296 



SUBANTARCTIC ISLANDS OF NEW ZEALAND. 



[Echinoderttu. 



of A. regtUaris, so common on our coasts, so that I collected only three specimens ; 

 but on examining these in my laboratory I perceived at once that they present several 

 important differences from this species, and I believe them to belong to Perrier's 

 species, for reasons that will be mentioned below. 



Dimensions. — The dimensions are given below, in millimetres : — 



In shape they are pentagonal, with well-rounded angles and slightly excavated 

 sides, thus differing from A. regidaris, in which the tip of the arm is much more 

 acute and the excavation relatively greater. They also differ in the following details : 

 (a) The arrangement of the plates and spines on the abactinal surface, (b) and on 

 the actinal surface, (c) and in the adambulacral armature. 



The abactinal surface is covered with nearly regularly distributed rounded or 

 oval plates, each bearing a compact group of many small spines, some 30 to 40 in a 

 group; the plates are separated by isolated large papulae (fig 2). In the smaller 

 individuals, at any rate, there is, nearly in the centre of the disc, a circle of 5 

 curved somewhat crescentic plates, radial in position, alternating with 5 smaller 

 rounded interradial plates (fig. 3). On the median portion of the ray the plates are 

 arranged in longitudinal series, diminishing in size towards the tip ; but in the 

 largest specimen the 4 median rows of plates are slightly transversely extended 

 and feebly curved. This arrangement is in marked contrast to the distinct 

 crescentic plates, each bearing only a few marginal spines, of the arm of A. 

 regvlaris. 



Each of the actinal plates bears in its centre a group of 8 or 10 or more long 

 slender spines, leaving the rest of its surface bare (figs. 4, 5). These plates are 

 regularly arranged in regard to the ambulacral groove ; they form curved rows, 

 starting at right angles to the groove and curving gently outwards as the inter- 

 brachial area is reached, and getting smaller as the margin is approached. The 

 plates, and, of course, the groups of spines, are also in rows parallel to the groove 



(fig-1)- 



The margin of the disc is fringed with similar groups of spines. 



The adambulacral armature is formed of a row of 4 spines in a longitudinal 

 row on each plate, set slightly obliquely to the groove (fig. 5). These spines are 

 stouter and longer than those of the actinal plates, and differ entirely from the 

 adambulacral spines of A. regidaris. 



So far as the limited literature allows me to judge (for I have been unable 

 to obtain from the libraries of New Zealand or Sydney several of the works 

 enumerated in the bibliography at the end of this article), this little starfish 



