232 
Another method which I adopted in order to activate the gas 
gland was to sink fish enclosed in a cage in the open sea to a depth 
of some 30 fathoms (to the east of the Eddystone lighthouse) for 5 
hours. The fish employed in this experiment were Conger, Wrasse 
and Siphonostoma. I admit that the swollen appearance of the bladder 
is the only evidence I possess in connection with this experiment that 
the gas gland had been active. 
The Structure of the Gas Glands in the Fish employed. 
Little need be said concerning the structure of the gas glands 
in the fish employed in view of the general description supplied in 
preceding pages. I will merely mention that the gas gland of the 
Pollack is a large diffuse red mass situated on the interior of the 
ventral floor of the bladder at the extreme anterior end. In general 
form it closely resembles the gas gland of the Perch (see 15, fig. 69, 
pl. IX). The rete mirabile, instead of being a bipolar compact ovoid 
mass such as is found in the eel and represented in the preceding 
diagram (Text-fig. 1) is, on the contrary, broken up into numerous 
small unipolar tufts, each supplying a small area of the diffuse 
glandular epithelium. The gland epithelium is of the massive type 
(see 15, text-figs. 61, 62) and more developed than in the Perch. 
The gas glands of the Mullet and of the Wrasse are of the same type; 
those of the Conger and Siphonostoma have already been described 
by me in the paper mentioned (15). 
Experimental Tests of the General Theory of Gas Production. 
The principal questions which I set myself to answer were three 
in number, viz. 1) do the spherical vacuole-like spaces to be found 
inside and the globules to be found outside the cytoplasm of the gas 
gland cells in serial sections of presumably active gas glands represent 
gas bubbles in process of formation, as originally affirmed by 
BykowskI, NUSBAUM, Reis and myself, or do they merely represent 
semi-liquid bye-products of the gland’s activity, as affirmed by JAEGER?; 
2) does haemolysis of the blood occur in the capillaries of the rete 
and the gas gland, as supposed by JAEGER, Bykowsk1 & NUSBAUM 
and myself? — a supposition again based upon balsam preparations 
(serial sections); 3) is there any evidence that the gas gland cells 
absorb dissolved oxyhaemoglobin from the blood? I will now discuss 
the evidence afforded by my experiments at Plymouth which compels 
