In footnotes he adds:

Particular interest attaches to Wilkinson’s account of the tradition, not only because of its being the earliest, and therefore less likely to be coloured by imagination than later versions, but also because of the remarkable fact that although some of the other places named in the above extract were unknown to European geographers at the time, they have all since been discovered; “Gebabo” and “Tazerbo” by Rohlfs, and “Rebeeana” by Mrs. Forbes and Hassanein Bey.

The weak point in the account is the loose Arab way of stating directions. Gebabo (Kufra) is not west of the Baharia-Farafra road, but south-west. If we amend the bearing in the first account accordingly, and take Zerzura at about midway between the Baharia-Farafra road and Kufra, it must lie near the intersection of the parallel of 26° with the meridian of 26°, or some 200 kilometres east of the Mehemsa Hattia, and some 150 kilometres north-west of Rohlfs’ “Regenfeld” camp. In the second account given by Wilkinson, the words “due west” suggest that the bearing is more certain. Two or three days (say 120 kilometres) due west of Dakhla would put Zerzura in about latitude 25½°, longitude 27½°, or about 30 kilometres north of Rohlfs’ “Regenfeld.” There is thus a difference of more than 100 kilometres between the two positions indicated by the accounts given to Wilkinson.

Rohlfs evidently knew of the traditions regarding Zerzura before making his attempt to reach Kufra from Dakhla in 1874; and since such an intermediate oasis, if it really existed, would be an immense aid to him in attaining his objective, he made careful inquiries concerning it before leaving Dakhla. But he found that though every one in Dakhla knew the names of Zerzura and Kufra, no one could tell him where Zerzura was, nor how far it lay from Dakhla.[27] Evidently Rohlfs placed little faith in its existence, for during his journey he gave the name “Zerzura” to a locality, about 120 kilometres west-south-west of Dakhla, where he found nothing but very sparsely scattered vegetation.[28] Ascherson, who was with Rohlfs in Dakhla, was however informed by Hassan Effendi, one of the principal inhabitants of Mut, that about a hundred years before (i.e. about 1770) there had been frequent raids on Dakhla by Arabs from the south-west; and that in order to stop these raids the Mameluke Government of the time installed a military colony, called the Surbaghi, in the village of Qalamun. These Surbaghi went out and destroyed all the wells for seven or eight days’ distance along the road by which the raiders had come. This road, which up to that time had served as a trade route from Darfur, before the road through Kharga was opened, was in consequence forsaken. The road was said still to exist, and to lead into the desert from Mut. At a day-and-a-half’s journey from Mut there were said to be two “pillars,” half an hour’s distance apart, which Hassan Effendi’s guard likened to minarets. In the neighbourhood there were said to be large stone-quarries, and about thirty years previously an iron instrument had been found; the instrument was still in Hassan Effendi’s possession, and some wonderful stories had grown up as to the manner of its use.[29]

Ascherson himself believed this information of Hassan Effendi’s to be substantially correct, but he thought the “pillars” might be merely pinnacle-shaped natural rocks. I am likewise inclined to believe it trustworthy, but I think the “wells” may have been merely water-dumps—that is, collections of jars of water. My reason for this view is twofold. In the first place, the ground-levels in that part of the desert are so far above the static water-level, that if wells were bored they would have to be very deep, and the water would not rise anywhere near the surface (there can, of course, be here no question of local rainfall as a source for the water in the “wells”); and in the second place, it seems very probable that the accumulation of large broken earthenware jars which I discovered in 1916 at the place I named “Pottery Hill” (latitude 24° 26′ 27″, longitude 27° 38′ 54″) is one of the dumps in question.[30]

The importance of the above interpretation of Hassan Effendi’s story, if accepted, as I think it must be, lies in the proof which it furnishes that there cannot possibly be an undiscovered oasis anywhere near a point seven or eight days from Dakhla in the direction of Kufra. A consideration of the static water-levels and the contours of the ground would lead us to this conclusion if the views I have advanced on the underground-water question are correct; but the fact that the raiders would never have made such a large water-dump if water could have been readily got from a well in the vicinity makes the conclusion almost certain independently of my hypothesis, and thus incidentally tends to support the latter. Is it possible that “Zerzura,” which I am told signifies in Arabic a starling (but is commonly applied to any small bird), is here a corruption of some other name derived from “zeer,” a water-jar, and that instead of “the oasis of the blacks,” Zerzura was really “the water-depôt of the blacks”?

