IN RATS AND GUINEA-PIGS. 33 



such animals have borne more pigment than did their parents. A similar 

 result would doubtless follow selection of self-pigmented rats obtained from 

 Irish parents. No doubt many of the young would bear some white fui. 

 With selection of less extreme variates, regression less extreme may possibly 

 occur, though our statistical observations do not show any regression in 

 the case of rats.* 



If regression does occur, can we with propriety consider the effects of selec- 

 tion permanent? De Vries has answered this question in the negative on 

 the basis of his selection experiments with maize, striped flowers, double 

 buttercups, and other similar material. It seems to us, however, that the 

 answer should be qualified. The final result will depend upon the amount 

 and the persistency of the regression. In DeVries's experiments with maize, 

 as in those of Fritz Miiller ('86), the regression grows less with each selection. 

 If this continued, the regression should ultimately become a negligible quantity. 

 After repeated selection for a desired extreme condition, the race should 

 become stable at a condition only a little less extreme than that selected. 



De Vries's fine series of selection experiments with the buttercup (Ranun- 

 culus bulbosus) seems to the writers scarcely to justify the conclusion that 

 selection has no permanent effects. Starting with a one-sided or "half- 

 Galton" variation curve, with a range from the modal number, 5, upward 

 to 13, De Vries was able by selection for an increased number of petals to 

 raise the mode to n, the average to 8.6, and the upper limit of variation to 

 31, and to obtain a two-sided, or Galtonian, variation curve with only a 

 moderate amount of skewness, and with greatly diminished regression. All 

 this was accomplished within five generations. 



We consider the selection question still an open one. Further experiments 

 and longer continued ones are needed. Our own observations, so far as 

 they go, and those of Fritz Miiller and De Vries, lead us to think that selection 

 is a most important factor, not only in the isolation of discontinuous varia- 

 tions, but also in their production. 



Further, we are far from convinced that all evolutionary progress is to be 

 attributed to discontinuous variations, any more than to Mendelian inher- 

 itance. The distinction between continuous and discontinuous variations 

 is a useful one, just as that between alternative and blending inheritance, 

 but a sharp line of division can be drawn in neither case. The hooded and 

 Irish coat-patterns of rats are recognized discontinuous variations, alter- 

 native in inheritance, yet our lot M of hooded rats is as nearly intermediate 

 between typical hooded and typical Irish rats as anything that can well be 

 imagined. The coat-patterns of fancy rats, though discontinuous as they 

 ordinarily occur, can be transformed into continuous variations. Concerning 



*April, 1907. In this year's experiments we are getting some evidence of the expected 

 regression. 



