Cereals. 115 



also give rise to constant races. Within each cukivated 

 variety the number of differentiating characters may 

 be small, but the large number of elementary forms may 

 be due mainly to the vast number of ways in which they 

 may be combined. Probably these subspecies have 

 arisen in the fields by casual mutations and subsequent 

 free crossing; such crosses being, as a matter of fact, 

 very rare, but occurring often enough to give rise in the 

 lapse of years to almost all possible combinations of 

 characters. 



The diversity among these elementary forms is so 

 large, that it answers almost all the claims of practice. 

 This fact is of the highest practical interest, since now 

 it is only necessary to search for the desired types, to 

 isolate them, to estimate their value and to multiply them 

 to the amount required. No slow improvement, no re- 

 peated selection is necessary ; the fluctuating variability 

 within the elementary species being moreover so small, 

 as hardly to afford material for such work. 



This search for useful elementary types must be car- 

 ried on in the fields at the time just before the harvest. 

 The ripe ears are selected and their grains sown sepa- 

 rately for each parent-plant. From these the new races 

 are started. At the time mentioned manv valuable char- 

 acters can be judged, but others cannot, or cannot be 

 estimated with a sufficient degree of accuracy. This 

 difficulty, however, has been met by another discovery 

 of NiLSSON,^ viz., the correlation between botanical and 

 agricultural characters. On the ground of this, the first 

 selection is made by the aid of botanical marks, which 

 are discernible in the ripe ears, and the subsequent com- 



^ For more details see Plant-Breeding, Comments on the Experi- 

 ments of Nilsson and Bnrbank. Chicago, 1907. 