Schweinfurth has recorded[31] a story which was told him in Kharga Oasis, that in 1872 some Arabs of a Darfur caravan, who had missed the road, found a small oasis about 1½ days’ journey to the west of Beris. The oasis was said to contain pools full of wild geese, date-palms, and a temple. People who went out of Beris to find the place returned unsuccessful. I am, however, rather inclined to think that the story has some basis of fact, because in 1898 I myself found springs and traces of ancient conduits in the dunes north-west of Beris, and there may be other water-sources beyond the point which I reached.[32]

A writer in the ninth edition of Murray’s ‘Guide to Egypt,’ published in 1896, gives the following different statements of Arabs as to the position of Zerzura:

(1) Some days south of the Dakhla Oasis.

(2) Five days west of Farafra Oasis.

(3) Three days west of Dakhla Oasis.

(4) Two or three days west of Selima Oasis.

The first of these statements would correspond sufficiently well with Bir Terfawi, which is about 280 kilometres due south of Dakhla Oasis; the second would place Zerzura about in latitude 27°, longitude 26°; the third corresponds with the second account given to Wilkinson; while the fourth would place it far to the south-west, in latitude 21½°, longitude 28°, or rather less than halfway between Bir Terfawi and Merga.

Mr. Harding King, who in his journeys in the Libyan Desert in 1909 and 1911 devoted much attention to the collecting of native information regarding its geography, thinks that “Zerzura” may possibly be only a generic name applied to any mythical or undiscovered oasis.[33] He heard the name applied to the following localities:

(1) Rohlfs’ “Sersura.”

(2) The “Egyptian Oasis,” said to have been seen by an Arab from the top of a high black hill lying in the dune-belt ten long days by ordinary caravan from Kharga. Another Arab is said to have seen what is possibly the same place eight days somewhere to the south of Dakhla. Both these accounts agree that the place is a large oasis lying at the foot of a scarp and containing olive trees.

(3) A stone temple eighteen hours’ journey west of Gedida in Dakhla Oasis.

The information given to Mr. Harding King as to the position of the “Egyptian Oasis” would place it about in latitude 23°, longitude 28½°, or only some 50 kilometres north-west of Bir Terfawi; in fact, considering the vagueness of the information, it might correspond fairly well with Terfawi itself, except that there are no olive trees at Terfawi, nor is that place overlooked by any escarpment or hill. The “high black hill in the dune-belt” may just possibly be one of those I mapped near the farthest point I reached with Moore in 1916, in about latitude 24°, longitude 26°, though this would imply a direction south-west of Dakhla, not south. It is not likely to be Gebel Kamil, as that mountain lies much nearer to Merga than to Dakhla.

The “stone temple,” according to the information given to Mr. Harding King, would be about 80 kilometres west of Dakhla Oasis, in about latitude 25½°, longitude 28°, or not very far from the second of the positions indicated by Wilkinson for Zerzura. But I think it is likely that the statement really refers to the “Deir el Hagar,” a well-known temple ruin much nearer to Dakhla.

So much for the various statements as to whereabouts Zerzura may be. Let us now inquire where it is not. On the map I have indicated the principal routes followed by explorers of the southern and western parts of the Libyan Desert during the last fifty years. We may be tolerably certain that Zerzura is not on, or very close to, any of these routes; for although a depression within a kilometre or so of one’s track may easily be passed without notice, the existence of a large oasis, such as most of the traditions make Zerzura out to be, would almost certainly be betrayed to an explorer by animal-tracks leading to it from considerable distances, except possibly in places where the ground was very sandy.

