Chapter VII: Mediæval Geography--Seistan and Afghanistan (2)

The Arabs entered India and occupied the Indus valley through Makran, and throughout their writings we find, strangely, little reference to any of the Indian frontier passes which we now know so well. But in the north and north-west of Afghanistan, in the Seistan and the Oxus regions, they were thoroughly at home both as traders and travellers; and with the assistance of their records we can make out a very fair idea of the general network of traffic which covered High Asia. The destroying hordes of the subsequent Mongol invasions, and the everlasting raids of Turkmans and Persians on the border, have clean wiped out the greater number of the towns and cities mentioned by them, and the map is now full of comparatively modern Turkish and Persian names which give no indication whatever of ancient occupation. There are, nevertheless, some points of unmistakable identity, and from these we can work round to conclusions which justify us in piecing together the old route-map of Northern Afghanistan to a certain extent. This is not unimportant even to modern geographers. The roads of the old khafila travellers may again be the roads of modern progress. We know, at any rate, that the Arabs of 1000 years ago were much the same as the Arabs of to-day in their manners and methods. Their routes were camel routes, not horse routes, and their day's journey was as far as a camel could go in a day, which was far in the wider and more waterless spaces of desert or uninhabited country, and very much shorter when convenient halting-places occurred. These Arab itineraries are bare enumeration of place-names and approximate distances. As for any description of the nature of the road or the scenery, or any indication of altitude (which they possibly had no means of judging), there is not a trace of it; and the difficulties of transliteration in place-names are so great as to leave identification generally a matter of mere guesswork.

One of the most interesting geographical centres from which to take off is Herat, and it may be instructive to note what is said about Herat itself and its connections with the Oxus and Seistan. Herat, says Idrisi, is "great and flourishing, it is defended inside by a citadel, and is surrounded outside by 'faubourgs.' It has many gates of wood clamped with iron, with the exception of the Babsari gate, which is entirely of iron. The Grand Mosque of the town is in the midst of the bazaars.... Herat is the central point between Khorasan, Seistan, and Fars." Ibn Haukel (tenth century) mentions a gate called the Darwaza Kushk, which is evidence that Kushk was of importance in those days, though no separate mention is made of that place; and he adds that the iron gate was the Balkh gate, and was in the midst of the city. The strategical value of the position was clearly recognized.

That grand edifice, the Mosalla, with its mosques and minars, which stood outside the walls of Herat and was the glory of the town in 1883 (when it was destroyed in the interests of military defence), had no previous existence in any other form than that which was given it when it was built in the twelfth century.

Both Ibn Haukel and Idrisi mention a mountain about six miles from Herat, from which stone was taken for paving (or mill-stones), where there was neither grass nor wood, but where was a place (in Ibn Haukel's time, but not mentioned by Idrisi) "inhabited, called Sakah, with a temple or Church of Christians." Idrisi says this mountain was "on the road to Balkh, in the direction of Asfaran." This would seem to indicate that Asfaran, "on the road to Balkh," must be Parana (or Parwana), an important position about a day's march north of Herat. Ibn Haukel says nothing about the road to Balkh, which can only be northward from Herat, but merely mentions that the mountain was on the desert or uncultivated side of Herat, where was a river which had to be crossed by a bridge. This could only be _south_ of Herat. Asfaran is also stated to be on the road to _Seistan_ and to have had four places dependent on it, one of which was Adraskand; and the route to Asfaran from Herat is further described as three days' journey (Idrisi). Ibn Haukel also describes Asfaran as possessing four dependent towns, and places it between Farah and Herat, or _south_ of Herat. As Adraskand[6] is a well-known place between Herat and Farah, we must assume that this is either another Asfaran, or that Idrisi has made a mistake in copying Ibn Haukel. It might possibly be represented by Parah, twenty-five miles south-west of Herat, although the limited area of cultivable ground around renders this unlikely. Subzawar would indicate a far more promising position for an important trade centre such as Asfaran must have been, and would accord better with the three days' journey from Herat of Idrisi, or the itinerary from Farah given by Ibn Haukel, while the extensive ruins around testify to its antiquity. Asfaran was almost certainly Subzawar.

Considering the interest which may once again surround the question of communications from Herat to India, it may be useful to point out that the route connecting Farah with Herat 1000 years ago remains apparently unchanged. The bridge called the Pul-i-Malun, over the Hari Rud, must have been in existence then, and there was another bridge over the Farah River one day's march below Farah, on the highway between Herat and Seistan. To the west of Herat, on the ruin-strewn road to Sarakhs, we have one or two interesting geographical propositions.

