marginal furrow at about 75°. Genal angles not produced, subrectangular, armed with short sharp subcylindrical spine directed straight backwards and situated rather inside outer angle, which is subtruncate. Eyes elevated, rather large, slightly oblique, extending back from level of first to third lateral furrows, with anterior end nearly touching glabella. Free-cheeks with broad shallow marginal furrow separating off rather wide gently convex border. Surface of head-shield covered with numerous closely-placed, small, low, rounded, irregularly-disposed pustules.

Thorax with broad strongly-convex axis bearing short vertical spine; pleurae gently arched down, strongly furrowed. Whole surface of thorax coarsely pustulated like head-shield. Pygidium triangular? mucronate, and with broad cylindrical prominent convex axis, annulated for nearly whole length with 9-10 rings, having blunt tip not reaching posterior margin; pleural lobes narrow, rather strongly arched down, rapidly decreasing in width posteriorly, composed of 5-6 low, rounded pleurae, with scarcely any trace of pleural furrows.

Remarks.-There are two slabs of a rather micaceous greenish flaggy slate in the Sedgwick Museum from an unknown locality, presented in 1906 by the South African Museum, containing portions of 7 head-shields and 2 pygidia of the species. In the South African Museum there is another similar slab and its reverse (7200 S.A. Mus.) containing several head-shields, a pygidium and portions of the thorax, and in the Durban Museum there is a slab of identical lithological character with a head-shield showing the genal angle and a pygidium. It is on these specimens that the above definition is based. But there is also a head-shield (3934 S.A. Mus.) from Hottentot's Kraal in the South African Museum, and another (433 Pret. Mus.) from Boschluis Kloof in the Pretoria Museum, and another (139 Cape

Univ.) in the Cape University collection, which seem to have precisely the same characters.

The general characters of the glabella and head-shield recall D. falklandicus Clarke,* and the pygidium is almost identical in character; but the presence of the meso-occipital spine is distinctive, and the pleuro-occipital ring seems to be wider towards the genal angles. Clarke, moreover, does not mention or depict any axial spines on the thoracic segments.

The head-shield is much like that of D. africanus, but the glabella is more pentagonal, the two pairs of anterior furrows are stronger and deeper, the eyes are larger, and the ornamentation is much coarser. The genal spines are developed in rather an unusual way, and are well seen in the specimen (181 Cape Univ.) in the collection of the Cape University. Their shape and position are just as in Schwarz's fig. 4, pl. x (op. cit.), of the posterior edge of a head-shield attributed to Ph. acacia, which, he states, is a composite figure drawn from several casts of fragments. The thoracic segment (11285) in the British Museum, figured by Salter (op. cit., pl. xxv, fig. 3) as belonging to D. africanus, may belong to D. Baini, and the pygidium from Gamka Poort (222 S.A. Mus.) figured by Lake (op. cit., pl. xxiv, fig. 8), also as D. africanus, may be likewise referable to it. The thoracic segments of D. Baini seem to have shorter, smaller, axial spines than D. africanus, but only isolated and imperfect portions of the thorax are associated on the slabs with the head-shields and pygidia.

Dalmanites (Corycephalus ?) capensis sp. nov.
(Pl. VIII, figs. 5, 6.)

Head-shield semi-elliptical to semi-circular, gently convex, with the anterior border in front of the frontal lobe of the glabella furnished with a few short blunt denticulations. Glabella large, subpentagonal to subclavate, widening anteriorly, with straight sides and rounded anterior end; frontal lobe large, swollen, transverse, about half the length of the glabella, projecting slightly at sides, with median pit at base; second and basal lobes depressed, small, subequal, transverse; anterior lateral furrows strong, oblique, long; second and third lateral furrows represented by isolated transverse deep horizontal pits, scarcely connected with axial furrows, the third pair better marked. Axial furrows straight, diverging anteriorly. Occipital ring simple,

*Clarke, op. cit., 1913, p. 105, pl. v, figs. 1-4.

well defined by deep furrow; meso-occipital furrow arched forward gently in middle; pleuro-occipital furrow wide, deep, meeting marginal furrow at about 60°. Pleuro-occipital ring rounded, widening slightly to genal angles. Lateral border raised, rounded, defined by definite furrow, wider and shallower than pleuro-occipital. Genal angles sharply rounded, not produced back. Facial sutures with posterior branch curving forward strongly in its outward course and then arching back to cut lateral edge some distance in front of genal angles. Eyes large, sharply bent, reaching from first to third lateral furrows, nearly touching axial furrows at front end, composed of 24-26 vertical rows of lenses. Eye lobe depressed, with swollen edge. Surface of glabella and border coarsely granulated. Cheeks covered with large, low, closely-placed, rounded tubercles, especially towards genal angles. Doublure in front ornamented with triangular group of similar tubercles; rest coarsely granulated.

Remarks. There is one fairly-well preserved head-shield (H 76 Stell. Mus.) in the Stellenbosch Museum from Klein Straat Station, with three thoracic segments attached, but the whole specimen is rather crushed. The axis of the thorax is broad, and there are lateral swellings on each ring; the pleurae have a rather strong but rounded fulcrum at about half their length, and a deep, gently-sigmoidal, diagonal furrow. The marginal projections on the front of the headshield are indistinctly seen.

The other specimen is a better head-shield (216 Cape Univ.) in the collection of the Cape University from Touws River Road. It shows clearly the marginal denticulations of the border in front of the frontal lobe, and is not much crushed; the peculiar ornamentation of the cheeks and median part of the doublure is well preserved, the rest of the surface of the head-shield being only coarsely and uniformly granulated.

