ii DEVELOPMENT AND ENVIRONMENT 53 



Another form of the same phenomenon is presented 

 when we examine data for plants sown at varying times.- 

 This was first recognised by the author when plotting 

 data for sowing and picking- dates in a series of years, 

 provided from various large estates in Egypt. The graph 

 was interesting as showing the influence of district on 

 sowing-date, and on picking-date in consequence, but 

 more especially so because it indicated in spite of the 

 many sources of error that it is useless to sow before a 

 certain critical date in any given district. However early 

 or late the sowing may be, the crop "arrives" at the 

 same mean time until this limiting date is exceeded, after 

 which a delay in " arrival " results, following the delay in 

 sowing. Ultimately, of course, the paraboloid " arrival 

 curve" will leave the "sowing curve," since plants sown 

 absurdly late will not flower at all. 



The same phenomenon was shown on a smaller scale, 

 though with greater precision, by weekly sowings on a single 

 site in a single season. The mean date of the first 

 flower was actually slightly later in the earliest sowing, on 

 account of the larger number of stunted plants. Three 

 Upland sowings indicated the same behaviour as the 

 Egyptians, but within a shorter period. Sowings made 

 on the critical date were not merely superior in arrival, 

 but also in height, in yield, and in all respects. 



We cannot, of course, lay down rigid rules as to the 

 date on which to sow in any place, since the weather 

 varies a little from year to year. Nevertheless it seems 

 clear that the tendency in Egypt is towards sowing too 

 soon. Thus, March 24 was the critical date for Giza in 1911. 



The variations in night-temperature from year to year 

 during the six weeks before flowering probably constitute 

 the chief cause of variations in first-flower date, and hence 

 in crop arrival, since the growth-rates of all the branches 

 are closely correlated until June. There have recently 

 been two striking examples of this in the early crop of 

 1910 and the late one of 1911. 



