VISCERAL ARCHES OF THE GNATHOSTOME FISHES. 303 
m part perforates the cartilage, the muscle strands of the 
adductor there being directly continuous with those of the 
musculus interbran chialis of the arch. The adductor is thus 
here not yet fully cut off from the primitive constrictor of its 
arch, and if the epibranchial were similarly perforated in 
younger stages it is certain that these particular strands 
would be innervated by a nerve that traversed the perforation 
of that cartilage. 
If this be the explanation of the conditions in the Plagio- 
stomi, it is quite certain that the conditions in the Ganoidei 
were not derived from them. The conditions in these latter 
fishes are associated with, and quite undoubtedly correlated 
to, the presence of the straight form of branchial bar instead 
of the sigma form, and to the absence of cartilaginous 
branchial rays. Where these latter rays are found, the con- 
strictor muscle of an arch could slip, or project, over the 
anterior edge of the related branchial bar, as described by 
Dohrn in plagiostoman embryos, and so give rise to an 
adductor muscle, but it could not so slip over the posterior 
edge of the bar. The ganoidean adductores could not, 
accordingly, have been developed in a fish already possessed 
of cartilaginous branchial rays, and it would even seem a< if 
they could not have been developed in a fish already pos- 
sessed of the cartilaginous or osseous rods that support the 
branchial filaments in all the Teleostomi. These osseous rods 
I have already described in Scomber (Allis, 1903), and I now 
find similar supporting rods, of cartilage instead of bone, in 
Amia, Polyodon, and Polypterus. In Amia they are not 
evident until the fish is over 12 mm. in length. These rods 
are found in two series, one along the anterior and the other 
along the posterior edge of the branchial bar, and it would 
seem as if a constrictor muscle, which must primarily have 
occupied a position between their lines of attachment to the 
branchial bar, could not, after their development, have 
slipped over either edge of the bar. The Ganoidean adduc- 
tores must accordingly have been differentiated before these 
supporting rods were developed. In a specimen of Ceratodus 
