subject of genetics, or even to strengthen and further illustrate conclusions already accepted. Such studies naturally follow in the train of the systematist. The systematist is the pioneer; he discovers and describes his species, he makes known what material is available, and then the evolutionist attempts to show from a study of relationships and environmental conditions how the different forms have come to be, that is, their method of transformation.

The object of the present paper is to discuss in a preliminary manner the various characters which are relied upon by systematists for diagnostic purposes within the geometrica-group, to endeavour to trace how the characters have been evolved, and how by their combinations they have given rise to types which are, or may ultimately become, distinct species; we have to consider, first, the variations of the individual specific characters, and, second, the origin of the specific types. With the acquisition of still further material, it is hoped that a more complete account will be possible, illustrated by coloured drawings.

The claim for specific recognition of the ten forms hitherto described will be discussed later. For present needs the following may be allowed to be sufficiently well differentiated to warrant their acceptance as distinct species or, perhaps better, as sub-groups, namely geometrica, oculifera, tentoria, and trimeni. As regards the other species described, it is, to my mind, very doubtful whether they can be considered as representing more than one complex assemblage, the members of which are as yet very fluctuating in nearly all their characteristics. The evidence for this conclusion will be presented later, and in the meantime I shall consider them as grouped around verreauxii.

The following are the principal taxonomic characters which are relied upon for distinguishing the various specific forms: (a) Coloration of carapace and plastron; (b) axillary and inguinal shields; (c) margin of carapace; (d) margino-costal angle; (e) nuchal shield; (f) antebrachial shields; (g) femoral tubercle; (h) form of shields of carapace and plastron; (j) beak; (k) frontal and prefrontal shields.

I. 1. INDIVIDUAL SPECIFIC CHARACTERS.

a. COLORATION OF CARAPACE.

As already indicated, the coloration of the shell is the most distinguishing and conspicuous character of the geometrica-group, and, from its general similarity throughout, is the one feature which at first sight suggests the close relationship of the different members. of the group. It consists essentially of a series of yellow rays on a black or dark brown background. The rays on each shield radiate all the way from the central areola to the margin of the plate, though sometimes they extend only a part of the distance.

The yellow ravs vary much in number, width, length, arrangement, and tint in different forms. Sometimes they are so broad that the lighter colour may be regarded as constituting the background on

which narrow black bands are displayed. The following tints are found to occur: Pale yellow or straw colour, lemon, light and dark orange, reddish yellow, yellowish brown, and liver red. Usually any one tint is characteristic of an individual, but occasionally the distal part of the ray is darker than the part near the areola. The colours are rarely very clear and bright on the natural shell, but when the latter is wetted or rubbed with oil they are displayed in a brilliant manner.

Though the yellow rays present very varied appearances in the different sub-groups, a general underlying plan can be traced throughout, and followed in its greater and greater elaboration through the series. The simplest condition is that found in tentoria, and the most complex in oculifera. In the former a number of simple yellow bands radiate from the areola, which is also yellow; some of the rays extend all the way from the areola to the margin, while others are shorter, due to incompletion either proximally or distally. The width of the ray also varies much, and may be the same throughout or increased distally; frequently a broad ray is bifid distally. In general, the rays of one shield show no correspondence in position with those of the adjacent shields, so that no complex geometrical pattern for the whole shell is possible.