Idrisi mentions a place possessing considerable local importance "before Herat had become what it is now," about 9 miles west of Herat, called Kharachanabad. This can easily be recognized in the modern Khardozan, a walled but very ancient town, which is about 8½ miles distant. Between it and the walls of the city there is now no place of importance, nor does it appear likely, for local reasons, that there ever could have been any. Another place, called Bousik, or Boushinj (Pousheng, according to Ibn Haukel), is said to be half the size of Sarakhs, built on the flat plain 6 miles distant from the mountains, surrounded with walls and a ditch, with brick houses, and inhabitants who were commercial, rich, and prosperous, and "who drink the water of the river that runs to Sarakhs." This indicates a site on the banks of the Hari Rud. The only modern place of importance which answers this description is the ancient town of Zindajan, which is about 6 miles from the mountains, and which (according to Ferrier) still bears the name of Foosheng. This name, however, was not recognized by the Afghan Boundary Commission. "To the west of Bousik are Kharkerde and Jerkere. One reckons two days' journey to this last town, which is well populated, smaller than Kuseri, but where there is plenty of water and cultivation. From Jerkere to Kharkerde is two days' journey." These two places are obviously on the road to Nishapur. There is an ancient "haoz," or tank, below the isolated hill of Sangiduktar, near the Persian frontier, which might well represent what is left of Jerkere, and Kharkerde lies beyond it, on the road to Rue Khaf (itself a very ancient site, probably representing Rudan), near Karat. Another place which has a very ancient and troubled history is Ghurian, about thirteen miles west of Zindajan. This is readily identified as the Koure of Idrisi, which is described as twelve miles from Bousik, on the left of the high-road westward, and about three miles from it.

This corresponds exactly with Ghurian, and proves that the high-road has retained its position through ages. Koure is described as an important town, but there is no mention of walls or defences. Another place, second only in importance to Bousik, is Kouseri. It is in fact said to be equal to Bousik, and to possess "running water and gardens." There can be little doubt that this is Kuhsan (or Kusan), one of the most important towns of the Herat valley.

This great high-road, intersecting the plain from the north-west gate of the city, is a pleasant enough road in the spring and summer months. For a space it runs singularly free from crowded villages and close cultivation, and the tread of a horse's hoof is amongst low-growing flowers of the plain, a dwarf yellow rose with maroon centre being the most prominent. Then, as one skirts the Kaibar River as it runs to a junction with the Hari Rud from the northern hills, cultivation thickens and villages increase.

The road next hugs the Hari Rud, and, passing the high-walled town of Zindajan to the south, runs, white and even and hard, with the scarlet and purple of poppies and thistles fringing it, between long gravel slopes of open dasht and the twin-peaked ridge of Doshak, to Rozanak and Kuhsan. Kuhsan is a little to the south of the Kaman-i-Bihist. It was here that the British Commission of the Russo-Afghan Boundary gathered in the late autumn of 1884, one half from England and the other half from India. The drab squares of the cultivated plain were bare then, in November, and the poplars on the banks of the river were scattering yellow leaves to the blasts of the bitter north-west winds of autumn which sweep through Khorasan and Seistan, making of life a daily burden. But there came a marvellous change in the spring-time, when the world was scarlet and green below and blue above; when the sand-grouse began to chatter through the clear sky; then Kaman-i-Bihist (the bow of Paradise) justified its name. The old Arab of the trading days who wandered northward to Sarakhs must have loved this place.

Stretching Sarakhs-ward are the hills, rocky and broken along the river edge, but gradually giving place eastward to easy rounded slopes, softened by rain and snow, and washed into smooth spurs with treacherous waterways between which become quagmires under the influence of a north-western "shamshir." The extraordinary effect of denudation which yearly results from the heavy rain-storms which are so frequent in spring and early summer in these hills must have absolutely changed their outlines during the centuries which have elapsed since the Semitic trader trod them. A summer storm-cloud charged with electricity may burst on their summits, and the whole surface of the slopes at once becomes soft and pulpy. Mud avalanches start on the steeper grades and carry down thousands of tons of slimy detritus in a crawling mass, and spread it out in fans at their feet. It is not safe to say that the modern passes of the Paropamisus north of Herat--the Ardewan and the Babar--were the passes of mediæval commerce, although the Ardewan is marked by certain wells and ruined caravanserais which show that it has long been used. It seems possible that these passes may have shifted their positions more than once. There was undoubtedly a well-trodden route from Bousik, which carried the traveller more directly to Sarakhs than would the Ardewan or even the Chashma Sabz Pass. This road followed the river more closely than any railway ever will. It turned the river gorge to the east, and probably passed through the hills by the Karez Ilias route, which runs almost due north to Sarakhs. The only certain indication which we can find in Idrisi is the statement that the "silver hill" (_i.e._ the hill of the silver mine) is on the road from Herat to Sarakhs. The Simkoh (silver hill) is still a well-known feature in the broken range of the Paropamisus, near that route. But it is difficult after centuries of disturbing forces, natural and artificial, to identify the sites of many of the towns and markets mentioned by Idrisi, who places Badghis to the west of Bousik, and gives the "silver hill" as one of its "dependencies." There were two considerable towns, Kua (or Kau) and Kawakir, said to have been near the silver hill, and there is mention of a place called Kilrin in this neighbourhood. Probably the ruins at Gulran represent the latter, but Kua and Kawakir are not identified. Gulran was one of the most fascinating camps of the Afghan Boundary Commission. On the open grass slopes stretching in gentle grades northward, bordered by the line of red Paropamisan cliffs to the south and west and by the open desert stretching to Merv on the north, it was, during one or two early months of the year, quite an ideal camping-ground.

It was here that the wild asses of the mountains made a raid on the humble four-footed followers of the Commission, and signified their extreme disgust at the free use which was made of their feeding-grounds; thus witnessing to the condition of primeval simplicity into which that once populous district had subsided after centuries of border raid and insecurity. The remains of an old karez, or underground irrigation channel, not far north of Gulran, testified to a former condition of cultivation and prosperity.

