64 SMITHSONIAN MISCELLANEOUS COLLECTIONS VOL. I46 



Considering probable changes of an accidental character that may 

 have happened in the long interval of 22 years 9 months, I do not 

 think a fair-minded critic should object because the latter coefficient 

 is smaller than the former. For, after all, the correlation coefficient 

 for the interval 1942-1948 is almost 50 percent and nearly 8 times 

 its probable error. 



CONCLUDING REMARK 



In the present paper I have summarized 13 aspects of the depend- 

 ence of weather on solar variation. But without unduly expanding 

 the paper, I could not give half of the evidence that supports these 

 positions. My best course is to give in the Appendix references to 

 many papers where, by tables, illustrations, and text, the evidence 

 is amplified for those who may be interested. I would like to call 

 attention particularly to Publications 4088, 4090, 4103, 4135, 4211, 

 4213, 4222, 4352, 4390, 4462, and 4471, where much that I have not 

 crowded into this paper will be found. I have suggested there one 

 theoretical hint as to the nature of the connection between solar 

 variation and weather (P. 4211, pp. 10-11). Also the atmospheric 

 conditions that prevent discovery of the family of regular periods 

 in weather by cursory tabulations are more fully explained. I am 

 still hopeful that meteorologists in America will at length see that a 

 useful measure of long-range forecasting, even to a half century 

 in advance, can be attained by using the records of the past with due 

 attention to atmospheric conditions. 



