166 President's RemarJcs on Rainfall. [Nov., 



have at our command the registers of at least three or four sun-spot 

 cycles and of a much larger number of stations. And I think it would 

 be premature therefore to discuss Mr. Pearson's conclusions from any- 

 other point of view than that of the validity of the grounds on which 

 they rest. Mr. Pearson's data are taken from a paper by Mr. Hill, and 

 comprise the registers of 20 stations scattered over Northern India, and 

 extending altogether over four sun-spot cycles. But the first of these 

 is represented by 1 station only, the second by 2, the third by 5, and 

 the fourth by 17. In computing the mean variation, the figures of 

 each cycle are weighted according to the number of stations represent- 

 ing it, and hence it follows that the resulting variation is mainly that 

 of the last cycle. 



" How far the rainfall variation of Northern India may be considered 

 as fairly represented by 20 stations, may be judged, in some measure, 

 from the following comparison of the variation of five years in portions 

 only of Northern India, as computed from a smaller and larger number 

 of stations in each case ; (in all but one, from more than 20 stations). 



Province. 



Number of 

 Stations. 



Nominal 

 Average. 



1877 



Per Centage variation. 

 1878 1879 1880 



1881 



Punjab 



C 29 

 (112 



214 



20-1 



+ 14 

 + 11 



+ 18 

 + 11 



— 21 



— 24 



— 22 



— 16 



+ 1 

 + 5 



N. W. Provinces C 45 

 and Oudh ( 230 



356 

 34-6 



— 43 



— 44 



— 5 



— 8 



+ 34 

 + 37 



— 26 



— 28 



— 3 



— 1 



Behar 



C 13 

 I 27 



43-4 

 43*4 



— 22 



— 20 



— ]2 



— 10 



+ 22 

 + 21 



+ 8 

 + 6 



+ 6 

 + 8 



Lower Bengal 



C 29 



( 83 



66-7 

 69'0 



+ 7 

 + 5 



+ 7 

 + 7 



— 6 



— 8 



+ 11 



+ 8 



+ 4 



+ 2 



" Now it appears that, in 20 years, the mean annual deviation of the 

 rainfall of any one year from the general average, in each of these four 

 provinces, is as follows : 



Punjab 



± 



13 per cent. 



N. W. Provinces 







and Oudh 



+ 



23 ,, „ 



Behar 



+ 



18 i 9 ,1 



Lower Bengal 



+ 



10 „ „ 



and as far as any conclusion may be drawn from the comparison just 

 given, it would seem that, assuming the figures obtained from the more 

 numerous stations as true values, the mean error and maximum error 

 of the result for any one year, as derived from the smaller number of 

 stations, are in 



