' TRANSACTIONS OF SECTION I. 547 



to form the capillaries supplying the gas gland, the venous capillaries uniting 

 in a corresponding manner, or unipolar, the arterial capillaries of the rete 

 directly supplying the gas gland cells and the corresponding venous capillaries 

 returning from the gland at once mingling with the arterial capillaries without 

 first uniting into a few large veins. The essential feature of the rete mirabile 

 is the intimate intermingling and juxtaposition (not inter-communication) of 

 the two sets of capillaries carrying blood in opposite directions. The author 

 has published 1 a full account of the structure of these gas glands and their asso- 

 ciated retia mirabilia in various teleosts, and also certain suggestions as to the 

 use of these organs, based upon the facts described by Bykowski, Nusbaum, 

 Reis, himself, and other investigators. In brief, the general theory proposed 

 was as follows : — 



In the supply of oxygen to the ordinary tissues of the body the gas is, of 

 course, combined with the haemoglobin of the red blood corpuscles, and is 

 only liberated from this combination into the blood plasma as the partial 

 pressure of the oxygen already dissolved in the plasma is lowered by absorption 

 of the oxygen by the tissues. Now the cells of the oxygen gland differ from 

 the other tissues of the body in that, being already employed in pumping 

 oxygen into the bladder, they are in no special need of it for metabolic 

 purposes, and even if they were able to abstract the oxygen from the blood 

 plasma in the same way that other tissues do, this method of abstraction would 

 be quite inefficient in view of the fact that the oxygen required by the gas 

 gland must be abstracted both in relatively large quantity and at a relatively 

 rapid rate. The necessity for this rapid production by the gas gland of large 

 quantities of oxygen will be evident on realising the further fact, proved by 

 numerous experiments, that oxygen (doubtless in virtue of its abundance in 

 arterial blood and its ready absorptibility) is the only gas employed both for 

 the sudden increased production of gas in the bladder required when a fish 

 sinks in water and so experiences great increase of pressure, and for the sudden 

 abstraction of gas (the oxygen re-entering the blood stream through the ' oval ' 

 when this structure is present in the bladder wall) necessitated by the rising 

 of a fish, the body of which thereby experiences diminution of pressure. 

 Available evidence thus points to the fact that the presence of oxygen in the 

 bladder connotes the habit on the part of the fish of migrating in a vertical 

 direction in deep water; in the case of those fishes which do not change their 

 depth to any appreciable extent (e.g., the marine flying-fish and many fresh- 

 water fish, including a few living at considerable depths) the bladder is filled 

 with gases which do not require to be produced quickly — nitrogen and carbon 

 dioxide — and could not be if required owing to the small amount present in th« 

 blood, the oxygen of the blood in these cases being reserved for respiratory 

 purposes. It is considered extremely probable by some authorities that the 

 method by which the gas gland forcibly abstracts the oxygen from the arterial 

 blood is by the secretion of a ' toxin ' (Jaeger) on the part of the gas gland 

 cells. This toxin is poured into the blood traversing the capillaries of the 

 gland, where it effects partial haemolysis, the erythrocytes being broken up and 

 their contained oxyhemoglobin dissolved in the plasma. This dissolved oxy- 

 haemoglobin is then abstracted by the gas gland cells, which thus seize upon the 

 prime source of the oxygen in the blood in place of the weak solution of oxygen 

 in the plasma supplied to ordinary tissues. In the author's opinion the rete 

 mirabile was for the purpose of enabling the toxin to act upon the arterial blood 

 before it comes into contact with the gas gland, so that by the time the blood 

 reaches the gland the oxyhaemoglobin shall be already dissolved in the plasma 

 and so available for abstraction by the gland cells. On this view the gland 

 pours the hypothetical toxin into the blood, which, returning from the gland, 

 subsequently traverses the venules of the rete ; during its course in the rete the 

 toxin diffuses from the venules into the adjacent arterioles where it produces 

 haemolysis of the arterial blood, a process which effects the solution in the plasma 

 of a quantity of oxyhaemoglobin by the time the arterial blood reaches the gas 

 gland. 



Recent experimental work, however, conducted at the Plymouth Marine 



1 Proc. Zool. Soc. Lond., vol. i., 1911. 



n n2 



