196 DI'o W. N. F. WOODLAND ON THE 



ill the usual way, the epithelium lining the rest of the bladder 

 remaining squamous in character. It is, however, noticeable 

 that in the rete of Gobius the fine capillaries do not unite 

 too'ether at all to form larger vessels before supplying the 

 o-landular epithelium (see fig. 27). In Syngnathus it has been- 

 mentioned that the reunion of the fine capillaries is very slight, 

 but here it is quite absent. 



The o-landular epithelium is of the same type as that found in 

 Syngnathus, the cells only, as may be seenfrom figure 27, being 

 somewhat more columnar in form. The epithelial cells of Gobius 

 are relatively large, and, as just sfcated, columnar in form and in 

 my preparations faintly granular. They are folded in the simple 

 manner already described for Syngnathus and Anguilla, large 

 capillaries being situated in the interior of the folds *. In figure 27 

 are represented three epithelial folds of the gas glancl, underlying 

 which are a few of the multitudinous fine capillaries composing 

 the rete mirabile. Even in this figure it can be observed that 

 several of the cells contain spherical clear spaces and that two 

 small capillaries have encroached on the space previously occupied 

 by the cytoplasm of two cells of the gas gland. These intra- 

 cellular bubbles and capillaries and the intracellular lumina, which 

 1 shall also shortly mention, are cytological fea,tures which have, 

 within the last year or so, been studied in great detail by Jaeger 

 (44-47), Bykowski &, Nusbaum (24), and Reis & Nusbaum (54, 

 55 62-66) and they are of considerable interest from several 

 standpoints. I shall discuss their significance later; for the 

 present I may remark that I think I am justified in redescribing 

 these structures, especially in view of the controversy concerning 

 the mode of gas-production considered in Part II. That the fine 

 capillaries of the rete mirabile often find their way deep into the 

 substance of the glandular cells, as well as between them, is proved 

 in Gobius by a veiy moderate amount of inspection under a high 

 power of the microscope (PI. IV. figs. 28, 29). It is also noticeable 

 that the protoplasm of that border of the glandular cells in contact 

 with the capillary develops a dark and often striped appearance, 

 which Nusbaum & Eeis suggest is due to difTusion streams of 

 nutritive matter passing from the blood into the gland-cells, but 

 which I believe is intimately concerned with the production of 

 the gas contained in the bladder (see Part II.). Concerning the 

 presence of vaciiolar spaces in the cytoplasm of the cells of the 

 o-as gland, nobody can doubt their existence who has examined 

 well-preserved preparations of gas glands which are in an active 

 condition. These vacuolar spaces are with good reason assumed 

 to represent bubbles of gas being generated within the cell- 

 substance (see Part II. and Appendix B), but it is evident that only 

 on occasion will the gas gland be fixed and preserved when active ; 

 in the majority of cases, preparations ot gas glands will not 

 exhibit these intracellular gas bubbles, simply because the gas 



* Eels (64) states that tlie folds branch and anastomose in G-ohius; in all my 

 nrenarat'ions of Gobius nic/er, Gobius paganellus and Gobms minutus (also in 

 Gobms ca}rito—Cos:g\, 25) the folds are quite simple, as described in the text. 



