38 



to cut out most of the normal )3 rays, and, indeed, some of 

 the soft y rays as well. The Dolezalek electrometer was fitted 

 with a fine quartz fibre : one volt moved the scale-image 

 through four metres. The zero was fairly steady, and conse- 

 cutive readings generally agreed to three per cent. ; the diffe- 

 rences were by no means wholly due to the electrometer. As 

 appears to be usual in the measurement of such small cur- 

 rents, we were somewhat troubled by unaccountable distur- 

 bances, but on most days these were not enough to interfere 

 with the general results. 



Having set up the apparatus in this way, we proceeded 

 first to compare the incidence and emergence radiation of 

 various substances. This was done by the measurement of 

 the current under three different arrangements : — 



(a) When those parts of the top and bottom walls of the 



chamber through which the y rays passed were made 



of the thinnest Al foil (2),p and q,q, in the figure) ; 

 (h) When a plate of the substance, of proper thickness, 



was so placed at p,p that the rays passed through it 



into the chamber ; 

 (c) When a plate of the same substance was made to 



form part of the top wall, at q,q, so that the y rays 



struck it after crossing the chamber. 

 We took h—a as a measurement of the emergence radiation, 

 c—a as a measurement of the incident. Some results which 

 we obtained in this way are contained in the following 

 table, in which the numbers refer to the movement of the 

 scale in 30 seconds, 10 ==1 mm. : — 



Table. 

 Comparison of emergence and incidence radiations. 



The figures here given show the very large want of symmetry 

 between the radiations on the two sides of a plate. Our pre- 

 vious experiments proved the existence of this want of sym- 

 metry ; but it is now shown more clearly and satisfactorily. 



The resi;lts still require correction before they can be 

 considered accurate. It will be clear that h —a, the emergence 