From Gulran (which is connected with the Herat plains directly by the pass called Chashma Sabz) roads stretch northwards and north-eastwards, without obstacle, to the open Turkistan plains, where ancient sites abound. Idrisi's indications, however, are but a very uncertain foundation for identifying most of them. The "dependencies" of Badghis are said to be Kua, Kughanabad, Bast, Jadwa, Kalawun, and Dehertan, the last place being built on a hill having neither vegetation nor gardens; but "lead is found there, and a small stream."

The great trade centres of Turkistan, north of the Paropamisus, in mediæval days were undoubtedly near Panjdeh, at the confluence of the Kushk and Murghab rivers, and at Merv-el-Rud, or Maruchak. Two or three obvious routes lead from the passes above Kaman-i-Bihist, or above Herat, to Panjdeh and Maruchak. One is indicated by the drainage of the Kushk River, and the other by that of the Kashan, which is more or less parallel to the Kushk to the east of it, with desolate Chol country in between. From Herat the most direct route to Panjdeh and Merv is by the Babar Pass, or by Korokh, the Zirmast Pass, and Naratu. Korokh (Karuj) is mentioned both by Ibn Haukel and Idrisi as being situated three marches from Herat, surrounded by entrenchments, and in the "gorge of mountains," with gardens and orchards and vines. The Korokh of to-day is between the mountains, but only some twenty-five miles from Herat. This modern Korokh has, however, many evidences of great antiquity, and it is on the high-road to an important group of passes leading past Naratu to Bala Murghab and Maruchak. The most remarkable feature about Korokh is a grove of pine trees closely resembling the "stone" pine of Italy, which mass themselves into a dark blotch on the landscape and mark Korokh in this treeless country most conspicuously. There are no other trees of the same sort to be found now in this part of Asia, but I was told that they once were abundant in the Herat valley, which renders it possible that the "arar" trees, mentioned by Ibn Haukel as a peculiar source of revenue to Bousik, may have been of this species. Naratu, again, is very ancient, and its position among the hills (for it is a hill-fortress) seems to identify it with Dahertan. Undoubtedly this was one of the most important of the old routes northward, and it is a route of which account should be taken to-day.

In the Kuskh River more than one ancient site was observed, Kila Maur being obviously one of the most important, whilst in the Kashan stream there were evidences of former occupation at Torashekh and at Robat-i-Kashan. Whilst there is a general vague resemblance between the names of certain old Arab towns and places yet to be found in the Herat valley and Badghis, it is only here and there that it has been possible to identify the precise position of a mediæval site. The dependencies of Badghis, enumerated by Idrisi, require the patient and careful researches of a Stein to place them accurately on the basis of such vague definitions as are given. We are merely told that Kanowar and Kalawun are situated at a distance of three miles one from the other, and that between them there is neither running water nor gardens. "The people drink from wells and from rain-water. They possess cultivated fields, sheep, and cattle." Such a description would apply excellently well to any two contiguous villages in the Chol country anywhere between the Kushk and the Kashan. Those rolling, wave-like hills, with their marvellous spread of grass and flowers in summer, and their dreary, wind-scoured bareness in winter, are excellent for sheep and cattle at certain seasons of the year; but water is only to be found at intervals, and there are much wider distances than three miles where not even wells are to be found.

Writing again of Herat, Idrisi says that, starting towards the east in the direction of Balkh, one encounters three towns in the district of Kenef: Tir, Kenef, and Lakshur; and that they are all about equally distant, it being one day's journey to Tir, one more to Kenef, and another to Lakshur (Lacschour). Tir is a rich town where the "prince of the country" resides, larger than Bousik, full of commerce and people, with brick-built houses, etc. Kenef is as large, but more visited by foreigners; and Lakshur is equal to either. They are all of them big towns of commercial importance, Lakshur being bounded on the west by the Merv-el-Rud province, of which the capital is Merv-el-Rud.

Assuming for the present that Maruchak, on the Murghab, represents Merv-el-Rud (Merv of the River), where are we to place these three important sites, so that the last shall be east of the Maruchak province and only three days' journey from Herat? The distance from Herat to Maruchak is not less than 150 miles, and it is called by Idrisi a six days' journey. Starting towards the east can only refer to the Balkh route already referred to, _i.e._ _via_ Korokh and the Zirmast Pass. It cannot mean the Hari Rud valley, for that leads to Bamian rather than Balkh. By the Korokh route, however, it is possible to follow a more direct line to Balkh than any which would pass by Maruchak or Bamian. There is on this route, east of Naratu and south-east of Maruchak, a place called Langar which might possibly correspond to Lakshur, and it is not more than 70 to 80 miles from Herat. From Langar there is an easy pass leading over the Band-i-Turkistan more or less directly to Maimana and Balkh, and it seems probable that this was a recognized khafila route. Tir is an oft-repeated name in the Herat district. The river itself was called Tir west of Herat, and there is the bridge of Tir (Tir-pul) just above Kuhsan. The mountains, again, to the north-east are known as Tir Band-i-Turkistan, and the Tir mentioned as on the road to Balkh must certainly have been east of Herat. Of Kenef I can trace no evidence. It must have been close to Korokh.

