STRt'CTURE OF THE EXCRETORY OR(T\XS OF AMFHIOXUS. 193 
tlie canal of tlie nephridium gives off two or three short 
diverticula, which are turned away from the coelom towards 
the middle line. The numerous solenocytes projecting from 
these diverticula lie in a sort of pocket or chamber (figs. 1, 2 , 
3, 23), which only communicates with the coelom by means of 
a dorsal opening, over which pass a large number of soleno- 
cyte tubes. In one region the inner wall of this chamber is 
fortued by the skeletal rod of the gill-bar (figs. 3, 23). Some- 
what similar pockets are found occasionally in connection 
with other parts of the nephridium, as, for instance, the 
anterior limb of the canal. The cavity in the chambers is, 
I believe, rather of the nature of a lymph space than of a 
true ccclomic cavity. 
The Relation of the Nejihridium to the Ccclomic 
Epithelium. — It is important to determine exactly what is 
the disposition of the coolomic epithelium in the neighbour- 
hood of the nephridium. Boveri ( 1 ) and Weiss ( 13 ) have 
already shown that the canal is covered by the ccclomic 
epithelium ; but this epithelium only clothes the outer or 
ccclomic surface (text-figure). It passes on to the nephri- 
dium from the atrial wall, covering the canal and its blind 
branches to their extremity. Here it is not reflected so as to 
pass over to the inner or atrial surface of the organ, but ends 
abruptly near the base of the solenocyte tubes (figs. 2, G, 
8, 9). 
Thus the nephridium and the bloodvessels which accom- 
pany it maybe said to lie ‘‘morphologically” eiitirely outside 
the coelom; between the ccclomic epithelium and the atrial 
epithelium. The nephridium is, in fact, retroperitoneal. 
'I’liis is true, I believe, of the solenocytes themselves, though 
less easy to prove. For the ccclomic epithelium stops short 
where the solenocytes begin (figs. 6, 8, 2), passing neither on 
the inner side over the bloodvessel, nor outside them over 
their ccclomic surface. For a long time I was under the 
impression that a very delicate membranous extension of the 
epithelium covered over the coolomic surface of the soleno- 
cytes; but I am now satisfied that this is not the case, 
