them with the salt he had amassed for the convoy to convey to El Aghouat. However, an hour or two of labour would soon suffice to supply the loss. The Bedouin usually prefer the blue rock-salt to the pure crystallized deposit of the streams.

However inaccurate may be the statement of Herodotus with respect to the five mountains of salt in the interior of Libya being at exact intervals of ten days' journey, yet modern researches seem to have proved the truthfulness of the information of the father of history as to the main facts. How admirably does his expression "a ridge of sand,"+ rather than a plain, describe the edge of the northern Sahara! Again, we can undoubtedly discover hills of rock-salt at uncertain intervals, but averaging ten days' journey between each, through the whole ridge from the oasis of Ammon to the Atlantic.

The five mountains of salt mentioned by Herodotus have with the exception of the last been identified with tolerable certainty. About those of Ammon and Augila there can be no question. The third hill, in the country of the Garamantians, both Rennell and the later authorities place in Fezzan, and Rawlinson observes that it has a ruined city, Germa (Garama), once the capital. The fourth, Rennell would place at 'Ghadames, where, however, there have been no salt-hills yet discovered,

* The Hadjera el Mêhl is, without doubt, the spot described by Dr. Shaw (i. 97) as "the eminences and salt-pits of Zaggos," as he states them to be a little way beyond the Seven Sisters mentioned above, a few leagues to the northward of Saary (Dj. Sahari). They are, however, not salt-pits, but rocks. It may be noted that the rock-salt is of the grey or bluish colour, and with the same bitter taste which Shaw observes of the salt-rocks of El Outaia, near Biskra, and is in the same way prepared by the natives. The sal gem sold by the Beni M'zab, and mentioned by Shaw, is, so far as I could ascertain, really obtained at the Hadjera el Mêhl. † ὀφρύη ψάμμης, iv. 181.

and Rawlinson vaguely places it among the Touaregs of the Western Sahara. Admitting that Herodotus may in this, as he evidently was in the case of some of the other salt-hills, have been misinformed as to the distance, and has considerably understated it, we may place the fourth salt-hill station on the caravan route from Fezzan to Carthage either at 'Ghadames or elsewhere, and then in looking for the fifth salt-hill we shall find ourselves not far from Haddeffa (Hadyfa), the first salt-hill mentioned by Shaw, and which the Arabs call sixteen days' journey from 'Ghadames. As the tracks of the caravans have remained unaltered for ages, the fifth hill must either be looked for on this route or on the westerly route from 'Ghadames to Souf, which is also sixteen days' journey. This course is followed only by caravans to the western oases, and not to the coast, consequently we can scarcely conceive it probable that Herodotus had any direct information concerning it.

Again, though there is an oasis with many wells at Souf, there is no salt, and it is nine days' journey thence to El Outaia, the next known salt-hill. When I visited the Sebkha, the ancient lake Tritonis, on the western side of Hadyfa, the whole plain for two days' journey was encrusted with a thin crisp coating of salt deposit, thus again corroborating the description of Herodotus.†

* See map of the caravan routes.

†The Kabyles are in the habit of purchasing large quantities of the "sal gem" for the manufacture of saltpetre, and the French government have used the utmost vigilance to intercept their supplies. Their mode of manufacture is very rude. I have seen the M'zab engaged in the process of extracting saltpetre from the saline incrustation of the Chott, near Waregla. This they do by erecting large tanks of stone, with split palm-branches laid across them about half way up. Over these coarse matting is spread, the upper portion of the tank is filled with the saline substance, and water constantly poured upon it. This

