Discussion

The species is closely related to the Hawaiian Spatangus paucituberculatus A. Ag. and H. L. Clark, 1907, sharing with it the absence of tubercles from the anterolateral and posterolateral interambs. It differs conspicuously from the latter in the subambital periproct (posteriorly placed in paucituberculatus, where it is therefore not visible from below); other differences include the relatively robust test (excessively fragile in paucituberculatus, of which all known specimens are fragmentary).

In Mortensen's (1951) classification, the species without tubercles in the anterolateral and posterolateral interambs are referred to a subgenus Granopatagus Pomel, to which accordingly the present species would be admitted. However, I do not think this classification is justified. The distinctive subambital position of the periproct in beryl is a character not shared with paucituberculatus, but it is shared with S. thor which, in Mortensen's classification, would not be admitted to Granopatagus. As it seems unlikely that the subambital periproct could have evolved quite independently in two sympatric (and potentially interbreeding) species, it would seem wiser to abandon the subgenus Granopatagus.

Further, examination of Mortensen's excellent photographs of a specimen from Albatross Station 5565, in the Philippines (reproduced by him as Plate 2, fig. 3 and Plate 3, figs. 21 and 22, Mortensen, 1951), shows that the species in question cannot be identified as Spatangus paucituberculatus (as Mortensen believed). This specimen has a much larger anterior notch than in the holotype of paucituberculatus (H. L. Clark, 1917, Plate 157, figs. 7 to 9), and more important, the Philippines specimen has a subambital periproct, as in the New Zealand species, whereas the Hawaiian type material has a terminal periproct. The obvious differences between the Philippines specimen and S. beryl, which can be seen by comparing the illustrations here given with those published by Mortensen, prohibit the union of both under one name. Accordingly it is necessary to introduce page 6 a new name for the specimen from Albatross Station 5565, now in the United States National Museum, and for this Spatangus diomedeae is hereby proposed. Its diagnostic characters are set out in the general key to all known extant species of Spatangus which follows, but are also indicated in the present paragraph.

It is evident, thus, that Spatangus beryl is more closely related to S. diomedeae than to any other known species, but differs in having a more robust test, and a shallower and smaller frontal notch, and a more extensively dorsal and ventral tuberculation. Both species share with S. thor the character of a distinctly subambital periproct.

Spatangus beryl is named for Mrs Beryl Nielsen, who first forwarded a specimen of the species for examination, and subsequently was instrumental in obtaining other valuable material. The specific name is treated as a nominative in apposition.

Synoptic Key to The Known Extant Species of Spatangus

1 (8)	No primary tubercles in the anterolateral interambs, and usually none (occasionally 1 or 2) in the posterolateral interambs.
2 (5)	Periproct subambital, fully visible from below.
3 (4)	Peristome transversely oval, not overhung by labrum. Anterior notch shallow, not reaching midway to the anterior margin of peristome. Test robust	beryl n. sp.
4 (3)	Peristome transversely crescentic, overhung by labrum. Anterior notch deep and broad, reaching midway to the anterior margin of peristome. Test fragile	diomedeae n. sp.
5 (2)	Periproct terminal, not overhung by ambitus.
6 (7)	One or two primary tubercles in each posterolateral interamb. Subanal plastron obliquely subventral	paucituberculatus
7 (6)	No primary tubercles in posterolateral interambs. Subanal plastron terminal	inermis
8 (1)	A dozen or more tubercles in each posterolateral interamb.
9 (14)	No tubercles in the anterolateral interambs, and in the posterolateral interambs the tubercles are restricted to the area within the petals.
10 (11)	Sternum with no keel	lutkeni
11 (10)	Sternum keeled.
12 (13)	Periproct terminal, not overhung by ambitus, and not visible from below. Greyish lavender	pallidus
13 (12)	Periproct subambital, entirely visible from below. Deep reddish purple (large forms)	thor n. sp.
14 (9)	Numerous tubercles in all interambs.
15 (18)	Tubercles in the ambs, beyond the petals.
16 (17)	Test low, posterior end vertically truncate	multispinus
17 (16)	Test high, posterior end obliquely truncate	raschi
18 (15)	No tubercles in the ambs beyond the petals.
19 (20)	Subanal fasciole transversely elongate, with a re-entrant posterior margin	purpureus
20 (19)	Subanal fasciole shield-shaped, margin entire.
21 (22)	Three pore-pairs in either side of subanal plastron	altus
22 (21)	Two pore-pairs in either side of subanal plastron	capensis and californicus

