﻿37 



75 



Fig. 23. 

 CciUrisciix seuliipax. Right dorsal cuirass from inside. 1—3: plates 

 ol' upiier row; I-V: of lower row; (: ridge on plates I— III; n: 

 notoli for the transverse process of 3rd vertebra; b: for that of 4th. 



Tlie inner surface (fig. 23) of the dorsal cuirass is smooth and somewhat con- 

 cave in the ujjper plates and the two posterior of the lower row; bul on the first 

 3 large plates of the lower row there is a very prominent ridge (/); this is firmly 

 attached by means of dense connective tissue to the transverse processes on the 

 2nd, 3rd and 4th verlebræ. Examined more closely, the conditions are as follows: 

 the transverse process of the 2nd vertebra lies at its outer end in a notch, which 

 is formed by a shallow depression on the upper edge of the clavicle and by the 

 anterior margin of the ridge on 

 plate I; the transverse process of 

 the 3rd vertebra fits into an inci- 

 sion (a) almost in the centre of the 

 part of the ridge belonging to plate 

 II and the tip of the transverse 

 process of the 4th vertebra is simi- 

 larly situated on the ridge on plate 

 III. On close inspection of the 

 ridge we find that the long teeth 

 of the sutures on the one plate fit 

 into those on the others in such 

 a way that they almost reach to 

 the notches. 



There is thus no small resemblance to Amphisile; there can scarcely be any 

 doubt that the plates marked / — V in Plate I, figs. 1 and 3 are completely homolo- 

 gous. In both genera no. I is firmly attached to the posttemporal and is connected 

 with both the supraclavicle and the clavicle as well as with the transverse process 

 on the 2nd vertebra; nos. II and III are connected by means of a prominent ridge 

 with the transverse processes on the 3rd and 4th vertebræ, the ends of which fit 

 into incisions on the ridge; the plates IV and V are somewhat difi"erent, as they 

 have no ridge in Centriscus and no connection with verlebræ, whilst IV in Amphi- 

 sile, sometimes also V, possesses a ridge and is connected with the vertebral 

 column. Nor can there be any doubt that the plates of the upper row in the two 

 genera are also "general homologues" and represent each other; but there may be 

 some doubt, naturally, as to whether the plates in Centriscus marked 1, 2, 3 in 

 PI. I, fig. 3 and text-fig. 23, are strictly homologous with the plates of Amphisile 

 in fig. 1, PI. I indicated by the same numbers; their connection with the plates I, 

 II and III seems to be in favour of homology, but on the other hand, the posterior 

 plate 3 in Centriscus resembles plate 5 in Amphisile both in form and in the fact 

 that it is connected with the 3rd interspinous bone. If the numbers given, 1, 2, 

 3 are correct in Centriscus, compared with Amphisile, then the posterior plates, 4 

 and 5 in Amphisile, are not developed in Centriscus scolopax and C. gracilis (which 

 is in complete agreement with scolopax). For Centriscus humerosas Günther (14a, 

 p. 523) gives 4 plates in the upper and 4 in the lower row, but so far as I can 