But if we agree to place the Atlantes at Hadyfa, where the rock shines of a purplish pink hue, we must resign all attempts to identify the Atlas peak of Herodotus, which probably is a confusion between the Phoenician tradition of the Peak of Teneriffe and the Mount Zaghouan, the only conspicuous isolated mountain in the regency of Tunis, but which is too far north to be seen from Hadyfa. Still, being one of the great landmarks on the caravan route, an exaggerated description of it might easily have been here misplaced. But almost accurately true is the further observation of Herodotus, διήκει δ ̓ ὧν ἡ ὀφρύη μέχρι Ἡρακληΐων στηλέων καὶ τὸ ἔξω τουτέων· ἔστι δε ἁλός τε μέταλλον ἐν αὐτῇ διὰ δέκα ἡμερέων ὁδοῦ καὶ ἄνθρωποι οἰκέοντες (V. ch. 185). Twelve days' journey due west from Hadyfa is the salt mountain of El Outaia, and nine days further the Hadjera el Mêhl. Beyond this again are the Djebel Menes, and others, of which, however, I could gather no further precise information from the Arabs, except that Hadjera el Mêhl are to be found at intervals right across the south of Morocco.

filters into the bottom of the tank, and, after the process has continued for a few days, the matting, with all that remains upon it, is removed, the water below is allowed partially to evaporate, and then the sediment is boiled away until the saltpetre is crystallized. From the scarcity of fuel I presume they do not proceed further in the manufacture, but carry their saltpetre to the Kabyles, who, burning charcoal and in some way or other obtaining sulphur, produce a considerable quantity of inferior gunpowder.

-

CHAPTER V.

Roman ruins - Djelfa - A countryman in the Sahara - A dilatory Dragoman - A cold sponge- Dr. Riboud - Botany of the Desert Saharan pool - A young wife - Ain el Ibel- Forced colonization-Sheep-farming Rat à trompe- The general hoaxedBedouin shepherds Large camp-The commander's tent - Hospitable cook- Commandant Marguerite-Battle of the horsesHairbreadth escape Touareg embassy - Puits Enez - Honest companions-Arrival at El Aghouat.

FROM the Rochers to Djelfa, our next halting-place, was only a six-leagues ride up the course of the Wed Melah. The scenery was pretty, to our eyes charming, in contrast with the quaker-like drab which had clad the desert for so many miles; with a little cascade and low ranges. of thuya covering the hills on either side, which were broken and diversified. On the way was a curious collection of ruins, which tradition assigns to the Romans. They occupy the whole enceinte of a low isolated tablerock, by the side of a sparkling stream which emerges from a tunnel just above, and waters some rich cornfields below. If the ruins be Roman, admirable as is their strategic position, they are constructed with much less care than that people usually expended on their masonry. They may, however, have been hastily thrown up for the shelter of a distant outpost, or in the advance of Theodosius against Firmus and Igmazen.* At all events they are not Arabic, and must, I think, be ascribed to a period antecedent to the arrival of the races which now roam over the district. Their character is that of

Gibbon, ch. xxv.

small square dwellings surrounding a large oblong courtyard, and they must have been quite strong enough easily to resist a sudden attack of cavalry.

Five or six miles further on, a water-mill has been recently erected, chiefly from the materials of Roman remains. Many tombs have been discovered here, and a few unimportant inscriptions and coins, all of the later Empire. But near the same place have been found many tombs of a very different character, and exactly like Celtic remains of the Stone age. The graves are formed of three or four large stones, with a very massive top slab, and smaller stones heaped round them into the form of a cairn. We saw several stone arrow-heads which had been dug up in the same locality. Do not these traces indicate an occupancy prior to the Roman invasion, perhaps by Gætulians, in times more ancient than Herodotus, and before the Greeks had introduced the knowledge of bronze, or the Phoenician traders that of iron ? Other tombs resemble the ship-tombs of Scandinavia, oblong circles of stone, with larger pillars at the head and feet. But these latter approach much more nearly the modern Arab mode of sepulture.

Djelfa itself, the highest village in the Djebel Senalba, and a little above the Roman post, is 3400 feet above the level of the sea, cold and bleak. The barrack, or Bureau Arabe, is a square redoubt enclosing two arched and colonnaded hollow squares-one appropriated to man, the other to beasts and stores. There are a few French cottages outside, the inhabitants of which are employed in sawing timber in the neighbouring forest; and the black and white striped tents of a squadron of Spahis occupy the other slope; while opposite is another green mamelon with an unfenced cemetery, consisting of three nameless graves, and as many more recording the