^{page 557 note *} The length of the skull is obtained by measuring from the occipital crest to the alveolar point, i.e. to the base of the wedge-like piece which projects between the upper-central incisors. In living animals, the length of the skull is arrived at by measuring from the summit of the ridge across the top of the head to the edge of the gum which projects between the central incisors, and deducting 4 mm. for the skin over the occipital crest and the mucous membrane covering the alveolar point.

^{page 558 note *} Ewart, , “The Multiple Origin of Horses and Ponies,” Trans. Highland Soc. of Scotland, 1904.Google Scholar “The Tarpan,” etc., Proc. Roy. Soc. Edin., 1906.

^{page 560 note *} The methods of Czerski, Nehring, and others for measuring horse skulls are described in Salensky's work on Prejvalsky's horse. Translation by Hayes & Bradley. Hurst & Blackett, 1907.

^{page 561 note *} The difference in the relation of the face to the cranium in these two forms is perhaps accounted for by the fact that the elk is a short-necked Forest form adapted for feeding on shrubs and trees, i.e. for holding the head in a nearly horizontal position, while the sheep is a denizen of the mountains, adapted for holding the head when feeding in a nearly vertical position.

^{page 561 note †} For other indices, see Table 1.

^{page 561 note ‡} Flower and Lydekker, Mammals, Living and Extinct, p. 389.

^{page 562 note *} Professor Lankester, in his paper on the Okapi, points out that “the whole brain-case or post-orbital region of the skull of the Bovidæ appears to be bent down as on a joint across the junction of the cranial and facial portions of the skull”; and he adds, “there is good ground for connecting the presence of the deflection of the cranial cavity above noted with the mechanical conditions arising from the use of horns having the position and direction of those found in Bovidæ and the Giraffe” (“On Okapia,” Trans. Zool. Soc., vol. xvi. pt. 6). If, as Professor Lankester suggests, the deflection is connected with the use of horns, it should doubtless be regarded as due to the downward bending of the cranium on the face. If, on the other hand, the deflection is connected with grazing, with feeding on short herbage close to the ground, it might be more accurate to regard it as due to the bending downwards of the face on the cranium. There is no evidence that any of the ancestors of the Equidæ possessed horns, or that either in Prejvalsky's horse or the other recent Equidæ with a pronounced deflection is the forehead used for defence or attack. Moreover, in the Elk (Alces), notwithstanding the large horns, the face is nearly in a line with the cranium

^{page 562 note †} This view is supported by the 547 mm. Newstead skull agreeing in the frontal and other indices with the skull of an Iceland pony (Pl. I. fig. 2) of the Forest type; e.g. in the Newstead skull the frontal index is 61·29, and in the Iceland skull 61·30.

^{page 563 note *} The skull of a very young Giraffe is figured in Professor Lankester's paper on the Okapi (Trans. Zool. Soc., vol. xvi.). In the young Giraffe it certainly looks as if the cranium had been bent downwards on the face in the interest of the horns.

^{page 564 note *} In Neohipparion of the Miocene, Hipparion of the Pliocene, and E. Scotti of the Pleistocene, the face is strongly bent downwards on the cranium.

^{page 565 note *} The cranium of the skull represented in Pl. II. fig. 7, is probably abnormally small; perhaps small-brained members of the Steppe variety were more easily domesticated than individuals with a large brain, such as normally occurs in the recent Steppe horse, E. Prejvalskii.

^{page 568 note *} Ridgeway, Origin and Influence of the Thoroughbred Horse, p. 423.

^{page 569 note *} Ridgeway, Origin and Influence of the Thoroughbred Horse, p. 95.

^{page 569 note †} Tacitus mentions that at the battle of Mons Graupius the horse of one of Agricola's officers became unmanageable and carried its rider into the British lines. In such ways, and by capture, a few foreign horses would fall into the hands of the Caledonian and other tribes of Northern Britain.

^{page 570 note *} One of several skulls from Walthamstow, Essex—probably of Neolithic age—in the British Museum, closely agrees in its measurements with the 495 mm. Newstead Celtic skull, while a second Walthamstow skull agrees with the small Newstead skull which seems to have belonged to a Celtic pony with a strain of Forest blood. Even in Neolithic times intercrossing seems to have been practised, or at least possible.

^{page 572 note *} Mr Macdonald points out that too much stress must not be laid on the territorial names of the auxiliary regiments; these names undoubtedly indicate the districts in which the regiments were originally raised, but there was no organised system of territorial recruiting, and consequently (as the inscriptions show) the soldiers were often of different nationalities. This need not vitiate the inferences which I suggest; the name of the district in which a cavalry regiment was originally raised is probably a safe index to the foundation stock of its horses.

^{page 575 note *} A similar view is held by Mr Lydekker, Knowledge, Aug. 1904.

^{page 575 note †} In the Steppe variety the face is very long (PI. III. fig. 12), decidedly convex, and strongly bent on the cranium; there are only five lumbar vertebræ, the limbs are long and slender, the fetlock joints small, and the hoofs long and narrow. The mane is short and upright (hence there is no forelock), while the tail is mule-like. The dorsal band is narrow, and at the most only faint vestiges of stripes occur on the trunk and legs.

^{page 575 note ‡} In having the face nearly in a line with the cranium, a concave profile, six lumbar vertebræ, large fetlock joints and broad hoofs, a full mane and tail, a broad dorsal stripe, and stripes on the face, trunk, and legs, it profoundly differs from the Steppe variety.

^{page 575 note §} In the Plateau variety the face is dished, as in the Forest variety, but longer, and slightly more bent. In colour and markings and in the number of the dorso-lumbar vertebræ, it agrees with the Steppe variety. From the Steppe as well as the Forest variety it differs in having a very narrow face, and in being devoid of hind chestnuts and all four ergots. Though in the Plateau, the hoofs are broader than in the Steppe, variety, the fetlock joints are as small and the “bone” as flat.

^{page 576 note *} In the skull of a Shire horse (Starlight) in the British Museum, the frontal index is 51·1, i.e. in its frontal index Starlight is almost identical with Prejvalsky's horse and with the 560 mm. Roman skull from Newstead.

^{page 577 note *} Evidence of Celtic blood we have in the full eyes, small ears, and a more or less perfect tail-lock, and at rare intervals of the all but complete absence of hind chestnuts.

^{page 580 note *} Marshall, “The Horse in Norway,” Proc. Roy. Soc. Edin., vol. xxvi. pt. 1.

^{page 584 note *} Cortes, e.g., took with him nine chestnuts and three grey horses, one sorrel, one light-dun, one yellow-dun, and one dark-coloured horse. Ridgeway, Origin and Influence of the Thoroughbred Horse; p. 268.

^{page 586 note *} The frontal and orbital measurements and indices of these three skulls are taken from a paper communicated to the Royal Society, November 1906, by Prof. O. Charnook Bradley, D.Sc., Proc. Roy. Soc. Edin., vol. xxvii, pp. 46–50.