7 o DISCOVERY REPORTS 



This being so, euphysoclists simply require unipolar retia, but the eels are an exception to this 

 ' rule '. It should also be stressed that a bipolar retial system is not a characteristic of all paraphyso- 

 clists. Unlike the stomiatoids, other teleosts with this type of swimbladder (Synentognathi and 

 Microcyprini) have unipolar retia. 



Turning now to the composition of the retia, Krogh (1922) was the first to appreciate the extra- 

 ordinary extent of the capillary elements. In a cross-section of the two retia of the freshwater eel he 

 estimated that there were 88,000 venous and 116,000 arterial capillaries. As the two retia were 

 4 mm. in length this gave aggregate lengths of 352 and 464 m. for the two sets of capillaries. Krogh 

 also pointed out that the capillary elements are remarkably long compared to those in muscles, which 

 are otherwise among the longest in the vertebrate body (e.g. 4 mm. in eel retia against 0-5 mm. 

 in muscle). 



Similar data are given in Table 3 for various species of bathypelagic fishes (see PI. I, figs. 3 and 4 for 

 the appearance of a rete in cross-section). The significance of the rjv ratio will become apparent in the 

 text which follows. 



Table 3. Retial length : swimbladder volume ratios 



At first sight the figures for the total lengths of the retial capillaries seem iow compared with 

 Krogh 's 816 m. for the freshwater eel. But as the retia are an essential part of the gas-producing 

 mechanism, their numerical constitution will be best regarded in relation to the volume of the swim- 

 bladder. (The volume of the eel swimbladder was estimated from Fange's (1953) fig. 17, assuming 

 the secretory sac to be a perfect ellipsoid.) In view of this, an rjv ratio was calculated (r being the 

 total length of the retial capillaries, and v the volume of the swimbladder (ml.)). 



While these figures can only be rough approximations, it will be seen that, except for Polyipnus and 

 Vinciguerria, the rjv ratio of the eel is much the same as those of the bathypelagic species. And in 

 comparing these data, it should be remembered that during its reproductive migration the freshwater 

 eel becomes a deep-sea fish. The agreement between the ratios is thus not altogether surprising. But 

 before considering this problem further, reference must be made to Scholander's (1954, 1958) 

 theoretical study of the rete mirabile of deep-sea fishes. 



Following earlier suggestions, Scholander convincingly argued that the retia must form a counter- 

 current system allowing of gaseous exchange between the arterial and venous capillaries. Without 

 intimate contact between the two sets of vessels, the blood leaving the gas-gland would be continually 

 removing oxygen from the swimbladder. He wrote as follows in his 1958 paper (page 7): 



The atmospheric oxygen which is dissolved in the sea water has a gas pressure of no more than one-fifth 

 of an atmosphere at any depth, and the arterial pressure in the fish is slightly below this. So across the 

 thin swimbladder wall of a fish living at a depth of 2000 m. there is a drop in oxygen pressure of nearly 

 200 atmospheres. The swimbladder is a living organ and is circulated with blood. At a pressure of 200 

 atmospheres the blood becomes charged with ten times its own volume of oxygen by simple physical 

 solution, and if such amounts were to leave the swimbladder there would soon be no oxygen left. In more 



