A CONTINUOUS RECORD OF ATMOSPHERIC NUCLEATION. 211 



23. Observations during the first half of July (charts 46, 47) were sus- 

 pended. High rain values occtu- on July 16, 18, 27, 30. Thus the rain on 

 Jul\- 27 has but partially wiped out the maximum. On the other hand, the 

 rain niinima on July 23, 24, like those of August 4, are excessively low, being 

 among the lowest nucleations observed (1000-2000 per cub. cm.). Sunday nu- 

 cleations are not remarkable. 



24. The nucleations of August (charts 47, 48) are low at the outset, pre- 

 ceding the abrupt maximum of August 9. The temperature effect is vague. 

 During the middle of the month observations were suspended. The maximum 

 on August 31 suggests the approach of the winter nucleations. 



25. September nucleations (chart 48) present no new features and but 

 few are charted. The observations were suspended from September 13-27, 

 during mv absence in St. Louis. 



GENRR.\L INFERENCES. 



26. Efficiency of apparatus. — The experiments as a whole are in the first 

 place to be regarded as a severe test, an:iountmg almost to a strain of the method 

 emploved. Since both maxima and minima are registered with equal facility 

 at exceedingly low as well as at high atmospheric temperatures, the temperature 

 error of the method is not menacing. Rain minima, snow minima, and other 

 cxccedinglv low nucleations have all been recognized. The high maxima re- 

 ([iiire a long apparatus, since the diameters of the corona occup}' nearly half 

 a meter at the condensation chaml)er ; but the maximum of atmospheric nuclea- 

 tion does not, except in very rare cases, exhaust the limit of measurable 

 coronas. Nevertheless a higher pressure difference would make it possible to 

 confine the measurements to normal coronas exclusively — a desideratum, inas- 

 much as Chapters VII. and VIII. have shown that after the g-b-p corona is 

 reached, the data lying in the region of the cusp are estimates. Higher pressure 

 differences are somewhat less convenient, and in regions remote from cities it 

 is improbable that the normal coronas will ever be exceeded. Care must be 

 taken to have the influx ]oipe far enough from the walls to guard against ex- 

 halations from the building in smnmer. If this pipe is sufficiently long and 

 thin (influx current rapid), the air arrives in the condensation chamber almost 

 at room temperatiu-e and without appreciable loss of nuclei during the passage. 



27. Variations of nuclcation. — Mere inspection of the charts shows the 

 extreme variability of atmospheric nucleation, though quiescent periods are 

 also met with. A part of this must be a local effect, even if the changes cor- 

 respond to the weather. Work in cities where there is so much chance for the 

 pollution of the atmosphere is to some extent unsatisfactory, but the method 

 could hardly have been developed in the country. It is probable that even the 

 small variations of the chart (if they exceed 2000 nuclei per cub. cm.) are real. 

 If the nuclei were colored, the atmosphere would look like mottled soap with 



