222 PLANT-BREEDING 



either by isolation or by covering the flowers and fertilizing 

 them artificially. If this is done, the varieties prove con- 

 stant, with few exceptions. In practice, however, isolation 

 and artificial fecundation are often too cumbersome, and the 

 easier way by repeated selection is preferred. Gradually 

 it eliminates the effects of the natural crosses with the neigh- 

 boring varieties and in three to five years it ordinarily brings 

 the purity up to the degree required for commercial purposes. 



If, however, the varietal mark should show itself only 

 imperfectly in the beginning, being limited to a supernum- 

 erary petal, to a slight indication of a new color, to a fine 

 line of a different hue, or to a hardly perceptible fragrance, 

 selection has to improve it. In the case of pure and artifi- 

 cial pollination the ordinary variability of the new mark 

 would bring it to its main condition as soon as the sowings 

 reach the required extent, but in practice the process must 

 be combined with the described gradual elimination of the 

 effects of crosses, by which it is, of course, made much 

 slower. Four to five years, however, are, as a rule, sufficient 

 to reach the aim. 



The way in which all these varieties originate is best ex- 

 pressed by the technical term of chance seedlings. Fruit- 

 growers, nurserymen, and amateurs are always on the look- 

 out for such. Even among the commercial growers there is 

 a regular search for deviations in the direction of improve- 

 ment. Any seedling that chances to show more desirable 

 characters than the average will be noted and cared for. To 

 this discriminating search is due the superiority of some of 

 the leading varieties of fruit-trees, which have been recog- 

 nized as meeting special requirements and have been multi- 

 plied accordingly. Such chance seedlings occur every- 

 where, from time to time, but the cause of their appearance 

 is, as yet, wholly unknown. The only thing we know about 

 them is that among large numbers the chance of finding 



