Chemistry and Physics. = 315 
ee 
_ It may then well be a ked what proof can we have of purity 
pe purity. In the writer’s own experience poneeaies cy of results has_ 
frequently meant nothing bat constancy of impurity, and the illu- 
Sion caused by coine idences so-called has been forced a 
ef least squares, results of sae weight determtunbious made ies 
different chemical processes. ‘The writer undertook such a discus- 
PS sion, as stated in his early memoir on the relations of the atomic 
De eights, above referred to, but abandoned it simply because acei- 
u quite 
tors in the roblem. No useful conclusion can be < 
rent — of results, each 
e ssible 
of pro rebel a s 
__ st now 08 question is still cael : oan is Poca criterion — -. 
pore and definite compound? the only explicit answer that can _ 
: Siven i is that a substance obtained ae oad ete 
