[THOMPSON] EARLINESS AND LATENESS IN WHEAT 161 



If the sum of the number of factors which distinguish the succes- 

 sive pairs of parents is great, then the parental types in a cross in- 

 volving wide difference, should be recovered only among an extremely 

 large number of plants in F2. And yet the F3 results show that the 

 parental types were recovered among very few. The mathematical 

 considerations have been dealt with in the previous paper and need 

 not be repeated here. 



From the practical point of view the chief conclusion is that by 

 crossing an early with a later type races can be established from a 

 reasonable number of F2 plants, whose means will occupy any position 

 from that of the early parent to that of the later. The great majority 

 of races will occupy some intermediate position. Consequently, if 

 it is desired to combine earliness in as great a degree as it is shown in 

 the early parent with several desirable qualities of the later parent, it 

 will be necessary to grow a very large number of F2 plants. But if it 

 is sufîficient to combine an intermediate degree of earliness with 

 characteristics of the late parent, the task is comparatively easy. 

 When certain varieties are used in crossing it is apparently possible to 

 secure races earlier than either parent. 



Summary 



1. Records are given in this paper of the length of time from 

 planting to ripening in many families of the third generation of several 

 crosses between wheat varieties which differed by various degrees in 

 regard to the character in question. The second generation results 

 had been reported previously. 



2. In all cases families were obtained which correspond with the 

 parental types. In some cases families were obtained earlier than the 

 early parent. But most of the families were intermediate and all 

 possible intermediate positions were occupied. To what extent this 

 fact is due to environment is discussed. There was no evidence of 

 special groupings of families around certain centres. 



3. There was considerable difference in the variability of different 

 families and in the distribution of individuals within a family. Many 

 intermediate families showed no greater variability than the parental 

 types. 



4. It is impossible from the data to draw conclusions based on 

 the proportion of very variable families or the proportion of families 

 which occupy a given position (e.g. with one parent). 



5. The difïïculty in applying the multiple-factor hypothesis of 

 blending to the results as a whole is brought out. 



