198 Plant-Breeding 



5, row 2). Now SI should produce the types SL, SI, 

 SC, while. Sbl should produce all nine types again 

 (these nine types can be separated only into six by exter- 

 nal appearance). It is therefore seen that the group 

 represented by head 5, row 2, should produce all six types 

 again. Heads 8-13, row 3, show these types. Types 

 SbC and SC of the diagram are alike externally, and were 

 hence selected together last year. Of these SC should 

 produce only SC, while SbC should produce SC, SbC, and 

 EC. But since SC and SbC look alike, the progeny of 

 these two types should show only SC and BC. The last 

 two heads in row 3 show that this actually occurred. 

 "In the single set of heads shown, there were two easily 

 explained exceptions to theory. It will be seen that 

 heads 2 and 3, row 2, differ only in length ; now the group 

 represented by head 2 varied in length from that of 1 to 

 that of 3. In separating 2 and 3, it might easily happen 

 that some of 3 should be placed with 2. In this case 

 the progeny of 3 would show a few heads like 1, and this 

 was the case. I have shown in the photograph only the 

 heads called for by theory, for it would only lead to con- 

 fusion to include the exceptions which would probably 

 not have occurred if 2 and 3 of row 2 had been accurately 

 separated last year. Again, in the progeny of the group 

 represented by head 5, row 2, only five of the six types 

 shown (row 3, heads 8-13) were found in this particular 

 case, though all six were found in most of the others. 

 As the missing type should constitute only 4 per cent 

 of the group, and as it differed from one of the others 

 only slightly, it is possible that it was included with the 

 related type when the selections were made. 



