210 BULLETIN : MUSEUM OF COMPARATIVE ZOOLOGY. 



Inspection of this table indieiites that lugens is dominant over tau, 

 for when the two forms are crossed, in Generation III., the offspring 

 are apparently all of the lugens form, at least Standfuss does not 

 mention the occurrence of any tans. The resulting fourth genera- 

 tion hybrids, L in the table, but really L (T), when bred mter se, or 

 wheu crossed with normal tau, produce, as we should expect, both lugens 

 and tau forms. See Table III., crosses [IJ, [2], [3]. Likewise the lifth 

 generation lugens, obtained by intercrossing lugens of the fourth genera- 

 tion (cross [2]), produce when bred inter se both lugens and tau forms. 

 See Table III., crosses [4j, |]5]. We have, then, convincing evidence that 

 tau may be recessive (or latent) in lugens, but lugens is in no case shown 

 to be latent in tau. Accordingly we have here a case of simple domi- 

 nance of lugens over tau. The numerical proportions of lugens and tau 

 in the crosses between those two forms are close to those demanded by 

 the Mcndelian principles of dominance and segregation. See Table IV., 

 Generations III., lY. [1], and IV. [3]. But when hybrid lugens indi- 

 viduals are bred i?iter se ([2], [4], ami [•>]), considerable discrepancies 

 occur between calculated and observed results. These discrepancies, I 

 believe, arise from coupling — in tlie gametes produced by the h^'brids — 

 of the male character with the lugens character, and of the female char- 

 acter with the tau character. This explanation acco\ints at the same time 

 for the peculiar sex-distribution between lugens and tau forms observed 

 in all the crosses. 



Suppose that in the germ-cells of every hybrid individual, D (R), the 

 segregation of characters occurs in such a way that the male sex- 

 character passes into the same gamete as the dominant (lugens) form-charac- 

 ter. Then there will be produced only gametes D $ and II 9 • I <lo 

 not say that this is invariably so ; indeed, it clearly is not so for any of 

 the crosses in all cases. It occurs only in a certain number of cases in 

 each cross, but this number is large enough materially to affect tlie 

 result. The calculation, however, will be simplified if, for the time 

 being, we suppose the segregation to occur in all possible cases among 

 the gametes of hybrids. See Table IV. 



In Generation IV., crosses [1] and [3], a hybrid lugens^ I) (R), is 

 mated with a recessive wild tan, R. The two crosses are reciprocals, 

 but the outcome is substantially the same in both, so that evidently 

 whatever peculiarity is possessed by hybrid ova belongs also to hybrid 

 spermatozoa. Suppose, as suggested, that it be coupling of the male 

 character with the lugens character. Then we shall have gametes D ^ 

 and R 9 furnished by the hybrid parent, and gametes li $ and K 9 



