186 



additive nature. Its existence is a connecting link between 

 the atomic weight on the one hand and the atomic volume, 

 refractive power, etc., on the other. The preliminary paper 

 on this subject contained a table of stopping powers as found 

 up to that time. I have made several new measurements of 

 these constants, which are, I believe, an improvement on the 

 old. This is particularly the case with the metals Au, Pt, 

 Sn, Ag, Cu, and Al, since the specimens used were obtained 

 as pure from Messrs. Johnson, Matthey, & Co. I find that 

 if the stopping powers of S, CI, and I are calculated from 

 those of molecules containing them, on the assumption that 

 the additive law holds, then these fit in very well with the 

 metals. So also does Br fit in fairly well ; it is quite possible 

 that the dwergence is due to experimental error, since the only 

 measurement on a molecule containing Br was made at a very 

 early stage of this enquiry. The divergence from the exact- 

 ness of the square-root law, which I have previously pointed 

 out, seems to occur only in the atoms whose weights are 

 below 30 ; these have an abnormally low value, as may be 

 seen from the table below (in which s for the air atom = l). 

 It is curious that a similar effect should occur in the case of 

 the atomic heats : — 



One other point invites some consideration. Whilst the 

 saturated ionisation curve seems to be the same for all gases, 

 yet the effects of initial recombination vary from gas to gas 

 and from point to point on the curve. This fact can be ex- 

 plained by the consideration that the amount of the ionisa- 

 tion produced is an intramolecular effect, and is therefore 

 independent of the physical conditions of the molecule and 

 of the relations of one molecule to another, whilst the amount 

 of initial recombination depends on extramolecular relations, 

 on pressure, perhaps on temperature, and so on. The in- 

 crease of initial recombination towards the end of the path 

 of the particle may be due in part to the existence of a 

 greater number of molecules that have lost more than one 

 ion, since in such cases recombination would be harder to pre- 

 vent. This raises the question as to how the ionisation is 



