l6 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOr,. 71 



lations for the first day following' were omitted in 1914 and for the 

 first and second days in 191 5. 



The computations were based on 720 observations each year and 

 the resulting" correlations are shown in table VII and in the curves 

 in figure 4. In these plots, ordinates represent correlation coefficients 

 and abscissas days following observed values. Curves i, 2, and 3 

 show the coefficients for the years 1913, 1914, and 191 5. The 

 coefficients are not large and the curves differ from year to year. 

 However, the coefficients for 1914 show a remarkably regular seven- 

 day period and the curve for 191 5 shows an oscillation approximating 

 the period of a solar equatorial rotation. When the years were 

 divided into parts there remained evidences of periodicity, but the 

 results indicate that if such periodicities exist, they change rapidly 

 from time to time, first one class predominating and then another. 

 Thus, in the early part of 1915, there appeared to be a period of about 

 26 days with minor maxima at intervals of six to seven days. During 

 the month of June to August there appeared to be a period of about 

 22 days and from October to December a period of about 30 days. 

 In figure 4, plot 4 shows the coefficients for the interval January to 

 June, 191 5, and plot 5, the coefficients for the interval October to 

 December, 191 5. 



On the receipt of the measurements of solar radiation from 1913 

 to 1914 (Smithsonian Misc. Coll., Vol. 66, No. 5), the correlation 

 factors for solar radiation were computed in the same way as for 

 the temperature of Buenos Aires, and although the solar observa- 

 tions were very broken, the results indicate that the two sets of values, 

 solar variation and temperature in Argentina, showed similar varia- 

 tions. The curve for 1913 was published in Smithsonian Misc. Coll., 

 Vol. 68, No. 3. This conclusion is now much strengthened by 

 Dr. C. G. Abbot's investigation, " On Periodicity in Solar Variation," 

 Smithsonian Misc. Coll., Vol. 69, No. 6. In computing the correla- 

 tion factors for 1913, I divided the year into two periods, that is, 

 the deviations used in the computations were taken from two means, 

 one embracing more or less the first half of the period of observation 

 and the second the second half. This was done in order to eliminate 

 a long period oscillation resulting from a marked decrease in the 

 solar values during the other part of the year. In the other years 

 treated by Dr. Abbot this change was not necessary. Dr. Abbot's 

 coefficients of 1914 for solar variation are given in column a, 1914, 

 of table VII and are plotted in curve a of figure 4; while those for 



