History of Iran

The Ethnic of Sakas (Scythians)
By: I. P’iankov


The Sakas have long attracted the attention of scholars, as note a Russian monograph that appeared in 1871. However, new and constantly updated archaeological, linguistic, and other materials concerning them now make it possible to reinterpret the written sources in greater depth and to use information that has gone unnoticed. The present article traces the history of the Sakas as an ethnic community and the relationship of individual groups of Sakas and tribes associated with them to the modern peoples who are their direct descendants. I refer not to the Sakas as defined by linguists and archaeologists but to Sakas proper, i.e., tribes directly designated as such in historical sources.

The following study is based primarily on conclusions drawn in my 1968 article, in which almost all of the available primary data on the Sakas were used to outline the history of ancient descriptions of them. A generally coherent picture emerged.

The present article gives a brief overview of Saka tribes of the Amyrgians, Homodotes, and Komedes, who originally lived in the Farghana valley and the foothills of the northern Pamirs, and also the closely related Kaspians (Kaspirs), who inhabited territories to the south and provided some part of the ethnic substratum of the Sakas.
The Amyrgians
In my earlier article, the ancient written sources on the Saka Amyrgians were examined. The root of the word is the element murg-, mfg-. The Greek form of the name is Amyrgioi, and the Old Persian form, Haumavarga, an obvious reinterpretation, with the corresponding personal names Amorges and (H)omarges. In brief, the article concluded that the Amyrgians were the group of "Sakas proper" in closest proximity to Sogdiana and Bactria. These were the Sakas who clashed with Cyrus and then Alexander and who were first subjects and then allies of the Achaemenid dynasty and subsequently entered into an alliance with Alexander.

Scythians, as depicted by themselves on a steel bowl found in a Scythian grave site near the Black Sea.
Concerning the area in which the Amyrgians lived, one finds that the Persians and Greeks placed them beyond Sogdiana, across the Tanais River or the Jaxartes, because they encountered them after crossing Sogdiana and the Syr Darya in the approximate region of modern Khojend. The Amyrgians were evidently centered somewhere in that direction, across the Syr Darya. According to the account (which is completely reliable geographically) by Chares of Mytilene, who accompanied Alexander on his campaigns, the headquarters of "King Omarg" was located 800 stadia (150 km) from the crossing at the Tanais (frag. 5, Jacoby). That the Amyrgians' domain was not limited to territory on the right bank of the Syr Darya was recognized by the ancient writers when they repeatedly referred to the Amy rgian Sakas as Asiatic Scythians. In addition, Hellanicus (fifth century BCE) referred to them as inhabitants of the "Amyrgian plain [Amyrgion pedion] of the Sakas" (frag. 65, Jacoby). Litvinskii was correct in concluding that the Amyrgian plain referred to all of Farghana and also included the Alai valley.' The inclusion of the latter is based on the fact that the first century CE Qian Han Shu mentions a river valley, the "Migration of the Birds;" located within the Saka territory that is commonly considered to be the Alai valley; in the Chinese term, Herrmann succeeded in identifying an allusion to the same toponym as that reflected in the Greek Amyrgion.

Finally, the same element murg- is found in the modern name of the Murghab River, the lower course of which is called the Bartang. Although Murghab is a widely used hydronym, in this instance, in combination with other similar names the identification seems to bear some weight. Thus, these reports containing direct references to the Amyrgians permit us to place them in Farghana and the Alai valley and, more tentatively, in Qarategin, Darvaz, and probably on the upper Panj River approximately as far as the Bartang (the lower Murghab).

While most scholars place the Amyrgians in the eastern part of Central Asia, and in particular in the area of Farghana and the Pamirs,' there are other points of view. One, based on primary sources on the history and historical geography of the area, holds that the Amyrgians occupied the entire basin of the upper Gokcha, i.e., the main part of Badakhshan, and lived only there. This thesis was first suggested by Markwart [Marquart] in a series of studies;' it was subsequently accepted by A. Herrmann; and Grantovskii is inclined to agree.' Markwart bases his theory on the fact that Munji, an East Iranian language used by people who must be descendants of the Amyrgians, has been preserved in valleys of the upper tributaries of the Gokcha River, and he also cites other indirect data. While this does not in itself constitute absolute proof, subsequently it has been shown that ethnonyms associated with the Munjan people are related to the name of the ancient Amyrgians. Despite doubts about interpretations of some ethnonyms, their connection to the Amyrgians is incontestable."

