ON THE GENERAL MORPHOLOGY OF THE MYXINOID FISHES. 781 
The branchial skeleton is situated entirely posteriorly in relation to the fused gill 
ducts, and on the left side to the ductus cesophago-cutaneus also. To take the latter 
side first (fig. 11), we find in dissections that about half the skeletal apparatus in its fully 
developed form is connected with the fused gill ducts, and the other half with the ductus 
cesophago-cutaneus and the cesophagus itself. The latter portion lies on the external 
lateral wall of the cesophagus and the ductus, and passes generally downwards and 
backwards. Above, it has two processes—a dorsal posterior one (x’), which passes under 
the constrictor muscle on to the roof of the cesophagus ; and a ventral anterior one (x), 
which passes straight on to the ventral surface of the cesophagus. Below, it has one 
process (a*)—a posterior one which passes backwards towards the caudal wall of the 
ductus—and a vertical rod (x*), which connects it with the second portion of the apparatus. 
The latter commences with a horizontal rod (y’), which passes forwards in a curve over 
the branchial cloaca, bears a blunt process in front (y’), and then suddenly dips down, 
bends round under the ventral border of the first gill duct, to terminate in a lanceolate 
plate (y*), which turns upwards and forwards and lies on the inner surface of the 
first two efferent gill ducts. The plate in the specimen figured was pierced by two 
fenestree. 
We may now turn to fig. 13, which represents this apparatus as reconstructed from 
serial sections, and we note at once that it consists of two perfectly distinct parts. The 
dorsal one commences ventrally on the anterior external wall of the ductus, and below, 
it sends inwards a hook-shaped process underneath the ductus (x*), At about section 
3070 (cp. the chart) the process x has reached the cesophagus, on the outer wall 
of which it lies for the rest of its course. The process x* passes backwards and down- 
wards over the external surface of the ductus, and terminates a short distance beyond it. 
The ventral part begins anteriorly by a perforated plate (y’), which lies immediately 
internal, and is related, to the first four efferent gill ducts (including the fifth posteriorly), 
instead of the first two, as in fig. 11. The knob y’ is represented in the sections by a 
long blind process extending forwards for some distance external to the first four efferent 
gill ducts. The bar y’ bends round under the ventral wall of the branchial cloaca to 
fuse with y*; whilst above, it passes backwards external to the branchial cloaca, exhibit- 
ing a longitudinal fenestra, and behind ends blindly before the branchial cloaca fuses 
with the ductus. 
A comparison of these two figures renders it abundantly clear that the branchial 
skeleton of the left side is a complex of at least two parts (see below), the point of 
junction in fig. 11 being where x* meets y’. In fig. 13, therefore, the greater portion of 
x* has been lost, in this way separating the two sections—one of which clearly belongs 
to the efferent gill ducts, and the other to the ductus cesophago-cutaneus. This view of 
the branchial skeleton is borne out by its wide range of variation, and I have dissected 
two specimens in which each half of the apparatus was respectively missing. One of 
these variations is represented in fig. 12, which obviously represents the ductus portion 
of the skeleton, and corresponds exactly with the dorsal portion of fig. 13, except that 
TRANS. ROY. SOC, EDIN., VOL. XLI. PART III, (NO. 30). 115 
