NO. 3 SOLAR RADIATION AND THE WEATHER — CLAYTON 1 5 



takes place, probably owing to the predominating mass of the land 

 surfaces in the northern hemisphere, which being heated causes an 

 overflow of air to the south with rising pressure and falling tem- 

 perature. It is evident, however, from the data at hand that seasonal 

 changes are not the only causes of reversals and that complicated 

 movements of air masses are involved which cannot as yet be 

 explained. 



The correlations at eight to nine days later, shown in table VI, also 

 show an annual inversion, but in an inverted sense to those at three 

 and one-half days later; as shown by the following figures for the 

 correlations for eight and one-half to nine days following the solar 

 observations : 



Jan. Feb. March Apr. May June July Aug. Sept. Oct. Nov. Dec. 



—.49 —.20 -f .18 4 .23 -I-.33 . . . . -1-.20 +.26 —.23 —.29 —.33 



Study of Periodic Changes in the Temperature at Buenos 

 Aires by Means of the Correlation Factor 



The great interest in any possible periodicity in the weather as an 

 aid to forecasting has led me to try many different kinds of attack 

 on the problem. One of these was by means of the correlation factor 

 which has come much into use during the past few years. The 

 annual and diurnal periods were well known, and the problem was 

 to find any periodicities which might exist in the variations from these 

 known periods. A long series of years of observations at Buenos 

 Aires have given accurate mean values of the temperature for each 

 hour of the day and for each day of the year. These values show 

 very clearly the annual period in the temperature with the maximum 

 in January and the minimum in July and the daily period of tempera- 

 ture with the minimum near sunrise and the maximum about 2 P. M. 

 Subtracting the mean, or normal values, from the values observed at 

 8 A. M. and 8 P. M. of each day, shows the abnormal oscillations of 

 temperature. The method of research was to correlate the tempera- 

 ture departures from normal with those found one day later, two 

 days later, three days later and so on successively to 40 days later. 

 These computations were made for the temperature abnormals at 

 Buenos Aires for the years 1913, 1914, 1915. It was found that, 

 owing to the persistence of temperature effects, the correlation with 

 the following day was quite high, being 73 per cent in 191 3. As the 

 object of the research was to find evidence of periodicity, the calcu- 



