344 Mr. (1 T. Regan on the 



derived from the Annelid parapodia, muscular projections 

 used in progression, by increase in size accompanied by 

 hardening and segmentation of the exoskeleton, so do I con- 

 ceive the Asterolepid limb to have been derived from the 

 lobate Crossopterygian pectoral fin, already being used to 

 support the body and for ambulatory progression, by the 

 development of dermal plates on the ranscular lobe of the fin 

 at the same time that the anterior part of tire trunk became 

 armoured. The fixed spinous a])pendage of the Coccosteidse 

 seems to represent the pectoral limb of the Asterolepidre, so 

 that we niay regard the former as the more generalized in 

 the structure of the skull, the latter in that of the pectoral 

 limb. 



We now pass to the Cephalaspidte and the related forms 

 included in the Osteostraci. The reasons for regarding these 

 as allied to the Asterolepidaj have been given by Smith 

 Woodward, and they appear to me suflacient and convincing, 

 and may l^e briefly summarized here. In both groups we 

 have a similar caudal region, with a single dorsal fin in the 

 same position and with the caudal fin heterocercal, with a 

 well-developed lower lobe. Then, again, in two Osteostracan 

 genera, Tremutaspis and Didymaspis, the anterior part of the 

 trunk is enclosed in armour, consisting of a dorsal shield to 

 which a ventral shield is opposed, the dorsal shield being- 

 distinct from the head-shield in the former genus, but fused 

 with it in the latter. Since the head-shield is continuous, 

 the nostrils must have been inferior, as in the Asterolepidaj, 

 whilst the orbits are approximated and separated only by a 

 pineal ])late, as in that family. Finally, the exoskeleton is 

 composed of true bone in its inner layers, as in other 

 "Ganoid" fishes. Where I differ from Dr. Smith Wood- 

 ward with respect to this group is that whereas he looks 

 upon the genera which most nearly approach the Asterolepidaj 

 as the most specialized, J. regard them on that account as the 

 most generalized, and the loose pineal plate and the ganoine 

 layer of Tremataspis appear to me in favour of my view. 

 Conceived as specialized and degenerate Asterolepidaj, the 

 structure of the Osteostraci is easily explicable, but I cannot 

 reconcile the Asterolepid structure with the idea that they are 

 a further development of the Osteostraci or of anything like 

 them, whilst if the resemblances between Asterolepidie and 

 Coccosteidc\3 are due to convergence (as they must be if they 

 belong to difi:erent subclasses), then morphology has ceased to 

 be a guide to relationship. Finally, the Heterostraci must 

 be consideredj since they have often been associated with tlie 



