﻿50 12 



shining substance, obviously of the same nature as the rachides. This is readily 

 seen on holding a specimen up against the light, and on drying this part stands 

 out with a distinct boundary line against the remainder of the scale. If alizarin 

 is used, only this part becomes red along with the rachides of the ventral plates 

 as also all the plates of the dorsal cuirass and all the skeletal parts — with excep- 

 tion of a part (though not the margin) of the preoperculum, which assumes a 

 yellowish tinge like the rest of the ventral plates. Likewise coloured red are some 

 very fine lines which radiate out horizontally on each side from the lowest point 

 of the rachides, specially distinct on the scuta in front of the ventral fins. These 

 red-coloured parts are obviously ordinary bony substance, which is always coloured 

 red by alizarin. The rachides and the faint lines connected with them seem to 

 me to correspond to the almost similarly situated rachides on the ventral scales in 

 Centriscus. And it seems reasonable to suppose that the dense marginal parts of 

 the plates in Amphisile represent the unpaired ventral plates in Centriscus, which 

 all have an often fairly high, compressed keel. 



The first ventral plate in Amphisile may with a good deal of certainty be 

 regarded as in reality unpaired; this applies very probably also to nos. 2 and 3; 

 but it is possible that 4—7 have been formed by fusion, each of a pair of lateral 

 plates and an unpaired keel plate; similarly the 8th with a short keel plate, in 

 front of the ventral fins, and also the others with exception of the last. It is 

 certainly against this view that just the last, quite indisputably paired plate also 

 has a denser, ventral marginal part and further, that the unpaired keel plates in 

 Centriscus alternate with the paired ventral plates. The possibility cannot be ex- 

 cluded therefore, I think, that (with exception of the first ventral plate) all the 

 unpaired elements corresponding to the keel plates in Centriscus have fallen out in 

 Amphisile, whilst the paired ventral plates corresponding to the paired in Centriscus 

 have become greatly developed and fused together ventrally. Lastly, there is a 

 third possibility, that it is just the unpaired keel plates of Centriscus which have 

 developed so much in Amphisile, whilst the paired ventral plates have disappeared 

 with exception of the small scutum s under the anterior edge of the clavicle, which 

 has its definite, demonstrable homologue in Centriscus. 



I may add further regarding the structure of the ventral plates, that they are 

 smooth, without sculpture; the greater portion, excluding the parts coloured red 

 by the alizarin, is fairly soft though tough and dense and does not seem therefore 

 to be a true bony tissue; it does not dissolve however in potash, which completely 

 isolates the ventral plates from the dense connective tissue of the skin; under (he 

 microscope it shows very similar, fine concentric lines of growth to those known 

 in the scales of most bony fishes. The ventral plates are not articulated by sutures 

 but overlap each other ventrally, whilst their margins touch higher up. In the 6 

 ventral plates which lie behind the pectoral fins the rachides reach up to the 

 lower margin of the dorsal cuirass or sometimes in under this. The upper parts 