There is another method now available to us by which we may narrow down the search for Zerzura or other “lost” oases; and that is, by a consideration of the general contours of the country and those of the static water-surface. Every oasis in the Libyan Desert must owe its existence to one or other of two conditions: either it must depend on springs fed by local rainfall, in which case, like the oases of Owenat and Arkenu, it probably lies near to mountains of considerable height; or else it must depend on underground supplies, and must therefore lie in a depression wherein the ground-level and static water-level are practically coincident, as in the cases of Kharga, Dakhla, Farafra, Baharia, and Siwa. It is highly doubtful whether any mountain masses at all approaching the altitude of Arkenu and Owenat can remain undiscovered in any of the various areas in which Zerzura has been traditionally placed. Zerzura is immensely more likely to be in a depression, and indeed Wilkinson’s name Wadi Zerzura almost conclusively points to that view. The depression is more likely to be shallow than very deep, for it presumably lies in the sandstone country of the south-west, and all the known depressions of great depth are confined to the limestone country of the north-east. It is quite easy to trace out on our new map the areas wherein the static water-surface would be reached by a depression of say 50 or 100 metres below the general ground-level as indicated by the contours. We must, however, remember that our ground-contours are liable to be considerably in error in areas where observations have been few—that is, in precisely those areas where the depression, if it exists, is most likely to be found. So we must allow a good margin for our depth, and I have chosen 100 metres on this account. On the map I have drawn the “locus” of points in the southern part of the Libyan Desert where the static water-surface is 100 metres lower than the general ground-surface[34]; and I have edged with red the only areas in the west and south in which the two surfaces are within 100 metres of each other—that is to say, the areas within which comparatively shallow depressions with underground springs must lie, if our contours of the two surfaces are drawn with even approximate correctness. It will be seen from the map that this restricts the search very considerably. In fact, if Zerzura is within the frontiers of Egypt, it lies in all probability either to the west of longitude 26° 20′ and north of latitude 26°, or to the east of longitude 27° and south of latitude 23° 30′. We may dismiss from our field of search all the broad tract of rising ground which extends from near the Dakhla escarpment south-westwards towards Owenat; for unless Zerzura is in a depression of great depth, it cannot possibly lie within this tract. This consideration shows that most of the previous rough estimations of the possible position of Zerzura must have been erroneous; and indeed it is remarkable how few of the various explorers’ tracks traverse the only two areas within which Zerzura, if it exists, almost certainly lies. The northern area has only been crossed by the Rohlfs expedition of 1874, and that near its eastern edge; while the western part of the southern area has only been crossed by Prince Kemal el Din’s expedition of 1925. Of all the Arab traditions, only those which would place Zerzura far to the west of Farafra, or far to the south-south-west of Dakhla, can now have any high degree of credibility. It can hardly, I think, be doubted that the various traditions refer to more than one place; and in view of the almost totally unexplored state of the only two areas in which our new contour-maps indicate the possibility of unknown oases existing, it is quite conceivable that at least one oasis may lie within each of them. In the northern area, Siwa would probably make the best starting-point from which to conduct a search, as it is easily reached by motor-car from Alexandria or Cairo, and exploratory journeys southwards from Siwa would mostly lie along the direction of the dune axes. In the southern area, Terfawi or Sheb would form the best base for exploration; at each of these places there is a good water-supply. I am inclined to think that the southern area is more likely to yield results than the northern one. Water was found to exist at a point some 15 kilometres west of Terfawi, and though no other source was observed on the way to Owenat, it was impossible to see very far on either side of the track, so that such a source might have been passed within a few miles without notice. Terfawi itself is very difficult to find, being inconspicuous even when one is fairly close to it; and as it is in a sandy area, tracks leading to it are soon obliterated. Another factor which favours this southern area is that, the country being all Nubian sandstone, the depth from the surface to the actual water-bearing beds (as distinguished from the depth to the static water-surface) is certain to be much less than in the northern area, where Cretaceous strata may overlie the sandstone; hence there is more likelihood of the existence of natural springs in the south than in the north.

9. Can Travel in the Great Tracts of the Libyan Desert which are now Waterless be facilitated by the Sinking of New Wells?

It has sometimes been asked whether tracts which it is now difficult or impossible to cross by camel, owing to the non-existence of wells or natural springs within them, might possibly be opened up to transport by the sinking of new wells along a proposed route, as has in fact been done by the Senussi in the case of the well at Sarra, between Wanianga and Kufra. Hitherto there have been no data from which to form an opinion on this question; but a study of the new map will enable at least a qualified answer to be given.

To be a practicable proposition, a new well must fulfil two conditions: the first, that the boring must not be required to descend to a very great depth; and the second, that the water when struck must rise in the bore to within a reasonable distance, say 20 to 30 metres, of the ground-level at the place. The first of these conditions will be satisfied if the geological horizon of the water-bearing bed is comparatively near the surface; the second, if the level of the static water-surface in the locality is within 20 or 30 metres of the ground-level.

As regards the first of these conditions, we know that the water-bearing beds underlying the Libyan Desert are situated within the geological formation called the Nubian sandstone. We may therefore eliminate from our consideration all tracts in which this formation is overlain by any great thickness of younger rocks, for all these rocks will have to be passed through in order to tap the water-bearing strata.

Concerning the second condition, our map at once informs us that the areas in which the static water-surface is within 20 or 30 metres’ depth below the ground-level are decidedly limited. Even if we suppose that there may be unknown depressions extending to a depth of 70 metres below the general level of the country indicated by the contour-lines, we see that the localities in which the second condition would be fulfilled are confined to the tracts edged with red on the map. Outside these tracts, not only have we no chance of discovering oasis-depressions, but we also have no prospect of being able to bore wells in which water would rise to within a reasonable distance of the ground-level. Thus the broad tract of rising ground which extends south-westwards from Dakhla to Owenat must always remain a waterless waste.

