The scale of Polyacanthus appears to be much more specialised than that of Anabas; the present-day geographical distribution of the two genera also shows that the latter is certainly a more ancient form. Anabas (including the genera Spirobranchus Cuvier and Ctenopoma Peters) is found in south-eastern Asia and Africa, while Polyacanthus is only found in Ceylon, Singapore, Java, Sumatra and Borneo.

So far as I am aware no fossil Labyrinthid remains have hitherto

been described.

Order PERCOMORPHI.

Series KURTIFORMES.

Family SERRANIDAE.

(Plate 17, fig. 5; text-fig. 8.)

Material. Specimen No. K29/629 (from Deothan).

The scale is represented by an impression of a part of its area. The entire scale must have been about 12 mm. in length and 11 mm.

in width. The nucleus is subapical and, though small in area, is clearly defined. The apical ctenoid area is also well marked, showing several rows of small, clearly defined, sharp spines. The circuli are very fine and compactly arranged. There are about 18 delicate radii, all of which start from near the nucleus and extend towards the periphery for a considerable distance; some even reach the

margin.

Remarks. In Cockerell's (1913, p. 143) identification table of the Acanthopterygian scales, the following characters are noted for a Serranid scale:

(i) Basal radii spreading.

(ii) Basal margin not deeply lobed.

(iii) Basal area not beset with linear, spine-like structures.

(iv) Apical margin provided with many sharp and small teeth. Judged by these features, the fossil scale impression is referrable to the family Serranidae. Though the earliest known SerranidProlates Priem-dates from the Upper Cretaceous, numerous fossil representatives occur in the Tertiary formations of Europe and North America. I have indicated above (vide supra, p. 275) that Eoserranus is probably a very primitive representative of the family. The specimen under report possesses a highly specialised structure, so is probably younger than Eoserranus in the scale of geological history.

Incertae Sedis.

There are two other types of scales in the collection which, though quite distinct from those of the Serranidae, may be described in this place.

TYPE A.

(Plate 18, fig. 5; text-fig. 9.)

Material. Specimen No. K40/286 (from Deothan).

This is a small, almost complete scale. Excluding the apical spines, it is almost as long as broad. The apical teeth seem to form a single row of about 15 teeth. A well formed tooth is about three times as long as broad at the base. The nucleus is well defined and subapical. The scale is provided with a large number of fine

circular striae, a few of which are concentrically arranged round the nucleus. There are only six, well-defined, basal radii which

are widely spaced and seem to run almost parallel to one another. The lateral margins of the scale are incomplete but in general outline it appears to have been circular.

Remarks.--In the present state of our knowledge of lepidology, I am unable to refer this scale to any Acanthopterygian family From the arrangement of the basal radii and the nature of the apical teeth, this scale appears to be of a somewhat primitive type. It shows a certain amount of resemblance to the scales of the family Mullidae.

TYPE B.

(Plate 18, fig. 6; text-fig. 10.)

Material. Specimen No. K40/283 (from Kheri).

This is only a fragment of the basal portion of a large scale about 9.6 mm. in width. The basal radii are delicate and few in number. The circular striae are very fine and closely set.

TEXT-FIG. 10.-Basal part of a large, fossil, Acanthopterygian scale (K40/283). ×101.

Remarks. It is very difficult to express an opinion about the relationships of this fragment. It is described and figured here merely because it appears to represent a large scale of a very distinct type, differing from all the other scales in the collection.

Family NANDIDAE.

Material. Specimens Nos. K29/628, K40/287 (from Deothan): K40/285 (from Kheri).

These are small, subquadrate scales of the Percoid type with a subapical nucleus and strong basal radii. The number of radii varies from scale to scale. The nuclear area is fairly large and the radii, all of which are complete, arise at short intervals from the basal margin of the nucleus. The circuli are fine and closely set. There are several rows of short, blunt spines in the apical region.

1 In the case of the genera included in the family Nandidae, Jordan (1923, p. 202) remarked that the group may require further division. By certain authorities it is now divided into the family Nandidae for Nandus C. V. and the family Pristolepidae for Pristolepis Jerdon, Badis Bleeker, etc. I have followed the latter arrangement.

Remarks. Superficially these scales appear to resemble those of Anabas Cuvier (vide Cockerell 1913, p. 164), but the main difference lies in the disposition of the circuli. In Anabas "the basal

TEXT-FIG. 11.-A fossil and a present-day Nandid scale.

a. Fossil scale (K29/628). x24; b. Scale of Nandus nandus (Ham.). ×20. circuli are dense, the lateral ones rather widely spaced and those at the sides of the apical field strong and very far apart". In the fossil scales they are uniformly distributed. Moreover, the ctenoid patch of Anabas is of a totally different nature. In comparing the fossil scales (text-fig. 11 a) with those of the living forms I found a great general similarity between them and the scales of Nandus Cuvier and Valenciennes (text-fig. 11 b). The only difference lies in the fact that in Nandus the basal radii are not so strong and some of them may be incomplete.

Though the Nandidae, as restricted at the present day, are found in the Great Sunda Islands, Malay Peninsula, Burma and India, their close allies of the family Polycentridae are found in West Africa and South America.

So far as I am aware, no fossil remains of the families Nandidac and Polycentridae have been described so far.

Family PRISTOLEPIDAE.

(Plate 17, fig. 1; text-fig. 12.)

Material. Specimens Nos. K29/629, K32/157, K40/288, K40/289, K40/290, K40/292, K40/295, K40/296 (from Deothan); K31/546, K40/284 (from Kheri).