That this route, through the Korokh valley and across the water-parting by the Zirmast Pass to Naratu, was the high road between Herat and Balkh I have very little doubt. It was the route selected for mail service during the winter when the Afghan Boundary Commission camp was at Bala Murghab, on the Murghab River, and it was seldom closed by snow, although the Zirmast heights rise to over 7000 feet, and the Tir Band-i-Turkistan (which represents the northern _rebord_ or revetment of the uplands which contain the Murghab drainage) cannot be much less. The intense bitterness of a Northern Afghan winter is more or less spasmodic. It is only the dreaded shamshir (the "scimitar" of the north-west) which is dangerous, and travelling is possible at almost every season of the year. The condition of the mountain ways and passes immediately above Bala Murghab is not that of steep and difficult tracks across a rugged and rocky divide. In most cases it is possible to ride over them, or, indeed, off them, in almost any direction; but as these mountains extend eastward they alter the character of their crests. From Herat to Maruchak this is not, however, the direct road; the Kushk River, or the Kashan, offering a much easier line of approach.

All our investigations in 1884 tended to prove beyond dispute that Maruchak represents the famous old city of Merv-el-Rud, the "Merv of the River," to which every Arab geographer refers. Sir Henry Rawlinson sums up the position in the Royal Geographical Society's _Proceedings_ (vol. viii.), when he points out that there were two Mervs known to the ancient geographer. One is the well-known Russian capital in trans-Caspia, the "Merv of the Oasis," a city which, in conjunction with Herat and Balkh, formed the tripolis of primitive Aryan civilization. It was to this place that Orodis, the Parthian king, transported the Roman soldiers whom he had taken prisoners in his victory over Crassus, and here they seemed to have formed a flourishing colony.

Merv was in early ages a Christian city, and Christian congregations, both Jacobite and Nestorian, flourished at Merv from about A.D. 200 till the conquest of Persia by the Mahomedans. Merv the greater has as stirring a history as any in Asia, but Merv-el-Rud, which was 140 miles south of the older Merv, is altogether of later date. This city is said to have been built by architects from Babylonia in the fifth century A.D., and was flourishing at the time of the Arab invasion. All this Oxus region (Tokharistan) was then held by a race of Skytho-Aryans (white Huns) called Tokhari or Kushan, and their capital, Talikhan, was not far from Maruchak. Now, Merv-el-Rud is the only great city named in history on the Upper Murghab, above Panjdeh, before the end of the fourteenth century A.D. After that date, in the time of Shah Rokh (Timur's son), the name Merv-el-Rud disappears, and Maruchak takes its place in all geographical works, the inference being that, Merv-el-Rud being destroyed in Timur's wars, Maruchak was built in its immediate neighbourhood. This surmise of Rawlinson's is confirmed by the appearance of Maruchak, which is but an insignificant collection of inferior buildings surrounded by a mud wall, with a labyrinth of deep canal cuttings in front of it and a rough irregular stretch of untilled country around. Merv-el-Rud must have been a much greater place.

There are, however, abundant evidences of grass-covered ruins, both near Maruchak and at the junction of the Chaharshamba River with the Murghab some 10 miles above Maruchak. Sir Henry Rawlinson points out the strategic value of this point, as the Chaharshamba route leads nearly straight into the Oxus plains and to Balkh. At the point of the junction of the two rivers the valley of the Murghab hardly affords room enough for a town of such importance as we are led to believe Merv-el-Rud to have been, even after making all due allowance for Oriental exaggeration. It is only about Maruchak that the valley widens out sufficiently to admit of a large town. It seems probable, therefore, that the site of Maruchak must be near the site of Merv-el-Rud, although it does not actually command the entrance to the Chaharshamba valley and the road to Afghan Turkistan.

On this road, some 30 miles from the junction of the rivers, there is to be seen on the slopes which flank the southern hills, the jagged tooth-edged remains of a very old town (long deserted) which goes by the name of Kila Wali. It is here, or close by, that the Tochari planted their capital Talikan, at one time the seat of government of a vast area of the Oxus basin. There is, however, another Talikan[7] in Badakshan to the east of Balkh, and there are symptoms that some confusion existed between the two in the minds of our mediæval geographers. Ibn Haukel writes of Talikan as possessing more wholesome air than Merv-el-Rud, and he refers to the river running between the two. This is evidently in reference to the capital of Tocharistan at Kila Wali. Again when he writes of Talikan as the largest city in Tocharistan, "situated on a plain, near mountains," he is correct enough as applied to Kila Wali, but this has nothing to do with Andarab and Badakshan with which we find it directly associated in the context.

On the other hand the Talikan in Badakshan was one of a group of important cities whose connection with India lay through Andarab and the northern passes of the Hindu Kush. Between Maruchak and Panjdeh, along the banks of the Murghab, are ruins innumerable, the sites of other towns which it is impossible to identify with precision. There can be little doubt, however, that the remains of the bridge which once spanned the river at a point between Maruchak and Panjdeh marked the site of Dizek (or Derak, according to Idrisi), which we know to have been built on both sides of the river, and that Khuzan existed near where Aktapa now is (_i.e._ near Panjdeh). The name Dizek is still to be recognized, but it is applied to a curious sequence of ancient Buddhist caves which have been carved out of the cliffs at Panjdeh, and not to any site on the river banks.

