1905.] Rainfall and the Yield of Wheat of the Following Ymr. ;153 



" computed yield,"' a comparison with the actual yield for the 21 years 

 shows that the computed yield agrees with the actual yield within 

 half a bushel in 7 years out of the 21. In 14 years the agreement 

 is within 2 bushels ; in the remaining 7 years the difference between 

 computed and actual yield exceeds 2 bushels. The extreme variation 

 of yield in the 21 years is 9 bushels, from 26 bushels per acre in 1892 

 and two other years, to 35 bushels per acre in 1898. 



Of the 7 years for which the formula gives yields differing from the 

 actual by upwards of 2 bushels, 1896 is the most conspicuous, its 

 actual yield exceeds the computed yield by 4*5 bushels. 



These 7 years all show anomalous seasons. Taken stiiatim, they 

 are 1887, 1888, 1893, 1895, 1896, 1899, and 1903. 



In 1888 and 1903 the crops were washed away by 10 inches of rain 

 in the summer; 1893 is the year of phenomenal drought and the 

 crop was below the computed figure by 2 '5 bushels. The years 1892 

 and 1899 are interesting, because though the amounts of rain were 

 up to the average, the former had 8 dry weeks and the latter 10 dry 

 weeks out of the 13 included in the conventional autumn. They 

 were thus dry autumns, the average amount of rainfall being made 

 iip by a few exceptionally wet weeks. The yields correspond with 

 dry autumn values. They are above the average and above the 

 computed figures by some 2 or 3 bushels per acre. 



There remain 1895 and 1896. 1895 was the year of remarkably cold 

 weather, and in that year the yield fell short, but in the following 

 year the deficiency was made up by a yield as much above the com- 

 puted value as the previous one fell short. It would appear that in 

 this instance the productive power not utilised in the year of the great 

 cold was not lost but stored. On the other hand, it must be remarked 

 that 1896 had the advantage of a specially dry winter. 



It appears from these considerations that the dryness of autumn is 

 the dominant element in the determination of the yield of wheat 

 of the following year. The averages of yield and of rainfall are taken 

 over very large areas, and it may be taken for granted that the 

 investigation of the question for more restricted areas would introduce 

 some modification in the numerical coefficients, if not in the form of 

 the relation. 



The data for making such an investigation are not yet in an avail- 

 able form. A comparison has been made between autumnal rainfall 

 for " England, East " and the average yield for the counties of Cam- 

 bridge, Essex, Norfolk and Suffolk, which shows a similar relation 

 but a magnified effect of autumnal rainfall upon the crop, and also 

 two exceptional years which have not yet been investigated. 



