﻿7 45 



downwards at an obtuse angle to the horizontal axis of the body; the anal (in lies 

 immediately in front of the caudal. About half way along the ventral aspect we 

 find the generally small ventral fins. The last are thus abdominal, lying far behind 

 the pectorals, as is usual in the Physostomi. This form is however aphysostomous. 

 The swim bladder is present; this can be seen easily by holding a specimen up 

 against the light. It is noticed at the same time, that the sharpest part of the 

 belly is devoid of contents and forms a thin, transparent region, a veritable knife- 

 edge. It is said, from the observations by Willey (34 p. 719) and Townsend (25a 

 p. 318), that the fish swims in a vertical position, cutting the water with its belly 



— but I can hardly believe that this position is the normal one for the fish". 



Most of the fish is armoured; the part not covered by scutes is quite naked, 

 without scales. 



The head is movable, a little up and down; otherwise the whole body is 

 quite stiff and only the tail and fins are flexible. 



Exoskeleton. 



The armour of the body consists of a dorsal and a ventral cuirass. The 

 dorsal cuirass on each side is composed essentially of 2 rows of slightly alternating 

 plates, each row consisting of 5 parts; namely, an upper, dorsal row of scuta (PI. I, 

 (fig. 1, 1—5) and a lower, lateral row (PI. I, fig. 1, I— V). The four anterior dorsal 

 plates are elongated, narrow and meet the corresponding plates of the other side 

 in the middle line of the dorsum in a simple, straight suture; the unpaired scutum 

 covering the dorsal spine is wedged into the middle line of the dorsum behind 

 no. 4 and the 5th, hindmost, dorsal plate is thus excluded from meeting its fellow 

 of the opposite side. This plate is elongated, triangular, with its posterior point 

 running out alongside the dorsal spine, whilst anteriorly it spreads down on the 

 side more than the other plates in the same row. In the two species A. strigata 

 and punctulata there is still another, small, unpaired plate anteriorly in the middle 

 line of the dorsum, pushed in between the posterior ends of the first pair of plates 

 and the neighbouring, anterior part of the 2nd pair. This unpaired plate, which 

 belongs in reality to the endoskeleton, seems to have been hitherto quite over- 

 looked, though its absence in A. scutata should be one more, easily observable 

 character to distinguish this species from the two others. 



Of the lower row of lateral plates the first (I) is the smallest and narrow; 

 the remainder are of good size, especially III and IV. A distinct longitudinal line 



— sometimes somewhat depressed like a furrow — runs across these plates; in 

 the first it goes right along the ventral border, in the three following nearer to 

 the dorsal border; on the posterior plate it runs to the ventral border almost 

 through the centre and parallel to the dorsal margin. On the naked portion of 



* In one point the observations do not agree: wliile Willey represents the fish swimming with 

 the head upwards, Townsend (teste Regan) says that those of his specimens which were "sufficiently 

 alive when dredged to swim in a tub of water" swam head down. 



