Clayton — 27-Day Auroral Period and the Moon. 85 



northern declination of the moon. As shown in column 12 

 the excess of auroras on the fifteenth day, that is at the time 

 of greatest southern declination of the moon, is 46 per cent 

 greater than the mean ; while at the times of minima on the 

 seventh and twenty-first days the frequency of auroras is 20 

 per cent less than the mean. This range of 66 per cent in 

 the frequency of auroras in the totals of three hundred suc- 

 cessive periods is entirely too large to be considered accidental. 

 Each group of auroras of 50 periods given in columns 1 to 

 6 shows the same general distribution of auroras as the totals. 

 In order to see whether the period remains the same through 

 the different seasons of the year, two groups of data were 

 classified into periods, as described before, and summed for the 

 winter and summer half-year separately. The data for the 

 two groups were taken from tables found arranged conve- 

 niently for the purpose, the first by Dr. Veeder in the Pro- 

 ceedings of the Rochester Academy of Science, vol. ii, Plate 

 II, and the other by the editor of the TJ. S. Monthly Weather 

 Review in that publication. The first covers the four years 

 1885 to 1888, and the other the three years from October, 1893 

 to October, 1896. The results are given in columns 11 to 14, 

 Table II. These columns show that there is a strong indication 

 of the reversal of the period between the winter and summer 

 half-years. In the winter half-year the maximum is found 

 near the time of the greatest southern declination of the moon, 

 while in the summer half-year the maximum is found near 

 the time of greatest northern declination of the moon. In 

 attempting to discover the cause of this phenomenon, I noticed 

 that in my table of auroras arranged in periods of 27 days 

 there was a strong drift of the larger numbers to the right, 

 and of such an amount as to indicate a period of about 29 days 

 between them. This period I at once inferred to be the same 

 length as the synodic revolution of the moon, and its cause to 

 lie in the fact that more auroras were seen in the dark of the 

 moon than in the opposite phase. In order to determine to 

 what extent the frequency of reports of auroras was thus 

 affected, the intervals from 1885 to 1888 and from 1894 to 1896 

 were divided into periods of 30 days, beginning each period 

 with the date of full moon as taken from the Nautical Alma- 

 nac. The number of stations in the United States reporting 

 auroras on each date was entered in a table in its proper place 

 beginning with the date of full moon, and the totals were ob- 

 tained for each day of the period during the two intervals 

 mentioned. The results are given in Table III. In column 1 

 are given the totals for each day of the period from 1885 to 

 1888, and in column 2 are given the totals for each day of the 

 period from 1894 to 1896. In column 3 are given the totals of 



