37 



discarded his first and correct description for one which does not agree with the facts. The 

 point is really one of importance; as without a proper understanding of the order of the arms 

 it is impossible to grasp the manner in which the food must pass to the mouth ; and moreover, 

 the possibility of comparing the collar-region of Cephalodisciis with that of other animals clearly 

 depends on an accurate knowledge of its structure. As I cannot accept Masterman's account, 

 it follows that I do not agree with his theoretical conclusions on the subject. 



As no two individuals of Cephalodisciis have their arms in identically the .same position, 

 it is clear that the arms can be moved at the will of the animal into an indefinite number of 

 positions. I do not therefore assert that Masterman's second diagram represents an impossible 

 arrangement, though I do not think it would be easy to find it. But the whole significance of 

 a diagram of this kind depends on shewing the relations of the arms to one another at their 

 bases, where the arrangement is invariable, except so far as it may be slightly modified by 

 muscular contractions. Any such diagram of the bases of the arms must accordingly shew the 

 first arm in the position of Masterman's sixth (98, 2); and, vice versa, the sixth in the 

 position of his first. It is perfectly certain that the arm which I call the first originates from 

 the front end of the dorsal part of the collar, in the immediate neighbourhood of the proboscis- 

 pores, and with its food-groove facing the flattened part of the proboscis. This is shewn, for 

 C. leviitseni^ in figs. iJ2, 113 and for C. dodecalophiis, in figs. 148 — 143. It may thus be 

 stated that in both these species the arms are arranged, as seen in frontal sections, as a 

 series of F-like structures, their dorsal angles radiating to a common point, and their food- 

 grooves being disposed round the periphery of a circle. The difficulty of getting sections which 

 shew this arrangement quite diagrammatically depends on the facts (1) that the arms may be 

 directed in any direction — forwards, backwards, outwards or inwards — as soon as they have 

 become free, and this movement may be associated with rotation round their longitudinal axes; 

 (II) that the arm-bases are not all in a single plane. The first difficulty is most easily overcome 

 by examining sections of C. levinseni, in which the arms are very commonly arranged parallel 

 with one another in a bundle which is directed straight forwards. The second difficulty cannot 

 be so easily surmounted. The arm-bases of each side in reality follow a spiral line, commencing 

 close to the proboscis-pore, and passing outwards and then dorsally, the end of the spiral 

 reaching the posterior end of the central nervous system. In C. gracilis (fig. 25) the arm-bases 

 follow a comparatively simple line, although even in this species no two individuals have 

 their arms in the same position. The arrangement in C. levinseni can be understood by 

 imagining the number of arms in C. gracilis increased by one, and by supposing the last 

 arm of the series to be carried a considerable distance inwards, or towards the middle line of 

 the animal. 



I have attempted to shew, in a diagrammatic figure (PI. XII, fig. 158), a po.sterior view 

 of the arm-bases of the left side of C. levinseni, based on the frontal sections which have 

 already been described. A similar dorsal view of the arm-bases and of the operculum is given 

 in fig. 160; while fig. 159 is what I believe to be a fair diagrammatic representation of the 

 free edge of the collar. In this last figure, in which the proboscis-stalk is supposed to have 

 been cut through, and the anterior end removed, it has been necessary to exaggerate the size 



