NUCLEATION AT PROVIDENCE AND BLOCK ISLAND. 131 



observed nucleation in question, which is still present and undergoing 

 phenomenal variations in places remote from the habitations of man. 



Moreover, the arithmetical character of the fluctuation of nucleation 

 in the lapse of time is the same at the two stations, however widely 

 the geometrical character may differ. It is thus probable that in both 

 cases there is superimposed on a local nucleatiou (large at Providence 

 and vanishing at Block Island), fairly constant for long periods during 

 the winter months, a specific effect, due to causes which are certainly 

 not local. In fact, the same character of variations was observed at 

 the two stations in spite of the fact that the air under examination is 

 necessarily quite different. 



The outstanding February maximum may be a distant land effect, 

 due to artificial causes, chiefly combustion, and nearly uniformly dis- 

 tributed over the whole inhabited territory ; but from the suddenness 

 of its appearance, its pronounced character, and the extended occur- 

 rence as instanced at both stations, one is tempted to regard it as an 

 actual invasion of the atmosphere on the part of some external radia- 

 tion or nuclei-producing agency. 



It would have been better if the pressure difference at the station, 

 where the air is relatively pure had been more nearly equal to the fog 

 limit of dust-free air (8^=22), i. e., decidedly above the fog limit for 

 ionized air (8/^ = 19); but at the outset it was thought wise to avoid 

 this complication. In such a case the coronas reached by the present 

 method (8^ = 17) would all have been obtained; but in the compara- 

 tive absence of ordinary nuclei, a response from ionized material might 

 be anticipated. The treatment of filtered air, however, for purposes 

 like the present or the interpretation of the results obtained is an 

 extremely precarious matter, as I shall point out in a subsequent 

 paper. Meanwhile we may note that the curve of average monthly 

 nucleations is apt to show a maximum and minimum, respectively, at 

 about the time of the winter and summer solstices, and that any defi- 

 nite fluctuation from this curve is due to causes which are at least 

 nonlocal in character. 



