590 PROCEEDINGS OF THE NATIONAL MUSEUM vol. 83 



segment is closely correlated with the variation in the structure of 

 the last trunk segment and the number of trunk segments. The 

 frequency of occurrence of these variations differs with the local stock. 

 These points are discussed below.) Trunk segments usually 10, 

 often 9, sometimes 11. Caudal segments usually 31 to 33, varying 

 30 to 34. Dorsal rays modally 12, varying 11 to 14. Pectoral rays 

 10 to 12. Tubercles usually quite conspicuous, often becoming very 

 low in full-grown males. Coronet comparatively high. Trunk 

 rather deep ; snout usually quite short. Presence of filaments vary- 

 ing with the individual and evidently also with age, oftener absent, 

 the specimens having filaments usually belonging to the smaller size 

 groups, filaments when present relatively short, often branched. 

 Color variously mottled with contrasting yellowish shades, often with 

 wliite and brown, without any definite color pattern, sometimes quite 

 dark all over, sometimes with whitish cross bands on tail ; dorsal with 

 a submarginal brown streak typically present, usually with one or two 

 rows of diffuse spots at the base; often entire fin nearly colorless. 

 (See table 5 for counts.) 



Variability in structure of region where trunk and tail meet, and its 

 correlation. — H. zosterae shows two main trends of variation which 

 are correlated with locahty to a considerable extent. In the majority 

 of specimens of the entire available lot representing all localities, the 

 first caudal segment is quadrangular and the last trunk segment 

 octangular. All such specimens have the single extra plate on the 

 last trunk segment, while the dorsal is situated on the last two trunk 

 segments and the number of trunk segments is 10, infrequently 11. 

 Very often the foUowing important variation in structure occurs: 

 The first caudal segment is hexangular, and the last trunk segment 

 is also hexangular ; in other words, the extra plate is on the first caudal 

 instead of on the last trunk segment. In all such specimens the base 

 of the dorsal is situated over the last trunk and first caudal instead of 

 over the last two trunk segments, and the number of trunk segments 

 is 9 instead of 10. 



This latter variation may be easily conceived as having been derived 

 from the former by the last trunk segment losing the last lowermost 

 point of intersection and thus having changed to a caudal segment. 

 The probability that this is the correct explanation is increased by the 

 fact that in regulus, the near relative of zosterae, the former condition 

 is normal for the species almost without any exception. Further- 

 more, four specimens of zosterae out of 59 examined are asynunetrical, 

 one side of the fish showing one of the two general variations described 

 and the other side shomng the other variation, the probable manner 

 in which the change occurs thus being shown by the same individual 

 fish (see p. 592). In other words, in zosterae there is a very decided 