The neural and costal shields of tentoria are roughly quadrangular in their basal outline, and in some individuals there is a tendency for the rays passing to the four angles of the shields to be more conspicuous than those intermediate, and also for those of adjacent shields to correspond, thereby giving a more definite geometrical plan. All the types of the geometrica-group show this to be the fundamental disposition of the rays on both the neural and costal plates. The principal rays extend from the areola to the angles of the shields, and these four are the most conspicuous rays; midway between two adjacent principal rays may be another ray, and alternating with these two series may be others of lesser rank. The four principal rays will be spoken of as primary; the four intermediate, one or more of which may be wanting, will be secondary rays; and the third series, not always present, will be tertiary rays. On all the shields the four primary rays are always conspicuous, and, in the more complex patterns, such as those of oculifera, they are bifurcated distally. The secondary rays alternate with the primary, and vary much in the extent of their development; rarely they are all wanting on individual shields. Frequently the two longitudinal secondary rays of one shield correspond with those of the two adjacent shields, one in front and one behind, and thus give rise to a continuous yellow band, which is median along the neural shields and lateral along the costals on each side. The tertiary rays are generally imperfectly developed, and are rarely numerous except in certain tentoria. By the continuity of the two primary rays of one shield with the two primary rays of an adjacent shield there is produced a "diamond pattern" which, combined with a median band, gives an ocellated character to the carapace. This is very well shown in some specimens of the verreauxii sub-group (see Boulenger's

figure of T. fiskii, P.Z.S., 1886, pl. LVIII.), and more especially in oculifera. The middle of the ocellus is constituted by the lateral secondaries, and, if these are bifurcated, still more distinctive

66

ocelli" are produced.

When the simple ray type of coloration found in the tentoria sub-group is compared with the complex ocellated pattern of some of the oculifera sub-group, the two would seem to be very divergent and unrelated, and it is only by studying intermediate stages that the pattern of oculifera admits of comparison with that of tentoria. The members of the verreauxii sub-group are particularly instructive for this purpose. They include all stages in the prominence of the primary rays and their correspondence with those of adjacent shields. so as to strengthen the geometrical plan. Ocelli are also very clearly indicated in some of the extreme members. The geometrica subgroup also shows the same tendency towards the regularity and correspondence of the rays, but to a much less degree than most verreauxii; indeed the colour pattern of some of its members cannot be distinguished from that of tentoria. In trimeni the pattern seems to have become more fixed and regular, and is transitional in character; rarely any of the tertiary rays are present, and usually not all the four secondaries.

Taking all the representatives of the geometrica-group into account, it becomes an easy matter to arrange a continuous series showing all the transitional stages from the simple ray pattern of tentoria to the very complex, ocellated display of oculifera. No one can question that all the colour variations are genetically related in a direct line from such a form as tentoria to that of oculifera, and that they represent a strictly continuous type of transformation. Corresponding rays can be traced all through the series, the modifications. consisting in the closer relationship of those of one shield with those of adjacent shields, and in minor changes such as bifurcations. In certain of the sub-groups one transitional stage appears as if becoming fixed, while in other sub-groups another stage predominates, but probably nowhere do we get two individuals exactly alike. specimens of any sub-group from the same locality there are great Even in differences in the number of complete and incomplete rays present, and in the regularity of the pattern. While conforming closely to the general plan of the sub-group, the minor variations are innumerable.

b. COLORATION OF PLASTRON.

The plastron also affords a very complete gradational series in the production of an elaborate colour pattern. The coloration of certain members of the tentoria sub-group may be regarded as the extreme type of simplicity, while that of oculifera represents the extreme of complexity. Between these two there is a complete varietal series, showing that all are genetically related.

In what may be considered as a typical tentoria, the whole of the central part of the plastron is uniformly dark brown or nearly

black, while the sides are a light yellow, the boundary between the two being sharply defined. In some specimens from the same subgroup the beginnings of a breaking up of the central patch into separate radii are represented, and triangular rays also begin to appear on the yellow sides. The latter variation is well marked in specimens of tentoria from the Hex River Valley. The same stage is also characteristic of trimeni, an excellent representation of which is given by Boulenger, 1886, pl. LVII. In geometrica the middle. patch is likewise partly broken up into definite rays, and rays are also developed along the sides. Most of the rays in these instances are densest at the margin of the plastral shields, and start from the anterior suture, though some originate from the middle suture. A close comparison shows that the rays in the different types correspond with one another, ray for ray.

