84 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 94 



of weather. These are found to occur at integral multiples of ii^ years 

 measured from January 1819. 



These phenomena, if accepted as facts of Nature, propounded to us 

 several problems : 



A. Why should the sun, a gaseous body, emit complex pulsations 

 of radiation which are of the nature of a fundamental and 1 1 or more 

 overtones? A violin string may do this, but why should a gaseous 

 sphere ? 



B. Why should the terrestrial responses to these pulsations show 

 changes of phase, form, and amplitude at intervals intimately related 

 to the fundamental period of 23 years ? 



C. Are the terrestrial responses of an order of magnitude reason- 

 ably corresponding to the solar impulses? 



For question A, I confess that I have no suggestion to offer. I must 

 leave its solution to those theorists who may be convinced by section T 

 of this paper that there is a real body of facts which prove the existence 

 of complex solar variation. 



As for question B, the most natural hypothesis is to assume that the 

 phases and amplitudes of the solar periodicities themselves change 

 from time to time at intervals related to 11 7 years. Solar-constant 

 observations are not yet of long enough standing to verify this. I have 

 therefore sought to find some helping clue in a regularity of behavior 

 regarding changes of phase among the different stations. In this in- 

 quiry I have compared the changes shown by the 8-, 11-, 21-, 25-, 

 and 68-month periodicities in temperature as presented by the various 

 stations Berlin, Copenhagen, Helsingfors, Greenwich, Cape Town, 

 and Adelaide. It seemed superfluous to examine the precipitation which, 

 as meteorologists are aware, is loosely dependent on temperature. 



I have devised a sort of shorthand adapted to exhibit the results of 

 this comparison. It is shown in figure 38. At the left of each sub- 

 figure will be found the approximate dates of beginning and end of 

 each ii^-year interval for which tabular computations of periodicities 

 were made. Under the names of the stations appear symbols which 

 are designed to represent the types of curves found during the 

 various intervals of 11^ years. These symbols are five in number, 

 but may be combined to indicate that the first half of a curve is of 

 one type, and the second half of another. The symbols are as follows : 

 Numbers i and 2 are vertical and horizontal lines. They represent 

 inverted phases of curves of approximately the same form. Numbers 

 3 and 4 are lines inclined at 45° respectively to the left and the right. 

 They also represent inverted phases of curves of approximately the 

 same form, but of a form essentially differing from that represented 



