Devonian Ichthyodorulite^ Byssacanthus. 143 



B]'itish Museum (no. P. 7031), these two cavities are entirely 

 separated. The sh'ghtly hollow spine, in fact, is fixed upon 

 a ridji'e-shaped plate, and the concavity of the latter is not 

 directly continued into tlie central canal of the former. Other 

 specimens afford still further information, and one of the best 

 at Jurjeff is roughly outlined in the accompanying diagram. 



JByssacanthus crenulatus, Agassiz ; outline of imperfect plate, («) from 

 below, (b) from the side, of the natural size. — Devonian ; Li — — 

 [University of Jurjeff (Dorpat).] 



vonia. 



This fossil, though wanting the upper part of the spine and 

 imperfect on one side of the base, proves that Byssacanthus 

 was originally fixed to (and continuous with) the keel of a 

 plate, which had well-defined boundaries for articulation with 

 adjoining plates. It seems to have been bilaterally sym- 

 metrical, though this is not absolutely certain ; and its outline 

 when viewed from below (fig. a) bears a remarkable resem- 

 blance to that of the anterior median dorsal plate of the Astero- 

 lepidai. If there is any justification for this comparison, the 

 spine is shown to have been directed backwards. The super- 

 ficial ridged ornament of the spine terminates below and 

 passes into a tubercular ornament on the basal plate. 



The ridge of the plate is scarcely thickened, and the tissue is 

 very dense in all parts of the fossil, exhibiting only minute 

 vascular spaces when viewed in section with a lens. The 

 visceral aspect of the plate is quite smooth, though pierced with 

 minute vascular canals, and fractured surfaces show that its 