This species is allied to D. Drevermanni Thomas* (non Cryphaeus * Thomas, Zeitschr. deut. geol. Gesell., vol. lvii, 1905, p. 247, t. xi, figs. 1–3.

Drevermanni Richter, 1909) of the Devonian of the Argentine, and to D. Boehmi Knod of Bolivia, especially its variety boliviensis described by Kozlowski.† It marks an early stage of the development of the denticulated or crenulated border of Corycephalus, and the same stage is seen in the North American Lower Devonian species, D. stemmatus Clarke, D. Dolbeli Clarke,§ and others, as Clarke || has pointed out. If a distinctive subgeneric name is required for this group of species we may suggest Eocorycephalus.

The head-shields of D. capensis and D. caffer are difficult to distinguish when the anterior margin and ornamentation are not preserved, as the glabella and general characters seem to be identical.

Dalmanites (Hausmannia) Dunni sp. nov.

(Pl. XI, fig. 5.)

1904. Dalmanites sp. Lake, Ann. S. Afr. Mus., vol. iv, pt. 4, p. 212, pl. xxv, fig. 7 (38 S.A. Mus.).

The fragmentary pygidium from Gamka Poort (38 S.A. Mus.), which Lake briefly described and figured without a specific name, deserves further mention. The whole pygidium is flattened and apparently broadly semilanceolate in outline, and probably pointed behind; the pleural lobes are flat and horizontally extended, and the portion of the left one which is preserved shows 10-11 simple pleurae which are flat and do not possess any pleural furrow; the interpleural grooves are sharp and deep. The pleurae diverge backwards and outwards from the axis and curve gently back, becoming more and more directed backwards; the last one is parallel to the axis. There are 10-11 pleurae preserved, of which the last 7 correspond to the axial rings, and probably this is the case also with the anterior ones, but the front part of the axis is missing. The axis is very elongated and conical, tapering very slowly, and shows 16-18 incomplete rings, the interannular grooves being very faint or obsolete across the middle. The whole surface is ornamented with rather coarse regular granulation. The actual margin of the pygidium is not preserved, but the

* Knod, Neues Jahrb. f. Miner. Geol., Beil. Bd. xxv, 1908, p. 508, t. xxi, fig. 1. † Kozlowski, op. cit., 1923, p. 36, pl. ii, fig. 1.

Clarke, Mem. 3, New York State Mus., 1900, p. 15, pl. i, figs. 6-16; pl. ii, figs. 1, 2.

§ Clarke, Mem. 9, New York State Mus., pt. 1, 1908, p. 121, pl. viii, figs. 1–7. Clarke, ibid., p. 134.

length of the portion of the axis preserved is 37 mm., and of the pleural lobe 46 mm., and it may be estimated that when complete the pygidium measured at least 55 mm. Whether it had a terminal mucro or not is uncertain, but it was probably pointed behind. We may compare it with D. Andii Kozlowski* from the Devonian of Bolivia, but more especially with D. Clarkei Ulrich † from the same country, also founded on a pygidium.

Groth has compared Clarke's D. maecurua from Brazil § and Bolivia with this unnamed species of the Bokkeveld Beds figured by Lake, but Kozlowski (op. cit.) doubts if Knod's Bolivian specimen with its incomplete axial rings is referable to Clarke's D. maecurua, and Steinmann's examples from the Cordillera Real are attributed by Kozlowski to a new species, D. Andii, while Groth's Brazilian specimen above mentioned is considered to resemble D. Clarkei Ulrich ** rather than D. maecurua. Knod (op. cit., p. 500) had observed that this South African form as figured by Lake was very similar to the D. Clarkei Ulrich from Bolivia. In view of the special peculiarities of our specimen we may give it a distinctive specific name, Dunni.

In another larger incomplete specimen, also from Gamka Poort (3949 S.A. Mus.), which probably belongs to the same species, the first six rings at the front end of the axis (which consists of 22-24 segments) are complete, and the posterior ones (16-18 in number) are incomplete across the middle. The last two pleurae arise from the sides of the axis some distance (7 or 8 rings) from its tip; all of the pleurae end a little inside the margin, as in D. Clarkei, with which our species seems undoubtedly most closely allied.

The species undoubtedly belongs to the subgenus Hausmannia as used by Hall and Clarke, Etheridge and Mitchell,†† and the author,‡‡ but the obsolescence of the pleurae before reaching the margin of the pygidium is peculiar.

*Kozlowski, Ann. Paléont., vol. xii, 1923, p. 38, pl. ii, figs. 3, 4.

Ulrich, Neues Jahrb. f. Miner., Beil. Bd. viii, 1893, p. 19, t. i, fig. 13.
Groth, Bull. Soc. Geol. France, ser. 4, vol. xii, 1912, p. 608, pl. xix, fig. 4.

§ Clarke, "Trilob. Grez de Erere, Brazil," Archiv Mus. Nac. Rio de Janeiro, vol. ix, 1890, p. 23, pl. ii, figs. 1-3, 6, 7, 10, 15.

|| Knod, Neues Jahrb. Miner., Beil. Bd. xxv, 1908, p. 500, t. xxi, fig. 3; Lake, Quart. Journ. Geol. Soc., vol. lxii, 1906, p. 429, pl. xl, fig. 11 ?.

¶ Steinmann, Neues Jahrb. Miner., Beil. Bd. xxxiv, 1912, p. 205, t. ix, figs. 1-3.

** Ulrich, Neues Jahrb. Miner., Beil. Bd. viii, 1893, p. 19, t. i, fig. 13.

†† Mitchell, Proc. Linn. Soc. New South Wales, vol. xliv, pt. 2, 1919, pp. 440–446. ‡‡ Reed, Geol. Mag., Dec. v, vol. ii, 1905, pp. 172-178 and 224-228.