70 



Mr. Gr. M. Whipple. Results of an [Jan. 15, 



II. "Results of an Inquiry into the Periodicity of Rainfall." 

 By G. M. Whipple, B.Sc, F.R.A.S., Superintendent of the 

 Kew Observatory. Communicated by Robert H. Scott, 

 F.R.S. Received January 8, 1880. 



The exceptionally heavy rainfall of the past spring and summer 

 directed a large amount of attention to the records of rainfall in this 

 country, and more than one investigator stated that he had found a 

 certain periodicity existing in the quantity of rain annually collected. 



Dr. Meldrum, Professor Balfour Stewart, Mr. Hennessey, Professor 

 Stanley Jevons, Dr. Hunter, and others, have also widely published 

 theories based upon the assumption that the variation in the yearly 

 amounts of fall depends in some manner upon solar phenomena as 

 exhibited by the changes in the appearance of the sun's surface, thereby 

 indicating a cycle of approximately ten or eleven years' duration ; but 

 even among the supporters of this so-termed " sun-spot " theory of 

 rainfall there are differences of opinion as to the exact nature of the 

 influence an increase of sun-spots would exert upon the rainfall of any 

 locality. 



Mr. G. J. Symons, in "Nature," vol. vii, p. 143, has partially in- 

 vestigated these theories, and shown the ten-year period does not 

 obtain universally. 



After reading a paper by Professor Stewart, in the " Proc. Roy. 

 Soc," vol. xxix, p. 106, " On a Method of detecting the unknown 

 Inequalities of a Series of Observations," it occurred to me to try the 

 method he employed upon terrestrial magnetic changes, upon the 

 annual values of tbe rainfall, in order to deduce from the observations 

 themselves tbe true periodicity, if one existed. 



With a view of dealing with the largest mass of material possible, I 

 took the long series of rainfall observations made at Paris from 1689 

 to 1875, published by M. Marie Davy, in the " Annuaire de l'Observa- 

 toire de Montsouris." 



These I proceeded to discuss on Professor Stewart's plan, but finding 

 it entailed great labour, and held out little prospect of eventually 

 giving a good result, I soon abandoned it, and after several experi- 

 ments, adopted the following method, which would detect at once the 

 presence of a cycle of an integral number of years in duration. 



Starting with an assumption of a period, which I first made five 

 years in length, and subsequently extended, I grouped all the observa- 

 tions together, first in five-year groups, then in six-year, then in seven- 

 year, and so on, year by year, until I reached thirteen years. The 

 means of these furnished a set of curves, showing the variation from 

 the mean in the amount of annual rainfall for each of the years com- 



