f. TESTUDO SMITHII, Boulenger, 1886.

Boulenger established this species in 1886 upon a single specimen with the rather indefinite locality "South Africa." He distinguishes it from verreauxii mainly by the presence of the large femoral tubercle, but as it is practically certain that verreauxii possessed these structures, the distinction can no longer hold, and there appears no other valid character which can separate them. Boulenger's figures of the upper and lower aspects of the shell present conditions frequently occurring within the verreauxii sub-group.

Siebenrock identifies four shells from Great Namaqualand as this species, distinguishing them from verre auxii by differences in coloration; in the latter yellow and red rays occur, while in smithii there are only yellow rays; differences in the relative sizes of the shields are also noted. None of these characters can be regarded as of specific value, now that we know how very variable are the members of the geometrica-group.

Neither Boulenger nor Siebenrock, in my opinion, produces characters which would serve to separate smithii from the many varietal forms of verreauxii obtained from a single locality; none of the characters are sufficiently definite to indicate a new type, or an extreme modification in any direction.

g. TESTUDO FISKII, Boulenger, 1886.

This is another species which Boulenger founded upon a single specimen. Fortunately, in this case the exact locality from which it was obtained, namely, De Aar, not far from Hopetown, is known. The District of Hanover, whence came one large collection of over a hundred verreauxii, includes the town of De Aar, and I have also secured, through Mr. Cronwright Schreiner, 20 specimens from De Aar itself. Among them are many exactly corresponding with the description and figures given by Boulenger, so there can be no question of the correct identification; for a long time I regarded all my forms from Hanover as fiskii, but further study leaves no doubt that verreauxii was the name originally given to this type of the geometrica-group.

The color pattern in fiskii is certainly a very distinctive feature when taken by itself on any single specimen, but when it is stated that among 150 specimens there are no two specimens exactly alike in this respect, it is seen how comparatively unimportant is the feature. With the single specimen before him, and the few related types for comparison, it will be admitted that no other course was open to Boulenger than to describe the type provisionally as a new species; now, however, when an abundance of specimens from the same locality is available, we can better appreciate the value to be placed upon many of the characters hitherto utilized for specific purposes. For his studies on the four species-verreauxii, smithii, fiskii, and seimundi-- Boulenger had only three specimens, and Smith's description of the first; the verreauxii of Smith was retained, and

each of the other specimens was made the type of a new species. It is now seen that Smith's species is sufficiently variable to include them all.

h. TESTUDO STRAUCHI, van Lidth de Jeude, 1893.

Its

This species was founded by Dr. T. W. van Lidth de Jeude (Notes from the Leyden Museum, Vol. XV., 1903) for a single specimen of tortoise found in the Leyden Museum, collected at the Cape of Good Hope by the late Kuhl and van Hasselt." author considers it to have much in common with T. gometrica, but differing in the absence of the nuchal shield and the elongated form of the gulars. Comparing the figures and rather brief description with the several forms of geometrica in my possession, it is evident that the Leyden specimen is a true geometrica, except for the absence of the nuchal shield. In my examples this shield has become so very narrow that it is easy to conceive how it may disappear altogether in individual specimens. Everywhere the relative size of the gulars is found to be so very variable that the establishment of its absence in many specimens would be necessary before it could be regarded as a distinguishing specific character.

Further collections will certainly be necessary before the specific recognition of strauchi is assured. Even if a group of forms devoid of the nuchal should be forthcoming, it would still be a matter of dispute whether the character is sufficient to warrant the members being regarded as distinct from geometrica. In the present paper I can only regard strauchi as a form of geometrica in which the nuchal is wanting.

i. TESTUDO SEIMUNDI, Boulenger, 1903.

This species was established by Boulenger for a single specimen picked up near Deelfontein, in the District of Richmond. The district is adjacent to the District of Hanover, whence, as already stated, I have secured about 150 specimens, all belonging to the verreauxii sub-group. Compared with these, seimundi has no distinguishing characters whatever, and unquestionably it belongs to the same sub-group. The strong grooving of the marginals, the incurved supracaudal, and the size and shape of the gulars have their exact counterparts in the Hanover specimens. The yellowish brown horn colour above, in contrast with the black background, occupies more superficial area than usual, but introduces no new feature. The shells of older specimens frequently reach this stage. The plastral coloration compares even more closely with other members of the verreauxii sub-g -group.

Now that we know the variations of the verre auxii sub-group much better, it is clear that seimundi cannot be retained as a distinct species. Moreover, there is nothing in all the characters given which indicates that it represents the beginning of even a new type or variety; with the exception of the dominance of the yellow colour

of the carapace, there is nothing to distinguish it from an ordinary member of the verreauxii sub-group. Siebenrock (p. 308) has also expressed the opinion that, judging from Boulenger's figure and description, the species is much nearer verreauxii than tentoria.

j. TESTUDO BOETTGERI, Siebenrock, 1904.

This is a species lately founded by Siebenrock for a single shell from Greater Namaqualand, Boettger having previously determined it as T. smithii. The latter considered the habit of the shell to ally it with smithii, but the colour of the upper surface to be different; Siebenrock holds that the coloration is most nearly allied to that of trimeni, only there is wanting the small, yellow spot at the junction. of adjacent costal plates shown in Boulenger's figure of trimeni. According to Siebenrock also, its habit brings it nearest to smithii. but the upper surface of the shell is flatter, particularly in the vertebral region; the chief difference, however, lies in the comparative sizes of the plastral shields, the gulars being much longer than in smithii and verreauxii. The relative proportions between the free portion of the margin of the carapace and that united with the bridge are also different from usual. The black spots on the areola employed by Siebenrock as an important diagnostic feature for this and other species are very inconstant in specimens of verreauxii from the same locality.

