TRANSACTIONS OF SECTION D. 779 



arose independently of that in the Dipnoi, as did likewise a certain amount of 

 resemblance in their dentition. But those who from embryological grounds 

 oppose any notion of the origin of the Dipnoi from ' Ganoids ' might here say, if 

 they chose, If so, why should not also the same lorm of limh have been inde- 

 pendently evolved in Crossopterygii ? 



Accordingly, while philosophic palreontology is much indebted to M. Dollo for 

 his brilliant essay, and though we must agree with him in many things, such as 

 that the Crossopterygii were not derived froin the Dipnoi, and that the modern 

 representatives of the latter group are degenerate forms, yet as to the immediate 

 ancestry of the Dipnoi themselves, and the diphyletic origin of the so-called 

 archipterygium, we had best for the present keep an open mind. 



In his * Catalogue of the Fossil Fishes ' in the British Museum (vol. ii. 1891) 

 Dr. Smith Woodward, following the suggestion of Newberry in 1875, classified the 

 Coccostsans or ' Arthrodira ' as an extremely specialised group of Dipnoi. At first 

 I was much taken with that idea, but after looking more closely into the subject 

 I began to doubt it extremely. My own opinion at present is that the Coccosteans 

 are Teleostomi belonging to the next order, Actinopterygii ; but Prof. Bashford 

 Dean, of New York, will not have them to be even ' hshes,' but places them in a 

 distinct class of * Arthrognatha,' which he places next to the Ostracophori 

 ( = Ostracodermi), even hinting at a possible union with them, whereby the old 

 ' Placodermata ' of McCoy would be restored. It will, therefore, be better to leave 

 them out of consideration for the present, pending a thorough re-examination of 

 their structure and affinities. 



We come then to the great order of Actinopterygii, to which a large number 

 of the fishes of later Palaeozoic age belong, as well as the great mass of those 

 of Mesozoic, Tertiary, and Modern times. Of these we first take into considera- 

 tion the oldest sub-order, namely, the Acipenseroidei or Sturgeon tribe, in which 

 the dermal rays of the median tins are more numerous than their supporting 

 ossicles, while the tail is, in most, completely heterocercal. And the oldest family 

 of Acipenseroids with which we are acquainted is that of the Pakeoniscidaj, 

 which, in addition to well-developed cranial and facial bones, has the body 

 normally covered with rhombic ganoid scales furnished with peg-and-socket articu- 

 lations. Of this family one genus, Cheirolepis, appears in the same Devonian strata 

 (Orcadian series) with the earliest known Crossopterygii, and of its immediate 

 ancestry we know no more than we do of theirs. Cheirolepis is a fully evolved 

 palseoniscid, as shown by its oblique suspensorium, wide gape, and other points 

 of its structure. In the Lower Carboniferous rocks of Scotland, where the family 

 attains an enormous development, we find one or two genera, e.g. Canobius, 

 which appear less specialised, as the suspensorium is nearly vertical, and the 

 mouth consequently smaller. 



This family endures up to the Purbeck division of the Jurassic formation, and 

 in the Carboniferous Cn/phiolepis, the Lower Permian Trissolepis, and the Jurassic 

 Coccolepis we find the same degeneration of the rhombic scales into those of a 

 circular form and imbricating arrangement, which we find repeated in other groups 

 of ' Ganoids.' In fact, in one Carboniferous genus, Thanerosteon, the scales 

 disappear altogether with the exception of those on the body prolongation in the 

 upper lobe of the caudal fin, and a few just behind the shoulder-girdle. 



And in these Palaeozoic times we notice also a side branch of the Palseoniscidae, 

 constituting the family PJatysomidae, in which, while the median fins acquire 

 elongated bases, the body becomes shortened up and deep in contour. The scales 

 become high and narrow, their internal rib and articular spine coincident with the 

 anterior margin ; the suspensorium, too, instead of swinging back as in the typical 

 Palseoniscidae, tends to be directed obliquely forward, while the snout becomes 

 simultaneouslv elongated in front of the nares. 



A most interesting series of forms can be set up, beginning with Euri/notuf, 

 which, though it has the platysomid head contour and a long-based dorsal, has 

 only a slight deepening of the body, and still retains the palseoniscid squamation 

 and a short-based anal fin. In Mesolepis, which resembles Eurynotus in shape, 

 being only slightly deeper, we have now the characteristic platysomid squamation, 



