.N'(J. I SOLAR VARIATION AND HURRICANKS ABBOT 3 



Although the depressive effect is very clear in the mean values, not 

 all the sequences of solar-constant values show a depression on the day 

 of first report of a hurricane. In fact, of the 46 cases used to form the 

 means, there are 14 in which on the day of first report of the hurri- 

 cane the solar constant was observed as high, and in a few cases even 

 slightly higher than on the days preceding and following. I doubt 

 if this discroi)ancy should he attrihutrtl to acciilciital error. 1 Icnce we 



Table 2. — Dates of fust report of major West Indian hurricanes, /y.'j to 1946 



I'fjj: Sept. 24. 



1924: .\ug. lO, 27; Sept. 14. 



1925: Nov. 29*. 



igzb: July 22; Aup. 21 f; Sept. 6; Oct. 14. 



u)2S: Sept. 6 f. 



ig2[): .Sept. 22. 



1930: Aug. 31. 



1931 : Sept. 6*, 9t. 



1932: Aug. II, 30*; Sept. 26*; Oct. 31 ♦. 



103.^ : June 27 : July 25 * ; Aug. 17 *, 28 t, 31 t ; Sept. 10, 16 ; Oct. i. 26. 



• 934: June 4: Sept. 5; Oct. i ; Nov. 21. 



1935 : Aug. 31 * ; Sept. 23 f ; Oct. 19, 30 *. 



1936: July 27; Aug. 28*; Sept. 8, 24. 



1937: Sept. 14, 20. 



1938: .-Vug. 14, 23; Sept. 16. 



1939: Oct. 12, 29. 



1940: .\ug. 5, 30; Sept. II. 



1941 : Sept. 18, 23 * ; Oct. 3. 



1942: Aug. 21 ; Nov. 5. 



1943: July 26; Aug. 30; Sept. 11 ; Oct. 11 *. 



1944: July 30; Aug. 16; Sept. 8*, 19*: Oct. 13*. 



1945: June 20; Aug. 24; Sept. n *; Oct. 2*. 



i04^»: Sept. 12; Oct. 6 *. 



• Rejected because of fragnienlary sequences of solar-constant values. 



♦ Omitted because of almost complete lack of solar-constant data. 



are not to infer that a depression of the solar constant is alwavs nec- 

 essar}' to bring on a hurricane. Nevertheless frequently it appears to 

 be the iminilsc which starts the cataclysm. 



Lest some critic should suspect that the ly dates rejected for frag- 

 mentary solar-constant sequences might have been unfairly rejected, 

 and, if included, would lead to a different conclusion, I have thought 

 well to take the mean values for all of the 17 rejected sequences, frag- 

 mentary though they are. The means are given in table 4 with the 

 numbers of values entering into each mean. The means of these 

 values are very divergent, partly because the observations are few, 

 but more because parts of these sequences lie at different levels of 



