258 TRANSACTIONS OF THE [JUNE 4, 
parts only—the unmodified parts disappearing. The attachment 
of external gills to the pectoral arch of Lepidosiren is very sig- 
nificant in this connection. The branching of the gills is note- 
worthy here. In most of the larve I had the branching was 
unilateral, but in a few there were traces of bilateral branching. 
In Salamandra atra, as figured by Wiedersheim,’ the gills are 
very long, and the bilateral branching is perfect, giving them 
the appearance of plumes. From such a gill as this the deriva- 
tion of the Dipnoan type of limb would be very simple, while by 
the suppression (or non-development) of the branches on one 
side there would be a uniserial type from which other forms 
could be derived. 
In my specimen of amblystoma, the branches appeared first at 
the distal extremity of the gills, and developed in regular suc- 
cession toward the proximal. ‘The limbs appeared as lateral 
buds close behind the gills, and had the circulation well estab- 
lished by the fourth day after hatching. By the fifth day the 
limb was elongated, and a slight bifurcation of the distal ex- 
tremity was noticeable, which was very marked by the seventh 
day, and the first and second digits were well separated. So far 
as my observations on the living specimen are concerned, they 
go to show that these digits are formed before there is any trace 
of differentiation of the humerus, radius, or ulna. 
If the derivation of the limbs suggested here is admissible, 
the presence of a lateral fold in the Cyclostomata larve does not 
necessarily imply that this group formerly possessed lateral fins, 
as Prof. Dohrn supposes, but might only indicate the former 
possession of external gills, which extended backward through 
the body segments. ‘The branchial clefts were formerly far 
more numerous than now, as is shown by Amphioxus, and I 
think it probable that, in the earliest forms, they extended the 
entire length of the animal.* Some of the suggestions offered 
here I hope, in a later paper, to develop more fully. 
1 Wiedersheim, ‘‘ Lehrbuch der Verg]. Anat.,” p. 611, f. 475, C. 
° Prof. John A. Ryder has kindly allowed me to state that he find 
in the embryos of the shad rudimentary gill clefts, posterior to those 
which develop in the adult. Mr. Ryder has also called attention to 
some facts which do not seem quite in accord with the lateral fin 
theory. See ‘‘ Annual Report Commissioner of Fisheries,” 1884, pp 
1,044-1,046. 
