210 BULLETIN OF THE BUREAU OF FISHERIES. 
Development and general structure of the gills—The embryonic development of 
the gills of Teredo has been observed by Hatschek (1880) in the viviparous larva 
of the unidentified species he studied. Here the rudimentary gill of each side is a 
fold in which perforations appear in succession, new ones being added posterior to 
those already formed. In the newly attached larva of X. gouldi, the gills have 
advanced but little beyond the stage described by Hatschek. On each side there 
are two slits and the rudiment of a third. The slits, however, have so increased in 
size as to occupy most of the space on the upper sides of the foot and the gill-fold 
has fused to the sides of the foot by its ventral edge. In this way, the gill-slits 
come to separate bars or filaments attached at both ends (fig. 24); and as the fold, 
when it appears, is attached dorsally along the line of attachment of the mantle 
on the sides of the body and the ventral edge fuses with the upper part of the foot, 
the gill-bars or filaments lie almost horizontally in the mantle chamber. 
This mode of differentiation of the gill by the formation of gill-slits in a fold 
whose ventral edge fuses continuously, at first with the sides of the body and 
visceral mass (fig. 2) and later with its fellow of the opposite side (fig. 3), is kept up 
during life. Beginning, however, with a stage still less than 1 mm. long, the pro- 
cess is modified as follows: In specimens less than 1 mm. long (fig. 2) the gill of 
either side consists of a membrane with a single series of gill-slits which decrease 
in size from before backwards. When, however, there are about fifteen slits in the 
series, a perforation in the gill-fold or membrane appears opposite and internal to 
the tenth (rarely ninth or eleventh) slit of the first series. New ones are added 
in succession posterior to it, so that a second series of slits comes to be formed 
internal to the first (fig. 3). At the posterior end the slits of the inner series always 
lag slightly behind those of the external in their development. As shown in figure 
3, there are no slits in the inner series internal to the ten first formed in the outer 
series, and none ever appear, a fact of significance, as will be shown in describing 
the gill of the adult, where the first formed part of the gill, with its outer series of 
ten gill-slits, becomes widely separated from the rest of the gill. 
The gill-fold and gill in the young Teredo represent the internal half of the 
molluskan ctenidium. From the resemblance of the mode of development to that 
in Cyclas (Ziegler) and Mytilus (Lacaze-Duthiers), it is seen that the slits of the 
first formed series separate the descending limbs of the lamellibranch gill filaments, 
and that the second series separate the ascending limbs. The anterior ten filaments, 
then, never develop the ascending limbs. Likewise, the other (the outer) half of 
the ctenidium is never developed in Teredo—contrary to the belief of Deshayes and 
Quatrefages, who believed the whole ctenidium, or ‘‘pair’’ of gills, to be present on 
either side of the body. 
The term ‘‘gill-fold”” I have used to designate the posterior end or growing 
point of the gill, and ‘‘gill-filaments”’ the elements that are formed from it. In 
later stages, however, soon after that shown in figure 2, the growing point forms a 
more or less cylindrical hollow tube, filled by a blood space, which fuses continu- 
ously on the midline with its fellow of the opposite side, and dorsally over a wide 
area (between the two points indicated in fig. 37) with the mantle. In this way 
the epibranchial cavity is separated from the rest of the mantle cavity. Meanwhile, 
