NO. 6 MAGNETIC STORMS — ABBOT 5 



St. Katherine are so much more accurate than any others that the 

 resiihs from other stations must be ignored in a study of small changes 

 of this kind. That restriction cuts ofT a great many dates, because the 

 sequences of solar-constant values, at and near the storm dates, were 

 often too incomplete to be used. With the utmost liberality of selec- 

 tion, I could find but 53 storm dates from 1923 to 1946 when solar- 

 constant sequences observed at Montezuma or St. Katherine were 

 complete enough to be fairly used in the tabulation. Even among those 

 sequences retained, many were so imperfect as hardly to deserve em- 

 ployment. I therefore made two reductions, one employing the whole 

 group of 53 dates, the other employing 22 of them, when the se- 

 quences were at least two-thirds complete, and were not broken badly 

 near the zero dates. However, the mean results of the complete tabu- 

 lation of 53 and the tabulation of tlie 22 most satisfactor}^ sequences 

 are in almost perfect agreement. Hence it may be said that two inde- 

 pendent tabulations, one of 22 cases, the other of 31 cases, yield prac- 

 tically identical results as to the influence of severe magnetic storms 

 on the solar constant. 



My solar-constant data, 1923 to 1939. are taken from table 24, 

 volume 6, Annals of the Astrophysical Observatory of the Smith- 

 sonian Institution. From unpublished daily results, those of 1939 

 to 1946 were kindly put at my disposal by Director L. B. Aldrich 

 of the Observatory. In quoting from the Annals I have used the 

 direct mean values from Montezuma or St. Katherine, and not the 

 "preferred" values. I have come to distrust the method used to 

 obtain "preferred" values, and it has not been used in the reductions 

 of 1939 to 1946. Furthermore I have ignored "grades." They are 

 more or less liable to personal bias, and especially to a tendency to 

 discredit apparently wild values. It is very clear from the present 

 research, and from another I have made on hurricanes, that some 

 wild values are caused by cosmic conditions, not by errors of 

 observation. 



EFFECT OF MAGNETIC STORMS OX THE SOL.\R CONSTANT 



With these explanations given, I now ask attention to table 1. It 

 enumerates the 53 dates retained.- Corresponding to each one is a 

 sequence, more or less complete, of solar-constant values from Monte- 

 zuma or Mount St. Katherine. It extends from 10 days before to 10 

 days after the date marked zero, when the magnetic storm appeared 

 to be at its height. The table has 53 more or less complete lines of 21 



'The phenomenon of March 22, 1920, is of so much prcatcr ranpc of severity 

 that I have treated it separately above, and do not include it in table i. 



