NILSSON'S DISCOVERY 81 



and selected for some peculiar quality, it will repeat the same 

 splittings in its progeny and thus be found not to give rise 

 to a pure and uniform race. This, however, is no drawback. 

 On the contrary, it often affords a means of acquiring new 

 and useful varieties. The selection has only to be repeated, 

 the hybrid group being treated in the same way as the 

 cultures of the original fields. For each type one ear must 

 be selected, and its kernels must be sown separately. Accord- 

 ing to the ordinary rules of hybrids, some of these separate cul- 

 tures will at once prove uniform, but others will repeat their 

 splittings. Among these, the choice must be repeated once 

 more, and by continuing this process we may finally succeed 

 in getting all the possible combinations in constant and uni- 

 form races. These, though of hybrid origin, have definitely 

 lost the character of variability, which at the beginning dis- 

 tinguished the progeny of the original cross. They are 

 further cultivated and tested in exactly the same way as all 

 other separate cultures, and may yield as valuable agricul- 

 tural varieties as these. 



The work of Svalof is not connected with the origin of 

 the elementary species which are observed in the field. 

 This is a question of purely scientific interest. Two possibil- 

 ities offer themselves. Either the high degree of variability 

 is old and the same elementary types which are now existing 

 have already existed for centuries, or the production of new 

 varieties is steadily going on, affording a cause of increasing 

 variability or, at least, of a changing group of units. In the 

 first case the mixtures would be constant and only exposed 

 to accidental losses by the crowding out of some of their 

 rarer constituents. In the latter case, however, the process 

 must be assumed to be a slow one, and the existence of 

 hundreds of types is no proof of a high degree of change- 

 ability. This conclusion will easily be arrived at from the 

 following considerations. The ever-occurring crosses must 



