CYCLE PROBLEM AND LONG-RANGE FORECASTING 141 



The exact outcome of this mozaic is not important since we have not 

 finished testing our hypothesis; but the advance of technique developed by 

 carrying forward this process is most important. Such questions as these 

 arise: In casting forward the observed cycles, should we give more weight 

 to the latest cycle data than to past series of cycles? What is the practical 

 rate of increase of weight from the past toward the present? How far can 

 we rely on recognition of cycles that may be just beginning? What is the 

 last word on the expected number of repetitions of each cycle operating? 

 Is this number constant through history or not, and how does it change? 

 We see a need for investigating dominating cycles: Should a subtraction 

 process be used in isolating cycles; what value should be given to weak cycles? 

 Why do some curves give most satisfactory sets of cycles and others not? 

 Answering these questions in a more or less complete way will rest in further 

 studies of the long chronologies which were largely developed for that purpose. 

 We have even planned a definite method of attack on these questions in a 

 process which we call "laboratory prediction." Experience has shown that 

 as these methods go forward, further improvements will develop. These 

 long chronologies also supply data which we believe will enable us to locate 

 phases of the sunspot cycle in past history. We recognize also many studies 

 to be made on the relation of solar changes to tree growth since the invention 

 of the telescope in 1610. 



By outlining the process as done herein, we find the data that need to be 

 observed more carefully. We shall gain knowledge and skill both from 

 historical studies of the past and more intensive work on the present. In 

 the course of time, the action of these climatic cycles will be known in more 

 and more detail, and long-range forecasting will improve. It will not be 

 good sense to look for accuracy at the start, and for some years verification 

 of error must be taken as merely so much scientific information that merges 

 into and improves the method. 



