m RA*S AND GtJINeA-Pie§. i^ 



SSl,ECTlON FOR STRIPS OP INCREASED SIZE. 



Selection in an opposite direction, for a wider and more continuous stripe, 

 leads to similar conclusions. In this series of experiments, the wide-striped 

 parents used were obtained as a result of crosses between hooded and Irish 

 individuals. Such crosses tended to widen the stripe of the hooded offspring, 

 the widening effect being permanent, as the sequel shows. 



Hooded individuals, either parents of wide-striped young, or such as 

 themselves had wide stripes, were selected from lots B and C, and mated 

 together. The average grade of the parents was 12. Their young (lot /, 

 fig. 5), III in number, had a higher average grade than the parents, namely, 

 15.8. The curve was similar in form to the curve {A) produced by the 

 original hooded stock. 



A second group of wide-striped individuals, of higher grade than that 

 just described, though of similar ancestry, produced a group of 70 young 

 (lot K, fig. 5), like themselves in width of stripe. The average grade of 

 the parents was 23, that of the young (lot K) was 22.2. The variation 

 curve (K, fig. 5) is a very flat one, similar to curve B, fig. 3. 



A third group of wide-striped parents, of still higher grade, was taken 

 chiefly from lot D. Their average grade was 40.7. They produced 64 

 young (lot L, fig. 5), having an average grade of 36.2. The form of variation 

 curve given by lot L is similar to that given by lot D, flat and bimodal. 

 The upper mode, however, is relatively taller, indicating that the upper 

 (or wide-striped) group is increasing at the expense of the lower, as a result 

 of the selection, and that the original two-peaked condition of curve D was 

 due, not to the heterogeneity of the material included in lot D, but to the 

 fact that part of the gametes formed by the cross-bred individuals trans- 

 mitted a modified (wide-striped) condition. This wide-striped condition 

 was now (in lot L) in process of segregation through the action of selection, 

 and in a fair way to form a stable wide-striped race. Indeed, in the best 

 wide-striped pairs of lot L this seems already to have been accomplished; 

 for certain extremely wide-striped individuals (partly of lot L and partly 

 of a later generation), which had an average grade of 50, produced the 19 

 young included in lot M, fig. 5, which had an average grade of 53. The 

 mode for this curve is close to the mean, being 54.5, and the curve itself 

 is nearly symmetrical, indicating approach to a condition of stability. 



The theoretical importance of the results of these selection experiments 

 is evident. Characters can be permanently modified by selection, contrary 

 to the view of De Vries that such modification is impossible. 



The objection may be raised that the permanency of the modification has 

 not been fully established, for later generations might show regression to 

 a narrower-striped condition. While admitting that this objection has 



