218 Barus — Ionization of Water and Phosphorus Nuclei. 



B = 10- 6cm U= -125 cm /sec 



E — I0" 5cm TI— -012 cm/sec 



if I = 50 cm . 



Since R = 10" Bcm is the superior limit of size* found from 

 subsidence considerations, it appears that IT is probably larger 

 than -012 cm./sec, if I — 50 cm , and hence this value of U com- 

 puted purely from viscosity is in agreement with the value of 

 U just deduced from purely electrical data. 



Furthermore, the order of size and the velocity deduced are 

 compatible with my work on the structure of the nucleus, 

 ~k = -001 to *03 being here the specific velocities in the absence 

 of the electric field. Hence k is small compared with TJEf 

 {7\— r /\\ the condition supposed. 



3. The behavior of the phosphorus emanation is next to be 

 treated. Here the value of h, or the specific velocity of the 

 nucleus in a given cardinal direction, was found by direct 

 experiments to be k = *3 cm./sec. Hence the value of the 

 exponential factor, 1— e —a , even when E=0, is about '96 and 

 may therefore be neglected as experiment showed. Conse- 

 quently in this case 



as heretofore stated, where, moreover, for want of direct knowl- 

 edge, U=k was assumed. This is virtually supposing, since 

 here R U — *125XlO~ 6 , that the size of the phosphorus nucleus 

 is R= -42xl0" 6cm , a datum which, provisionally at least, is 

 admissible. 



If the nuclei are larger than this, the factor JJ/Tc must be 

 evaluated in computing n Q from 2*3 a = {dE/dt)/ ' E, as 



n = (k I U) (2-3 a C (r 2 -r J) / 16-7 eV. 



Now it is suggestive to observe that the datum for n for 

 tubular condensers was always low as compared with the same 

 datum for plate and for spherical condensers. 



Thus from plate condensers, n = 33X10* 



from spherical condensers, n = 39X10* 

 from tube condensers, n = 8X10* 



I attributed this with some misgiving wholly to want of satu- 

 ration in the latter case since an influx tube is inevitable. 



But in the value of n for plate and spherical condensers the 



value of k does not enter, and the results depend simply on U. 



Hence what was supposed to be wholly undersaturation was 



probably, in part, the neglect of the factor k/ U. Put, there- 



* There would be coincidence for R = 8'6 x 10 — 6 . 



