506 GENERAL MORPHOLOGY OF PLANTS 



carried too far in certain directions, when it will be necessary to 

 breed back to the best state, as was the case with a variety of 

 walnuts bred by Burbank. He was breeding for a nut with a 

 thin shell. He succeeded so far as to produce a nut with a shell 

 so thin that the birds could break through it and eat the meat. 

 It was then necessary to breed back a thicker shell, at the same 

 time retaining the other good qualities of the nut and the general 

 vigor of the tree. 



678. Crossing different species for the production of new 

 varieties. Where species not too distantly related possess cer- 

 tain characters which it is desired to bring into one variety, they 

 are often crossed for this purpose with excellent results, but some 

 plant breeders advise against the crossing of species as a rule. 

 Some of the reasons for not crossing species are that the hybrids 

 of species are apt to be sterile or at least to possess a high per- 

 centage of sterility. The more distantly related the species 

 crossed, the greater, in general, will be the percentage of sterility. 

 Thus little or no seed is obtained which makes the propagation of 

 the new variety unprofitable if it is to be propagated from the 

 seed. If a valuable variety is obtained in this way which has 

 a low percentage of fertility, the percentage of fertility can often 

 be improved by crossing the hybrid with one of its parents. 

 On the other hand hybrids of varieties are apt to possess a 

 high percentage of fertility. Crossing of species, if they are 

 not closely related, is apt to be more or less guess work, and 

 in the large number of cases does not lead to success. Now 

 and then, however, very valuable varieties are obtained in this 

 way. 



679. Relation of MendePs law to plant breeding.* 

 Because of the approximate precision with which the hybrids, 

 which follow Mendel's law, come out with dominant and recessive 

 characters, it has been thought by some it would be possible to 

 predict the nature of varieties or hybrids which one could obtain 

 by the crossing of closely related species or varieties. In cross- 

 ing varieties differing in respect only to one character, which 



* For special assignment. 



