No. 601] MENBELIAN FACTOR DIFFERENCES 43 



of functional Mendelian reaction systems. This high de- 

 gree of mutual incompatibility of the two reaction sys- 

 tems exhibits itself in the high degree of sterility of the 

 Fi hybrids. As Goodspeed (1912, I. c.) has shown, how- 

 ever, this sterility is only particil and a few good ovules 

 are formed which produce viable seed in the case of open 

 pollination or when crossed back with the parents. In 

 evidence which is presented elsewhere (Goodspeed and 

 Ayres, 1916) it has been shown that, while it is experi- 

 mentally possible to modify the behavior of the plants 

 in such a manner that the fruits without pollination are 

 retained for a considerable period rather than falling 

 soon after anthesis, the percentage of good ovules pro- 

 duced can not be appreciably modified. This is an im- 

 portant point, for it indicates that the number of good 

 ovules produced is a function of the chromatin behavior 

 and not to be influenced by environmental factors, and 

 that they should, therefore, exhibit a consistent behavior 

 and lend themselves to a logical interpretation. In fact, 

 evidence at hand indicates that the small percentage of 

 functional ovules represents the Tahacum and sylvestris 

 extremes of a recombination series, and that, therefore, 

 the middle members of the series, which are made up of 

 relatively high proportions of both Tahacum and sylves- 

 tris elements, fail to function because they produce in- 

 compatible reaction systems. This is shown clearly by 

 back crosses which have been made with the parents, 

 although here portions of our evidence are not so well 

 controlled as we would prefer. When back crosses are 

 made with sylvestris as the pollen , parent there is pro- 

 duced a variety of forms many of which are highly ab- 

 normal, but among them there is a considerable propor- 

 tion of plants which are pure sylvestris in all characters. 

 These plants are fertile and have bred true to the sylves- 

 tris type for three generations. A number of the remain- 

 ing plants resemble sylvestris, but show contamination 

 with other elements presumably derived from the Taha- 