The confusion which occasionally exists between places bearing the same name in mediæval geographical annals is very obvious in Idrisi's description of Merv. The greater Merv (the Russian provincial capital) is clearly mixed up in his mind with the lesser Merv when, in describing the latter, he says that Merv-el-Rud is situated in a plain at a great distance from mountains, and that its territory is fertile but sandy; three grand mosques and a citadel adorn an eminence and water is brought to it by innumerable canals, all of which is applicable to Merv but not to Merv-el-Rud. He then continues with a description of the greater Merv, which is quite apropos to that locality, and makes it clear incidentally that Khiva (not Merv) represents the ancient Khwarezm. Again, he enumerates towns and places of Mahomedan origin which are "dependent" on "Merv." Amongst them we find Mesiha, a pretty, well-cultivated place one day's journey to the west of Merv; Jirena (Behvana), a market-town 9 miles from Merv, and 3 from Dorak (? Dizek), a place situated on the banks of the river; then Dendalkan, an important town two days from Merv on the road to Sarakhs; Sarmakan, a large town to the left of Dorak and 3 miles farther, Dorak being situated on the banks of the river at 12 miles from Merv in the direction of Sarakhs; Kasr Akhif (or Ahnef), a little town at one day's distance from Merv on the road to Balkh; Derah, a small town 12 miles from Kasr Ahnef where grapes were abundant. Here, says Idrisi, the river divides the town in two parts which are connected by a bridge. It is quite impossible to straighten out this geographical enumeration, unless we assume that it refers to Merv-el-Rud and not to Merv. Then Mesiha becomes a possibility, and might be looked for among the ruined sites on the Kushk River--possibly at Kila Maur. Dorak, at 12 miles from Merv in the direction of Sarakhs, and Dendalkan at two days' journey in the same direction, would still be on the river banks. Kasr Ahnef we know to have been built after the Arab invasion in the valley of the Murghab, about 12 miles from Khuzan (identified by Rawlinson with Ak Tepe) and 15 from Merv-el-Rud, and must have been situated near the Band-i-Nadir, where the desert road to Balkh enters the hills. Ak Tepe must once have been a place of great importance, both strategically (as it commands the position of the two important highways southward to Herat, the Kushk and the Murghab valleys) and commercially. But apparently its importance did not survive to Arab times. Dendalkan was certainly near Ak Tepe.

In making our surveys of this historic district it was exceedingly difficult to associate the drab and dreary landscape of this Chol (loess) country and its intersecting rivers with such a scene of busy commercial life as the valleys must have presented in Arab times. The Kushk is at best a "dry" river, as its name betokens, an unsatisfactory driblet in a world of sandy desolation. Reeds and thickets hide its narrow ways, and it is only where its low banks recede on either hand as it emerges into the flat plains above Panjdeh that there is room for anything that could by courtesy be called a town. The Murghab River shows better promise.

Below Maruchak, where towns once crowded, it widens into green spaces, and the multiplicity and depth of the astonishing system of canals which distribute the waters of the river on its left bank leave no room to doubt the strength of the former population that constructed them. Where the pheasants breed now in myriads, in reedy swamps and scrubby thickets, there may lie hidden the foundations of many an old town with its caravanserais, its mosques, and its baths. The economic value of the Murghab River is still great in Northern Afghanistan. No one watching the sullen flood pouring past Bala Murghab in the winter time and looking up to the dark doors of the mountains from whence it seems to emerge, could have any idea of the wealth and fertility and the spread of its usefulness which is to be found on the far side of those doors. From its many cradles in the Firozkohi uplands to its many streamlets reaching out round Merv and turning the desert into a glorious field of fertility, the Murghab does its duty bravely in the world of rivers, and well deserves all that has ever been written in its praise by past generations of geographers.

Amongst the many high-roads of Northern Afghanistan which are mentioned by the Arab writers, none is more frequently referred to than the road from Herat to Balkh, _i.e._ to Afghan Turkistan. Intervening between Herat and Afghan Turkistan there is immediately north the easy round-backed range called by various names which have been lumped under the term Paropamisus, down the northern slopes of which the Kushk and Kashan made a fairly straight way through the sea of rounded slopes and smooth steep-sided hills which constitute the Chol. But this range is but an extension of the southern rampart of the Firozkohi upland, which forms the upper basin of the Murghab and overlooks the narrow valley of the Hari Rud.

The northern rampart or buttress of that upland is the Tir Band-i-Turkistan, the western flank of which is turned by the Murghab River as it makes its way northward. So that there are several ways by which Afghan Turkistan may be reached from Herat. Setting aside the Hari Rud route to Bamian or Kabul, which would be a difficult and lengthy detour for the purpose of reaching Balkh, there is the route we have already mentioned _via_ Korokh, Naratu, and Langar, and thence over the Band-i-Turkistan, or down the Murghab. But there is another and probably the most trodden way, _via_ the Kashan to the Murghab valley at the junction of the Chaharshamba River, and up that river to the divide at its head, passing over into the Kaisar drainage, and so, either to Andkhui and the Oxus, or to Maimana and Balkh. This was the route made use of generally by the Russo-Afghan Boundary Commission, and the existence of ancient tanks (called "Haoz") and of "robats" (or halting-places) at regular intervals in the Kashan valley, testifies to its use at no very ancient date.

