Cereals. 117 



method of repeated selection and slow improvement may 

 be explained in a way which shows its complete agree- 

 ment with the theory of mutation.^ 



Such a selection was begun by choosing a handful of 

 ears from the best specimens of the variety which it was 

 intended to improve. From these a race was started, in 

 which the selection was repeated every year on the same 

 principles. But according to the discoveries of Nilsson 

 the first handful of ears must have been far from uni- 

 form; in fact it must have contained almost as many dif- 

 ferent types as there were ears. For even after a selec- 

 tion by the elaborated methods now in use at Svalof, two 

 ' apparently similar ears may give rise to wdiolly different 

 races. In former years the similarity of the best ears 

 chosen from a field, was therefore erroneously expected 

 to give a uniform result. 



From this point of view the real meaning of the sub- 

 sequent repeated selection at once becomes clear. Every 

 year, from among the motley group, those ears were 

 chosen which most closely approached the ideal. In this 

 way, slowly but surely, the mixture must become purified, 

 till at last, after perhaps 10 or 20 years, it was reduced 

 to only one of the constituent types of the first choice. 

 As soon as this stage was reached, the race was pure and 

 constant and independent of further selection, which now 

 only played the role of permanently keeping it in a pure 

 condition. 



The slow improvement of agricultural races, which 

 played such a large part in the selection theory of Dar- 



^ Elementary Species in Agriculture, Proceed. Americ. Philos, 

 Soc, Vol. XLVT, 1908, p. 149. Cf. also: Aeltere und neucre Sclck- 

 tionsmethodc, Biol. Centralbl. XXVI, No. 13-15, 1906; La thcorie 

 darwinienne et la selection en agriculture, Revue scientiflque, 5e Serie, 

 Tome V, p. 445, 1906, and Neiv Principles in Agricultural Plant- 

 Breeding, The Monist, Chicago, 1907, p. 209. 