In the Egyptian portion of the Libyan Desert there are, as will be seen from the map, only two areas in the west and south in which new wells might successfully be bored: one extending for some 300 kilometres or so to the south of Siwa Oasis in the neighbourhood of the western frontier, the other extending for some 400 kilometres westward from the Nile in the neighbourhood of the southern boundary. Fortunately, however, these areas are so situated that wells sunk within them might be of considerable use in opening up the outermost parts of Egypt to exploration. A well near the western frontier about on the parallel of 26°, for instance, might make it just possible for cars or camels to reach Owenat directly from Siwa, since it would divide the present waterless stretch of 830 kilometres into two stretches of about half that length; while one or two wells near the southern frontier between the meridians of 27° and 28° would render Owenat fairly accessible to camels coming from the Nile viâ Terfawi or Sheb, and might be of great use to explorers or geologists desirous of making a detailed investigation of the Owenat and Arkenu region. From the little I saw of Gebel Owenat during my visit to it in 1925, its geology must be of extreme interest. The south-western part of the mountain appears to be entirely composed of granite and other crystalline rocks, while the eastern part presents huge cliffs of sandstone, with crystalline rocks showing only at the foot; there is evidently a great fault traversing the mountain mass, with a downthrow to the east; and the thickness of Nubian sandstone exposed on the eastern precipices is greater than that at any other place I have seen.[35] Gebel Kissu probably resembles the south-western part of Owenat in being entirely composed of crystalline rocks.[36] East of Gebel Owenat there are many igneous hills, some of which exhibit bands of a dark brown colour. These bands, though probably mostly igneous dykes, may possibly in some cases be the gozzany outcrops of veins containing metallic minerals; I had no opportunity of examining them at close quarters, but I saw enough to make me long for facilities to undertake such an examination.

It may be remarked that a well in the northern area would probably have to be very deep, owing to the comparatively high geological horizon of the surface rocks there; but in the southern area, where the Nubian sandstone forms the surface rock, no great depth of boring would be likely to be required in order to tap the water-bearing beds.

Apart from the question of new wells in the more remote parts of the Libyan Desert, a study of the map gives us some hints which may be of value should it ever be desired to sink wells in places nearer to the oases and to the well-known tracks. There appears, for instance, to be no reason why wells should not be successfully bored at intervals along the Darb el Arbain between the south end of Kharga and Lagia, or on the south side of the Qattara and other depressions of the northern part of the desert. In the southern area, the best sites to select for wells will be depressions wherein the ground-level approximates most closely to the static water-level; and readings of an aneroid barometer, compared with corresponding readings at a place of known altitude, such as Sheb or Terfawi, would enable the most favourable sites to be determined. Observations of the geological structure will also be of importance; for anticlinal folds, by bringing the water-bearing strata nearer to the surface, would diminish the depth to which borings would have to be carried to tap the beds; while faulting might likewise introduce favourable conditions by producing cracks and fissures along which the water could rise. It may be remarked that tamarisk-bushes are generally a sign that water exists at no great depth. In regard to wells in or near the northern depressions, it is obvious that sites on the south side of the lakes and salt-marshes should be selected; for not only is the artesian static level higher in the south, but the water is less likely to be contaminated by salts derived from the lake and marshes.

10. The “Tortoise Marshes” of Ptolemy.

The passage in Ptolemy’s ‘Geographia’ (lib. IV. cap. 6, sect. 4) in which the position of the “Tortoise Marshes” is given may be translated as follows[37]:

There are two great rivers running into the Mediterranean; one of them is the Gir, joining Mount Usargala with the Garamantic narrows, from which, changing its course, the river is located in long. 42°, lat. 16°, and makes the Tortoise Marshes (Chelonitides Paludes), whose position is long. 49°, lat. 20°.

The information on which Ptolemy based this statement, at least as regards the river Gir, must have been very fragmentary, if not, indeed, grossly erroneous; for there is no river flowing to the Mediterranean anywhere near the positions he gives. But there can be little doubt that his names Gir and Chelonitides Paludes refer to real places, and there has been much speculation among geographers as to their identification.

Thus, for instance, Knoetel[38] suggested that the Gir may have been the Wadi Djedi, to the south of Biskra, and the Tortoise Marshes the modern Lake Melghir (lat. 34°, long. 6°)[39] Dr. William Smith[40] thought that the Gir was really a branch of the Niger, and the marshes the modern Lake Fittri (lat. 13°, long. 18°). On the map of Africa on the 2-million scale published by the Geographical Service of the French Army in 1899 the Tortoise Marshes are depicted as lying in about latitude 19° 20′, longitude 27°, with a note stating that this delineation is taken from an earlier map of the Nile Basin by Miani, who regarded the marshes as being connected with the “dry river” which was then supposed to run northwards through Dakhla Oasis. Colonel Tilho has lately suggested[41] the lowlands to the north-east of Lake Chad (lat. 18°, long. 17°) as a possible site for the marshes; while still more recently Mr. Harding King[42] has thought that they might perhaps be identified with the salt lake of Merga (lat. 19° 3′, long. 26° 19′).