From all this it is evident that boettgeri has no definite characteristics, constant in nature, which serve to separate it from the variable sub-group which includes trimeni, smithii, fiskii, and verreauxii; moreover, the shell was obtained from the area whence trimeni was first obtained.

The foregoing account fully establishes that we have in the geometrica-group of tortoises a remarkable series of allied forms, including a certain number which may be considered as extreme types, with fairly fixed characteristics, connected by a large number of intermediate forms with variable characteristics. It must be acknowledged that if the members of the extreme types only were known there would be no hesitancy in according them full specific rank. Ever since their discovery no one has questioned the true specific value of the forms embraced under geometrica, oculifera, tentoria, and verreauxii, and it is only the acquisition of a large number of related individuals which renders their standing doubtful. As regards the other species, I am of opinion, now the degree of variation is better known, that reliable, distinguishable characters are not available; founded, as they were, mostly upon single specimens, it was impossible at the time to determine the true taxonomic value of the characters presented. The conclusion may be stated in another

way. I am convinced that if all the tortoises belonging to the geometrica-group at present living in South Africa could be gathered together no sharp distinction would be found between the extremes; there would be an almost imperceptible passage from one so-called species to another; there would be numerous specimens, of which it would be impossible to say to which of two species they should belong.

In any discussion of specific types, the conclusions reached will depend much upon the conception entertained of the term species. If we consider a species to include a group of individuals having certain distinct characters in common, and find that they breed true to these characters, and continue to do so, then we must recognize a large number of species within the geometrica-group; for almost each locality has its group of individuals which differ from others in features which might fairly be considered as constant. With such slow breeding creatures as tortoises it would be difficult to prove this experimentally, yet there can be little hesitation in holding that such as oculifera, tentoria, verreauxii, trimeni, and geometrica would breed true.

When, however, we find that all the groups are connected with one another by transitional forms, the above conception of a species is insufficient. Probably the clearest expression upon the bearing of transitional forms on taxonomy is that recently given by Ortmann (Science, May 11th, 1906): "Of course, it is generally admitted that species should breed true; but this is also, a necessary character that belongs to the concept of variety. What distinguishes species from varieties is the fact that a species is not connected by intermediate or transitional forms with the most closely allied forms. This latter principle is the one made use of exclusively (if possible) by systematists, botanists as well as zoologists. In many cases, indeed, it cannot be used on account of the insufficiency of our knowledge; but under such conditions new species are always described with the tacit understanding that the demonstration of the existence of intermediate forms will reduce them to the rank of varieties."

The theory of evolution by gradual modification implies that all living organisms are genetically related, having had ancestors in common. Were all these available for study there would be no such assemblages as those involved in the concept of species. It is the breaking of the chain, the extinction of intermediate forms, which gives us species. The concept of species implies that a number of individuals have sufficiently distinct and constant characters to separate them from their relatives, the transitional forms originally connecting them and these nearest relatives being no longer existent.

From this point of view, the results of the discussion upon the taxonomic standing of the ten different species of the geometricagroup may be summarized as follows: Both oculifera and geometrica have numerous, well-defined, distinguishing characters, many of which represent extreme variations of features transitional in others. Regarded as combinations of characters, the two must for the present

be considered as well worthy of specific rank, as no closely transitional forms are yet forthcoming. The individual characters are related to those of others, but the combinations are very distinct. ndividually, the wo types are widely separated, both geographically al by the stages reached in the transformation of their separate chacters. Efforts are being made to secure specimens from distīts intermediate between those whence oculifera and geometrica have en obtained, and the areas occupied by their nearest allies; such spimens may possibly be transitional in character, and, according to th、conception here followed, this would affect the specific standing ofthe two species. Strauchi I regard as a geometrica in which the nual has disappeared.

Tentoria ah erreauxii present very fluctuating characters, especially dependen upon the locality from which they are obtained. The members of the tentoria type pass by such gradual transitions into those of the veeauxii type that it becomes impossible to say where one species els and the other begins. When studying the extremes there is no difficulty in according them specific rank, but in passing to the boler forms no distinction can be maintained; geographically, they also pas into one another. Verreauxii can only be regarded as a variety of theolder tentoria.

Trimeni seems to have fairi, constant, generalized characters, but these are so closely transition! with those of tentoria and verreauxii that it cannot be considere as having attained specific separation. It may be regarded as a varie ty of tentoria.

distinct from

Smithii, fiskii, seimundi, and boettgeri" questionably are not sufficiently differentiated from the verreauxii subgroup to warrant their recognition as distinct species. The combinaton of characters employed to distinguish each one is such as may be found among a large collection of verreauxii from any one district. They do not seem to indicate the formation of any new type, and, verreauxii, cannot be accorded even varietal rank. For the time being, we may accord separate spe of oculifera and geometrica, the transitional forms betwee others have either disappeared or are not yet forthcomie, tentoria others, with the exception of strauchi, must be merged with the following varieties or sub-species: tentoria (sensi verreauxii, and trimeni.

b. GEOGRAPHICAL DISTRIBUTION.

wit

ank to

n and all the

trictu),

The different members of the geometrica-group appear to a very well-defined distribution over South Africa.

have

ng

ՂՏ

Tentoria occurs along the southern and eastern coasts, extendr inland for a considerable distance. It reaches at least as far the Hex River Valley in its western extension, and to Natal in the north-eastern direction. It varies very little near the coast, the strongly knobbed type predominating; more inland the shields are usually flatter, the yellow rays more dominant at the angles, and the femoral spines better developed. Towards Beaufort West it passes insensibly into verreauxii, and towards the Hex River into trimeni.