Barus and Watson — Air Stored over Water. 393 



Two reasons may be suggested for the cleansing effect of 

 rotation : It is snpposable that eddy currents are produced by 

 the rotation, as a result of which all parts of the confined air are 

 in succession washed on the surrounding surface of water. 

 Such currents, however, could hardly persist but for a short 

 time after the beginning of the rotation. In the second place, 

 if there is churning of the pure water connected with the 

 rotation, the enclosed air must pass in small bubbles through 

 the water many times. This seems to be the more potent 

 cause for the absorption of nuclei. The details of the experi- 

 ments with X-ray nuclei did not conform very closely with 

 either of these hypotheses, and we therefore decided to inves- 

 tigate the question thoroughly with a larger type of coronas 

 and in a large glass vessel, through which the whole behavior 

 could be better seen. 



3. The following results were obtained with a glass cylinder 

 about 35 cm long and 25 cm in diameter (part of an old electrical 

 machine). It was nicely mounted axially and capable of very 

 rapid rotation, enabling the observer to catch up the whole of 



Table II. — Air nuclei removed by rotation, dp = 17 cm . 



resp. from chamber. 



lamp 85 cm and 250 c 



Time of 

 successive 

 Remarks. rotations, 



min. 





 3 

 3 



4 



Water in a smooth 

 cylindrical shell ; 

 fixed to glass walls. 



Turns 



per 

 minute. 



?450 



Water in verv turbu- 

 lent motion. 



Water in turbulent 

 motion. 



Time loss of nuclei 

 without rotation. 



Slight subsidence of 

 the successive fogs. 



350 



150 



s 

 cm. 



10 

 6*4 

 6'1 

 5-6 



7-6 

 3*3 

 2*1 

 1-2 



8'8 

 5'6 

 5-4 



8-7 

 6-8 

 6-2 

 6-2 



Goniometer and 



n 

 nuclei 



per 

 cuo. cm. 



(200,000) 

 85,000 

 74,000 

 57,000 



140,000 



12,000 



3,000 



500 



(160,000) 

 57,000 

 51,000 

 24,000 

 13,000 

 7,000 

 3,000 



(160,000) 

 100,000 

 77,000* 

 77,000 



Note : Corrections for periodicity were not applied as it did not seem neces- 

 sary. The results are thus somewhat irregular. The numbers in paren- 

 thesis are estimated. 



