May 7, 1903] 



NATURE 



rainfalls of Mailand, Padua, and Klagenfurt, and 

 found a well-marked recurrence of the wet and dry 

 periods every thirty-five years, the mean epochs of the 

 former being^ 1808, 1843, and 1878, and of the latter 

 1S23, 1859, and 1893. 



In determining^ the variation of rainfall over such 

 long periods as that of thirty-five years, it is necessary, 

 if possible, to smooth the curve representing the varia- 

 tion from year to year, 

 for this curve, as a rule, 

 displays large fluctua- 

 tions from the normal in 

 the course of a very few 

 years, and it is not easy 

 for the eye to grasp the 

 longer periods of varia- 

 tion ; these long periods 

 may to some extent be 

 rendered more apparent 

 by coupling up together 

 the mean values of the 

 rainfall for several years, 

 and forming another 

 mean, but somewhat fic- 

 titious value, for each 

 successive year. Thus, 

 for instance, the mean 

 for one year, say 1870, 

 might be computed from 

 the means of the five 

 years 1868 to 1872, or the 

 means for 1871 from the 

 mean of the years 1869 

 to 1873 ; instead of a five- 

 year mean, a ten-year or 

 a fifteen-year might be 

 chosen. 



In the figure here 

 given, five-year means 

 have been adopted, and 

 the curves resulting from 

 these have been further 

 smoothed by drawing 

 freehand another curve 

 to eliminate as far as 

 possible the smaller 

 fluctuations of short 

 period that still exist, 

 even after still minor 

 changes have been elim- 

 inated. The stations, 

 the rainfall curves of 

 which are here given, 

 have not been specially 

 selected, but simply 

 taken as the data for 

 them were easily avail- 

 able, and they afforded 

 long records for the study 

 of such variations as are 

 here discussed. The 

 short curve for the 

 British Isles is attached 

 so that not only can a 

 comparison be made of 

 this record of the Meteor- 

 ological Office with that obtained by the late Mr. 

 .Symons, but that the actual variation over the islands 

 taken together can be compared with two widely 

 separated stations in them, as Greenwich and Rothesay. 

 The European continent is here represented by Brus- 

 sels, the epochs of the maxima and minima of the 

 rriinfall curve of which can be compared with the values 

 Iven by Hann and referred to in a previous paragraph. 



NO. 1749, VOL. 68] 



Two stations in India, Bombay and Madras, one station 

 in South Africa, Cape Town Observatory, and lastly 

 three stations in the United States of America repre- 

 senting the rainfall of the Upper Ohio Valley, com- 

 plete the rainfall information here given. 



A general collective glance at these curves shows that 

 there is an undoubted long period variation in all the 

 stations here brought together. Further, that the 



ItOO 



19 00 



— Curves showing the relation between the 35-year sunspot period and that of the Briiclcner rainfall cycle. Each 

 of the rainfall curves is determined from the means of five-years, and the.se curves are smoothed by freehand 

 drawing in order t_i show the long period variation of rainfall. The smoothed curve through the eleven-year 

 sunspot curves indicates the epochs of the long period sunspot variation. 



periods of greatest rainfall occur generally in the years 

 181 ^, 1845, ^nd 1878-83, while those at which the rain- 



fall is decidedly deficient are about the years 1825-30, 

 i860, and 1893-5. 



With the existence of these very definite fluctuations 

 it is important to notice that the last minimum or dry 

 period which is most apparent in the case of the curves 

 representing the British rainfall seems now to be just 



