ATOMIC WEIGHTS 447 



The values used in reducing these ratios are — 



H = 1.00779, ± .00001 



In making the computations, the oxalate ratios of Schneider are com- 

 bined, in order to avoid the uncertain hydration of the compounds. That 

 is, in each set of ratios, instead of calculating from the percentage of 

 nickel or cobalt found, the cross ratio is taken, 2C02:ISri or Co, as the 

 case may be. So much assumed we obtain the following values for Ni : 



From ratio 5 Ni = 58.027, ± .0345 



" 1 and 2 .58.044, ± .0319 



" 3 and 4 58.064, ± .0648 



" 6 58.085, ± .0441 



"11 58.661, ± .0033 



"12 .58.661, ± .0028 



" 8 58.062, ± .0017 



"13 58.685, ± .0009 



" 7 58.708, ± .0381 



"16 58.789, ± .0068 



"17 58.917, ± .0172 



" 9 .58.958, ± .0131 



"10 .59.027, ± .0141 



"14 59.030, ± .0127 



"18 59.031, ± .0231 



"15 .59.456, ± .0250 



General mean, Ni = 58.G82, ± .00074 



This mean lies within the limits of variation of Richards and Cush- 

 man's determinations, and must be regarded as satisfactory. Theii 

 work and Zimmermann's dominates the entire combination. 



For cobalt we have twenty ratios, as follows : 



(1). Per cent, of Co in CoC,04.2H„0, 32.5555, ± .0149 

 (2). Per cent, of CO, from CoC„0,.2H,0, 47.7475, ± .0213 

 (3). Per cent, of Co in CoO, 78.6324, ± .0002 



(4). Per cent, of Co in purpureocobalt chloride, 23.G045, ± .0018 

 (5). Per cent, of Co in phenylammonium cobalticyanide, 11.8665, ± .0124 

 (6). Per cent, of Co in ammonium cobalticyanide, 21.943, ± .029 

 (7). Per cent, of Co in brucia cobalticyanide, 3.7437, ± .0036 

 (8). Per cent, of Co in strychnia cobalticyanide, 4.5705, ± .005 

 (9). Per cent, of CoO in C0SO4, 48.287, ± .0135 

 (10). 2Ag:CoCl„: : 100: 60.1975, ± .0006 



