ON THE GENERAL MORPHOLOGY OF THE MYXLNOID FISHES. 731 



dorsal one of which turned upwards to become attached to the fascia at the posterior 

 end of the longitudinalis linguae, whilst the ventral and larger one bent downwards to 

 effect an extensive connection with the internal surface of the rectus muscle near its 

 outer margin. This termination just described is, however, by no means of general 

 occurrence. The same remark as to the occasional presence of non-ventral fibres applies 

 to the second as well as to the first loop. 



Third Loop (3). — Immediately behind the origin of the anterior dorsal limb 

 the main body of the constrictor despatches ventrally a third loop. This loop, in all 

 the specimens I have dissected, includes a greater or lesser number of the non-ventral 

 fibres. These fibres were very numerous in the present case, and by fanning out in 

 front and behind largely covered the external surfaces of gills 4 and 5. The whole of 

 the fibres of the loop were marked off at their origin from the major portion of the 

 muscle by a ligamentous partition, but this is by no means of universal occurrence. 

 The ventral portion of the loop passes external to the origin of efferent gill duct 4 

 almost straight downwards, but slightly forwards, then slips under the efferent ducts 

 3, 2, 1 in the order mentioned, and finally goes over into the ventral longitudinal tract 

 in much the same way as loop 2. 



The last two loops, owing to their origin from the internal surface of the now con- 

 tinuous posterior sheet of the constrictor some distance above its ventral border, 

 are consequently not entirely seen until the covering sheet has been turned dorsally 

 or its ventral portion removed. Hence in fig. 3 the latter has been done in order 

 that all the loops may be included. The behaviour of these two loops also varies 

 considerably. 



Fourth Loop (4). — Passes external to the origin of efferent duct 5, and then down- 

 wards internal to the ducts 4, 3, 2 1. Ventrally it may go over into the ventral 

 longitudinal tract, the fibres diverging to accompany the latter backwards and forwards 

 (as shown in fig. 3), or it may assist in forming the internal portion of the ventral 

 insertion (as shown in fig. 13, 8). In the latter case the loop distinctly bifurcated, one 

 limb passing into the ventral insertion, and the other coursing backwards as described 

 under the ventral insertion, 2. 



Fifth Loop (5). — Courses external to the origin of the efferent duct 6, and. then 

 ventrally internal to ducts 5, 4, 3, 2, 1. I have never seen this loop connected with 

 the ventral longitudinal tract except by a very few fibres which it occasionally despatches 

 to it (but cp. the description of the sections). It always passes external to the tract 

 and on to the inner portion of the ventral insertion (figs. 3, 13, 8), but some of them 

 turn backwards as described under the ventral insertion, 2. 



The last two loops are very differently arranged in Bdellostoma, according to 

 J. Muller. Here they bend right round and fuse in the mid- ventral line over the 

 cardiac aorta on the ventral surface of the constrictor cardiae, so as to form a distinct 

 sphincter system of their own. Also in Myxine itself, according to Muller, the last three 

 loops fuse up to form on each side a ventral longitudinal tract, which, however, courses 

 TRANS. ROY. SOC. EDIN., VOL. XLV. PART III. (NO. 26). 104 