In the verreauxii sub-group the plastron is at times only a faint, dirty brown, perhaps a little darker in the middle. This variation is represented by Boulenger in his figure of T. fiskii, 1886, pl. LVIII. Frequently, however, in verreauxii the plastron is strongly rayed with black or dark brown in a regular fashion, the individual rays corresponding with those of the forms just mentioned. Boulenger's figure of T. smithii shows this stage (B.M.C., pl. IV.), while that of T. seimundi (1903, pl. XVII.) displays it in an incipient condition. So far, then, as the plastron in verreauxii is rayed, the arrangement of the rays is exactly like that in other sub-groups.

The number of the dark brown rays on the plastral shields, and the complexity of the pattern produced thereby, reach their climax in the oculifera sub-group. Here the whole plastron has a yellow background on which are dark brown or black rays, arranged in a more or less regular manner, so as to produce a geometrical pattern. Each specimen presents a similar plan, with many variations in such details as the width of the rays, their number, and curvature. But each of the principal rays has its exact counterpart in verreauxii and the others; the number of rays may be increased by the partial or complete splitting of individual rays, but not by the intercalation of any new rays. The extreme rayed condition displayed in oculifera is foreshadowed in a complete gradational series, beginning with the uniform black and yellow of certain tentoria.

Thus, as regards the coloration of the plastron, we reach the same results as from a study of the coloration of the carapace. Starting with the simple condition found in certain members of the tentoria sub-group, we obtain a transitional series of variations leading to the very complex condition characteristic of the oculifera subgroup. Exactly corresponding rays can be traced in both carapace and plastron from an incipient stage to a very complex pattern of the same elements. In the plastron, however, we commence with a single uniformly black patch, and then trace its differentiation into separate rays, accompanied by the appearance of new rays on the sides, while in the carapace no new elements are introduced at any stage.

As regards the coloration of both the carapace and plastron, no additional type of variation is anywhere introduced. We have corresponding series of rays in each, and evolution consists in the elaboration of features common to all at one stage or another. The passage from one extreme to the other is by direct and gradual variation. The great variation in details shown in different individuals from the same locality indicates that the plan of coloration is not yet wholly fixed and permanent in any sub-group, though within varietal limitations it can be considered as characteristic.

C. AXILLARY AND INGUINAL SHIELDS.

Siebenrock, in his paper, "Die südafrikanischen Testudo-Arten der Geometrica-Gruppe s.1., has established as a character of some diagnostic value within the geometrica-group the occurrence of either one or two axillary shields; in fact, in his Synopsis of the species, he employs it as a means for separating the members of the group into two series. The first series includes geometrica and oculifera, both of which have only one axillary shield on each side, and the second all the remaining species, in which the axillary shield is divided into two small shields. The distinction holds as a general rule for all the specimens I have examined, but the acceptance of the character must not be taken as necessarily implying the direct relationship of the forms presenting the one or the other condition, for in nearly all other respects geometrica and oculifera are far apart.

The actual condition of the axillary shield within the geometricagroup is not so sharply defined as Siebenrock's use of it would imply. Both the axilliary and the inguinal shields represent the last remains of a series of numerous plates which in some Chelonia are intercalated on the bridge between the marginals and plastrals, and are known as inframarginals. Different members of the geometricagroup are found to vary as to the degree to which the reduction in number of these inframarginals has taken place. Both geometrica and oculifera, which in some other respects represent extreme types of variation, have the axillary shields reduced to one, and in nearly all the specimens of oculifera examined by me the shield is united with the femoral without any hint of a suture. In the other members of the group the reduction has not been carried so far, and a longitudinal series of two, three, or even four or five inframarginal shields may occur, which diminish in size in passing forwards. Individuals of tentoria generally show more than two shields; three or four are usually present, and occasionally one or two more, though not always in the same straight series. In verreauxii there are rarely three, usually there are two, the posterior somewhat larger than the anterior, and occasionally there is only one, while, as we have already seen, the latter condition is characteristic of geometrica and oculifera. Occasionally in verre auxii the number on one side is different from the number on the other.

Thus within the geometrica-group we have all variations represented in the reduction of the number of anterior inframarginals