138 ANNUAL REPORT SMITHSONIAN INSTITUTION, 1935 



fail and a warning will then be of great value. But those who pre- 

 pare formulas by the selection, based merely on the closeness of their 

 apparent relationship, of a few out of many factors must remember 

 that they cannot expect the value of all these factors to be main- 

 tained ; and if they have a forecasting formula which on paper works 

 out with a coefficient of, say, 0.75, they must realize that this is in 

 reality probably not more than 0.6, or in some cases even 0.4. And I 

 would plead for a much severer standard in handling questions of 

 periodicity. If these views are right, no anticipations should be 

 published except on the strongest evidence of excess or defect until 

 the experience of 15 or 20 years has justified a less cautious policy. 

 Finally I would express the hope that the subject may, by its 

 potential value to the race, and by the many-sided nature of its 

 interests, enlist the services of some of my hearers who are qualified 

 to unravel some of its intricacies. 