The entrance to the Chaharshamba valley is very narrow, so narrow as to preclude the possibility of any large town ever having occupied this position; but it opens out as one passes the old Kila Wali ruins where there is ample space for the old capital of Tocharistan to have existed. On the north, trailing streams descend from the Kara Bel plateau (a magnificent grass country in summer and a cold scene of windy desolation in winter), and their descent is frequently through treacherous marshes and shining salt pitfalls, making it exceedingly difficult to follow them to the plateau edge. To the south are the harder features of the Band-i-Turkistan foothills, the crest of the long black ridge of this Band being featureless and flat, as is generally the case with the boundary ridges and revetments of a plateau country. Over the Chaharshamba divide (at about 2800 feet) and into the Kaisar drainage is an introduction to a country that is beautiful with the varied beauty of low hill-tops and gentle slopes, until one either by turning north, debouches into the flat desert plains of the Oxus at Daolatabad, or continuing more easterly, arrives at Maimana, the capital of the little province of Almar, the centre of a small world of highly cultivated and populous country, and a town which must from its position represent one or other of the ancient trade centres mentioned by Idrisi. Here we leave behind the long lines of Turkman kibitkas looking like rows of black bee-hives in the snow-spread distance, and find the flat-roofed substantial houses of a settled Uzbek population, with flourishing bazaars and a general appearance of well-being inside the mud walls of the town.

Idrisi writes that Talikan is built at the foot of a mountain which is part of the Jurkan range (Band-i-Turkistan), and that it is on the "paved" route between Merv and Balkh. This at once indicates that route as an important one compared with other routes (there being a desert route across the Karabel plateau from near Panjdeh in addition to those already mentioned), although there is no sign of any serious road-making to be detected at present. Sixty miles from Talikan, on the road to Balkh, Idrisi places Karbat, a town not so large as Talikan but more flourishing and better populated. The distance reckoned along the one possible route here points to Maimana, which is just 60 miles from Talikan, but there is no other indication of identity. Karbat was a dependency of the province of Juzjan (or Jurkan, probably Guzwan), and 54 miles to the east of it was the town of Aspurkan, a small town, itself 54 miles from Balkh. Now Balkh, by any possible route, is at least 130 to 140 miles from Maimana, but if we assume Aspurkan to have been just half-way (as Idrisi makes it) between Maimana and Balkh, we find Sar-i-pul (a small place indifferently supplied with water, and thus answering Idrisi's description of Aspurkan) almost exactly in that position. In support of this identification of Aspurkan with Sar-i-pul there is the name Aspardeh close to Sar-i-pul. Other places are mentioned by Idrisi as flourishing centres of trade and industry in this singularly favoured part of Afghanistan, where the low spurs and offshoots of the Band-i-Turkistan break gently into the Oxus plains. He says that Anbar, one day's march to the south-west of Aspurkan, was a larger place than Merv-el-Rud, with vineyards and gardens surrounding it and a fair trade in cloth. There, both in summer and winter, the chief of the country resided. Two days from Aspurkan, and one from Karbat, was the Jewish colony of Yahudia, a walled town with a good commercial business. This colony is also mentioned by Ibn Haukel as situated in the district of Jurkan. From Yahudia to Shar (a small town in the hills) was one day's march. The main road south-west from Sar-i-pul has probably remained unchanged through the centuries. It runs to Balangur (? Bala Angur) and Kurchi, the former being 10 miles and the latter 30 from Sar-i-pul. Either might represent the site of Anbar. Twenty miles from Kurchi is Belchirag, and Belchirag is about 25 from Maimana. It would thus represent the site of the ancient Yahudia fairly well, whilst 25 to 30 miles from Belchirag we find Kala Shahar, a small town in the mountains, still existing. Jurkan is described as a town by Idrisi (and as a district by Ibn Haukel), built between two mountains, three short marches from Aspurkan, and Zakar is another commercial town two marches to the south-east. I should identify Jirghan of our maps with Jurkan, and Takzar with Zakar.

All this part of Afghan Turkistan is rich in agricultural possibilities. The Uzbek population of the towns and the Ersari Turkmans of the deserts beyond Shibarghan are all agriculturists, and the land is great in fruit. They are a peaceful people, hating the Afghan rule and praying for British or any other alternative. Shibarghan is an insignificant walled town with a small garrison of Afghan Kasidars; always in straits for water in the dry season. The road between Shibarghan and Sar-i-pul is flat, skirting the edge of the rolling Chol to the east of it. Sar-i-pul itself is but a small walled town in rotten repair, sheltering a few Kasidars and two guns, but no regular Afghan troops. There are a few Jews there who make and sell wine, and a few Peshawur bunniahs (shopkeepers).

From Sar-i-pul a direct road runs to Bamian and Kabul _via_ Takzar to the south-east, and strikes the hill country almost at once after leaving Sar-i-pul. It surmounts a high divide (about 11,000 feet), and crosses the Balkh Ab valley to reach Bamian. There is another route up the Astarab stream leading to Chiras at the head of the Murghab River and into the Hazara highlands; but these were never trade routes except for local purposes. The Hazaras send down to the plain their camel hair-cloth and receive many of the necessities of life in exchange, but there is no through traffic.

