214 BULLETIN OF THE BUREAU OF FISHERIES. 
among the epithelial cells (fig. 48), the filaments are arranged lengthwise; soon they 
take on a vertical position (fig. 49), enlarge, and become covered by the caplike 
membrane (fig. 49-51). If this derivation be the correct one, then the rods in 
figure 51 have been formed by the enlargement of the filaments of the dentritic 
structures. In the lamina the rods project toward the blood space; in the afferent 
branchial vein, away from the blood space. Why the difference I do not know. 
The development of the gland of the adult, so far as I can determine, is as 
follows: When the small ship-worm has been in the wood for a day or so, the gland 
of the larva sends out processes which invade the surrounding ectodermal tissues (the 
mantle, sides of the body). As the side of the body becomes enlarged, it fuses with 
the dorsal sides of the gill filaments (fig. 25). From the first there is close association 
between the gland and the gill. As the latter grows, the filaments become invaded 
by the gland; and, as the anterior ten filaments become separated from the rest of 
the gill, the two parts of the gland thus differentiated remain connected by a long, 
narrow duct which accompanies the epibranchial canal and lies in the afferent 
branchial vein (fig. 26-30). With the separation of the two parts of the gill, the 
intervening part of the gland disappears in Y. gouldi and T. navalis, but persists 
in T. dilatata. As the gland enters each gill lamina, it remains connected by a 
small duct with the main duct, and may send the granular cells into the latter. 
The main duct may become gorged with granular cells (fig. 37). In most cases, 
however, there are few cells in it, and I am inclined to regard the main duct (at 
least in Y. gouldi) as degenerate and perhaps essentially functionless. Likewise, 
T am inclined to regard the formation of the spherical cells in the one part as not 
the chief function of this part of the gland. The origin and fate of the cells I have 
not been able to determine. Their contents suggest that they may be modified 
mucous gland cells. 
What the special function of this remarkable structure is I am not able even 
to guess. The rudimentary character of the anterior part in the “head” of 7. 
navalis indicates that it can not be the formation of a secretion to soften wood. Its 
development in the gill, in small as well as large individuals, in male and female, 
and in forms that do not retain the eggs in the gills, proves that this part can nbt 
be for the nutrition of viviparous embryos. The close connection with the gill 
indicates, it seems to me, that its function may be the elaboration of some internal 
secretion or other material for whose formation the presence of both blood and 
water is necessary. 
CIRCULATORY SYSTEM. 
The circulatory system of the shipworms is highly modified in relation to the 
peculiar form of the body. The growth of the visceral mass ventrally at first, and 
afterwards its great elongation posteriorly, along with the elongation of the rest of 
the body, accounts for the changes that have taken place. Doubtless the ancestors 
of ship-worms were lamellibranchs with typical circulation, in which on either side 
in the pericardial cavity lay an auricle lateral to, and emptying into, the median 
ventricle which surrounded the intestine; and from the ventricle the anterior aorta 
passed forward above the intestine and the posterior aorta backward below the 
intestine. In ship-worms the pericardial cavity, with its contained parts, has come 
