[45] THE EVOLUTION OF THE FINS OF FISHES. 1025 



specialized PhysocUsti have either generally Lad tlie rudiments of the 

 pelvic fins translocated during development, until in some Ophidioids 

 they are submental in position, or have had them quite aborted. The 

 development of the pelvic fins, as well as that of the air-bladder in the 

 PhysocUsti indicates that they are more differentiated forms than the 

 Physostomi, from which they have probably been evolved. 



It may also be pointed out that the shifting of the position of the hind- 

 ermost pair of fins is quite in harmony with the views of Balfour and 

 Dohrn, and the mode in which the antero-posterior concrescence of 

 homodynamous elements occurs at the bases of the fins, as well as 

 their more or less extensive rotation, ajjparently over their own inser- 

 tions, goes very far towards giving us an insight into the way in which 

 the complex brachial and sacral plexuses have been developed by the 

 anastomosis through the antero-posterior concrescence of paired spinal 

 nerves. 



We now come to the consideration of another set of phenomena which 

 are of great importance in the development of a theory of the fins. 

 What is now referred to is the peculiar way in which the spines or 

 unjointed rays of certain forms develop. It is unfortunate that we are 

 not familiar with enough forms to enable us to generalize with more 

 certainty; but, thanks to the researches of A. Agassiz, we are enabled 

 in the cases of two species which have the anterior dorsal spines form- 

 ing a disconnected series, namely iop/wMS and Gasterosteus, to reach the 

 interesting conclusion that such spines are developed in a great meas- 

 ure independently of the continuous folds and in a special way, that is, 

 by the precocious outgrowth of a hollow terminally blind diverticulum 

 of the epiblast, in which such spines are formed from the mesoblast. 



In Lophius the first s])ine of the anterior dorsal api)ears as a de- 

 pressed conical mesoblastic thickening overlying the front end of the 

 spinal cord, as in Fig. 1, Plate XI. In the next stage the epiblastic 

 1 diverticulum is pushed out and the first indications of the first "dorsal 

 f ray have appeared, as in Fig. 2, Plate XI. The vertical median fin-fold 

 I now atrophies more and more, and in the next stage, Fig, 3, Plate XI, 

 I the medullary portion of the process seems to be differentiating in 

 I which subsequent ossification is to occur. In the next stage, Fig. 4, 

 1 Plate XI, two dorsal rays are visible, and in the next, Fig. 5, four may 

 j be distinguished. 



In Gasterosteus aculeatus the researches of Agassiz show that the 

 I rigid anterior spines are also formed as more robust distinct papilliform 

 outgrowths of the epiblast into which mesoblastic tissue is at once 

 j j)ushed out, while the hinder rays of the dorsal are formed in a contin- 

 ', uous fold in the usual way, embryonic rays being first formed. The 

 spines, however, do not in either Gasterosteus or Lophius seem to be 

 preceded by embryonic rays. Such spines are therefore probably de- 

 veloped from the basement membrane, which doubtless forms in the 

 ' mesoblast which grows into these epiblastic i)apill£e or diverticula, 

 H. Mis. 68 65 



