448 



SMITHSONIAN MISCELLANEOUS COLLECTIONS 



VOL. 54 



(11). 2AgCl:CoCL: : 100: 45.307. ± .0011 

 (12). 2Ag: Co:: 100: 27.705, ±.0050 

 (13). 2AgCl : Co :: 100: 20.687, ± .0033 

 ( 14 ) . CL : Co : : 100 : 83.220, ± .0082 

 (15). 2 Ag:CoBr„:: 100: 101.407, ±.0018 

 (10). 2AgBr:CoBr„: : 100: 58.2549, ± .0007 

 ( 17 ) . CoBr, : Co : : 100 : 26.9508, ± .0008 

 (18). 2 Au:3Co:: 100: 45.151, ±.025 

 (19). Co:H2::100:3.4110, ± .0009 

 (20). I,: Co:: 100: 23.462, ±.0027 



Hence, for tlie atomic weight of cobalt. 



From ratio 9 Co = 58.706, ± 



" 3 58.880, ± 



15 

 10 

 16 

 11 

 17 

 14 



5 

 19 



6 



13 



18 



20 



12 



1 and 2 . 



957, ± 



965, ± 



966, ± 

 968, ± 

 971, ± 

 017, ± 

 042, ± 

 091, ± 

 093, ± 

 100, ± 

 183, ± 

 203, ± 

 305, ± 

 .378, ± 

 556, ± 

 770, ± 

 006, ± 



.0286 

 .0006 

 .0040 

 .0014 

 .0027 

 .0032 

 .0019 

 .0059 

 .0630 

 .0156 

 .0810 

 .0647 

 .0048 

 .0570 

 .0095 

 .0330 

 .0069 

 .0108 

 .0384 



General mean, Co = 58.915, ± .0005 



It is evident that in this combination, ratio 3, representing principally 

 the work of Zimmermann, receives excessive weight. For that reason, 

 and also on chemical grounds, the final mean is probably too low. If, 

 however, we arbitrarily as.^ign to ratio 3 the " probable eri'or " and 

 weight of the next best ratio. No. 10, the general mean then becomes 



Co = 58.961, 



.0008 



This is probably not far from the trnth: but the change thus effected 

 serves to illustrate the fact that the rigorous mathematical combination 

 is not always conclusive. Although the mathematical method is most 

 useful, it cannot do away with the exercise of judgment as based upon 

 other knowledge than that shown in the mere figures. 



That the atomic weight of cobalt is higher than that of nickel clearly 

 appears from the evidence. xSTevertheless. attempts have been made, and 



