THE CRASPEDOTE MEDUSA OLINDIAS AND SOME OF ITS NATURAL ALLIES. 5 



aspect on the inner (oral) and outer (aboral) sides. On the former the cells are nearly- 

 cubical, stain deeply, and contain numerous fine granules, while on the latter side 

 the cells are exceedingly tall and are at places almost clogged with large granules, 

 which are exactly similar to those found in the lower walls of the radial canals (PL 

 II, Figs. 15, 16; PI. Ill, Figs. 17, 18). In some examples these cells are so tall as 

 to project into and considerably narrow the lumen of the canal. 



There are numerous centripetal canals of varying lengths in each sextant, lying, 

 as in most other cases where these are found, close to the ectoderm of the subumbrella. 

 They increase in number with the age of the medusa, and the longest lie midway 

 between the radial canals, the next longest between the longest and the next radial 

 canal, and so on. As the centripetal canals increase in number, however, this law 

 is apt to be disturbed. In the specimen represented in Figure 6, for instance, this 

 law holds good only for a short way; and even in small specimens the centripetal 

 canals are not frequently found so regularly graded as in Figure 5. In one and the 

 same specimen the number of centripetal canals varies from sextant to sextant, 

 although within a limited range. To give only two examples, in a specimen with the 

 disk of 75 millimetres in diameter, which is about full grown, the maximum number 

 of centripetal canals was 23, and the minimum 18; in another of 15 millimetres in 

 diameter the centripetal canals varied between 11 and 14; the total number for the 

 former being some 120, for the latter 78. 



The centripetal canals are, in structure, repetitions on a smaller scale of the radial 

 canals, and their endoderm presents exactly the same features as in the former. 



The radial, circular, and centripetal canals are connected with one another and 

 with the central stomach by the endodermal, or vascular, lamella. In younger exam- 

 ples this is a layer of strictly one cell in depth, but in a very large one of about 100 

 millimetres in diameter the nuclei lay without any order, forming irregular tiers across 

 its thickness. The cell boundaries could not be detected usually, except in very young 

 specimens. The endodermal lamella runs close to the subumbrellar ectoderm, but 

 is separated from it by a thin layer of jelly, and meets the radial, circular, and cen- 

 tripetal canals along the line that separates the different kinds of endodermal cells 

 described above. 



There are two sets of tentacles different in structure and position, and presumably 

 also in their main function. From their position I shall call them velar and exum- 

 brellar; both are very numerous. 



The exumbrellar tentacles may arise from the exumbrella at any level, from 

 very near the apex to a short distance from the velum, but they are provided with 

 endodermal roots, traversing the jelly and connecting them with the circular canal. 