This brings up the question of whether one should accept the placement of the Amyrgians in Badakhshan, ignoring the direct reports by ancient authors, or whether there is, in fact, no contradiction between the two sets of evidence? The key to solving the problem is whether reports of mass migrations of Saka tribes in the second century BCE can be linked to other information about the Amyrgians. According to Strabo (11.8.2), the Asii (Asiani), Tokhari, and Sakaraucae, the nomads who took Bactria from the Greeks, "migrated from beyond the Jaxartes, the territory occupied by the Sakas." Nothing is said about the fate of the Sakas themselves, and one can understand why: the territory through which the nomads passed was beyond the geographical ken of the Greeks. The only clear point is that, since we are discussing the migration of nomads to Sogdiana and Bactria (Pompeius Trogus, prol. xli), the Sakas of the trans-Jaxartes region are also the Amyrgians of the ancient tradition.

Two versions of missing data can be drawn from Chinese sources. According to Sima Qian, Zhang Qian (ca. 128 BCE) reported that the Great Yuezhi (the Asii and Tokhari) moved west through Da Yuan (Farghana) and subjugated Da Xia (Bactria). These reports correspond to information about the migration of the Ash and Tokhari across the Saka territories on the Jaxartes and beyond. Zhang Qian, who did not mention the Sakas, called the area Da Yuan, the common name of the time; a Greek writing somewhat earlier would have called it "Sakasena by the Jaxartes" (cf. Stephanus of Byzantium, s.v. "Alexandreia"). Another version of the events can be reconstructed from separate reports in the Qian Han Shu: in the west, the Great Yuezhi defeated the Saka king, occupied his territory, and then migrated farther west and subjugated Da Xia; the Saka king went south beyond the Hanging Passage (mountain passes in the Hindu Kush) and became the ruler of the country of Jibin (Kashmir). Two events are confused here.
  1. The Sakas were displaced from Semirechye and the Tien Shan by the Great Yuezhi. They then moved west along the usual route of migration for nomads of this area, but this event took place beyond the view of the Chinese authors and was presented by Greek authors as a migration of the Sakaraucae.


  2. Subsequently, the Sakas, i.e., the Amyrgians, were displaced from Farghana by the Great Yuezhi, an event connected with the migration of the Yuezhi across Da Yuan, i.e., Farghana.
This second event is of decisive importance in explaining the data about the Amyrgians. The "king of the Sakas" (compare with the Greeks' King Omarg/Amorg), driven out of Farghana, transferred his headquarters to the south across the Hanging Passage leading to Jibin. His retreat from Farghana took place in the mid-second century BCE, and by approximately the mid-first century BCE Saka kings appear in India. It is significant that the first Saka king known in India (Maues on coins; Moga in epigraphic evidence) had the same name as the king of the Amyrgian Sakas (Mavakes) during the time of Alexander the Great. Apparently this king figures in Chinese sources as Yinmofu, the ruler of Jibin. It is not by chance that Ptolemy records the Kaspirs (= Jibin) as occupying a vast territory from the Bidaspes (Jhelum) River to the mountain of Quindion (Vindhya), including the town of Modura (Mathura) (7.1.47); this evidently reflects the situation during the early period of Saka dominion in India when Kashmir was still regarded as the center of the kingdom. Apparently the king's horde, drawing after them the rest of the Amyrgians, moved from Farghana and the Alai valley, at first through areas where the Amyrgians were already assimilated, via Qarategin, Darvaz, and the western Pamirs, and then through the passes of the Hindu Kush. Individual groups of the Amyrgians might have taken routes farther to the west (although still within the limits of the eastern, mountainous area of Central Asia). In this way, the ancestors of the Munjans penetrated into the Gbkcha valley and Badakhshan. It is also possible that the Parsii mentioned in Classical sources ended up in the Bamiyan region, and the Parsyetae, the ancestors of the Afghans, on the upper Kurram. Thus, one may conclude that the Sakas who appeared in India and the eastern part of the Iranian plateau in the second-first centuries BCE were Amyrgians from Farghana and that they migrated across the western Pamirs and adjoining western mountainous areas, rather than, as is commonly assumed, through the eastern Pamirs.

Among the living East Iranian languages, Pashto and Munji share a distinctive feature, the shift of ancient Iranian d to 1. This is also typical of a dialect, traces of which have been preserved in personal names and toponyms of Indo-Scythia (the region of the lower Indus and Kathiavar), Sakastana, and Arachosia, and which was probably brought there or adapted by Sakas moving into the area from northern India.

