354 Prof. W. H. Bragg on the Ionization of 



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 very well ; it is quite possible 

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

 measurement on a molecule containing Br was made at a very 

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

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

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

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

 seen from the table below, in which the " air-atom " = 1. 

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

 the atomic heats: — 



H. C. N. 



s -24 -85 -94 



Vw 100 3'47 374 



* VwXlO 3 ...240 246 251 



Cu. Br. Ag. 



s 2-46 2-60 3-28 



Vw 7-96 8-93 1037 



WwXlO 3 ...309 291 31G 



[Note added Nov. 8. — The figures in this table will no 

 doubt be somewhat modified by more accurate measurements, 

 and I am revising my methods for this purpose. The values 

 for C and N are certainly, perhaps 4 per cent., too low. As 

 regards C, see also the accurate measurements of Kucera 

 and Masek (Phys. Zeit. 1906, p. 634). The value of H may 

 be in error as this gas really requires a special apparatus. 

 Good sheets of Fe and Ni are hard to obtain. The value for 

 Br requires redetermination. I think the rest are fairly 

 accurate, and I trust the average error in this table will not 

 in the end be found to exceed one or two per cent. The 

 values of s for Sn and Ag certainly show some variation from 

 the square-root law as indicated by the difference of the 

 ratios 316 and 326. 



The results of Kucera and Masek are a very welcome con- 

 firmation of the square-root law. Only on one point there 

 seems to be still an obscurity. These authors claim to have 



0. 



Al. S. 



01. 



Fe. 



Ni. 



1-05 



1-495 1-76 



1-78 



229 



2-44 



4-00 



5-20 5-65 



5-96 



7-48 



7-65 



262 



287 312 



299 



307 



319 



Sn. 



I. 



Pt. 



Au. 



Pb. 



3-56 



3-44 



4-14 



4-22 



4-27 



10-9 



11-2 ■ 13-95 



14-0 



1435 



326 



307. 



297 



301 



298 



