HISTOLOGY: GAS-GLAND 393 



The cells have been found to occur in clusters (islets) of from three to eighteen; in some cases, they 

 are distributed around a narrow but well-defined intercellular space. Like the giant cells, they occupy 

 a sub-epithelial position. They have sharp outlines, by virtue of their condensed and fibrous cytoplasm, 

 and a small, densely staining nucleus near the base of the islet. The nuclei show varying degrees of 

 staining intensity with iron haematoxylin. Some are so dark that the nucleolus (which normally stains 

 much more deeply than the rest of the nucleus) is not separately distinguishable. In Feulgen prepara- 

 tions, the nuclei stain brilliantly. This is sometimes a sign of pycnosis (Alfert, 1955), but there are no 

 other symptoms of degeneration in the islets. 



The islets occur without apparent regularity throughout the gland. They are frequently, but not in 

 the example shown in PI. XXVIII, fig. 2, associated with the giant cells, clustering around the latter. 

 Mitosis has not been observed in them. From their frequent association with the giant cells, it 

 might be thought that they contribute in some way to the latter's formation, but there is no direct 

 evidence for this. 



With regard to the physiological aspects of gas-secretion in the Siphonophora, very little is known. 

 In certain forms, such as Physophora, there is abundant evidence that gas-bubbles can be emitted 

 from the float, causing the animal to sink in the water. The gas can be regenerated in a few minutes 

 (Keferstein and Ehlers, 1861, and subsequent workers). Jacobs (1937), in an important paper, 

 described a similar process in Stephanomia {=Nanomia) bijuga. Gas-secretion was observed by this 

 worker in an isolated float after some of the contents had been discharged ; it involved the appearance 

 of small bubbles in the region of the gas-gland, their rapid increase in size and fusion together, and 

 their final merging with the air already in the float. 



It has yet to be shown which of the various cells of the gas-gland are actually responsible for the 

 secretion. Schneider (1902) maintains that the giant cells in Rhizophysa and Physophora are respon- 

 sible, but Dahlgren and Kepner's observations on Physalia point to the columnar epithelium as the 

 source. The present investigation has given some support to the latter view. The chemistry of gas 

 secretion, and the character of the stimulus or stimuli evoking it or inhibiting it are completely 

 unknown. 



Some workers have held that in Physalia, as in Rhizophysa, etc., gas may be emitted from the float 

 causing the animal to sink below the surface. Observations in the Canary Islands in no way support 

 this view. Specimens were never observed to liberate gas spontaneously. It is very difficult to squeeze 

 even a few small bubbles out through the apical pore by manual pressure. Nothing in the structure of 

 Physalia suggests that it is adapted for submarine existence. It is a robustly built animal adapted for 

 sailing on the surface, and there is no shred of evidence to show that it ever leaves the surface during 

 its adult life. 



The float probably loses gases slowly by diffusion in spite of its insulation (the pneumatocyst), and 

 as the animal grows the volume of gas in the float must steadily increase. A steady demand for gas is 

 therefore to be expected, but there is no reason why sudden demands for large quantities of gas need 

 normally occur. The persistence of the gas-gland in the adult, and its growth with the rest of the float 

 tissues, as well as its small size relative to the float as a whole are just about what one would expect to 

 find. 



A series of experiments was undertaken to test the capacity for gas-secretion. The volume of the 

 float was estimated by water displacement in a measuring cylinder. Known values of gas were then 

 withdrawn with a syringe, and the volume of the float measured again at intervals. However, this 

 rather clumsy technique failed to yield any significant data. It was found, moreover, that the speci- 

 mens deteriorated quickly in the laboratory, and both controls and experimental animals tended to 