If one accepts the hypothesis that these languages spread as a direct result of migrations of the Amyrgians, the same feature should be present in the latter's language. The small measure of direct information about the language of the Farghana Amyrgians includes this particular trait. Demodamas (who not long after Alexander renewed the campaign against the trans-Tanais Scythians, i.e., the Amyrgians) established that the Jaxartes (which Alexander and his soldiers had mistaken for the Tanais) was called Silis by the Scythians (Pliny 6.49; Solinus [mid-third century CE] 49.5; compare Eustathius ad Dionysius Periegetes 14). This Scythian word has long been regarded as an example of a typical d > 1 shift: the most convincing explanation of the word derives it from the name of the Syr Darya (*Sida = ?), the most ancient form of which is Sanskrit Sita.l6 In Indian texts, the Sita is mentioned together with the Ganges, Indus, and Oxus (Vakhsh; Amu Darya) as the either four or seven major rivers that trace their source to the mythical Mount Meru.

The Homodotes
Classical sources contain the earliest, if somewhat obscure, reports of another Saka group proper, of the Homodotes, locating them in the Emod and the headwaters of the Oxus. While the Emod had been known to the ancient world since the time of Alexander the Great's campaigns and almost invariably figured in Classical tradition as a mountain range in northern India, there is a certain duality in mentions of the area. Megasthenes noted that the Emod separated India on the north from "Scythia inhabited by the Scythians known as the Sakas" (frag. 4, p. 35, 1, Jacoby). In another passage, regarding the countries that surrounded India, he lists the Scythians and Bactrians together (frag. 4, p. 37, 5, Jacoby). It appears that Megasthenes created a single entity from two different Emods: the genuine Indian Emod and the Scythian Emod, an area near that inhabited by the Sakas. For example, Megasthenes may have based his location of the Sakas and the Bactrians on the fact that upstream the Oxus flowed through lands of the former and downstream passed through Bactria. In fact, it was thought that the Oxus and its tributaries cross Bactria (Polybius 10.48.4). Dionysius Periegetes [fl. CE 124] agrees, but he is more definite: the Oxus descends from Mount Emod (748). The name Emodon appears twice in Dionysius' text, and in the second instance it obviously denotes the Indian range (1146). It is significant that the first mention is given as Oimodon. Perhaps the name should be read differently as well (cf. the remark of a Byzantine commentator on variants of spelling of the term [Eustathius ad Dionysius Periegetes 747]). This would imply that Dionysius was making a distinction between the Scythian and Indian Emods and that his information originated in different sources. Avienus (4th century CE), who translated Dionysius' text into Latin, transcribes Hemodi as Hemodoontis (1351). The contents of the translation demonstrate that Avienus used additional sources of some sort (perhaps the map that sometimes accompanied Dionysius' work). Might Avienus have substituted for the name of the mountain range that of the tribe connected with it, which in the source he used could be read approximately as *Hemodontes? Then, what was the original source from which Dionysius drew his report about the Scythian Emod?

In Pliny's list of Scythian peoples, based on materials in Demodamas, he lists the Homodoti (6.50) in a group of tribes, including the Astacae, Rumnici, and Pestici, which inhabited areas along rivers that flow into the Caspian Sea from the north and east." If the concept of the Pestici as a people who lived near the mouth of the Oxus (Pomponius Mela [1st century CE] 3.39.42) is correct, one may assume that the Homodoti were listed directly after the Pestici because they inhabited the region at the headwaters of the Oxus. In his Argonautica, written as a narrative of the arrival in Persia of allied troops from Scythia, Valerius Flaccus describes the Scythian tribes in terms reflecting the source used by Pliny. "Emoda [other readings are: Eumeda; Oemeda] joins forces" with the Scythian hordes (6.143). In all probability, Demodamas (who himself traveled to Scythia beyond the Jaxartes) was the original source of the information about the Scythian Emod in later works by Dionysius, Pliny, and Valerius Flaccus, although none of them made direct use of his account.

Scythian Gold Pectoral
Thus, information about the Scythian Emod is found in works by Megasthenes and Demodamas dating from the late fourth-early third centuries BCE Megasthenes did not differentiate it from the Indian Emod and used the customary term Emodon for it. Demodamas may have denoted it separately with a name whose Greek prototype can be reconstructed approximately as *(H)oimoda. In this tradition, the Oxus rose in the Scythian Emod. Finally, the Scythian people living in this region were simply called Sakas by Megasthenes, and by Demodamas more specifically by a name which can be reconstructed approximately as *Ho(i)modotoi. The reconstructed toponym and ethnonym attest to a dialect in which -d- was preserved and which, in this respect, differed from the dialect of the Amyrgians. It is generally accepted that the Pamir languages of the ShughniYazghulami group stem from the language of the Sakas, and thus the word indicates that the dialect of the Homodotes, rather than that of the Amyrgians, is the Saka proto-language of this group.

