730 MR FRANK J. COLE 



(fig. 3, 1 and °2), but both loops were in the specimen now described, and as shown in 

 fig. 3, separated from the limb by thin ligamentous septa, which, however, judging 

 from my other dissections, are of exceptional occurrence, in which case the fibres of 

 the loop are directly continuous with those of the limb, a number of fibres in the latter 

 turning sharply downwards at right angles to form the loop. In one specimen all the 

 loops were exceedingly reduced ; nevertheless the ventral longitudinal tract (figs. 2, 3, 

 13, ?) was well developed. This shows that the latter tract has an existence independent 

 of the loops — which is to be expected, seeing that the first loop, in some cases the 

 fourth and fifth (cp. fig. 13), and generally the fifth, do not contribute any fibres to the 

 tract. In the same specimen I found that the first loop, however, did unite with the 

 ventral tract, some of the fibres passing backwards with the tract, and the others 

 passing forwards with it to the usual insertion of the loop on the longitudinalis linguae. 

 Incidentally, we may here note that the second loop of Bdellostoma, corresponding to 

 the first of Myxine, does not, according to J. Muller, reach the longitudinalis linguae. 



First Loop (1). — Passes downwards and forwards in a curve over the external face 

 of gill 2, then under the first efferent gill duct, not far from its origin from the gill, 

 and finally bends straight forwards to fan out on the lateral fascia of the longitudinalis 

 linguae. In one specimen this insertion was ligamentous, and not muscular, as it generally 

 is. In Bdellostoma the loops are said to course in the depressions between the gills. 

 I have frequently seen this in Myxine also. A very striking difference, however, is 

 that in Bdellostoma, according to J. Muller, there is no ventral longitudinal tract, 

 but all the loops are attached to the ventral belly-wall opposite the external branchial 

 apertures. In many specimens of Myxine the loop included a number of fibres (as 

 shown in fig. 3) which did not pass underneath the efferent gill duct, and which fanned 

 out in the fatty tissue covering the gills ; but in other individuals the first loop 

 consisted solely of fibres for the longitudinalis linguae. We do, however, find the non- 

 ventral fibres of the loops, when present, more and more spread out as we pass 

 backwards, so that the continuous posterior sheet may be regarded as having been 

 formed by these fibres of two or more loops fusing together. 



Second Loop (#). — Passes downwards and slightly forwards over the external face 

 of gill 3, crosses over the origin of its efferent duct, then dips underneath first the 

 second and then the first efferent ducts, to finally pass over into the ventral 

 longitudinal tract. The latter varies considerably, and it may be very greatly reduced 

 on one or both sides so as not to reach the second and third loops, which consequently 

 terminate in the fatty tissue below their respective gill ducts. The formation of this 

 ventral tract will be described subsequently. Most of the fibres of the second loop 

 pass forwards with the tract, but a few of them turn backwards along it. This, 

 however, here and in the other loops, is subject to variation, and the fibres of the 

 loop may be equally divided, one half passing forwards and the other half backwards, 

 and so on. The ventral tract, after receiving loop 2, continues its course straight 

 forwards, and in the specimen now described terminated in two ligamentous fans, the 