Of the various localities which have been suggested as possible sites for the Tortoise Marshes, the only one which I have visited is Merga. I found the salt lake at that place to be very small, covering only some 10 acres; it lies at an altitude of 509 metres above sea-level, and though it is situated in a rather wide shallow depression, I saw no traces of any extensive salt-marshes around the lake, while the configuration of the surrounding country appeared to me to be such that the depression cannot possibly have formed part of a continuous drainage-channel.

I do not know whether it has hitherto been suggested that Kufra Oasis may be the site of the Tortoise Marshes; but on correcting Ptolemy’s figures for the errors in his adopted position for Alexandria and in the length which he assumed for a degree of latitude, I find Kufra is in very much closer agreement with them than any of the places named above. In Ptolemy’s day, even the latitudes of but few places had been astronomically observed (Alexandria, where he himself resided, was supposed by Ptolemy to be in latitude 31°, instead of its true 32° 12′), and as the only method at that time known for astronomically determining differences of longitude was by the observation of eclipses, the number of observed longitudes was smaller still. The process by which Ptolemy deduced his position for the Tortoise Marshes and other places in the interior of Libya was most probably that of first estimating their distances south and west of Alexandria from travellers’ itineraries, then converting these distances into degrees of latitude and longitude, and finally subtracting the differences thus found from the latitude and longitude of Alexandria. But Ptolemy made the great mistake of assuming the length of a degree of latitude (or of equatorial longitude) to be 500 stades, instead of the 700 stades which it really is. Thus all his dead-reckonings resulted in differences of latitude and longitude which were too great in the proportion of seven to five. If we correct Ptolemy’s position for Alexandria, and his dead-reckoning for the erroneous assumption which he made regarding the size of the Earth, as follows:

	Latitude.		Longitude.
Ptolemy’s position for Alexandria	31°		60° 30′
Ptolemy’s position for the Tortoise Marshes	20°		49°
Ptolemy’s difference of lat. and long.	11°		11° 30′
These differences reduced in the proportion of 5 to 7 become, in true degrees and minutes	7° 52′		8° 12′
The true position of Alexandria is	32° 12′		29° 53′
Whence the corrected position for the marshes in our modern system becomes	24° 20′		21° 41′

we get lat. 24° 20′ and long. 21° 41′ for the Tortoise Marshes. Comparing this position with that of Kufra, we find that the latitude agrees very closely with the 24° 14′ which Hassanein observed at Taj, the principal village of that oasis; while the longitude, though more than a degree and a half west of that of the principal village, is almost exactly correct for Taiserbo, the north-western oasis of the Kufra group.

Besides this remarkably close agreement as regards position, Kufra presents several natural features which would tend to support the view that it may be the locality which Ptolemy meant by the “Tortoise Marshes.” Not only is Kufra an extensive tract of relatively low-lying ground with numerous lakes and salt-marshes of very considerable size, but it has distinctly the form of a valley. Hassanein Bey repeatedly refers to it as a valley in his description[43]; Mrs. Forbes also speaks of the “Wadi of Kufra,” and mentions that as a result of a ride westward from Taj the “wadi” was found to have no definite ending to the west.[44] What more natural, therefore, than that some ancient traveller should have imagined that Kufra was a series of marshes formed by a river coming from the south-west? And is it not possible that Ptolemy, in endeavouring to piece together the scraps of information he could get, may have mistakenly inferred that this river ultimately reached the Mediterranean, and also have confused the account of it with those of other streams further south, which may have been branches of the Niger?

11. The Sand-Dunes.

The sand-dunes of the Libyan Desert have been the subject of careful studies by Mr. Beadnell and Mr. Harding King, and practically all that was known concerning them up to the outbreak of the Great War is contained in the excellent papers by them which were read and discussed at meetings of the Royal Geographical Society in 1910, 1916, and 1918.[45]

With the commencement of the Senussi campaign in Egypt in 1915, the sand-dunes sprang into new and unexpected importance, from the fact that they formed one of the principal hindrances to the free movement of troops, and more especially of military motor transport, across the desert. Every line of dunes of any considerable extent had now to be carefully mapped and examined for possible car-passages through it; the result was to add greatly to our knowledge of the distribution and extent of these remarkable features of the Libyan Desert, especially in the northern parts, where many long lines of dunes were found of which the existence was previously unsuspected by geographers, though they are familiar landmarks to the Bedouin of the region, and all except the smallest bear Arabic names. After the cessation of hostilities, interest in the distribution of dunes was maintained, because of the increasing use of motor-cars in place of camels for transport across the desert; and this circumstance has recently led to light being thrown on the distribution of the dunes in those southern regions which lay outside the field of operations during the war.

