ATOMIC WEIGHTS 465 



Hence Pd = 106.47. Wh}' these analyses should give low values is un- 

 explained. 



The various series of figures for the percentage of palladium in pal- 

 hidiammonium chloride now combine thus: 



Keiser, first series 50.360, ± .0008 



Keiser, second series 50.359, ± .0028 



Bailey and Lamb 50.171, ± .0099 



Keller and Smith, electrolytic 50.508, ±.0014 



Keller and Smith, hydrogen series 50.388, ± .0043 



Keiser and Breed, first series 50.351, ± .0023 



Keiser and Breed, second series 50.352, ± .0026 



Amberg, first series 50.550, ± .0110 



Amberg, second series 50.388, ± .2064 



Amberg, third series 50.395, ± .0015 



Krell 50.3945, ± .0050 



Woernle 50.396, ±: .0012 



Kemmerer 50.332, ± .0025 



General mean 50.3882, ± .0005 



Like Haas, Gebbardt ' also made analyses of palladiammonium bromide, 

 and by the same method. His figures, with vacuum weights, are as 

 follows : 



Mean, 35.491, ± .0007 



Hence Pd = 106.68. 



The work of Haas and Gebhardt was done under the direction of 

 Gutbier, who has combined their material in a memoir bearing their 

 names in joint authorship with his.'' In this memoir ten additional 

 analyses of the bromide are given, but six of them are rejected by 

 Gutbier as unsatisfactory. I prefer, however, to include them in this 

 discussion, but with low weight. The ten determinations I have divided 

 into two sets, one containing the four preferred analyses, the other the 



■'Sitz. phys.-med. Soz., Erlangen, 40, 65. 1909. 



2 Gutbier, Haas and Gebhardt, Journ. prakt. Chem. (2), 70. 4.57. 1909. 



