2 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I ID 



constant data, or because of very closely overlapping sequences. Cor- 

 responding to each of the remaining 63 dates I wrote out the values 

 of the solar-constant as observed at Montezuma, Chile, from 10 days 

 before to 12 days after the dates when the hurricanes were first re- 

 ported. On 17 of these remaining dates the 21 -day sequences of solar- 

 constant values, especially in the neighborhood of the hurricane dates, 

 were so fragmentary that I disregarded them. This left 46 dates to 

 be considered with more or less complete solar-constant sequences 

 closely adjacent to the hurricane dates. 



I should remark that I used the solar-constant values as observed, 

 not those marked "preferred" in table 24, volume 6, Annals of the 

 Astrophysical Observatory of the Smithsonian Institution. I also dis- 

 regarded all marks of "grade." I have become convinced by several re- 

 searches that the procedure used to obtain "preferred values" is un- 

 desirable, and that the "grades" of volume 6 were sometimes assigned 

 with prejudice against wild values which after all may have been good. 

 The daily solar-constant values from 1939 to 1946 are as yet unpub- 

 lished but were made available to me. Mr. Aldrich informs me that 

 when assigning "grades" to them he and his colleagues were very 

 careful not to be influenced by the wildness of a value. Hence I felt 

 justified in rejecting a very few discordant values of low grades in this 

 later work. I may add that I rejected the solar-constant values of 

 September 9, 10, and 12, 1930, which fall far out of line, because they 

 preceded a spell of bad weather from September 11 to 24 when only 

 2 days of observation at Montezuma occurred in that whole interval, 

 and these also gave values abnormally low. I also rejected the solar- 

 constant value of September 25, 1923, because it is very abnormally 

 high. 



Table 2 gives the dates of hurricanes, including the entire 69, of 

 which the 17 rejected for fragmentary sequences of solar-constant 

 values are starred, and the 6 omitted as stated above have daggers. 



Table 3 gives the values of solar constants for the 46 remaining 

 hurricanes from 10 days before to 12 days after the date of first report. 

 The number of solar-constant values included in each of the means is 

 given just above the means themselves. The mean values are plotted 

 in figure i. As given in table 3 they should be understood as prefixed 

 by 1.9 calories. 



There appears a gradual descent of the solar constant prior to the 

 date of reports, amounting to 0.0016 calorie. Then comes a sudden 

 drop of 0.0031 calorie to a sharp minimum on the actual day of first 

 reports. After that the solar constant recovers on the third day, but 

 not quite to its former level. 



