THE STRUOTTTRK OV TlIK NIUU.KITS. 117 



of cleaning the apparatus free from niicleation, after the charge of strong HCl has 

 been removed. Nuclei are thcreaftei' s[)ontane()Usly produc^ed in the apparatus 

 itself. Neither precipitation nor aspiration will remove them. They survive with 

 undiminished intensity even after twenty-four hours. If all parts of the apparatus 

 are washed out witii flowing watei', liovvever, this nucleation usually disappears. It is 

 probalily due to traces of IIC\, reacting on the sul[)hides produced fi'om the I'uhber 

 tubing through the intervention of metals. Cnriously enough, these nuclei only 

 diffuse into the neutral atmosphei'e above solutions; they do not appear in the 

 acid atmosphere above the 3 % HCl solution, for e.\am[)le. 



The marked difference between the early and the later data for IK'! leads 

 one to expect that some occult or undiscovered influence is in acti(m. The same 

 unaccountable displacement of data is met with again in the case of the dilute 

 solutions of sodic sulphate, §§ 2(), 27, below. In l»otii cases, the data at a given 

 time of observation are determinate. 



1 1 . Puie wafer. Supfirxcttiirdtinn.' — Table 2 contains data for five samples of 

 water, all of nominally tiie same degree of purity. The first experiments follow 

 the work with HCl, and in spite of the rinsing and piecipitations which the vessel 

 had experienced, show the eff'ect of lingering traces of HCl in the relatively per- 

 sistent coronas. The second filling of water is strikingly pure by contrast. The 

 third and fourth charges show similarly lingering traces of the sodic sulphate from 

 the work which had preceded the experiments. The fifth charge is promiscuous. 

 Other examples for water will be found below, § 35. The influence of traces 

 found in this way is remarkable, and suggests the sensitiveness met with in electro- 

 lytic phenomena. It recalls the impossibility of dealing with absolutely pure 

 water in glass vessels. 



To determine the number of nuclei per cub. cm. in an average case, 

 ■it. = G4 ?/?,s-^ Xl'»^ and the data for pressure diffei'ences 16 and 20 cm. are pref- 

 erable. From these one may deduce, successively (since m = 73/108 and 92/108), 

 ,j _ 47 ,53 and 11 = 59 s'\ respectively. With these constants the pressui'e effects 

 are easily computed, since d is now .34 -^m. The results for d and n are sum- 

 marized in the tal)le, and the gi'aph figure 6 smooths the observed data except, as 

 usual, at low pressure differences, where, fi-om the evaporation of water globules, 

 the corona seen is too small. 



Similarly, the data for n at different pressure differences show too few particles 

 at the lower exhaustions. 



12. Time fo.s\ses.— Computed as above fi-oiu n = 47 X•s^ the results are given 

 in table 2 and the graphs, figure 6'. The feature of these results is the excessively 

 small number of particles in action, often falling below 5 per cub. cm., and the 

 corresponding faintness of the coronal display. Measurements ai'e mei-e estimates. 



Several interesting results need mention. The fleetness of the first sei'ies 

 which followed the work with HCl is 1.6 particles per minute. In case of the 

 purer water following, 6 particles vanish per minute. In series 3 (after sodic 

 sulphate), 40 particles vanish per minute; in series 4, 23 particles vanish per 

 minute. There is no immediate correspondence here between the size of the 



