8 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 69 



about 27. days, no doubt associated with the solar rotation. There was 

 evidently during this season a tendency toward a hot and cold side of 

 the sun, which persisted during several solar rotations but diminished 

 at the latter end of the season. Such a result is evidently a new proof 

 that the variations we find are truly solar, for they have a well-known 

 solar period in 1915. 



(4) Not less extraordinary is the result for 1916. The 27-day peri- 

 odicity seems to be no longer present, but 11-J full periods, as reg- 

 ular as the time intervals of 24 hours between observations per- 

 mit, occur in 40 days. This periodicity is then approximately 3.5 

 days. It is unique among the whole series of years. If the range of 

 the correlation factors was smaller I would regard it as surely due to 

 accidental error. But the range averages more than 50 per cent from 

 crest to trough in correlation factors whose probable error is only 

 about 8 per cent. It is really a most extraordinary result. 



(5) The years 1909, 1910, and 1914 show a similarity in the march 

 of correlation factors. From strongly marked positive values during 

 the first week the coefficients fall to minimum negative values after 

 about 18 days, and then, on the whole, tend to approach zero towards 

 the end of our 40-day period of investigation. In the seventh 

 curve of the figure, corresponding with the last column of Table II, 

 I give the mean of correlation factors from all three years. This 

 curve brings out in addition to the tendency just noted, a fairly well 

 marked indication of a periodicity of 7| days. 



(6) The results for the remaining years 1908, 191 1, and 1913 

 differ from all the others and from each other, but on the whole if 

 they stood alone would give less ground for a belief in the periodicity 

 of solar variations than the group of three years we have been dis- 

 cussing, and much less than the years 191 5 and 1916. In the sixth 

 curve, corresponding to column 10 of Table II, I give the mean values 

 for these three years. 



(7) To sum up the investigation, we find in 1915 a well-marked 

 hot and cold side of the sun persisting through several solar rotations. 

 This occurred in a year near sun-spot maximum. The years 1909, 

 1910, 1914, either of moderating or of slowly increasing solar activity, 

 show tendencies toward periodicities of solar variation, not very 

 marked, but somewhat in common over the three seasons. The years 

 1908, 191 1, and 191 3 yield little of interest. The year 1916 yields a 

 unique and extraordinary result. No definite periodicity in solar 

 variations of short interval persists year after year. 



