H. M. Leake 207 



certain of the races recognised by them have directly arisen by the 

 intercrossing of other extant races. It is not proposed to enter into 

 this question here, and it will suffice to say that abundant proof that 

 crossing is of frequent occurrence has been forthcoming since the 

 author's first note (11) on this subject'. Under the circumstances, 

 however, a word as to the procedure adopted in the present experiments 

 is necessary. 



The seed received from all sources has invariably given a crop con- 

 taining numerous, and frequently most diverse, forms. The seed of 

 those appearing recognisably distinct is collected and sown separately^ 

 From the similarity or dissimilarity of the offspring the purity or the 

 reverse of the parent can be determined. If the parent appears to be 

 pure the most typical individuals among the offspring are selected and 

 the flowers of these are protected. The form is, in future generations, 

 raised from the seed of flowers thus self-fertilised only. The unit of the 

 parental series on which these experiments are based is, therefore, a 

 series of individuals derived by repeated self-fertilisation through a 

 greater or less number of generations, from a single individual. Such 

 a unit may form the sole representative of a type but, in the majority 

 of cases, the type, as defined, includes several such units usually dis- 

 tinguishable by some small difference in one of the characters. The 

 unit is, therefore, comparable to the " pure line " of Johannsen. 



In all cases when it is proposed to make a cross, the flower of the 

 seed parent is emasculated in bud after removal of the petals and, after 

 emasculation, both before^ and after fertilisation, protected by a paper 

 bag for two days by which time the stigma has usually dropped. The 

 flower from which the pollen is to be obtained is also protected in like 

 manner before the bud opens. The manipulation is simple and among 

 several hundreds of plants in the F^ generation not a single case of 

 accidental selfing has occurred. The parental types have been grown in 

 each successive season from the seed of protected flowers. In the 

 majority of instances the bud is simply covered with a bag which is 

 removed after two days. This method has been found, however, to lead 

 to a considerable degree of sterility and in some cases it has been 

 necessary to adopt the method of intercrossing different plants of the 



' Vide also Balls (2) whicb has appeared since the above was written. 



^ Since several characters are not recognisable until ripe fruit is developed, it is 

 usually impossible to select these plants until it is too late in the season to obtain seed 

 from them by self-fertilisation. 



^ The flower is never fertilised at the time of emasculation. Compare Hartley (23). 



