ON THE GENERAL MORPHOLOGY OF THE MYXINOID FISHES. 299 



spokes of a wheel. These are easily seen in an injected gill coursing quite close to the 

 outer or more convex surface of the gill from the centre to the edge. The radial arteries 

 in their turn give off a number of vessels at right angles which travel in the core of 

 the gill lamella between the epithelial sheets above mentioned, and join up to form, on 

 the inner or less convex (sometimes concave) surface of the gill, another series of radial 

 or spoke-like arteries, which all open into a further circular vessel round the afferent 

 gill duct. From the latter circle the two efferent branchial arteries {ef.br.) arise, and 

 behave as described and figured in my Part IV. 



The actual structure of the gill, however, is not as simple as this. The radial epithelial 

 evaginations, for example, often dichotomise at least twice, and sometimes five times — 

 twice in the lateral regions of the gill {i.e. near the afferent and efferent surfaces), and 

 five times in the central region. This produces at the circular margin of the gill a 

 large number of diverticula instead of a few. Then the radial vessels connected with 

 the rings round the afferent and efferent gill ducts anastomose and branch. Again, the 

 respiratory surface of the gill lamellae does not remain simple and smooth. A number 

 of ridges or pleats are thrown out which themselves dichotomise several times along 

 their long axes so as to produce in transverse section a dendritic appearance — not, 

 however, over the whole of the gill lamella, for, axially and especially laterally, i.e. nearer 

 the greater gill surfaces, the gill lamella remains smooth. The same result would, of 

 course, be produced by a series of ingroivths of the epithelial covering of the lamella, 

 and this is perhaps how the dendritic structure seen in transverse section has arisen. 



The ridges or pleats stretch straight across the gill lamella from one surface of the 

 gill to the other, at right angles to the greater surfaces of the gill. Hence, they are cut 

 longitudinally in transverse and horizontal sections of the gill, and transversely in 

 vertical sections. Hence also the dendritic structure is only obvious in vertical sections. 



A vertical section of a well-injected Myxinoid gill is a beautiful but deceptive object. 

 One sees what appears to be a number of radial vessels converging inwards from the 

 periphery to the centre of the gill. Proximally these vessels are simple in structure, 

 but as they approach the centre of the gill they become highly dendritic. Lateral 

 branches are given off which bifurcate a number of times. There is apparently no 

 question of arteries and veins. Each branch lodges only one vessel, and therefore seems 

 to be a vascular cul-de-sac. 



A closer examination and comparison with sections in other planes reveals the 

 explanation. Proximally each lamella exhibits in vertical section a linear series of 

 vessels cut in transverse section. These are often connected up by vertical anastomoses, 

 and this conveys, in vertical section, the wholly false impression of a radial vessel. The 

 dendritic structure is again equally misleading. Here also, so far from having vascular 

 culs-de-sac, we have in reality a large number of transverse sections of very small vessels 

 coursing at right angles to the plane of the section from one large surface of the gill to 

 the other, and connected up at right angles by innumerable branching anastomoses. In 

 a transverse section these anastomoses show very well as extensive aggregations of 



