64 ANNALS NEW YORK ACADEMY OF SCIENCES 



If we now compare more closely the curves of Kazan, Warsaw, and 

 Bucarest with the Arequipa curve, we notice sufficient similarities to 

 grant that tlie primary cause of the Kussian variations is most probably 

 the same as that which produces the equatorial variations. The main 

 cause of the complications in the geographical distribution of the excess 

 and deficiency of temperature has to be ascribed to the perturbations of 

 atmosplieric circulation and transport of water vapor. We have to admit 

 that if the value of the solar radiation changes, the temperature at the 

 earth's surface must change; but the total atmospheric pressure remains 

 the same. Consequently, a rise or fall of temperature must produce 

 abnormal changes in the distribution of atmospheric pressure. These 

 changes will affect the winds, the rainfall, and also the temperature. 

 The normal, or let us say the Arequipa, variation of temperature must 

 therefore undergo, in different regions, all sorts of modifications due to 

 the local conditions of atmospheric circulation. 



This fact explains the coexistence, and mutual dependence, of pleions 

 and antipleions and explains also, to a certain extent, the persistence and 

 more or less progressive displacement of the pleions from one region to 

 another. 



On the other hand, some, at least, of the brachypleions may be con- 

 sidered as peripheric trepidations of the pleions. 



At present this interpretation is evidently but a simple working hy- 

 pothesis for investigations yet to be made. It will, however, be suffi- 

 cient to compare the curves of Aachen and the Scandinavian stations 

 with those of figures 59 and GO to arrive immediately at the conclusion 

 that the brach^'pleionian oscillations are not at all a particularity cliai- 

 acterizing the purely maritime climate of oceanic islands, as at first one 

 would have been inclined to think.- 



The temperature scale not being indicated on the diagrams (Figs. 

 10 and 11)1 give in the following table (Table III) the values of the 

 highest and lowest consecutive means and their differences. These fig- 

 ures are °(*. It would have meant too much work to reduce all the fig- 

 ures utilized to draw the curves into mean temperatures and into °C. 

 The utilized figures are simply totals of twelve monthly means. In the 

 ca.se of °('. T added fifty to all figures in order to avoid the negative 

 values of the winter months. For totals of °F. the figures of course give 

 an apparently greater amplitude of variation to the curves. The pre- 

 ceding table will serve to make comparisons possible in case anyone would 

 like to examine the amplitudes of individual crests. For my present 

 purpose such comparisons are unnecessary. 



