And this incongruous bit of Greece or Rome, in the Arabian wild, kept its mystery to the last: the more we looked at it, the less we could explain its presence. Not a line of inscription, not even a mason's mark—all dark as the grave; deaf-dumb as "the olden gods."

The site of the Gasr is in north lat. 25° 55' 15";[70] and the centre of the Libn block bears from it 339° (mag.). It stands upon the very edge of its Wady's left bank, a clifflet some twenty-five feet high, sloping inland with the usual dark metal disposed upon loose yellow sand. Thus it commands a glorious view of the tree-grown valley, or rather valleys, beneath it; and of the picturesque peaks of the Tihámat-Balawíyyah in the background. The distance from the sea is now a little over three miles—in ancient days it may have been much less.

The condition of the digging proves that the remains have not long been opened: the Baliyy state less than half a century ago; but exactly when or by whom is apparently unknown to them. Before that time the locale must have shown a mere tumulus, a mound somewhat larger than the many which pimple the raised valley-bank behind the building. A wall is said to have projected above ground, as at Uriconium near the Wrekin.[71] This may have suggested excavation, besides supplying material for the Bedawi cemetery to the south-west. The torrent waters have swept away the whole of the northern wall, and the treasure-seeker has left his mark upon the interior. Columns and pilasters and bevelled stones have been hurled into the Wady below; the large pavement-slabs have been torn up and tossed about to a chaos; and the restless drifting of the loose yellow Desert-sand will soon bury it again in oblivion. The result of all such ruthless ruining was simply null. The imaginative Nájí declared, it is true, that a stone dog had been found; but this animal went the way of the "iron fish," which all at El-Muwaylah asserted to have been dug up at El-Wijh—the latter place never having heard of it. Wallin (p. 316) was also told of a black dog which haunts the ruins of Karáyyá, and acts guardian to its hidden treasures. Years ago, when I visited the mouth of the Volta river on the Gold Coast, the negroes of Cape Coast Castle were pleased to report that I had unearthed a silver dog, at whose appearance my companion, Colonel de Ruvignes, and myself fell dead. But why always a dog? The "Palace" is a Roman building of pure style; whether temple or nymphæum, we had no means of ascertaining. The material is the Rughám or alabaster supplied by the Secondary formation; and this, as we saw, readily crumbles to a white powder when burnt. The people, who in such matters may be trusted, declare that the quarries are still open at Abú Makhárír, under the hills embosoming Abá'l-Marú. We should have been less surprised had the ruin been built of marble, which might have been transported from Egypt; but this careful and classical treatment of the common country stone, only added to the marvel.

It must have been a bright and brilliant bit of colouring in its best days—hence, possibly, the local tradition that the stone sweats oil. The whole building, from the pavement to the coping, notched to receive the roof-joists, is of alabaster, plain-white and streaked with ruddy, mauve, and dark bands, whose mottling gives the effect of marble. Perhaps in places the gypsum has been subjected to plutonic action; and we thought that the coloured was preferred to the clear for the bases of the columns. The exposed foundations of the eastern and western walls, where the torrent has washed away the northern enceinte, show that, after the fashion of ancient Egypt, sandstone slabs have been laid underground, the calcaire being reserved for the hypaethral part. The admirable hydraulic cement is here and there made to take the place of broken corners, and flaws have been remedied by carefully letting in small cubes of sound stone. There are also cramp-holes for metal which, of course, has been carried off by the Bedawin: the rusty stains suggest iron.

The building is square-shaped, as we see from the western wall, and it evidently faced eastward with 25° (mag.) of southing. This orientation, probably borrowed from the Jews, was not thoroughly adopted in Christendom till the early fifth century, when it became a mos. The southern wall, whose basement is perfect, shows everywhere a thickness of 0.95 centimetre, and a total length of 8 metres 30 centimetres. At 2 metres 87 centimetres from the south-western corner is a slightly raised surface, measuring in length 2 metres 15 centimetres. Mr. James Fergusson supposes that this projection, which directly fronts the eastern entrance, was the base of the niche intended for the image. On each side of the latter might have been a smaller colonette, which would account for the capital carried off by us to Egypt. Thus, adding 2 metres 87 centimetres for the northern end swept into the valley, we have a length of 7 metres 89 centimetres; and the additional half thickness of the east wall would bring it to a total of 8 metres 30 centimetres.

