420 



Journal of Agricultural Research 



Vol. XIX, No. 9 



The uniformity of the Aberdeen seasons and the accuracy of the method 

 of sampling used is nowhere more evident than in figure 12. In this 

 figure the dry matter per kernel is plotted in 12-hour periods. For the 

 first 14 days neither the error of sampling nor the differences in rate of 

 growth of individual spikes, separately or together, exceeds the growth 



/YS/Z 



30. 



^ao 



< 

 4 f 



' ^ ^^ 



/O.O 



S.O 



^e/^r/s/? /s /s /7 /ff /i> ^V4?/^^^3,^^^J'^s^7^<9^s303/ /'\^ ^ ■*«■ v5" e- ^S s 



Fig. II.— Graph showing dry matter per kernel from date of flowering to near maturity in 1916 (dotted 



line) and in 1917 (solid line). 



in 12 hours. There is an apparent reversal of the curve in the fifth and 

 seventh days after flowering, but the larger of these losses is less than 

 0.2 mgm., and in each case is due to the abnormalities of a few kernels 

 on the spike. When this curve is plotted from the data of the more 

 representative sixth, seventh, and eighth kernels, these irregularities dis- 

 appear. It is only when the fourteenth day is reached that fluctuations 



