ign] SHU LL REVERSIBLE SEX-MUTANTS 331 



generation, and two explanations seemed possible: (i) these two 

 types of hermaphrodites might be respectively homozygous and 

 heterozygous in regard to a modifying factor H, whose presence 

 was assumed, on the suggestion of CORRENS, as possibly necessary 

 for the change of a normal male into a hermaphrodite; (2) the 

 hermaphrodites of the second type (C and Z)), which gave first 

 generation progenies equivalent to those produced by normal males, 

 might owe their hermaphrodite character to some accident of 

 development which affected the soma alone, leaving the germ cells 

 unchanged. In this case they might be appropriately called 

 " somatic hermaphrodites," to distinguish them from those of the 

 first type (A and B) which transmitted the hermaphrodite char- 

 acter to their male offspring and which are therefore to be recog- 

 nized as " genetic hermaphrodites" or true hermaphrodite mutants. 



Neither the character of the females nor the relationship of the 

 two types of hermaphrodites could be determined from the results 

 of the first generation, but it was obvious that at least a partial 

 solution could be expected from the second generation. To attain 

 this end a large number of crosses were made in 1909, by using 

 hermaphrodite individuals and their derivatives in various com- 

 binations with each other, with unrelated females, and with normal 

 males. The offspring of these crosses were grown during the sum- 

 mer of 1910, and the 104 families produced from them included 

 6132 individuals which came to bloom and of which the sex was 

 recorded. These records were made in the writer's absence by Mr. 

 R. CATLIN ROSE, to whose energy, faithfulness, and care it gives 

 me pleasure to testify. 



In .order to comprehend fully the problems involved, it will be 

 advantageous to consider some assumptions which were permitted 

 by the results of the F t crosses, and whose availability is partially 

 tested in the F 3 families reported in the present paper. In this 

 connection it is also important to consider briefly the "presence 

 and absence" hypothesis, a full discussion of which, however, 

 would require too great a digression. Although this so-called 

 hypothesis is frequently referred to by students of genetics, I am 

 not aware that it has ever had a very definite formulation, and it 

 would undoubtedly be defined differently by different students. 