The shrine was not in antis, and the site hardly admits of a peristyle; besides which, excavations failed to find it. That it might have had a small external atrium is made probable by the peculiarity of the entrance. Two rounded pilasters, worked with the usual care inside, but left rough in other parts because they could not be seen, were engaged in the enceinte wall, measuring here, as elsewhere, 0.95 centimetre in thickness. Nothing remained of them but their bases, whose lower diameters were 0.95 centimetre, and the upper 0.65; the drums found elsewhere also measured 0.65. The interval between the lowest rings was 1 metre 63 centimetres; and this would give the measure of the doorway, here probably a parallelogram. Lying on the sand-slope to the north, a single capital showed signs of double brackets, although both have been broken off:[72] the maximum diameter across the top was 0.60 centimetre, diminishing below to 0.50 and 0.44, whilst the height was 0.40. The encircling wall was probably adorned with pilasters measuring 0.62 centimetre below, 0.45 above, and 0.11 in height: they are not shown in the plan; and I leave experts to determine whether they supported the inside or the outside surface. Several stones, probably copings, are cut with three mortice-joints or joist-holes, each measuring 0.15 centimetre, at intervals of 0.14 to 0.15.

In the tossed and tumbled interior of this maison carrée the pavement-slabs, especially along the south-western side, appear in tolerable order and not much disturbed; whilst further east a long trench from north to south had been sunk by the treasure seeker. The breadth of the free passage is 1 metre 92 centimetres; and the disposal suggested an inner peristyle, forming an impluvium. Thus the cube could not have been a heroön or tomb. Four bases of columns, with a number of drums, lie in the heap of ruins, and in the torrent-bed six, of which we carried off four. They are much smaller than the pilasters of the entrance; the lower tori of the bases measure 0.60 centimetre in diameter, and 0.20 in height (to 0.90 and 0.25), while the drums are 0.45, instead of 0.65. It is an enormous apparatus to support what must have been a very light matter of a roof. The only specimen of a colonette-capital has an upper diameter of 0.26, a lower of 0.17, and a height of 0.16.

Although the Meccan Ka'bah is, as its name denotes, a "cube," this square alabaster box did not give the impression of being either Arab or Nabathæan. The work is far too curiously and conscientiously done; the bases and drums, as the sundries carried to Cairo prove, look rather as if turned by machinery than chiselled in the usual way. I could not but conjecture that it belongs to the days of such Roman invasions as that of Ælius Gallus. Strabo[73] tells us of his unfortunate friend and companion, that, on the return march, after destroying Negrán[74] (Pliny, vi. 32), he arrived at Egra or Hegra (El-‘Wijh), where he must have delayed some time before he could embark "as much of his army as could be saved," for the opposite African harbour, Myus Hormus. It is within the limits of probability that this historical personage[75] might have built the Gasr, either for a shrine or for a nymphæum, a votive-offering to the Great Wady, which must have cheered his heart after so many days of "Desert country, with only a few watering-places." Perhaps an investigation of the ruins at Ras Kurkumah and the remains of Madáin Sálih may throw some light upon the mystery. In our travel this bit of classical temple was unique.

Mr. Fergusson, whose authority in such matters will not readily be disputed, calls the building a small shrine; and determines that it can hardly be a tomb, as it is hypæthral. The only similar temple known to him is that of "Soueideh" (Suwaydah), in the Haurán (De Vogüé, "Syrie Centrale," Plate IV.). The latter, which is Roman, and belonging to the days of Herod Augustus, has a peristyle here wanting: in other respects the resemblance is striking.

M. Lacaze photographed, under difficulties such as bad water and a most unpleasant drift of sand-dust, the interior of the building, the stones lying in the Wady below, and the various specimens which we carried off for the inspection of his Highness the Viceroy. Meanwhile we "pottered about," making small discoveries. The exposed foundations of the north-western wall, where the slabs of grit rest upon the sands of the cliff, afforded signs of man in the shape of a jaw-bone, with teeth apparently modern; and above it, in the terreplein, we dug down upwards of a yard, without any result beyond unearthing a fine black scorpion. The adjoining Arab graveyard, adorned with the mutilated spoils of the classical building, gave two imperfect skulls and four fragments. We opened one of the many mounds that lie behind the Gasr, showing where most probably stood the ruined town; and we found the interior traversed by a crumbling wall of cut alabaster—regular excavation may some day yield important results. A little to the south-west lies a kind of ossuary, a tumulus slightly raised above the wavy level, and showing a central pit choked with camels' bones: at least, we could find no other.