S July, 1908.] Improvement of Cereals by Seleciiori and Crossing. 41 1 



four seasons and surrounded by varieties susceptible to the disease it re- 

 mained free and showed no trace of infection. This wheat belonged to the 

 sub-species Triticum comf actum br Dwarf wheat. On the other hand there 

 was a type known as Michigan Bronze which was so' liable that not a 

 single individual escaped and very few ripe grains were obtained. 



On crossing these two types, the first generation consisted of plants 

 badly rusted wdthout exception, even the awns and grains being affected, 

 but fortunately a number of ripe grains were obtaim^d foT subsequent 

 sowing. Every available grain was sown to produce the second generation, 

 in plots alongside the parent varieties. The result was that while every 

 individual of the susceptible variety was affected and the immune variety 

 escaped entirely, the cross-breeds were mostly badlv rusted, but certain 

 individuals stood out perfectly clean, not showing signs of disease even 

 in the withering basal leaves. 



The second generation of the cross-breed was thus composed of plants 

 either badly rusted or immune, and the exact numbers were 1,609 diseased 

 and 523 immune, or a ratio of 3.07 to i. Further, the immune plants 

 bred true to that character, showing that they were fixed and the rusty 

 showed one part fixed and two parts unstable, so that the ratio of i : 2 : i 

 held good, or one part with one original character fixed, one part with the 

 other original character fixed and two parts unfixed. 

 The result is shown in the following scheme : — 



Parents — Rusty variety X Immune variety 



(Michigan Bronze) , {Triticum comprif turn) 



This practical experiment shows that the pair of characters — immunity 

 and susceptibility to rust^ — behave like other pairs of characters, and that 

 in three generations it is possible to obtain by crossing a variety immune 

 to rust with the desirable qualities possessed by either of the parents. 



By selecting from among the immune plants of the second generation 

 the individuals which combined the best characteristics of Michigan Bronze 

 and sowing the seed, the breeder is able to fix the type in the third genera- 

 tion. With the simple conception of unit-characters and their mode of 

 inheritance, the breeder can now proceed in his work with a definite aim 

 in view and with a certainity as to the means to achieve it. He can build 

 up character by character in the plant he is dealing with, until he has 

 obtained his ideal, and experience will teach how far he may go 

 without exceeding that amount of variation which is compatible with 

 stability. He can now base his practice upon fixed principles, in the one 

 case on the principle of selecting elementary forms or plants with indi- 

 vidual peculiarities to start with and in the other of breeding for separate 

 unit-characters and fixing his attention on one primary characteristic at a 

 time. 



Sports or Natural Crosses. 



In the careful experiments at Svalof where a large number of groups 

 were grown each descended from a single mother-plant, the offspring of 

 each parent was uniform. But there were occasional exceptions to this 

 rule and the differences could not be explained as the result of stray seeds 



