 tions and compute directly from the ratio between the salt and the metal. 

 But the ratio here chosen agrees best with the determinations made l)y 

 other observers, giving for this series the mean value Pd = 10.5.4-5.5, and 

 is, moreover, uniform with the data given by Berzelius and by Bailey 

 and Lamb. 



Joly and I^eidie also give two experiments made by reducing the 

 K.,PdCl,j in hydrogen, with the subjoined results : 



K.,Pda^. Pd. 2KCI. Ratio. 



2.4481 .7949 1.116S 71-177 



1.8250 .5930 .8360 70 933 



Mean, 71.055, zb .0823 



Combining these data with previous series, we have — 



Berzelius 71-233. ± -1066 



Bailey and Lamb 70-485, ± .0290 



Joly and Leidie, first 70.979, ± .0188 



Joly and Leidie, second 71-257, =b .0736 



Joly and Leidie, third 7I-055, ± -0823 



General mean 70.865, ± .0150 



In view of the discordance among the determinations hitherto cited 

 and because of the criticisms made by Keller and Smith, Keiser, jointly 

 with Miss Mary B. Breed,* repeated his former work, with some varia- 

 tions and added precautions to ensure accuracy. His general method 

 was the same as before, namely, the reduction of palladiammonium 

 chloride by a stream of hydrogen. First, palladium was purified by 

 distillation as PdCl,, at low red heat in a current of chlorine. From this 

 chloride the palladiammonium salt was then prepared. Upon heating 

 the compound gently in a stream of hydrogen, decomposition ensued 

 absolutely without decrepitation or loss of palladium by volatilization. 

 Neither source of error existed. The results obtained were these : 



* Am. Chem. Journ., i5, 20. 1894. 



