418 l^i-icofyloiis luicrs. 



difficult to raise one if it is not. Isolation can be effected 

 in two to three generations as a rule, in fact as soon 

 as the hereditary values of the plants raised from the 

 original mixture of seed have been determined. Very 

 little care and no artificial fertilization at all is necessary 

 for this purpose. It takes place as quickly in dioecious 

 wind-fertilized flowers, such as the hemp, as in self- 

 fertilizing species enclosed in bags or cages. 



In the choice of species and varieties one obviously 

 has not a free hand. We must first search for cases in 

 which the desired races ha])pen to be present, and for this 

 purpose sowings of considerable extent have to be made. 

 If lots of ten to twenty grams of seed, according to the 

 size of the individual grains, are sown, it can be seen 

 soon after germination whether the species is likely to 

 produce a tricotylous intermediate race or not. The oc- 

 currence of 1 to 2 tricotyls amongst about 10,000 normal 

 seedlings does not justify this expectation, but that of 

 one or more per thousand does as a rule. 



The cultures of my half races mentioned in § 4 (p. 

 392) had begun in 1892 or earlier, and in the spring 

 of 1895 there could be no doubt that I should not obtain 

 any intermediate races from them. I resolved therefore 

 to seek them elsewhere, and selected for this purpose 

 about forty species and varieties of garden plants, which 

 were chiefly annuals. The result of this sov/ing, which 

 was conducted on a large scale, has already been given 

 on page 380. Most of the sorts contained too few tri- 

 cotylous seedlings. Only 8 offered the prospect of giving 

 rise to an intermediate race, and of these I had to reject 

 three for various reasons. The remaining five, however, 

 fulfilled my expectations. 



Before I proceed to the description of my experiments 



