90 KXPERIMKNTS WITH IONIZED AIK. 



l.S. Observations for constant V ^^ and for comitant R, compared. — I sliall now 

 einlt'iivnr to compare tlie datii of the two secti<ins, by referring tliein either to the 

 same radii or to the same electric fields. The latter uiethod. is preferable, not only 

 as yielding a greater range of data but because the values of the currents Ss/St 

 and of ?ij^iiave not yet been regarded in this light. To begin with the formei-, 

 figure 15 contains the values of 8x/6t varying with tiie fields as taken from the tables 

 2, 3, 4, in which V ^f H varies by reason of varying It. These data, though vague, 

 lie within tlie limits marked by the curves R and R'. In the same chart, Fig. 15, 

 I iiave inserted the curve F, taken from figure 14, in which V^/R vaiies by reason 

 of varying F^,. It is the curve for K2 for which 7?=11.7 is largest, and thi- data 

 surest. The point of importance is clear at once : as the fields grow stronger the 

 curve 7*^ lies (piite above the curves R. If, therefore, high fields are produced by 

 diminishing the radius of the condensei-, the currents may be upwards of 10% or 30^ 

 too small, both because of the escape of ions around the stem and of the access of 

 air. Indeed this state of things is not unexpected, inasmuch as the chief object of 

 the investigation with splierical condensers wjxs the avoidance of such losses of ions 

 as occur in plate condensers. Returning for additional consideration to figure 12, 

 it appears that the defiintely low data corresponding to the I'adii 2 and 3 centi- 

 meters are erroneously much too low, whence it follows that the probability of a 

 constant ?/, A^is enhanced. 



The quantities next to be considered are the constants n^ K in relation to tin- 

 strength of field. This is done in figuie 13 for tlie case of the present section in 

 which potential F is varied directly while R is fixed. As the ol)servatii'ns must 

 here be considered individually, they ai'e maiked for each table by numerals show- 

 ing the order of serpience, and joined by straight lines. Turning first to the data 

 of table 6 for the inteiraediate size, R = 6.5 cm., it is seen at once that comjilicated 

 influences are at work, particularly in the case of low fieltls. The bfliavi'-r here is 

 not unlike ordinary galvanic polarization. Thus in the zigzag line 1 to 12, there 

 is successive partial exhaustion alternating witli partial recuperation of the conduc- 

 tivities. The observations 1 and 7 correspond to nearly the same field, for instance, 

 but with feuei' ions available in the latter case. If the fields are increasing, pro- 

 longed action decreases n^K ; and vice versa for table 8. On the whole, m,^, as the 

 fields increase indefinitely, seems to decrease (table 7), reniembcrini; always that this 

 decrease is of an order which may be attrii)uted to the ionizer itself. 



Finally if the data for n^ 7i and field strength as given in the tables 2, 3, 4, of 

 the last section, be graphically ivpresented, the results are (piite without suggestion 

 as to the character of the variation of // , K. in dillVrrnt fields. An average value of 

 this constant is oidy to be deduced. 



14. Condnsion. — Contrary to my e.\pectation and in spite of the lal)or sj)ent 

 upon them, the results for s[)herical condenseis li.i\c not enal)l('d me to give a de- 

 cisive answer to the question at issue. The difficulty encountered and which occurs 

 here in accentuated form is the .same which has hanqiered me throughout the 

 present research, namely the elusive variability of the ionizer. Moreover, as the 

 conditions determining it exist immediattltj at the surface of the active phospliorus, 