It is rather curious to note that although the sand-dunes form the greatest obstacle to motor-cars in the desert, yet it is chiefly by means of motor-cars that the true extent and distribution of the sand-dunes has been ascertained. Dunes are the most difficult of all desert features to map properly by ordinary reconnaissance methods with camel transport. Their smooth outlines provide no points on which intersections can be made, and no survey marks put on them will remain in place for more than a few hours, or at most a few days; they occur mostly in nearly level country, where it is impossible to find a station whence they can be overlooked; the absence of shadows on them renders it impossible to say whether one is looking at a single line of dunes, or at several lines, miles apart, one behind the other in echelon. The only sure way of mapping dunes is to traverse their sides along their entire length, and this is impracticable with camels owing to the enormous distances which would have to be covered without water. But with motor-cars one can run alongside them at 40 kilometres an hour instead of the camel’s 4, and their distribution can thus be rapidly and easily ascertained.

On the map I have shown the distribution of sand-dunes as far as it is known at the present day. A comparison of this latest map with that given by Mr. Beadnell in his paper of 1910 will show how large is the number of dunes discovered and mapped in more recent years; and it will be noticed how pronounced is the constancy of the general direction of all the lines in the Egyptian portion of the Libyan Desert. Though no new lines of dunes comparable in extent with the great Abu Moharik belt have been added to the map, some of the newly discovered lines are of very considerable length, and they show an even more remarkable ratio of length to breadth than the Abu Moharik belt; the Ramak dunes, for instance, extend from near Moghara south-south-eastwards in a straight line for more than 100 kilometres, with only a single small break, and their width nowhere exceeds 1 kilometre. Moreover, the linear arrangement in a direction from about 20° west of north to about 20° east of south is even more pronounced in the newly discovered northern dunes than it is in those previously known farther south. In the “great sand sea” of Rohlfs, to the west of Dakhla and Farafra, the same directional arrangement of the individual lines of dunes is very noticeable, at least in the part near “Regenfeld” where I have penetrated it; and I found the same general direction to hold for most of the lines of dunes which I crossed on the way from Terfawi to Owenat. During the war I received a number of reports of dunes extending nearly east and west, often in curved lines; but investigations on the spot showed practically all these reports to be erroneous, the commonest mistake having been that of sighting different lines of dunes from different places under the impression that they were a single one.

The general south-south-easterly direction which is so marked a feature of the lines of dunes in the Egyptian portion of the Libyan Desert is not, however, preserved in the extreme south-west of the country. The line of dunes which crosses the Egypt-Sudan frontier in longitude 25° 40′ has a direction of about 30° west of south, and the dunes which extend from near Gebel Arkenu to the west of Gebel Owenat run about 38° west of south. From a point about 80 kilometres to the south-west of Gebel Owenat, the dune-lines make a still further distinct bend to the west, changing their direction by some 20° to about 58° west of south, and this latter direction is maintained by the dunes in the neighbourhood of Sarra Well.[46] According to the maps of Rohlfs and Mrs. Forbes, the dunes in the sandy tract to the north-west of Kufra follow approximately the same bearing as those near Sarra. The suggestion has been made that this swinging round of the direction of the dunes in the south-western part of the Libyan Desert may be due to a deflection of the prevailing wind by the mountain-masses. It is almost certainly caused by differences in the prevailing winds—in fact, the dune-lines probably afford a very exact index to the general wind-direction in the areas where they occur—but I am inclined to think that the regional distribution of atmospheric pressure, rather than local deflection by mountain masses, is the cause of the differences of wind-direction. The dunes, especially those near Sarra and Kufra, extend into localities which seem to me to be too distant from the mountains for the influence of the latter on the prevailing winds to afford a satisfactory explanation.[47]

In certain areas, especially in the south, as for instance on the Arbain road from Kharga to Sheb, around Pottery Hill, and in the region between Owenat and Erdi, there are sand-covered tracts which can hardly be mapped as dunes, because there are no very marked crests. These tracts are easily crossed by camels, but give much trouble with cars, and would probably be classed as dunes by car-drivers; but they are really only vast undulating fields of drifted sand. On the old Darb el Arbain slave road between Kharga and Sheb, one travels for more than 100 miles over drifts of this kind. No footprint of camel marks the track, footprints being obliterated in an hour or two. But one is never in doubt about the way, for it is indicated by the skeletons of countless camels which have perished on the weary march.

