On the structure and classification ov the Trehataspidae. 9 



The original fossil was so badly woru by repeated handliug and uuskillt'ul attempts to 

 remove the matrix, that much of its former character was lost. Oii the other hand the 

 original cast was in a practically undisturbed condition, so that our wax impressions of it 

 gave a perfect reproduction of the untouched fossil. I hope to have these impressions re- 

 produced by photographs, so that the exact basis of the reconstructions will be accessible. 



I recognize in the oral region one unpaired, triangular plate; four rows of large paired 

 ones; and a row of four or five small plates just back of the anterior margin of the dorsal 

 shield. The latter is badly crushed and it is not possible to make out its original condition with 

 certainty. I have given on the left side an exact representation of the plates seen in the fossil. 



The marginal plate of the first row, not counting the row of small square plates, is 

 almost certainly a distinct plate, but whether the other pieces in this row are distinct plates, 

 or fragments of a large crushed one, could not be determined from this fossil. It is very 

 probable that a large isolated plate figured by Rohon '93, (PI. I. fig. 2.) and which most 

 probably comes from the oral region of Tremataspis, belongs in this row. It fits fairly well 

 into the place filled by the crushed fragments, and there is apparently no other place in the 

 oral region for such a large plate. This plate, however, cannot be the isolated «mittlerc 

 Seitenplattew as Rohon claims, for in the cast under consideration, the two plates lie nearly 

 side by side, and the difference in size and form between them is very striking. This large 

 plate appears to be movable or to have had some movable body attached to it, for it has at 

 one end a small rounded notch or joint. A determination of whether this notch was at the 

 median or lateral end of the plate would probably throw a great deal of light on the meaning 

 of all the oral plates. 



If three plates were present in this row, instead of two, as shown on the right side of 

 the model, the joint between the first two plates would come opposite the one between the 

 first two plates of the second row. On this supposition, however, there would be no place 

 for the large notched plates, unless they were placed lengthwise of the liead so that the 

 tooth-like process on the dorsal shield could fit into the notch on the ends of the plates. 

 This, however, would bring them into an unnatural position, overlapping the median plate 

 and part of the first two rows. 



Moreover, if the plates are left as they are in my n^storation, they correspond roughly 

 with the two large mandibular plates of P^6r/67<^%s. Traquair figures an incision on the 

 lateral margin of these mandibular plates that is supposed to be a nasal opening. But it is 

 hardly probable that the notch at the end of the plate in Tremataspis can have this 

 significance.') 



1) If this plate i3 placed transversely, as in the figures, 

 and with its notched end next the margin of the oral 

 region, the notch would fall in line with those on the poste- 

 rior marginal plates and would add still another marginal 

 opening to the nine already known. If we adhere to our 

 interpretation of these openings, this would mean the 



3aa. <tH3.-MsT. Otx. 



presence of a pair of small antenna-like appendages, just 

 in front of the pair of large oar-like swimming ouca. 

 However the position of this plate is as yet so doubtful 

 that we do not attach too much importance to such a 

 speculation. 



