Eeactions and Products in Interspecific Crosses 159 



cially color-characters. Each of the three lines has been bred out to at least Fg or 

 beyond, and no breeding or testing has been able to break them up into different 

 form-index groupings, the decemlineata form remaining stable and constant. 



At no point in the crosses of these two species has there been any linking or 

 correlation of the different color and pattern characters with these form-produc- 

 ing agents, and so in the three pairs that breed true for the decemlineata form 

 there was abundance of combinations of the color and pattern conditions. The 

 findings suggest the existence either of another instance of a masked heterozy- 

 gote for form with the decemlineata dominant, or the obliteration of the oblon- 

 gata form and other agents in the F^ combination. In crossing natural species 

 it has often seemed that the results of the interactions in F^ have been to produce 

 a total obliteration of some of the agents of one or the other parent, as far as 

 further ability to get them out of the combination is concerned. Whether they 

 have actually ceased to exist as agents, having been thrown out of the complex 

 produced in F^, or are held in fixed form not manifested or manifest in new 

 actions, is a problem for further investigation. 



About 60 per cent of the F^ (a) type decemlineata matings are heterozygous 

 in the ordinary sense and show in Fg only routine combinations of the different 

 characters and groups thereof that are dissociated from each of the two parental 

 complexes. From these, extracted types of each of the parental species as well 

 as many stable combinations are commonly derived, while other rarer combin- 

 ations and associations are erratic in their appearance. 



WITH THE FORM INTERMEDIATE. 



In every cross between these two species there are certain intermediate forms 

 as regards form-index, with variable associations of visible parental characters. 

 Breeding of these shows in all that have been tested that they are uniformly 

 typical Fi heterozygotes, and the F, arrays that are produced from them show 

 only routine separations into the parental form-types with differing combin- 

 ations of the color, pattern, and juvenile characters, so that out of the series one 

 can obtain a very considerable number of stable combinations of the characters 

 shown in the two species. The Fg arrays are in all instances irregular and the 

 ratios are not in accord with the expected arrays, owing to the chance non-dis- 

 sociation of two or more characters in one or both parents. At no point in the 

 series is there any indication of blending of these characters, and the reactions 

 are always clean-cut and precise, the uncertainty being entirely inability to 

 devote enough space and labor to any one series to fully analyze its composition. 



These F^ heterozygotes are interesting because of the irregular metathesis of 

 one or several characters which at other times or even in the same lines are dis- 

 sociated, and suggests agents present either within or without that act to 

 influence this association of factors and the subsequent behavior of them as a 

 unit. The direct result of this condition is that one can derive a large number 

 of races from the cross, which breeds true — i. e., is homozygous acting in genetic 

 lines — and these when crossed give simple Mendelian arrays, usually, those char- 

 acteristic of monohybrid or dihybrid reactions. Some notion of the extent of 

 the array that can be obtained is shown by the series of pairs of characters that 



