72 THE COTTON PLANT IN EGYPT CHAP. 



plot was not disturbed and always produced superior 

 plants during the three years over which the experiment 

 lasted, but the three terraces were quite comparable. 

 Each terrace carried not less than 300 plants per annum, 

 sown in ten ridges, under conditions which were otherwise 

 those of the field crop. The water-table in this site was 

 chiefly controlled by changes in the level of the Nile, at a 

 distance of nearly a kilometre, though there were variations 

 due to impervious patches of soil, even within this small 

 area. 



In the year 1909 the Nile rose abnormally early ; in 

 1910 it rose late; 1911 was rather later than 1910. 

 The curves of flowering and boiling in 1909 were taken 

 from daily observations of three ridges in each plot.* 

 They are plotted in Fig. 47, which shows that while 

 the flowering curves were identical within the probable 

 error on terraces A, B, and C, the later portions of 

 the boiling curves were dissimilar, and that this dis- 

 similarity was due in the shedding curve. The steady 

 progression from terrace C the lowest to terrace A, 

 becomes still more apparent when the curve of shedding 

 is plotted in the form of percentage of flowers shed to 

 flowers open. The amount of shedding thus appears to 

 be proportional to the depth of root system which is 

 submerged by the rising water-table. In 1910 the same 

 set of observations gave no result which could be dis- 

 cerned by the methods of observation employed ; this 

 was not unexpected, since the rise of the Nile shown by 

 the water-table on these plots began forty days later than 

 in 1909 ; as also in 1911 (Fig. 45). 



It is noticeable that the 1909 curves show two maxima 

 of shedding; one is coincident with the " critical " period 

 of maximum evaporation, the other with root-reduction. 

 The first maximum was very clearly shown by M. Audebeau, 



* Report 1910 Cotton Commission. 



