8 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. 68 



results for Denver and Chicago also showed a negative correlation, 

 but going- northeastward to St. Johns, Canada, and to Jacobshavn 

 on the west coast of Greenland, and northwestward to Dawson, 

 Yukon territory, a second band of positive correlation is found. Next 

 going southwards from Pilar, negative correlations are found at 

 Valdivia, and Punta Arenas, Chile, and a positive correlation at the 

 South Orkneys near the Antarctic circle indicating very clearly that 

 the tropical and subtropical land stations in the western hemisphere 

 are subject to temperature changes similar to the changes of solar 

 radiation, while the changes in the temperate regions are in the 

 reverse direction and those near the Arctic circle are again similar 

 to the solar changes. This arrangement corresponds closely with the 

 arrangement of the high and low pressure belts of the world. Com- 

 putations were next made for land stations in tropical Africa, for 

 Mauritius, for Manila, and for a station in the Fiji Islands. The 

 results are given in table 5 arranged according to different zones 

 shown in the table. 



The means of these results indicate a maximum positive correla- 

 tion in the tropics on the second day following the solar observations, 

 a maximum negative correlation in the temperate zone and sub- 

 tropical oceans on the third to fourth day following the solar obser- 

 vations and a maximum positive correlation on the second day in the 

 Arctic Zone. 



The later occurrence of the negative correlation indicates that 

 it is a secondary result of the solar action in other parts of the world. 



The most probable explanation is that tropical areas, and especially 

 the tropical land areas, are the parts most heated by the increase of 

 solar radiation. 



This heating causes an expansion of the air over the tropics and an 

 overflow toward the temperate zones, particularly towards the cooler 

 ocean areas in this zone. The final result would be a fall of pressure 

 in the tropics and a rise in the temperate regions causing an intensi- 

 fication of the normal pressure belts of the earth. 



That this is a process which actually takes place seems borne out 

 by the fact that the negative correlation factors are largest on the 

 west coasts of continental land masses near the 35th latitude as at 

 San Francisco and San Diego in the north and at Valdivia in the 

 south, where an increase of intensity in the oceanic centers of high 

 pressure would cause an increase of polar winds ; while on the east 

 coast of the land masses as at St. Johns, New Brunswick, the positive 