The characteristics of the Astarab road are typical of this part of Afghanistan. After passing Jirghan the valley is shut in by magnificent cliffs from 700 to 1000 feet high. The vista is closed by snow peaks to the south, which, with the brilliancy of up-springing crops on the banks of the river, form a picture of almost Alpine beauty. There is, curiously enough, an entire absence of forest in the valley, but blocks of a soft white clay mixed with mica lend a weird whiteness to its walls, dazzling the eye, and making patchwork of Nature's colouring. Snakes abound in great numbers, mostly harmless, but the deadly "asp-i-mar" is amongst them. There is a yellow variety which is freely handled by the Uzbeks, who call this snake Kamchin-i-Shah-i-Murdan. About eight miles beyond Jirghan the Uzbek population ceases. From this point there are only Firozkohis and some few Taimanis who have been ejected from the Hari Rud valley for their misdeeds. They are all robbers by profession, supporting existence by slave trading. They kidnap girls and boys from the Hazara villages of the highlands and trade them to the Uzbeks in exchange for guns, ammunition, and horses. These Taimani robbers are by no means the only slave dealers. Nearly every well-to-do establishment in Afghan Turkistan has one or two Hazara slaves. The prices paid, of course, vary, but 300 krans each was paid for two girls bought in 1883. Expert native authorities have a very high opinion of the handiness of Hazara slave girls. They are good at needlework, turning out most exquisite embroidery, and they are never idle.

The narrowness of the Astarab gorge renders it impossible to follow the river along the whole of its course. The road finally leaves the valley and strikes up to the plateau on its left bank. One remarkably persistent feature in these valley formations is the existence of two plateau levels, or terraces; that immediately overlooking the valley being sometimes 100 feet lower than the second platform which is thrown back for a considerable distance, leaving a broad terrace formation between the line of its cliff edge and that bordering the stream. Occasionally there is more than one such terrace indicating former geologic floors of the valley.

On gaining the plateau level a very remarkable scene opens out--a broad green dasht, or plain, slopes away to a sharp line westwards bordered by glittering cliffs and intersected by the white line of the road. In the midst of this setting of white and green are the remains of what must once have been a town of considerable importance, which goes by the name locally of the Shahar-i-Wairan, or ancient city. Such buildings as remain are of sun-dried brick; there appears to be no indication of the usual wall or moat surrounding this city, and nothing suggestive of a canal or "karez"; nothing, in short, but scattered ruins covering about one and a half square miles. The kabristan (or graveyard) was easily recognizable, and its vast size furnished some clue to the size of the city. All history, all tradition even, about this remarkable place seems lost in oblivion; but a city of such pretensions must have had a fair place in geography from very early times. It seems improbable, however, that it could have been more than a summer residence in its palmy days, for winter at this elevation (nearly 7000 feet) and in such an exposed locality would be very severe indeed. The only indication which can be derived from Idrisi's writings is the reference to the small town in the mountains called Shah (Shahar) one day's march from the Jewish colony of Yahudia. As already explained there is a Kila Shahar some 25 to 30 miles from Yahudia (if we accept the position of Belchirag as more or less representing that place), but the Shahar-i-Wairan is nearer by some 10 miles, and fits better into the geographical scheme. I should be inclined to identify the Shahar-i-Wairan with the ancient Shahar (or Shah) and the Kila Shahar as a later development of the same place. The point, however, to be specially noted about this geographical theory is that there is no route by which camels can pass either over the Band-i-Turkistan or the mountains enclosing the Balkh Ab from the district of Sangcharak southward. The province of Sangcharak, which corresponds roughly to the ancient district of Jurkan (or Gurkan), is rich throughout, with highly cultivated valleys and a dense population, but it is a sort of geographical cul-de-sac.

Communication with the plains of the Oxus and with Balkh (by the lower reaches of the Balkh Ab) is easy and frequent, but there never could have been a khafila road over the rugged plateau land and mountains which divide it from the basin of the Helmund.

From time immemorial efforts have been made to reach Kabul by the direct route from Herat which is indicated by the remarkable lie of the Hari Rud valley. It was never recognized as a trade route, although military expeditions have passed that way; and it has always presented a geographical problem of great interest. From Herat eastwards, past Obeh as far as Daolatyar, there is no great difficulty to be overcome by the traveller, although the route diverges from the main valley for a space. Between Daolatyar and the head of Sar-i-jangal stream (which is the source and easternmost affluent of the Hari Rud) the valley is well populated and well cultivated, with abundant pasturage on the hills. But the winter here is severe. From the middle of November to the middle of February snow closes all the roads, and even after its disappearance the deep clayey tracks are impassable even for foot travellers. In the neighbourhood of a small fort called Kila Sofarak, about 40 miles from Daolatyar, there is a parting of the ways. Over the water-parting at the head of the stream by the Bakkak Pass a route leads into the Yakulang valley, a continuation of the Band-i-Amir, or river of Balkh, which, in the course of its passage through the gorges of the mountains, here forms a series of natural aqueducts uniting seven narrow and deep lakes. Inexpressibly wild and impressive is the character of the scenery surrounding those deep-set lakes in the depths of the Afghan hills.

Near the lakes are the ruins of two important towns or fortresses, Chahilburj, and Khana Yahudi. On a high rock between them are the ruins of Shahr-i-Babar the capital of kings who ruled over a country most of which must have been included in the Hazara highlands, and was probably more or less conterminous with the Bamian of Idrisi. Between the Yakulang and the Bamian valley is a high flat watershed. Looking north-west a vast broken plateau, wrinkled and corrugated by minor ranges, and scored by deep valleys and ravines, fills up the whole space from the mountains standing about the source of the Murghab and Hari Rud to the Kunduz River of Badakshan.

So little is this part of modern Afghanistan known, that it may be as well to give a short description of the existing lines of communication connecting the Oxus plains and Herat with Bamian and Kabul, before attempting to follow out their mediæval adaptation to commercial intercourse.

