THE STKIICTIIKE OF TlIK NUCLEUS. 127 



The values of h for paraffine are also in sharp contrast to those for naphthalene. 

 After If) hours, for instance, neai'ly all nuclei have vanished, a I'esult which indi- 

 cates that the phenomenal [)ersislence of nuclei in the preceding ex|ieriiiieut, after 

 15 hours of waiting, must We regarded as I'eferable to a sublimate from the naphtha- 

 lene. Experiments with pure benzol weiv not made for this length of time. 



38. Benzine. Coiieenivation. Time losses. — Pure benzine (table 23) was 

 tested to show the larger absorption velocities, k, in the absence of a solute. The 

 rates do not, however, compare with those of water, from the difficulty (frequently 

 stated) of keeping hydrocarbon solvents pure in an apparatus containing rubber 

 tubing in any of its parts, even if not in contact with the solvent. Of. figure 17. 



These relatively small values of /; lead to an important ulterior deduction, viz., 

 that an electrical influence on persistence due to the frictional effect of shaking 

 need not be apprehended. Thei'e is quite as much reason why benzine as paraffine- 

 benzol should evolve electrically persistent nuclei ; for whatever charges are evolved 

 are in like manrier capable of being stored and distiibuted. And yet benzine fails 

 while the other liquid is intensely active. 



SYSTEMATIC COMPARISON OF THE DATA. 



39. Dependence of the number of particles on the concentration of the solution. 

 — It is next desirable to select mean values from the above data for comparison. 

 The character of the variations of n with the concentration, c, of different solutes 

 has not been identified. It does not depend on valency or molecular weight as is 

 frequently the case with other constants holding under the same general relations. 

 The powerful hygi'oscopic solutes show no exceptional behavior. The tendency 

 of some of the solutes to crystallize with water of crystallization has left no mark in 

 the results. Nor can any of the l)etter-known electrical or similar constants of the 

 solutions be associated with the variations of n or k obsei'ved. Until some thread 

 of this kind has been foun<l it will therefore be necessary to proceed empirically. 



In tabulating the data for n it is advisable to refer them to a common con- 

 centration, and the 1 % solution is a useful standard. The equation to be selected 

 for this pui-pose is not easily found from the large discrepancies inevitably pi-esent 

 in the data. Thus an equation like //. = n^lOt"" where c is the concentration and h 

 a constant would be I'oughly available, but pi'obably too sim[)le. By pi'operly 

 manipulating the gi'aphs, the equation c = ^ 10 •^""-"'" is found to be in reason- 

 able accord with the properties of most of the cui-ves and desei'ves trial. This 

 may be further simplified by putting n^ - 0, assuming that from absolutely pure 

 water no nuclei of appreciable pei'sistence could be produced by shaking. A trial 

 case for calcic nitrate may be shown, in which the first two observations are u.sed 

 to compute the constants because of theii' relatively much greater accuracy. 



