525 



In the case of snow unaccompanied by any but inconsiderable winds, the readings at the lower level 

 are, as we have seen, not sensibly increased; those at the upper level are, on the whole, diminished. In 

 one case indeed, the 15-feet reading was actually less than the 5-feet reading, indicating an apparent 

 reversal of the field above the 5-feet level. It must, however, be remembered that the observations at 

 the two points were not simultaneous, the high-level observations being taken immediately after those at 

 five feet. The possibility of defective insulation in the case of the high-level observations during a fall 

 of snow must also be borne in mind. 



4. DAILY AND YEARLY VARIATIONS OF POTENTIAL. 



Observations taken during the occurrence of drift must certainly be treated as abnormal, and are 

 therefore excluded from the data used in the investigation of possible daily and annual variations of 

 potential. The comparatively few other occasions when the potential exceeded 300 volts, as well as 

 cases of zero or negative potential, have also been excluded. 



Daily Variation. 



Only during the summer months, November, December, and January, were the observations numerous 

 enough to admit of the possibility of a daily variation being detected. These three months are alike in 

 giving evidence of the existence of a minimum in the middle of the day or early afternoon. In December, 

 1902, and January, 1903, observations were taken at 10 a.m., 4 p.m., and 10 p.m. 



The observations for November were more numerous and were distributed over the 24 hours. The 

 minimum occurs in the middle of the day, the maximum being in the early morning soon after midnight. 

 Dividing the day into six equal periods of 4 hours, and taking the mean of the potentials observed during 

 each of these periods, we obtain the results given in the following table : 



The only continuous series of hourly observations extending over so much as 24 hours was one carried 

 out in November, 1902. It extended over three days, from 8 a.m., November 22, to 8 a.m., November 25 ; 

 the weather conditions were fairly satisfactory, with no snow or drift and little wind, but with a good 

 deal of cloud. 



