On the structure and classification of the Tremataspidae. 19 



be confounded with irregular projections due to lateral pressure and a downward thrust of 

 fragments of the inner layer. 



The processes just described appear from their position and trend to have served for 

 the attachment of muscles that were directed sharply downwards and forwards, to some mo- 

 vable organs on the anterior ventral margins of the head. , 



I know no other Ostracoderms, in which structures comparable with these processes 

 have been described. In a good specimen of Pteraspis, that was kindly loaned me by Pro- 

 fessor Sollas from the Oxford collections, I have found two semicircular furrows bordered by 

 crests that correspond very well with those in Tremataspis. They lie on the outer surface 

 of the dorsal shield close to the median line, about opposite the peculiar marginal openings 

 (gill openings of Lankester). Transverse sections alone will enable us to demonstrate with 

 certainty the homology of these structures in Pteraspis with those of Tremataspis. 



In lAmulus, there are two rows of separate infoldings of the dorsal shield that serve 

 for the attachment of muscles. If they fused to form continuous plates, they would corre- 

 spond well with the entapophyses of Tremataspis. 



InLimidiis (Patten and Redenbaugh' 99), there are two sets of muscles that pass from 

 the entapophyses downwards and forwards, or downwards and backwards. One set goes to 

 the haemal, and to the posterior side of the plastron (Plastro, and mesoplastro-entapophysal, 

 and longitudinal abdominal) and the other, to the abdominal appendages (abductor muscles 

 of gills and operculum), 



The entapophyses of Tremataspis probably gave attachment to a somewhat similar set 

 of muscles, that is muscles attached to the walls of an eudocranium and to movable appen- 

 dages, either respiratory or locomotor. 



There are other muscles in lAmulus that arise from the inner surface of the cephalic 

 shield and pass downwards, some to the plastron (plastro-tergals), others (coxo-tergals) to 

 the coxae of the thoracic appendages. The attachment of the coxo-tergal muscles of the five 

 pairs of walking appendages to the inner surface of the dorsal shield produces five well de- 

 fined pairs of radiating bands or ridges, that form conspicuous features on the outer, as well 

 as the inner, surface of the cephalic shield. The two sets of markings produced by the mus- 

 cles belonging to the cheliceral and chelarial segments, i. e. the first and last ones of the 

 series, are much less prominent. I have already pointed out, in an address before the V. 

 International congress of Biologists in Berlin, that these ridges in Limulus may be compared 

 with the radiating ridges on the inner surface of the dorsal shield of Pteraspis (see Lan- 

 kester's fig. 1, PI. IV) and with the well known series of ridges on the outer surface of the 

 dorsal shield of Cyathaspis. 



These muscle ridges in Pteraspis and Cyathaspis furaish still further evidence of the 

 presence in the Ostracoderms of numerous pairs of cephalic appendages. 



The So Called Endolymphatic Ducts : This well known pair of small circular openings 

 in the dorsal shield have been regarded as the mouths of endolymphatic ducts by Rohon 



