IOI 
As the length of the polyp-stalk may vary notably within one colony, this cannot be 
regarded as an important feature, though in certain cases it may be a useful diagnostic feature 
that the polyps are, it may be, sessile or with very long stalks. 
In studying the possible relationships of different species, the following provisional 
assumptions may be made: 
(a) That the presence of very numerous small spicules in the anthocodial points is more 
primitive than a smaller number of large spicules. 
{d) That a supporting bundle with a large number of spicules tending to form a sheath 
is a more primitive condition than the differentiation of a bundle out of a few large spicules to 
form a buttress. And here it may be noticed that in a few species — e.g., D. clavata , the 
supporting bundle is so slight that the polyps may at first sight be mistaken for those of a 
Eunephthya, where there is no supporting bundle whatever. 
(c) That the more primitive forms show: 
(1) less variety of spicule, e.g. that spindles predominate throughout, and less differentiation of 
spicules in the anthocodial point; 
(2) less localization of particular forms of spicule in particular areas. Thus there can be no doubt 
that the presence of special interlocking spicules in the sterile stalk is a later differentiation ; 
(3) the presence of few spicules in the canal-wall. 
Thus D. piltteri with a multitude of canal-wall spicules is in other respects highly speci¬ 
alized. Conversely, it is interesting that D. clavata with no canal-wall spicules belongs to what 
may be regarded as the more primitive type. So careful a worker as Kukenthal attached great 
importance to the canal-wall system of spicules, and it is obviously important to inquire what 
correlation there is between the primitiveness of a species and the number and nature of the 
canal-wall spicules. The unsatisfactory feature here is that the examination of the canal-wall 
spiculation has not been made uniformly by the various investigators. Minute spicules which 
escape attention in a fragment of canal-wall when boiled down may be found by examining a 
piece intact. But one is not uniformly successful with the sample examined. 
As regards number and differentiation of anthocodial spicules it seems to us possible 
to refer all the species to one or other of the following six grades, ranging from the most 
primitive (I.) to the most specialized (VI.): (Plate XIX, figs. 1—6). 
VI. Only one pair of spicules, or even but a single spicule replacing all the others in each point 
above the pseudo-crown; (Plate XIX, Fig. 6); 
V. reduction of the point so that one pair (or, it may be, one of the uppermost pair) predo¬ 
minates over the others (not above three pairs in total number) which may form a pseudo¬ 
crown ; (Plate XIX, Fig. 5); 
IV. one pair or one spicule predominating over 3—4 others (Plate XIX, Fig. 4); 
III. about 4—6 pairs, the uppermost slightly specialized, (Plate XIX, Pig. 3); 
II. about 6—8 pairs, strong and uniform, (Plate XIX, Fig. 2); 
I. about 8—12 pairs in each point, numerous and small, (Plate XIX, Fig. 1). 
The suggestion has been made that the differences between the Glomerate, Divaricate, 
and Umbellate groups are not of taxonomic value, but depend on conditions of growth. On this 
view, the occurrence of a species with the same anthocodial armature in each of the three groups 
