(=Dapedoglossus Cope) is known from the Eocene Green River shales of Wyoming, U. S. A.; Brychaetus A. S. Woodw. from London clay and Sheppey and Scleropages from the Tertiaries of Sumatra.

In her monograph on the Tertiary freshwater fishes of Middle Sumatra, Sanders (1934) has published photographs of complete scales of Scleropages and of her new genus Musperia. The structure of the scales of these two genera differs in several respects, some of which are noted below.

In Scleropages there is a fairly large nuclear area in the centre (text-fig. 4a) which appears to be devoid of smaller pieces and only covered with moniliform striations; the mosaic pieces are relatively small and more or less ovoid or squarish in outline. In Musperia, on the other hand, there is no clear nucleus, and the rhomboidal, mosaic pieces are fairly large, especially towards one end of the scale.

A comparison of the text-figures 3 and 4 shows that the fossil scales obtained by Mr. Crookshank correspond with those of Musperia in all important respects noted above. In discussing the affinities of her new genus, Sanders has pointed out certain resemblances between Musperia and the North American Eocene genus Phareodus on the one hand and with the African genus Heterotis on the other. It would thus appear tentatively that the original stock which migrated from the north to the south may have consisted of forms like Musperia, which is now extinct and whose remains are known from Sumatra on the one hand and Peninsular India on the other. This genus was probably a descendant of Phareodus and in its further wanderings along two definite routes it gave rise to Scleropages in Siam, the Indo-Australian Archipelago and Australia in the south-east and to the African and South Ameri

can genera in the south-west. The same two routes of wanderings seem to have been followed by the Dipnoan fishes at an earlier period of the earth's history. It may here be noted that fossil remains of Ceratodus Ag. are known from the Kota-Maleri beds of India (C. hislopianus Oldham); these beds are in the Godavari valley and are placed in the Upper Gondwana period.

The Osteoglossidae are recorded here from India for the first time; and this is of special significance from a zoogeographical point of view. As indicated above, it points to a possible route of migration by which these large fishes may have spread from India to Africa and later to South America.

Series OSTARIOPHYSI.

Order EVENTOGNATHI.

Family CYPRINIDAE.

(Plate 17, fig. 6; text-fig. 5.)

Material. Specimen No. K32/149 and its counterpart No. K32/ 150 (from Deothan).

Though there is only one fragment of a Cyprinid scale, its essential structural features are very clear. It seems to have been oval in outline, about one and a half times as long as broad. The nuclear area is small and somewhat eccentric in position. The scale is provided with a large number of radii which are distributed in all directions. The circuli are very fine and compactly arranged.

TEXT FIG. 5.A fragment of a fossil Cyprinoid scale (K32/149). ×17.

Remarks. The structure of the scale strongly suggests its very primitive nature and leads to its inclusion in the subfamily Abramidinae. In the main, it corresponds with the scale of Leuciscus æningensis Ag. from the Upper Miocene figured by Zittel (1932, p. 157) after Winkler, but its detailed structure and the form are quite different. I have not been able to assign it definitely to any genus, as the scales of most of the living genera have not been so far described.

The Cyprinidae are at present found in great abundance in the fresh waters of the Old World and North America, but are totally absent from South America. In the fossil state they are found from the Lower Eocene to the Pleistocene. Most of the fossil genera are represented by living forms.

Superorder ACANTHOPTERYGII.

The Acanthopterygii, which include the large group of Perches, are abundantly represented in the existing fauna. In the fossil material collected by Mr. Crookshank there are at least seven types of scales which can be referred to this group of fishes. According to Cockerell (1913, p. 143) a typical Acanthopterygian scale,

is more or less quadrate, with the nucleus subapical, the basal circuli fine and transverse, the basal radii strong, spreading out like a fan, and the apical area covered with fine dentiform structures which can be counted in rows obliquely or transversely, and on the margin form a series of fine teeth."

From the above it will be noticed that the scales of Eoserranus hislopi described by A. S. Woodward (1908) from the Lameta beds at Dongargaon in the Central Provinces differ from a typical Acanthopterygian scale in the fact that they are cycloid and in them the basal portion, marked with well-developed radii, is much narrower than the apical portion. Woodward's description of the scale (text-fig. 6) is as follows:

The scales appear to have been cycloid and rather small. As shown in the type specimen, their exposed sector is ornamented with fine, radiating ridges. Their inner face as shown in impression is marked in the covered portion with radiating furrows.”

The Eoserranus-type of scale, therefore, seems to represent a primitive stage in the development of an Acanthopterygian scale. According to Cockerell (loc. cit.):

66

The toothed or ctenoid feature appears to be derived from the longitudinal apical circuli which become modified and segmented, the terminal segments

especially taking the form of teeth. It is this segmented arrangement which gives the apical area in Acanthopterygians its special character, resembling very much the arrangement of bracts in the heads of some composite flowers."

TEXT-FIG. 6.-A scale of Eoserranus hislopi A. S. Woodw. (From the type-specimen, No. 9365 G. S. I.). ×5.

In my preliminary report to Mr. Crookshank I referred a large scale (pl. 17, fig. 5, text-fig. 8) in his collection to Eoserranus, but a careful examination with proper illumination has shown that it is not a cycloid scale. Its general outline is also different from that of an Eoserranus scale. In fact, all the Acanthopterygian scales in Mr. Crookshank's collection conform to the general type noted above. There is such a great diversity of Acanthopterygian fishes that it is with great difficulty, and not without hesitation, that I have referred them to any definite genera. Their position in the families noted below, however, appears to be fairly certain.

Order LABYRINTHICI

Family POLYACANTHIDAE.

(Plate 17, fig. 4; text-fig. 7b.)

Material. Specimen No. K32/147 (from Deothan).

It is an almost complete scale (text-fig. 7b), with part of the apical portion covered by skin. Structurally the scale can be

TEXT-FIG. 7.-A Polyacanthid scale and an allied fossil scale (K32/147).

a. Polyacanthid scale. x8; b. Fossil scale allied to that of the Polyacanthidae. X81.

divided into three regions, (i) spiny apical portion; (ii) an area with fine radiating striae between the apical portion and the nucleus; and (iii) the basal area marked with fine circular striae which extend round the nucleus also. There are about a dozen fine, basal radii, of which eight extend almost to the nucleus while the remaining four are shorter. Some of the radii are interrupted in their course. The apical teeth are very small.

Remarks. I have not been able to assign any definite generic position to this scale. While investigating the structure of a large number of freshwater Acanthopterygian scales I found some points of similarity between the fossil scale and that of Polyacanthus (K. & v. H.) Cuvier and Valenciennes (text-fig. 7a). Cockerell (1913, p. 164) described the scale of the Labyrinthid genus Anabas Cuvier and found the same divisions of the scale into three areas. The apical area is beset with linear, spine-like structures, of which the lateral ones are often continuous at the base with the circuli. The apical radii are widely spaced and are about 15 in number.