﻿76 38 



judge from the figure given by Richardson (2(5, PI. 34, lig. 5), the number of plates 

 in this species are just as in the other two species; but there is the interesting 

 difference in arrangement, that no. IV of the lower row has become larger and 

 moved down in line with I — III (and at the same time almost certainly become 

 provided with an inner ridge and connected with the 5th vertebra); no. V is like- 

 wise larger and has become attached by suture to 3 in the upper row, so that an 

 extra, completely bounded interspace is present in addition to the two in the other 

 two species. In this peculiar species all the large plates of the skin (also the ven- 

 tral) are much more conspicuous than those in C. scolopax and gracilis, but except 

 that there are some enlarged scales along the dorsal edge of the trunk (3 rows 

 according to Günther), I see no fundamental differences. 



I may add also that in the direction of the teeth on the sutures between the 

 components of the dorsal cuirass there is great agreement with the condition in 

 Amphisile. There is however the great difference in the upper row, that it does 

 not contain any canal for the lateral line. On the whole I have been unable to 

 observe any lateral line on the trunk of Centriscus, but only on the head. 



The ventral cuirass in Centriscus is represented by (a) a paired row of 

 bony plates, one on each side of the lower edge of the belly and (b) an unpaired, 

 median row from the isthmus to the anus, broken only by the groove for the 

 ventral fins; it is this row which forms the above-mentioned, sharp ventral keel. 



(a). Of the paired plates, 6 form a row in front of the ventral fins (PI. I, 

 fig. 3, 1—6), connected ventrally with the unpaired row; a 7th (s) lies more dor- 

 sally, inserted above 2 and 3, close to the gill-opening. The largest plates, 3, 4 

 and 5 are the most apparent and have therefore often been seen; their upper 

 margin is somewhat lobed and a central lobe especially is larger and more distinct 

 than the others. Alongside the groove for the ventral fins there is a shorter row 

 of 3 plates, nos. 7 — 9, the last of which is the largest (plates 6, 7, 8 do not seem 

 however to be always distinctly developed). On the outer side the paired ventral 

 plates have a sharp, longitudinal ridge with other, shorter ribs radiating out from 

 its centre; one of these runs out into the above-mentioned marginal lobe on the 

 large plates. On the hindmost plate, 9, the median rib forms an oblique spine. 



(b). The median unpaired row is composed in front of the ventral fins of 5 

 narrow, compressed and sharply keeled, symmetrical plates (/ — V); with exception 

 of the first they alternate with the paired plates in the row above; as in the latter 

 the inner surface is smooth and concave, the outer provided with ribs which 

 radiate out from the centre of the base of the keel. Behind the groove for the 

 ventral fins there is a strongly keeled or almost spiny plate (VI). In contrast to 

 the dorsal cuirass the components of the ventral armour are not mutually con- 

 nected by sutures, but overlap each other at the margins; the keeled scales of the 

 belly overlap the ventral margins of the paired plates of both sides. 



It seems certain that the whole ventral armour of Centriscus may be regarded 

 as corresponding to that in Amphisile. One of the paired plates, at any rate, 



