1903.] MONSTROSITIES IX FISHES. 15 



a typical specimen of this group is given in PL III. fig. 19. (The 

 hepatic veins are not shown because, in the removal of the yolk- 

 sac, the liver had been damaged, obscuring the relations of its 

 vessels.) The whole of this double heart lies inside a large com- 

 posite pericardial cavity, which is prolonged a little forwards on 

 either side round the origins of the ventral aorta. The ventricles 

 are separate, the auricles communicate with one another, and 

 there is a single large sinus venosus opening by a wide ostivim. 

 into the auricles at their junction (PL III. fig. 19, V.Ait). The 

 sinus venosus receives blood {a) on either side from the duct of 

 Cuvier formed by union of the outer cardinal and internal 

 jugular veins of the twin embiyos {i. e. from the right duct of 

 Cuvier of the right embryo and from the left duct of Cuvier of 

 the left embryo) ; (b) from two separate middle jugular veins ; and 

 (c) from a large trunk formed by union of the inner or adjacent 

 internal jugular veins of the twin heads (PL III. fig. 19, Be'). 

 This last trunk obviously corresponds to fused adjacent ducts of 

 Cuvier which receive internal jugular veins only and have no 

 corresponding cardinals. 



The head-kidney in this specimen is illustrated by PL IV. 

 fig. 31, and corresponds to the description given on page 14. 



The notochords and spinal cords are still widely separate 

 opposite the pectoral region, and they remain separate for a con- 

 siderable number of somites behind it, but ultimately they fuse, 

 so that the posterior part of the body contains a single notochord 

 and a single spinal cord. 



The behaviour of the neural and haemal arches and of the 

 median muscular mass corresponds to the description on page 8. 

 These structures, however, may be studied to greater advantage 

 in the type at present under consideration, as the whole transi- 

 tional region is open for observation. The dorsal ends of the 

 adjacent fifth branchial cartilages are fused, but otherwise the 

 twin branchial skeletons are quite separate. The ventral ends of 

 the two coraco-scapular bars fail by a wide interval to meet each 

 other below the pericardium. 



In a typical specimen belonging to group (b) of Class II., in 

 which adjacent 23^ctoral fins iDere present but united and reduced 

 in size, fusion was more complete towards the posterior (radial) 

 border of the cartilage than towards the anterior border. Thus, 

 near the anterior border, there was only a small bridge of cartilage 

 between them ; further back they approached one another, and the 

 bridge was widei' ; while at the posterior border they were united 

 along their whole leng-th. As regards the adjacent coraco-scapular 

 bars, they were quite separate, except at their ventral ends, which 

 were fused together and projected downwards into a septum 

 between the two pericardial sacs. The ventral ends of the two 

 outer coraco-scapular bars were very widely distant from one 

 another. 



The same specimen may be used as a type for illustrating the 



