4 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. Ill 



January for instance — yields first a table of either four or five lines, 

 depending on whether some January days were required to complete 

 lines of December or February, or whether some days from one or 

 both of those months were required to complete lines of January. 



Mean values having been taken, as in table i, there resulted a new 

 table of 21 lines, of 7 columns each, for each one of the 12 months 

 of the year, the entire 12 tables covering the interval 1928 to 1948. 



FREQUENCY DISTRIBUTION 



From the 12 tables of mean monthly temperature departures were 

 now read off from their 21 lines the days (i to 7) on which maxima 

 and minima of temperature occurred. Where identical mean depar- 

 tures came on more than one day, the reading was appropriately 

 split up, as, for instance the same minimum occurring on days i and 

 7 was tabulated as ^ at i and ^ at 7. 



Terrestrial responses to solar impulses lag behind their solar causes. 

 Thus, for example, the coolest and warmest parts of the day occur 

 several hours after midnight and noon, respectively. The lag differs 

 from place to place, and from time to time, depending on terrestrial 

 complexities. So it should not be expected that maxima and minima 

 of temperatures at New York and Washington, due to a periodic 

 solar change, would necessarily be coincident, or that at either sta- 

 tion they would always be found at the same columns of the tables. 



If, as hitherto generally supposed by meteorologists, weather is 

 almost wholly governed by terrestrial influences, apparently acci- 

 dental as to timing, then a study of the frequency of maxima and 

 minima in my tables should show no marked preference for any 

 columns over the others. But if, as indeed will be shown below, the 

 regular solar periodic pulse is a principal cause of weather, then, 

 despite the interference by terrestrial complexities, there should be 

 found marked preferences for certain columns as seats of maxima 

 and minima. Chance is largely subordinated by multiplicity in 

 this study, for each month of the 12 monthly tables results from 



21 X — =96 recurrences of the period. 



It was presumed that the lag of temperature response might differ 

 in different parts of the year. Hence the 12 months were tabulated 

 separately. Moreover, a great collection of unpublished studies on 

 the terrestrial responses to 14 long-period solar variations has shown 

 that such responses are apt to differ greatly with sunspot frequency. 

 Hence, in each of the 12 monthly tables, the years 1928-30, 1936-41, 



