268 THE ATOMIC WEIGHTS. 



case of silver it has already been shown that Dumas' cor- 

 rection is unfavorable to it considered in its bearings upon 

 Front's law. Silver is the only element among those having 

 minus variations which could carry very much weight 

 against the hypothesis. 



Among the elements whose variations are plus, colum- 

 bium, uranium, and yttrium have been poorly determined. 

 Yttrium especially may be considered doubtful. The atomic 

 weights of aluminum, arsenic, barium, cadmium, lithium, 

 phosphorus, and sodium involve Dumas' correction to a 

 greater or less extent, and will be lowered by its application, 

 that is, brought nearer to whole numbers. For aluminum, 

 certain other causes for variation were pointed out in the 

 chapter upon that metal ; and it may be noted that the 

 direct ratio between it and hydrogen gives Al = 27.998^ 

 ±: .007. Here the variation is less than the probable error. 

 For calcium, and consequently for fluorine also, sources of 

 plus error were indicated in the discussion of their respective 

 atomic weights, and reiteration here is unnecessary. Cobalt, 

 iridium, iron, nickel, and tungsten all involve such errors 

 as may arise from the possible occlusion of hydrogen by the 

 metals after reduction from their compounds. For scan- 

 dium, the atomic weight, calculated with SO3 = 80, becomes 

 44.032, a variation much within the limits of experimental 

 error. For carbon and bismuth the variations are insignifi- 

 cant. In short, in the majority of instances the errors may 

 be diminished by corrections which are in all probability 

 needed, and which can be easily pointed out. The more 

 carefully we scrutinize the data the more probable Front's 

 hypothesis appears. 



Among the twenty -six elements whose atomic weights are 

 removed by more than a tenth of a unit from whole num- 

 bers, chlorine, rubidium, and strontium have values nearly 

 half multiples of that of hydrogen, and in each case Dumas' 

 correction will make the approximation still closer. Erbium, 

 gold, indium, lanthanum, rhodium, ruthenium, silicon, and 

 zirconium may be dismissed from consideration as too im- 

 perfectly determined to carry much weight in the present 