As regards the relation between dunes and the relief of the ground, there is a gradual rise in the general level of the country from north to south of about 1 in 1000, so that the southern ends of most of the Egyptian lines of dunes lie at higher altitudes than the northern ends; but I am fairly certain that this slight general inclination has nothing to do with dune-formation. The straightest and cleanest-cut lines of dunes are found in the flattest and most open parts of the desert, over which the wind can sweep without obstruction. When a line of dunes encounters in its southward progression a sudden fall in the ground-level, as for instance at the north end of Kharga Oasis and to the east of Pottery Hill, it usually breaks up; sometimes, as in Kharga, to resume its course as dunes in more open formation, and at other times, as near Pottery Hill, to spread out into a rolling plain of drifted sand. It is not uncommon for the south end of a line of dunes to split up, even on level ground, and the same can be seen, though less often, at the northern ends; but this is probably merely because the dunes are lower near the ends of the lines, and the width of the individual dunes is consequently less.

The great lines of dunes are probably extending southwards at a very slow rate, under the action of the prevailing wind. In this connection it is interesting to note that Arabs always speak of the south end of a line of dunes as its head, and of the north end as its tail; just the opposite of what one would instinctively call them from merely looking at the map, but correct if the dunes are known to have a southward progression. Mr. Beadnell found the average rate of progression of isolated small dunes in Kharga to be 15 metres per annum; but he remarks that large dunes move more slowly than small ones, owing to the greater mass of material to be transported. It is interesting to make a rough guess at the antiquity of the dunes from the length of certain of the longer lines and an estimated rate of progression. The Abu Moharik belt, for instance, has a total length of about 350 kilometres; hence, if its tail has remained stationary and its head has advanced southwards at an average rate of 10 metres a year, some 35,000 years must have been occupied in its formation.

As regards the height of the dunes, the only one I have carefully measured is that close west of Rohlfs’ camp at Regenfeld, which I found to be exactly 30 metres. Some of the dunes in the Abu Moharik and Ramak belts are considerably higher; but I doubt if any of the Libyan dunes rise much over 50 metres above ground-level, except in the “great sand sea” which commences some 14 kilometres to the north-west of Regenfeld, where Jordan estimated them to attain 100 metres or more.[48] The top of a high dune to the south of Melfa Oasis was found by triangulation to be 172 metres above sea; the level of the ground at its foot is unknown, but is probably not very much above the sea-level, so that Jordan’s estimate of the height of some of the dunes in the “sand sea” is probably by no means an exaggerated one.

Much has been written about eddies in connection with dune-formation. The only observations I have made in this connection have been when endeavouring to place temporary marks on dunes to form survey-points; and the observations seem to show that if eddies are artificially created at the top of a dune, a rapid lowering of the dune-crest results. At first I used to thrust a walking-stick or ranging-rod some 2 or 3 feet into the dune-crest. The stick or rod, though quite firmly fixed at first, was always found after a few hours lying halfway down the dune-slope. The same thing happened with tripods firmly pressed down into the crest. When I was carrying out a little triangulation to determine the position of Rohlfs’ cairn at Regenfeld in 1924, I had occasion to place a mark on a high dune, and I used an empty 4-gallon petrol tin filled with sand, bedding it well into the dune; but the next day it had disappeared from the crest, and was found, as the other things had been, halfway down the slope. These observations remind me of Mr. Barclay’s description of the method of dealing with sand-accumulations in the Peruvian Desert,[49] where, as soon as dunes appear and threaten to obstruct a railway, the local inhabitants turn out and scatter pebbles and stones on the dunes, and very soon the dunes are gone, having been carried away owing to eddies produced by the wind around the stones. This device for removing incipient dunes might, I think, be tried along the Kharga Railway, where the screens put up at Mr. Harding King’s suggestion have not been entirely successful in preventing the encroachment of sand on the line. I am the more disposed to think an experiment of this kind might succeed, because I have noticed that at the gap in the Ramak dunes (discovered by Col. Partridge and known during the war as “Partridge Gap”) there are great lumps of silicified wood, which may have led to the gap being formed.

