﻿Ionization by Alpha Rays. 473 



The values of both n' and n l are, like the other currents, 

 the arithmetical sum of the values obtained with the field in 

 opposite directions. 



10. Since the interpretation of the experiments is com- 

 plicated and perhaps open to doubt, the readings themselves 

 are given completely in Tables V.-XIX. The figures given 

 are those actually recorded in the note-book, the only cor- 

 rection introduced being the reduction of the pressures tc 

 15° 0. ; no correction is made for the decay of the polonium. 

 For two values of y for aluminium, two complete series of 

 measurements were taken on different dates to ascertain the 

 consistency of the results. 



11. An attempt was first made to interpret the experiments 

 on the assumptions that Townsend's formula (equation (7) of 

 the previous paper) was correct. Each of the sets of 

 measurements given in one of the Tables V.— XIX. was 

 treated as a set of equations of the form (equivalent to (7)) 



//€ = („,+ ":W- '4, .... (5) 



\ ol J ex, 7 



in which i,p,y,n were known and p, and ^ unknown. The 

 values of the unknowns were found by least-square calcula- 

 tion. It w-as found that the value of n l calculated was 

 always about 0'6 of that found experimentally. Various 

 sources of error were considered and investigated without 

 result, till at last the suggestion occurred that the theory 

 might be in error. As the result of examination, the error 

 pointed out in the previous paper was detected, and the 

 revised calculation given there made. Since the revised 

 formulae are far more complicated than (5), the former 

 method of calculation was deemed unsuitable and was 

 rejected. 



12. Though the measurements which have been recorded 

 will provide a test of the accuracy of the theory of ionization 

 by collision, our primary object is not to examine this theory, 

 but to obtain information about the delta rays. The pro- 

 perties of these rays are represented in our equations by )i x j 

 n-[ is measured, but it is undesirable to take it as one of the 

 variables known. For, if it were so taken, and only v and 

 V treated as unknowns, the values of which were to be 

 determined, it would be uncertain whether the discrepancies 

 between theory and experiment, which appear when the 

 calculated values are substituted in the equations, are due to 

 errors connected with n^ or to errors connected with some 

 other of the quantities involved, It is preferable to retain 



