476 H. H. NE^VMAN 



Pigment cells may be conveniently classed as 'body' and 'yolk' 

 chromatophores. Four kinds concern us in this study: black 

 body, black yolk, red body and red yolk. The red chromato- 

 phores differ specifically in form, size, color, number and dis- 

 tribution. They may readily be distinguished, even when the 

 color with transmitted light may appear brown instead of red, 

 by the fact that they have an opaque whitish appearance when 

 seen by direct light. The black chromatophores range from dark 

 brown to gray to black and differ specifically in the same respects 

 as the red cells. 



Bancroft ('12) has added very materially to our knowledge of 

 the inheritance of pigment characters in teleost hybrids. Using 

 the same two species of Fundulus that furnished the basis of 

 my previous experiments he has carried his analysis of this phase 

 of heredity much further than I did. In my new experiments 

 I have had occasion to repeat much of this work of Bancroft's 

 and find little to criticize concerning his observations. If I were 

 to attempt a general criticism of the work I would be inclined 

 to ask whether he has not allowed his tendency to find Mendelian 

 dominance in all pigment characters to obscure his vision for the 

 many evidences of incomplete dominance or blending. If one 

 is on the lookout for cases of dominance and is not too particular 

 about its purity one can find an abundance of examples of it; 

 but if one is on the alert for intermediate conditions or mosaics 

 of maternal and paternal units, he can find abundant examples 

 of them in every hybrid. Whether or not any particular signifi- 

 cance attaches to the more or less pure dominance of unit charac- 

 ters in Fi hybrids between Linnean species, is an open question 

 that we need hardly discuss here. Doubtless, were these hybrids 

 fertile and capable of intercrossing, the parental characters would 

 segregate out in the F2 generation of hybrids. Since no one has 

 been able to get a second generation of teleost hybrids and since 

 there is already a considerable literature on the inheritance of 

 Fi hybrids between teleosts, it seems worth while to try to clear 

 up the situation and to remove existing discrepancies as far as 

 possible. With this idea in view I have performed these addi- 

 tional breeding experiments and offer the following considerably 



