128 nr.l.l. SVSTFSf Tr.CIIXICAL JOVRWAl. 



one, and within certain ranges of field strength and pressure it is; 

 again, good. The next step is to calculate the values of B and Vn 

 which the curve imposes on the gas to which it relates. I quote the 

 values of In, the ionizing-potential, which Townsend presents: 



W'luii ihe first of these values were (ielenuiiieil. an more direct way 

 of measuring ioiiizing-potcntials was known. Xow that we ha\e 

 some values obtained l)y the direct methods sketched a few pages hack, 

 and fortified 1)>- indirect but very forcible evidence from spectroscopy, 

 it is possible and cjuite important to test some of these. The values 

 for argon and helium, although of the proper order of magnitude, are 

 certainly too low. This is not in the least surprising, considering how- 

 many of the collisions between electrons and atoms must be perfectly 

 elastic. It seems indeed rather nnsterious that the current- voltage 

 relation in either of these gases should ha\e conformed closely enough 

 to (4) to make it jiossible to define and ineasure a; but the electrons 

 no doubt entered into luany of the collisions with energy enough to 

 put the atoms into excited states, if not to ionize them; and it is 

 nearly always possible to take refuge in the assertion that the im- 

 purities may have been sufficient to distort the phenomena. As for 

 the other gases in the list, all of them diatomic or triatomic, Town- 

 scnd's values are too high — not ver\- much too high, however; iisiiallv- 

 a matter of one-third to two-thirds." 



It appears therefore that the theory I have just developed is too 

 sinijile, and must be amended. It seems natural to begin by dropping 

 the tentative assumption that a molecule is ionized whenever it is 

 hit by an electron having as much or more energy than is required 

 to ionize it, and adopt instead the idea once already suggested in 

 these pages, that it is sometimes but not always ionized by such a 

 blow; that there is a certain prohubility of ionization by a blow from 

 an electron having energy U, a probability which is zero wiicn I' < V 

 and is some yet-to-be-determined fimction of U when {'> V. Tiiis 

 would leave intact the conclusion that a p should lie a function of 

 X p, a conclusion which we have already foimd lo be veritied b\' 

 experiment; but it would relieve us of the necessity of assuming that 



" TownsfiKj's v,t1iics of B likewise corrospoiul to values of the effectivT cross- 

 licction of the niolcriih.-, the (iiLinlity .1 of equation (2), which are of the same order 

 of magnitude as the directly determined values oi A. ' 



