COTTON-GROWING 161 



Four acres of land cut up into 125 plots, twenty- 

 five being sown each week for five weeks grouped 

 around the probable date of sowing, with five different 

 spacings of five plots each. The data thus obtained 

 could be taken as being correct within the 9 : 11 

 extremes for spacing and sowing combined, and for 

 either spacing or sowing separately their extreme 

 possibility of error would be 95 : 105. Whether such 

 combined experiments were practicable, or whether each 

 point would be investigated separately, would depend 

 solely on the labour available. This class of experi- 

 mental work requires either careful supervision at the 

 times of sowing and picking, or else the training of a few 

 natives to act as observers, with a modicum of thought 

 in the arrangement of fool-proof methods for them to 

 follow in making the observations. 



By conducting experiments on these lines, so as to 

 obtain a reliable answer, much time and money is econo- 

 Utility of ft 1 * 26 *! i* 1 ^ ne following year, when the 

 Accurate results from the plots are applied on a 

 Results, larger scale. The case of sowing-time in 

 Egypt is very much to the point; the author has shown 

 that the early sowing of cotton before a certain date is 

 of no advantage, and may bring a loss, while sowing after 

 that date delays maturity; the cause of the existence of 

 this " critical date " would appear to lie in the tempera- 

 ture of the soil, which at depths of a foot or two undergoes 

 practically the same seasonal changes in temperature 

 every year. 



So far we have sketched the methods only by which 



11 



