C. Bancs — Changes of Colloidal Nucleation. 207 



depart widely from it ; but coincidence will usually occur 

 only when both accompany the same temperature effect. As 

 a rule there is a rise of nucleation from morning to afternoon, 

 suggesting the phenomenon due to external radiation discov- 

 ered by Wood and Campbell (§ 1), but in these cases tempera- 

 ture is also apt to rise coincidently. The rise in question fails 

 to occur but four times out of the thirteen observed in August, 

 but seven times out of the twenty-four observed in September, 

 two being neutral, and but 5 times out of the thirteen observed 

 in October. Unfortunately there is no room for these long 

 graphs here ; but essential features may be taken from iigures 

 2 and 3. 



6. Effect, of the baromete?\ — If we look more specifically at 

 the new data beginning with August 10, coincidences of minima 

 and maxima of the nucleation with maxima and minima of the 

 barometric pressure occur only on August 13, 25, 27, and these 

 are not pronounced. In September there is no detailed simi- 

 larity until September 16, but both curves have dropped some- 

 what toward the marked minimum. After September 20, 

 however, the apparent agreement of curves is conspicuous up 

 to September 21, and would be decisive if the run of tempera- 

 ture were not similar. During the remainder of the month 

 there is no agreement, rather an opposition, and the two curves 

 are remarkably at variance during the unusually low barometer 

 in the early part of October. The peak of the barometric 

 curves from October 4 to 8 has nothing to suggest it in the 

 nucleation curve. We may conclude, therefore, that a direct 

 barometric effect is absent, that such coincidences as seem to 

 occur are referable to other causes, and that the method used 

 for the elimination of barometer discrepancies is to the same 

 degree vouched for. 



7. Effect of temperature. — Throughout the observations the 

 tendency of temperature of the fog chamber to rise from 

 morning to afternoon is most probably to be regarded as the 

 cause of a similar tendency in the nucleation. There are 

 exceptions, most of which, however, may be explained away. 

 The curves show a similar general march from August 10 to 

 23 and from here to Aug. 29. From September 7 to 18 there 

 is much detailed agreement, as for instance on September 8 to 

 10, 15 to 16. The same is true after September 20, where 

 markedly coincident variation occurs. 



So in October the agreement of curves is apt to be very 

 close, as, for instance, the effect from September 30 to Octo- 

 ber 3, the general fall thereafter and the effect from October 

 7 to October 9. All of this will appear more strikingly when 

 the observations are averaged for several consecutive days, and 

 most of the lack of synchronism is doubtless due to the diffi- 

 culty of finding the true value of nucleation. 



