40 EXPKRnrE>-Ts with ioxized aie. 



Tilt* value of k is given for each series in the tables, computed from three 

 points of the observational data corresponding to ,7' = and the niaxiinuni and 

 mean len<,'ths. It will be noticed that l\ = .GO liter/miii is nearly the same for 

 all the al)Sorption tubes, as it shoukl be for initiallj- saturated air, and has beeu so 

 taken. From the value of k found for each tul)e I then computed the corresponding 

 curves, these being given in the last columns of table 1. The computed curves are 

 constructed in the chart, figure 3, to show the distiibution of the observations with 

 respect to them. The agreement is throughout sur[)rising]y good ; it would lie im- 

 possible to get a better interpretation of the observations in view of tiie difficulty 

 of color e.xperiments. If we compare the nuclear velocities I; with tlie ladii of the 

 absorption tubes, with which they were obtained, we find that they vary for the 

 wide tubes (gray rubber and lead) as much as for the narrower tubes (lead, pure 

 i-ubber, and glass). Hence k must be regarded inde|)endent of r, and the variations 

 found are observational ei'rors. 



I c(mcluiie, therefore, that the proposititm which considers decay (k') to be 

 relativelv and virtually negligible and the absorption effect of the tubes of velocity 

 I; or an ionic velocity Si; to be real, is one of great (nobability. The whole ionized 

 rei^ion is under volume expanding stress, much like an osmotic [)ressure. 



8. Cainpiitution for diluted emanation. — The case of the wide tubes of tin 

 plate (2/* = 5 cm.) is different in character, for here the different lengths correspond 

 to different initial densities, ti„ and n\, while the ladius of the tube and the velocity 

 of the air current are the same. One may assume that the initial densities are to 

 each other as the liters per minute ( V) of air saturated with phosphorus emanation 

 put into the tube at distances ,r and x' from the jet. Thus njn\^= V/V', and 

 therefore I = (rv/2(.v - x')) In ( T/ V). Since >t/2(x - x') is about .093, the vol- 

 ume ratio V/ V' should be between 15 and 20. The observed values rarely exceed 

 2, often falling much below this. The coefficients k are thus too small as com- 

 pared with the jireceding set. 



It is interesting to compare tlie degree of dilution here with the above cases. 

 The volume of saturated air added rarely exceeded 1 or 2 liters/min. The volume 

 of air traversing the tul)e and due to the jet is 120 liters/min. The dilution is thus 

 from 50 to over inO. Tlip above cases of dilution with narrow tubes would be 

 given by >i//i„ = e -*''->^'^ *5' ^ while in the tul)e. Otherwise, since n„ V„ = n V for a 

 given color, the chart gives 7i/n„ at once. The range of values does not exceed 11 

 (irray rubber), the other maxima being 7 (lead, rubber), 4 (lead, glass), terminating 

 with 1. Hence the orders of dilution in the two experiments are very different. 



The results foi' the tin tube are naturally much less certain because the colors 

 become dull after the dust has passed through great lengths (50 ft.) of tube, or they 

 rticker, and because v is not easily found. Still the preservative (pialities of dilu- 

 tion are undeniable, and my observations again lead me to disbelieve that diffusion 

 witiiiii this wide, eddying current (tested), can be in question. Ignoring it, it seems 

 most probable to conclude with others (J.J. Thomson, Rutherford, Townsend) that 

 new ions are being continually reproduced by collisions. From the present experi- 

 ments with wide tubes it would then follow, that when the ionized air is markedly 



