102 
would not mean that a similar specialization had occurred three times on different lines of evolution; 
it would merely mean that one and the same species occurred in three different vegetative forms: 
Glomerate, Divaricate, and Umbellate, these depending entirely on growth conditions. 
But against the interpretation of the three groups as merely vegetative forms dependent on 
environmental conditions, two considerations may be urged, (i) In many cases the Glomerate, 
Divaricate, or Umbellate structure is clearly expressed in young forms i — 3 cm. high. (2) When 
exactly the same anthocodical armature occurs in two or all of the three groups, the two or 
three species in question differ in other details, such as the nature of the supporting bundle, 
and the spiculation of the lower cortex. This surely implies that the species in question are 
radically different, and that the Glomerate, Divaricate, or Umbellate structure is more than 
vegetative and environmental. 
Starting with the Glomerates, the following may be regarded as illustrations of primitive types: 
1. In the savignyi-group : 
D. argentea with irregular arrangement — most primitive of all, with ventrally and laterally 
nothing more than minute cylindrical corpuscles; D. fusca with indefinite double rows of about 
6 pairs — on a slightly higher level, but also showing primitiveness in the numerous quite 
small elements; D. savignyi with indefinite steep double rows reduced to 6, — somewhat 
higher on the scale, but also showing, like the preceding, very numerous small cylindrical bodies. 
2. In the hemprichi- group : 
D. clavata , where there are 10 — 12 pairs in each double row and a very weak supporting 
bundle. It may be noted that this group shows a gradual reduction in the number of spicules 
composing each point, 5 — 7, 5 — 6, 4, 3, 2 and 1, reaching a climax in forms like D. punicea, 
where there are a couple of slightly hockey-club like spicules occupying the whole area which 
in D. clavata , for instance, shows 10—12 pairs. 
3. In the hideri- group : 
Here there is, on the whole, a more differentiated group, with forms like D. mayi , with 6 — 8 pairs 
in each point, to be ranked as relatively primitive compared with the highly specialized D. kollikeri. 
So with the Divaricates, on this view the series begins with forms like D. japonica with 
8 pairs in each point, — a very primitive arrangement — and ends with D. suensctni , which 
has but one pair. As intermediate between grade II., which is defined as having numerous pairs 
regularly arranged in each point (grade I possessing very numerous irregular pairs), and grade III., 
with 5 — 6 pairs, cases like D. armata occur, where the lateral points may have 9 pairs. There 
must also be noticed D. caerulea , inclining to be primitive in minuteness but not in number of 
spicules, while D. klunzingeri may be considered as rising out of grade II. because one of the 
uppermost pairs is markedly differentiated. 
It is interesting to find that in so many of Kukenthal’s minor groups the arrangement 
of the species ends in forms with one large pair in each anthocodial point. 
Again, in the rigida-group, D. microspiculata with 6 — 8 pairs, is undoubtedly primitive, 
though the uppermost again is slightly larger and projecting. 
In the third great division — the Umbellates — grade I. is represented by D. australis 
with its very numerous, small polyp spicules. 
