240 
many cases almost entirely. All these experiments therefore directly 
contradict my previous results and there is no alternative but to ac- 
cept the conclusion of the former. This conclusion is indeed almost 
implied by the now proved absence of haemolysis, unless the “toxin” 
assumed to be secreted by the gland may be supposed merely to dis- 
solve out the oxyhaemoglobin without detriment to the corpuscle bodies 
but even this view is untenable in view of the experiments described 
above, I not once having been able to obtain again the appearance 
of the blood which I figured in my previous paper (fig. 21, pl. IID). 
Two facts must however be mentioned. One is the extraordinary 
amount of granular matter always present in the capillaries of the 
active gland and rete (described by several of the authors named 
above), a granular matter which in many cases appears actually to 
be passing into the cytoplasm of the gland cells (and perhaps thence 
into the bladder cavity), and the other is the occasionally empty 
appearance of the red corpuscles, the body of the individual corpuscle 
remaining quite unstained in some of my preparations. I however 
lay no stress upon these facts, since there is no evidence that these 
numerous plasma granules have anything to do with dissolved haemo- 
globin and the empty bodies of the corpuscles are probably only an 
effect of the particular method of fixation and staining adopted. 
Conclusions. 
These conclusions (provisional in the sense that additional evidence 
is much to be desired) that haemolysis does not occur in the gas gland 
and rete capillaries and that the gland cells do not absorb haemo- 
globin relieves the theory of gas production of the necessity of making 
several assumptions. There is no necessity to assume that the gland 
cells secrete a lysin, and, since gas bubbles are not formed in the 
cells, also no necessity to assume that they secrete a substance to 
liberate the oxygen dissolved in the plasma. There is also no neces- 
sity either to account for the elimination of broken-down corpuscles 
and the débris resulting from haemolysis or to assume that on active 
reproduction of red blood corpuscles takes place in some part of the 
body — a process which would be very necessary if erythrocytolysis 
occurred on the scale required by the hypothesis. Under these circum- 
stances we must suppose that, as in normal metabolism, the plasma 
of the blood conveys all the oxygen derived from the oxyhaemoglobin 
to the gas gland cells and that these in some manner transmit it by a 
process of diffusion under great pressure into the bladder cavity. The 
blood-stream itself is, it may be remarked, under considerable pres- 
