488 HIROSHI OHSHIMA 
by him, then the dorsal arch (d) should lie above the antero- 
lateral rods (c). The two arms of the base of the postero- 
dorsal rod (f, y) are shown very well developed, and that the - 
serrated recurrent rods meet each other at the broadened 
posterior ends is also clearly drawn. His fig. 6, which is the 
ventral view, is somewhat difficult to understand. ‘There 
are two sets of rods which seem to correspond to the ventral 
horizontal rods, both overlapping each other at the end. 
Whether it is really an abnormal case, as in the right dorsal 
horizontal rod in my oldest larva (figs. 7 and 8, dh), or due to 
his misrepresentation cannot be decided at present. 
The further advanced stages are represented by a small 
number of eight- to ten-armed larvae with dissolved skeleton, 
and a single specimen of the twelve-armed stage. 
The fourth pair of arms to appear are the pre-oral, which are 
supported respectively by the direct prolongations of each 
end of the dorsal arch. The fifth pair are the postero-lateral, 
supported by the lateral prolongations from the base of the 
aboral spike. l'rom want of material showing any adequate 
stage I cannot decide whether the postero-lateral arms have 
from the beginning a skeletal support, as e.g. in Echino- 
cardium cordatum (MacBride, 4, p. 479), or not, 
as e.g. in Brissopsis lyrifera (Mortensen, %, 
pp. 147-8). Judging, however, from the faet that the arms 
soon develop to assume their typical shape, instead of remain. 
ing as ear-shaped lobes, | am strongly inclined to think that 
the arms in question of Spatangus purpureus do 
contain their skeletal support from their earliest stage. 
Owing to the remarkable increase in size of the stomach 
during the eight to ten-armed stage, that skeletal framework 
which formerly encircled the stomach must have undergone 
corresponding changes. This can be judged from the state 
seen in the twelve-armed specimen (figs. 7 and 8). Both the 
body- and recurrent rods are shortened at the posterior 
ends, their side-by-side connexion being broken. ‘The ventral 
horizontal rods of both sides are also separated from each other 
at the joint. This broken framework does not now encircle the 
