39<» 



THE AMERICAN NATURALIST [Vol.XLIX 



case, the results were found to be due, when carefully 

 tested, to the number of eggs laid and the promptitude 

 with which they are laid when the food is fresh. The 

 question will again come up in certain of the crosses- 



(5. J mount of Water in Food.— Normal cultures lose 

 much of their water as the brood of flies develops. It 

 was a fact noticed at the start that in "wet" bottles the 

 abnormal characters appeared to best advantage, and in 

 most of the work on linkage that knowledge was utilized. 

 But whether the wetness was only incidental to other 

 changes, or in itself the normal condition was not pre- 

 cisely determined. Under all conditions the air in the 

 bottles must be completely saturated with moisture so 

 that we must be dealing with the water taken in with the 

 food and not with the amount of water in the inspired 

 air. In three ways the effect of water was studied. (1) 

 Food that had been fermenting for two or three days in 

 the old acid medium was squeezed until freed of much 

 of its water. The solid part was then further dried su- 

 perficially by pressing between pieces of filter paper, and 

 finally put into a bottle with more dry filter paper. The 

 fluid squeezed out was diluted with an equal amount of 

 water, and put into another bottle. Virgin normal flies 

 and abnormal males of pure stock were set free in these 

 two bottles. The results at the end of nine days were 

 most striking. In the dry bottles the F, females were 

 all normal; in the wet bottles the Fj females were ex- 

 tremely abnormal. 



7. Changing the Adult from Wet to Dry and Vice 

 Versa.-ln this same series the old (P,) flies that had 

 been in the wet bottle were transferred to dry food, and 

 conversely the "dry" flies to wet food. Their progeny 

 showed the influence of the food that they were reared 

 upon, and no effect of the feeding in the previous bottle. 

 Once more they were changed, the wet to dry, the dry to 

 wet, and the results were the same as before, t. e., the 

 actual conditions, not the preceding ones, fully accounted 

 for the results that were obtained. 