It has been suggested that the first part of the name Homodoti cited by Pliny conceals the Iranian word Hauma-, which denotes a sacred plant of the ancient Iranians." But when this name is compared to (H)oimoda, another origin comes to mind: at its root is the Scythian name Emod (which might also originate from the name of the tribe). The plural formant, -t(a), is widespread in Scythian ethnonyms. Among modern languages of the Shughni-Yazghulami group, this feature is preserved only in Yazghulami as -a r9; however, according to linguists, the plural suffixes in modern Shughni dialects are relatively recent formations.'

The Scythian Emod and the headwaters of the Oxus can serve as points of orientation in determining where the Homodotes lived. Since scholars have always thought that Classical authors were describing only a single, Indian Emod, Dionysius' information regarding the Oxus rising in Mount Emod has been assumed to refer to the upper Panj.z° However, if one assumes that a completely different mountain range, a Scythian one, is meant, this interpretation comes into question. During the time when the entire Amu Darya was called the Vakhsh (= Oxus), its main source may have been the river that today bears the ancient name. In fact, in his report about the Oxus, Theophrastus, who was a contemporary of Megasthenes and Demodamas, definitely refers to the Vakhsh in the sense attached to the word today (Ps.-Aristoteles, De mirabilibus auscultationibus, 46; Pliny 31.75, 86).21 All this leads one to search for the Scythian Emod and the Homodotes of that region on the upper Qizil SuVakhsh. Chinese sources contain additional information that makes it possible to locate the Emod and the Homodotes more precisely. Here, it seems relevant to propose a correlation between the Scythian Emod and the Juandu (ancient form: iwan-d'uok), the nomadic confederation of "ancient Saka tribes." The basic source of information about Juandu is the Qian Han Shu. As for the location of this confederation, it is generally agreed that it occupied territories west and southwest of Kashghar, evidently in the basin of the Kashgharian Qizil Su and its tributaries.

On the west, its lands extended up to the crest of the Congling, probably up to the Taunmurun pass,za which is situated somewhat west of Irkeshtam at the headwaters of the Qizil Su-Vakhsh. Thus, the Homodotes might correspond to the Juandu; and Mount Emod, as the place where the Oxus rises, might refer to the Congling, by which in this instance the eastern areas of the Alai and Trans-Alai ranges are meant.

The Komedes
In sources dating from the fourth-third centuries BCE to the first-third centuries CE, mentions of the Homodotes always place them in the same location, the region of the northeastern Pamirs.

Comb with a Scythians in Battle,
Late 5th - early 4th century BCE
This refutes the idea that the Sakas of the Tien Shan and Semirechye took part in the Saka migration of the second century BCE and moved along the eastern Pamirs. In general, however, participation by the Homodotes and related tribes in Saka migrations of the second century BCE is entirely possible, as witness the Komedes, who are mentioned in the itinerary of Maes Titianus (1st century CE) used by Ptolemy in his treatise on geography. Ptolemy gives a relatively full account of this people: the Komedes inhabited the entire mountainous land of the Sakas (6.13.3). The headwaters of the Jaxartes (Gulcha is evidently considered its source) and its two left tributaries originate in the land of the Komedes (6.12.3). The "Gorge of the Komedes" (Qarategin) is mentioned (1.12.7-8; 6.13.2), and Ptolemy remarks that the mountainous region of the Lambates at the headwaters of the Koas (Kunar) "rises up as far as the Komedes' land" (7.1.42). From this, one may conclude that in the time of Maes Titianus, the Komedes inhabited the Alai Mountains, the Alai valley, and Qarategin, and that to the south their lands extended from Darvaz well up along the Panj, following the route leading to mountain passes on the upper Kunar (Chitral) River.

The badly damaged text of Julius Honorius' Cosmography also mentions the Komedes as the Traumeda (*Caumedae, A, 13, 38). The corresponding mountain, Caumestes (*Caumedes, A, 4), is described as the place where the Oxus rises (A, 7). Thus, the mountain appears to occupy all of the territory between the Oxus and the Jaxartes (compare with: A, 8; B, 5). Honorius' map belonged to the Agrippa tradition (first century BCE), but the Komedes are absent from other maps produced by that group and therefore probably were a later addition. However, the word Honorius uses, Caumedae, which preserves the diphthong au, is a more archaic form of the name given by Ptolemy as Komedai.

