In the first place one of the premises may be wrong. The three to one ratio in F_2 by no means proves that all conditions of hoodedness are due to one factor. The result shows at most that one factor that gives the hooded types is a simple Mendelian factor. The changes in this type may be caused by modifying factors that can show an effect only when hoodedness is itself present. That this is not an imaginary objection but a real one is shown by an experiment that Castle himself made which furnishes the ground for the second objection. Second. If the factor has really changed its potency, then if a very dark individual from one end of the series is crossed to a wild rat and the second generation raised we should expect that the hooded F_2 rats would all be dark like their dark grandparent. When Castle made this test he found that there were many grades of hooded rats in the F_2 progeny. They were darker, it is true, as a group than were the original hooded group at the beginning of the selection experiment, but they gave many intermediate grades. Castle attempts to explain this by the assumption that the factor made pure by selection became contaminated by its normal allelomorph in the F_1 parent, but not only does this assumption appear to beg the whole question, but it is in flat contradiction with what we have observed in hundreds of Mendelian cases where no evidence for such a contamination exists. Later Castle crossed some of the extracted rats of average grade (3.01) from the plus series to the same wild race and got F_2 hooded rats from this cross. These F_2 hooded rats did not further approach the ordinary range but were nearer the extreme selected plus hooded rats (3.33) than were the F_2's extracted from the first cross (2.59). Castle concludes from this that multiple factors can not account for the result. As a matter of fact, Castle's evidence _as published_ does not establish his conclusion because the wild rats used in the second experiment may have carried plus modifiers. This could only be determined by suitable tests which Castle does not furnish. This is the crucial point, without which the evidence carries no conviction. Furthermore, from Castle's point of view, these latest results would seem to increase the difficulty of interpretation of his first F_2 extracted cross, and it is now the first result that calls for explanation if one accepts his later conclusion. These and other objections that might be taken up show, I think, that Castle's experiment with hooded rats fails entirely to establish his contention of change in potency of the germ or of contamination of factors, while on the contrary they are in entire accord with the view that he is dealing with a case of modifying factors. [Illustration: FIG. 89. Races of Paramecium. (After Jennings.)]