From Balkh, or Mazar-i-Sharif, or from Deh Dadi (the new fortified position near Mazar) the most direct routes southward either follow the Balkh Ab valley to Kupruk and the Zari affluent, and then crossing the Alakah ridge pass into the river valley again, and so reach the Band-i-Amir and the head of the river at Yakulang; or passing by the Darra Yusuf (a most important affluent of the Balkh River) attain more directly to Bamian. Balkh and Mazar lie close together on the open plain, and about 10 miles to the south of them rises the northern wall of the plateau called Elburz, through which the Balkh River, and other drainage of the plateau, forces its passage. Thus the whole course of the Balkh River, from its head to within a mile or two of Balkh, lies within a deep and narrow ditch cut out from the plateau which fills up the space from the Elburz to the great divide of Central Afghanistan. East and west of the Balkh River the plateau increases in elevation as it reaches southward, culminating in knolls or peaks 12,000 and 13,000 feet high about the latitude 35° 30', and falling gently where it encloses the actual sources of the river. It is this plateau, or uplift, which forms the dominant topographical feature of Northern Afghanistan.

West of the Balkh Ab it is represented by the Firozkohi uplands, which contain the head valleys of the Murghab, bordered on the north by the Tirband-i-Turkistan from the foot of which stretch away towards the Oxus the endless sand-waves of the Chol, and by the highlands of Maimana and Sangcharak, and which trend northward to within a few miles of Balkh. At Balkh its northern edge is well defined by the Elburz, but between Balkh and Maimana it is more or less merged into the great loess sand sea, and its limitations become indefinite. East of the longitude of Balkh it is lost in a distance whither our surveyors have not traced its outlines, but where without doubt it fills a wide area north of the Hindu Kush, determining the nature of the Badakshan River sources and shaping itself into a vast upland region of mountain and deep sunk gully, and generally preserving the same characteristics throughout, till it overlooks the valley of the Oxus. That part of it which embraces the affluents of the Balkh Ab and the Kunduz is described as intensely wild and dreary, traversed by irregular folds and ridges which rise in more or less rounded slopes to great altitudes, hiding amongst them deep-seated valleys and gulches, wherein is to be found all that there is of cultivation and beauty. From above it presents the aspect of a huge drab-coloured, hill-encumbered desert where man's habitation is not, and Nature has sunk her brightest efforts out of sight. These efforts are to be found in the valleys, which are excavated by ages of erosion, steep sided, with precipitous cliffs overhanging, and a narrow green ribbon of fertility winding through the flat floor of them.

Across those dreary uplands, or else wandering blindfold along the bottom of the river troughs, run the roads and tracks of the country; some of them being the roads of centuries of busy traffic. A little apart from the obvious route supplied by the lower course of the Balkh Ab, and more important as leading more directly to the crest of the main divide, is the road from Mazar to the Band-i-Amir district which is practically the best road to Kabul. This strikes on to the plateau and crosses several minor passes over spurs dividing the heads of certain eastern affluents of the Balkh Ab before it drops into the trough of the Darra Yusuf. Following the course of this river, and skirting the towns of Kala Sarkari and Sadmurda, it strikes off from its head over a pass called Dandan Shikan (the "tooth-breaker") into the Kamard valley which runs eastwards into the big river of Badakshan--the Kunduz. From Kamard over three passes into the Saigan--another valley draining deeply eastwards into the Kunduz. From this again, two parallel routes and passes southward connect Saigan with the Bamian depression. Here the river of Bamian also runs east, parallel to Saigan and Kamard (the three forming three parallel depressions in the general plateau land), but meeting an affluent draining from the east, the two join and curve northward into the Kunduz.

This new affluent from the east is important, for it leads over the easy Shibar Pass into the head of the Ghorband valley and to Charikar. Finally, there is the well-travelled route from Bamian, leading southward over the Hajigak Pass into the Helmund valley at Gardandiwal, where it crosses the river and then proceeds _via_ the Unai Pass and Maidan to Kabul. Such is the general system of the Balkh communications with Kabul.

From Tashkurghan, east of Mazar, there are other routes equally important. There is a direct road southward, which starts through an extraordinary defile, where perpendicular walls of slippery rock enclose a narrow cleft which hardly admits the passing of a loaded mule to Ghaznigak and Haibak. From Haibak you may follow up the Tashkurgan River to its head and then drop over the Kara Pass into Kamard at Bajgah, and so to Bamian again; or you may avoid Bamian altogether and striking off south-east from Haibak over the plateau, slip down into the Kunduz drainage at Baghlan, and then follow it to its junction with the Andarab at Dosh. This position at Dosh gives practical command of all the passes over the Hindu Kush into the Kabul basin, for the Andarab drains along the northern foot of the Hindu Kush, and commands the back doors of all passes between the Chapdara (or Chahardar) and the Khawak.

The most trodden route to-day is that which is the most direct between Kabul and Mazar, _i.e._ the route _via_ Bamian and the Darra Yusuf. This is the route taken by the late Amir when he met his cousin Ishak Khan in the field of Afghan Turkistan and defeated him. It is not the route taken by the Afghan Boundary Commission in returning from the same field in 1885. They returned by Haibak and Dosh and deploying along the northern foot of the Hindu Kush, crossed by nearly every available pass either into the Ghorband valley or that of the Panjshir.