INTRODUCTION 



of populated areas will make these problems more and more urgent as time 

 goes on. 



In a general way there are two approaches to long-range forecasting. One 

 might be called the map method and the other the cycle method. The map 

 method applies to short-range weather forecasting and delineates the position 

 of storm areas, of polar fronts, and of air masses with definite characteristics. 

 The average motion of these masses is well known and each day upon receipt 

 of reports it is safe to make a short prediction by means of a series of maps 

 indicating the temporary position and condition of the several factors. As 

 these conditions become expressed for larger and larger geographical areas, 

 the predictions can extend (less precisely) farther into the future. A mete- 

 orologist who understands the known factors and the mechanics of the motion 

 of air masses can sometimes make predictions of value up to some months in 

 advance. 



Belonging to the same class, and yet depending on factors whose relation- 

 ships are less well known, there is a form of prediction depending upon ocean 

 temperatures. This basis of prediction is under special study at the Scripps 

 Institution of Oceanography on the coast of southern California. 



Long-range forecasting by a cycle method is highly desirable on many 

 grounds. In the first place, the use of the forms of analysis described in this 

 paper establishes the fact that an important proportion of climatic changes 

 may be expressed in terms of well-defined periods 1 that have four to twenty- 

 five or more repetitions. It is here believed that it is a mistake to neglect 

 the examination of climatic changes because they do not seem permanent 

 and exactly timed and are not molded in an artificial form like a sine curve. 

 It is believed that many earlier investigators have committed an error be- 

 cause they disregarded variations that could not be expressed in terms of 

 harmonic analysis. Observed variations have often been called accidental 

 with the implication that they were worthless for investigation. Such an 

 implication might be justified by the results of harmonic analysis, but should 

 be reconsidered in view of the methods here used. 



In the second place long-range forecasting by the cycle method is urged 

 because in cyclogram expression there are evidences of relationship between 

 climatic changes and solar variations. In the third place we naturally think 

 of atmospheric phenomena on the earth as full of random changes, but these 

 changes are found to agree over large areas of country and we conclude that 

 we are encountering something general and highly important. Similarities 

 over extensive geographical regions are among the important facts to be 

 considered in this book (Chapter V). 



But there is still another reason for investigating climatic cycles. We 

 have developed accurately dated chronologies 1900 and 3200 years long which 

 possess in some cases high climatic value. In these we are finding recurrent 

 phenomena or repetition according to some plan. This new investigation we 



1 The word "period" is used with a little stronger sense of precision and stability 

 than cycle. 



