556 PROCEEDINGS OF THE NATIONAL MUSEUM vol. 83 



subspecies, hudsonius and pundulatus. In the early part of my 

 study such specimens were tentatively identified as stylijer because 

 of their rather long spines. In putting my rough data in presentable 

 form, however, I noted that all such specimens, with one exception, 

 lacked any trace of a brood pouch, apparently being females or 

 sexually undeveloped males. None were over ^5 mm, and nearly all 

 were taken in comparatively deeper water or as pelagic specimens. 

 So far I have been unable to discover an}^ other characters to cor- 

 relate with these unusually well developed spines. In the characters 

 chiefly relied on for the separation of the species and subspecies, 

 counts and measurements, these specimens apparently differ in a 

 north and south direction, on a par with the difference between the 

 subspecies hudsonius and pundulatus. At any given latitude they 

 agree generally in these characters with the respective populations 

 taken in shallow water. The best explanation I have to offer is that 

 they represent the persistence of a juvenile condition with respect to 

 the development of the spines or tubercles. The absence of any trace 

 of a developing brood pouch in nearly all such specimens also suggests 

 the persistence of a juvenile condition in general. 



The color varies greatly with individual fish, but a characteristic 

 color pattern may be recognized, wholly or partly, in most fish 

 with color well preserved. The typical color pattern differs also with 

 age. Smaller and medium-sized fish, about 50 to S5 mm long, have 

 a characteristic blotched appearance, with lighter blotches against a 

 darker background (fig. 61). The light blotches are generally formed 

 around the tubercles and are more or less coalescent. The blotches 

 are often mottled with lighter and darker shades, sometimes with 

 strongly contrasting nearly white and black shades. Sometimes they 

 form figures somewhat resembling hourglasses in shape. In larger 

 specimens the typical, juvenile, blotched color pattern is usually 

 replaced, partly or wholly, by a striped pattern (figs. 62, 63). The 

 trunk has narrow dark brown or black transverse lines against a lighter 

 background. Similar lines are often present and arranged lengthwise 

 on opercle and are continued in a longitudinal direction on the anterior 

 part of the trunk, oftener at its lower anterior corner, where they 

 contrast sharply with the transverse lines. Sometimes these typical 

 lines on the trunk and opercle are broken up to form rows of elongate 

 spots. In most of the available full-grown specimens having the color 

 preserved, at least traces of the juvenile blotches may be discerned, 

 but in some the striped pattern entirely replaces the blotched pattern 

 of the young (as in fig. 62). Often large specimens have the blotches 

 very sharply marked, large in extent but few in number. A pair of 

 such large blotches, one above and one below, may be somewhat 

 confluent, forming a figure roughly suggesting an hourglass (fig. 63). 