There are two other physical characteristics of dunes which I do not remember having seen mentioned. One of these is their low thermal conductivity. On a hot day, one has only to thrust one’s hand a few inches into a dune to encounter cool sand, while on a bitterly cold day one can warm one’s hands by the same action; thus while the diurnal variation of temperature of the surface of a dune is frequently very great, it is almost insensible at quite a shallow depth in the interior. It follows that diurnal expansion and contraction of the grains composing the dune must be confined to the superficial layers, and the movements thereby caused may materially assist in consolidating the dune, by causing a “packing” of the grains. The other characteristic I have noticed is that the surfaces of the dunes always appear to be harder underfoot in the cool of the early morning than in the heat of the day. This may, of course, be an illusion owing to one’s being less sensitive to fatigue in the early morning, when one’s body has been refreshed by sleep, and when the air-temperature is comfortable; but I have often fancied there is more in it than that, and it may easily be that the “packing” which has taken place owing to the considerable fall of temperature in the night, to say nothing of the action of dew, may have caused a real compacting of the surface layers.

In his paper of 1910, Mr. Beadnell considered that the sand composing the dunes must be derived from arenaceous formations to the north, and he made special mention of the need for a careful survey of the region in which the dunes originate. A reference to the map will show how correct was Mr. Beadnell’s surmise, at least in the case of the Egyptian dunes which he had studied. The dunes originate in the great depressions which stretch from Siwa to the Wadi Natrun, where, as Mr. Beadnell inferred, the supply of arenaceous material from the loosely compacted Miocene and Oligocene beds is almost inexhaustible. We have here one of the clearest possible examples of the powerful action of the wind as an excavating and transporting agent. The Qattara depression has been excavated, largely by the wind, to a depth of over 130 metres below sea-level, and the excavated material has been carried southwards, some of it for nearly 1000 kilometres in distance and more than 500 metres in height, to form the dunes. We see not only the excavated hollow, but also the transported and piled-up arenaceous material removed from it; and the total quantity of rock thus removed and re-deposited must amount to hundreds of cubic kilometres.[50]

It is noticeable, however, that to the north of the “great sand sea” of Rohlfs, which extends for an unknown distance westward, the known depressions are much shallower; and yet the total quantity of sand in this region is probably greater than that of all the more easterly dunes put together. The country between Jarabub and Jalo is however almost unexplored, though it is known to be very sandy and to be the theatre of violent sandstorms. Whether similar deep depressions to that of Qattara have been excavated along, or to the north of, the Jarabub-Jalo road, or whether the area of loosely compacted arenaceous rocks exposed there is so enormous as to have yielded such a quantity of sand without any very deep depressions being formed, is a point that can only be cleared up by further exploration in Cyrenaica.

We now come to the most puzzling of all the questions connected with the sand-dunes, namely, why do they persist as sharply defined long narrow lines in certain places, while the ground elsewhere is almost absolutely free from sand? As Mr. Harding King puts it, the dunes appear to have a curious power of collecting all the sand in the neighbourhood.[51] The sheltering effect of ripples in the dunes, and a “shepherding” effect of winds a point or two off the normal, have been suggested as possibly affording some explanation of this apparent collecting-power,[52] but neither of these suggestions appears to be adequate. I think a more direct explanation is possible; namely, that the dunes really have the power of attracting sand from the air, and that this power is due to the well-known law of attraction between a conducting surface at zero potential and an electrically charged body. That air-borne sand-grains in the neighbourhood of a dune may carry electric charges was shown by Mr. Harding King’s experiments, recorded in his paper of 1916, and I have since found sand-grains to carry electric charges of high voltage during sandstorms in Cairo. As to how the grains become charged with electricity, it can hardly be due to friction against each other, for the sand is remarkably uniform in composition, consisting generally of rounded grains of almost pure silica coated with a thin translucent film of ferric oxide,[53] and for the production of frictional electricity it is generally regarded as necessary that the two bodies which are rubbed together should be of dissimilar materials;[54] nor is it likely that they are electrified by friction with the air, for they travel with it. But if the particles are carried into an upper air-stratum, they will gradually acquire its potential by conduction; and the ordinary potential gradient in the air in Egypt is over 100 volts per metre of height.[55] Hence a sand-particle carried along for some time in the air at a height of only 20 metres above the ground may become charged to 7 electro-static units of potential, or 2100 volts; and on approaching a dune rising to about that height it will be attracted to the dune, which being in connection with the earth is at zero potential. The attraction will be very small unless the sand-grain approaches very closely to the surface of the dune; from some calculations I have made, a spherical grain of silica half a millimetre in diameter charged to a potential of 2100 volts will not be attracted with a force equal to half its own weight until it approaches within a centimetre of the conducting surface, and the grain will have to approach within 7 millimetres of the surface before the attraction is equal to the weight of the grain. But the important thing to note is that the attraction is independent both of the sign of the charge and of the direction of the wind which carries the particle. Another way of looking at the matter is to consider the probable distribution of the equipotential surfaces in the air about a dune, as in the diagram below, which represents a section taken perpendicular to the direction of the wind and to the axis of the line of dunes: