138 CLIMATIC CYCLES AND TREE GROWTH 



characters and perhaps other factors cause different localities to emphasize 

 different individuals of this cycle complex. 



(2) For any locality the cycles existing in the last 100 to 300 years, or 

 more, are found and then each cycle still operating is projected forward to 

 construct a curve for a number of years in the future ; these averaged together 

 produce a mean which is to be regarded as the expected smoothed value of 

 a few succeeding years. 



(3) Less well established but still with some definite foundation in the 

 long chronologies there is the recurrence of several well-defined cycles at 

 intervals of about 270 years. In this recurrence we believe we are finding 

 signs of larger cycles or periods which will aid in removing the uncertainties 

 that result from discontinuity in our climatic cycles. 



PREDICTION TECHNIQUE 



We feel that prediction and verification in the popular sense can hardly 

 be a reliable guide as to validity of method at the present time. Percentage 

 success or failure is very deceptive when watched for a year or two and means 

 little in such a complex problem. The problem at this moment is not to 

 predict but to lay a basis of knowledge and method out of which conservative 

 prediction will develop. It therefore is best at this time to carry through a 

 prediction process in a crude preliminary way without any attempt to make 

 it exact in the sense of a prediction itself. It merely serves to call our atten- 

 tion to various features that require attention. 



Fundamentals of Prediction — The following three classes of prediction are 

 based on a consideration of errors or departures of the subsequent facts 

 from the predictions: 



(1) Prediction that the mean value will always occur in the future, let the 

 departures come as they may. This is the kind we find in climatic maps and 

 has its use as important general information. 



(2) Prediction on the average departure from the mean, figuring what 

 ought to come if it has not happened recently. This is straight probability 

 prediction from the known distribution of the various departures from the 

 mean, and the probability of any particular value coming next. It is exten- 

 sively used by engineers. 



(3) The third class of prediction is a probability prediction, not from a 

 mean straight line, as before, but from a sliding scale, a variable mean, which 

 embodies climatic cycles as well determined as possible and, if possible, in- 

 cludes a short cycle which we have called the two-year or "scatter" cycle. 



In carrying out our plan of illustrating the prediction process, we en- 

 counter at once two parts of the problem that need careful thought. At 

 first thought it seems perfectly easy to state the cycles that are operating 

 now. Then we realize that several repetitions are needed to cause recogni- 

 tion of a cycle. The average number involved in recognition is three or 

 four, the average total repetitions before change is eight to twenty. It is 

 a quite important matter to consider how soon an incoming cycle can carry 

 important weight. Some of the recent ring records in Arizona show a five- 

 year cycle operating since 1905. Should we give it full weight? Some cycles 

 are not perfectly constant in length or amplitude. Should we give more 



