20 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 7I 



out in the means of three days and there appear other maxima a, h, 

 c, d, at longer intervals apart. 



These maxima remain apparent in the five- and seven-day means, 

 grow faint in the nine-day means, and disappear in the ii-day means ; 

 there then appear maxima A and B at still longer intervals. In 

 obtaining the numbers for these mean curves the values observed in 

 the latter part of July were also used. The different classes of waves 

 appear even more distinct when the means for the longer periods are 

 subtracted from the shorter periods, so that the latter remain as 

 residuals. Thus, when the means of three days are subtracted from 

 the observed values the two- to four-day waves stand out distinctly 

 as shown in plot X in figure 5. The next set of waves shown in the 

 plot, in which the maxima are nine to ten days apart, are best found 

 by subtracting the means of 11 from the means of five and plotting 

 the residuals as shown in curve Y. The next set of maxima which 

 are usually some 25 to 30 days apart are best shown by the difference 

 between the means of 15 and 30 days. In the present case the data 

 for these differences are lacking in the first part of the month, so that 

 the dift"erence between the Ii-day means and a constant value of 1.950 

 is shown by the broken curve, 2, and the values of 1 5 minvis 30 days 

 are shown by a dotted curve beginning on the eleventh day. In the 

 curves x\ y', and s' are shown plots of the residual temperatures of 

 Buenos Aires obtained from the observed minus three-day means, 

 the five-day minus ii-day means and the 15-day minus 30-day means. 

 In these cases the plots are inverted, temperature below normal being 

 plotted above the line, it having been determiijed by preceding investi- 

 gations that the relation between temperature at Buenos Aires and 

 solar radiation is inverted in winter. 



The relation of the shorter waves of temperature to those of solar 

 radiation is not evident, but the numbers i, 2, 3, etc., are placed where 

 they appear to correspond with similarly numbered solar waves. In 

 the waves a, b, c, d, there is clearly a lag in the occurrences of the 

 maxima and minima of temperature with a mean of three and one- 

 half days and a similar lag in the long waves, A and B. 



In order to ascertain whether each class of waves thus outlined 

 followed the same periodic changes, correlation factors for successive 

 days were worked out for each successive set of waves. For this 

 purpose the temperature observed at Buenos Aires, at 8 A. M. and 

 8 P. M. were averaged by progressive means of 3, 5, 7, 11, 13, 15, 

 etc., and differences were obtained of observed minus three-day 

 means, of five minus eleven, etc., taking in each case the differences 



