24 -^rt- 9.-T. Terada. 



paragraphs, and on the otlier hand constitute the Ijasis of the 

 theoretical considerations to be propounded later on. 



It is the pleasant duty of the author to acknowledge here his 

 indebtedness to Messrs. T. Tatiiri, the late H. Suzuki, M. Nakata 

 and S. Hana who successively served as his assistants and com- 

 putors in examining and making toilsome measurements of the 

 records, and carrying out many tedious calculations, with in- 

 defatigable zeal and true scientific candour. 



Some parts of the statistical work wliich necessitated the 

 utmost care in measurements, foi' example the determination of 

 the phase differences of difïerent components etc. w^ere made by 

 the author himself. 



7. At first, the photographic record were examined sheet 

 by sheet, the conspicuous weaves constituting more or less regular 

 trains were selected and the approximate mean periods determined 

 by dividing the time interval by the number of waves contained 

 in the train. ^^ It is indeed difficult to decide wliich trains to take 

 and which to omit. It was therefore decided to take as many 

 trains as possible with different periods, provided that the trains 

 comprise at least three very regular simple waves with sensible 

 amplitudes. When during the course of an hour a very long 

 continuous regular train with nearly constant period occurred, 

 the period was determined only for one or two portions of the 

 train chosen arl)itrarily from the most conspicuous parts. 



The case is more difficult when the fundamental waves are 

 superposed with different " overtones." Very complicated waves, 

 liable to ambiguities, were omitted, even if it were possible to 

 analyse them into simple components, and the apparently regular 

 trains only were taken into account. The results of the ex- 

 amination wei'e systematically tabulated, separately for each of 

 the three components, according to the successive hours of day. 

 These tables formed the Imsis of some of the following in- 

 vestigations. 



1) Strictly speaking, the periods are sometimes variable even in an apparently reg-ular 

 train of simi^le waves. The inconstancy is too considerable to he explained Ijy the ir- 

 regularities of motion of the drum carrying the j^liotographio paper. 