Whether capensis and californicus are distinct is a question which requires further study. Mortensen's key (1951, p. 10), and subsequent discussion (1951, p. 17) seems to have been based on inadequate material of californicus. A representative series of californicus received from F. C. Ziesenhenne shows that the periproct in that species is normally subambital, as in capensis (not terminal as in the specimen illustrated by Mortensen (1951, Plate II) ).

So far as the other species are concerned, the forms with a subambital periproct comprise a linear sequence which can be arranged in order of steadily increasing tuberculation, as follows:

page 7

1.	diomedeae (all lateral interambs naked, and very few and inconspicuous tubercles in the posterior interamb and anterior amb-margins).
2.	beryl (similar, but tubercles more conspicuous)
3.	thor (anterolateral interamb naked, but tubercles present in posterolateral and posterior interambs)

Of this group, diomedeae (Philippines) is northern, and the other two forms are restricted to the Southern Hemisphere.

In the same way, the forms with a posterior periproct can be arranged in a corresponding sequence, as follows:

1.	inermis
2.	paucituberculatus
3.	lutkeni and pallidus
4.	altus and purpureus
5.	raschi and multispinus (tubercles present in all interambs, and also in the ambs beyond the petals)

Save for multispinus (New Zealand), every one of this second sequence is a Northern Hemisphere form.

Whether this geographic bias has any real significance cannot be established until richer collections are available from a greater number of stations.

There is, however, one additional piece of evidence which may be relevant in this context, for it is indicative that there is no hard and fast boundary between species (or forms) of the beryl-inermis type on the one hand, and of the thor-lutkeni-pallidus type on the other hand. As will be seen from the foregoing synopsis, typical paucituberculatus carries one or two primary tubercles in the posterolateral interambs, as if it were an intermediate form between the strictly naked group (beryl-inermis), and the group with a dozen or more tubercles in the posterolateral interambs (thor-lutkeni-pallidus). Of the New Zealand material examined, six specimens could be readily separated into groups, one of the groups (two specimens) comprising the material here assigned to thor, and the other group (four specimens) comprising beryl. There remains a seventh specimen, in the collection of the Dominion Museum, Wellington, obtained by Mrs Olga Sansom, from the oyster beds of Foveaux Strait in July, 1958. The specimen, when I examined it, was badly fragmented, but it has been possible to reconstruct the fragments so as to yield most of the characters of the specimen. It is illustrated, as restored, in Plate 5, fig. 14. In most respects it corresponds to beryl. The length is 126mm, breadth 122mm, height approximately 75mm. The plastron is 58mm long by 33mm wide. In the posterolateral interamb, however, there are three inconspicuous primary tubercles. It is thus intermediate between thor and beryl. The most reasonable interpretation of the data appears to be that this specimen must represent a hybrid beryl x pallidus; for if we were dealing with a single, variable species, we would expect to find a majority of intermediate forms, and a minority of extremes, whereas in fact the reverse appears to be the case, most specimens being either definite beryl or definite thor. The situation is thus comparable with that presented in the North Sea, where Spatangus purpureus and S. raschi are sympatric, and yield occasional hybrids. Whether this interpretation is correct or not is a question which must await the discovery of a richer material. The third species, S. multispinus, does not seem to range into Foveaux Strait, and no evidence of hybridisation (nor of marked variation) has ever been observed; it would seem to be genetically isolated, since no close congener is known in the Southern Hemisphere.