Information about the Komedes in Classical sources dates from the 1st century BCE-1st century CE, when the Komedes occupied the territory that had been inhabited by the Amyrgians before the migrations of the second century BCE. The Saka dialect of the Komedes, judging from the preservation of the ancient Iranian -d- in their ethnic name, is similar to that of the Homodotes and differs from that of the Amyrgians. Therefore, the Komedes might have originated among the Homodotes and related tribes; they occupied the territory outlined above by moving down the valley of the Qizil Su-Vakhsh and then into territories farther south after the Amyrgians abandoned those lands.

Chinese and medieval Islamic sources are helpful in tracing the further history of the Komedes. Chinese sources reflecting the situation during the 1st century BCE – 3rd century CE locate the Xiuxun, a nomadic confederation of ancient Saka tribes similar to the Juandu, in the areas where the Classical sources place the Komedes. They lived west of the Congling, i.e., of the Taunmurun pass-and they were centered in the valley known as the Migration of the Birds (the Alai valley), whose ancient name by then was only a toponymic relic of the Amyrgians who had formerly inhabited it. There are also indications that Xiuxun territory extended about as far south as the lands of the Komedes. It is evident that one and the same Saka tribe was called Xiuxun by the Chinese and Komedes by the Greeks. The latter name appeared in Chinese sources much later. Although the Chinese obviously knew only the southern part of the Komedes' territory (Darvaz and lands farther up the Panj), information concerning this area shows a tendency to diminish it: Xuanzang (seventh century CE) mentions yet another territory between the Komedes and Shughnan peoples.26 What medieval Muslim sources have to say about the Komedes (Kumed; Kumiji) reflects a later stage of settlement. They locate the Komedes only in the middle Vakhshab (Vakhsh) below Rasht (Qarategin). According to the more complete information from Hudud al-Alam, the Komedes were divided into two groups on either side of the Vakhshab.

This information may be interpreted as follows. The Komedes occupied the greatest territory during their initial period of settlement. Evidently, at that time they were ethnically homogeneous. Their dialect was that which most probably provided a basis for the development of the later languages of the Shughni-Yazghulami group; it is pertinent that the initial territory of the Komedes takes in almost the entire modern region of these languages. Between the third and seventh centuries CE, the community of the Komedes began to disintegrate, probably as the nomads gradually settled down. Shughnan was the first to separate, and other regions followed suit, leading to a disruption of ethnic unity. Evidently, only that part of the population that continued to lead a nomadic way of life preserved the name Komedes, but the territory of those Komedes was also reduced by new waves of nomads who came down the valley of the Qizil Su-Vakhsh. Eventually, the Komedes were left with only small islands of their former territory.

The Kaspians
An analysis of reports on the Kaspians (Kaspirs) of the sixth to second centuries BCE leads to the following conclusions. The (eastern) Kaspians were indigenous tribes of the Hindu Kush-Himalaya region whose direct descendants, under the name of the Burashki, still live in the center of that region in the valleys of the Yasin and the Hunza. From the 6th to the 2nd century BCE, the Kaspians occupied a vast territory that included Kashmir, Gilgit, Chitral, Badakhshan, and Wakhan. Somewhere to the north, they came into direct contact with the Amyrgian Sakas.

A Scythian archer with bow
and 'pointed hat'.
The Kaspians led approximately the same way of life as the neighboring Sakas, and therefore Classical authors sometimes called both groups Scythians. However, as well as herding livestock in the mountain pastures, in suitable places the Kaspians practiced agriculture to a greater degree than did the Sakas.

Regarding the situation in this region in the 1st century BCE – 1st century CE, Ptolemy, using contemporary sources (perhaps the itinerary of Maes Titianus) provides information that is detailed but difficult to interpret, as note the text of his Geography and his map. He places Kaspiria at the headwaters of the Bidaspes, Sandabal, and Adris 7.1.12, i.e., at the sources of the Jhelum, Chenab, and Ravi rivers, a region that generally corresponds to modern Kashmir. North of Kaspiria, an extension of the Imaum [Imaon] range continues to the west until it merges into the Caucasus. Ptolemy locates the territory of the Sakas north of the Imaum and the country of the Sogdians north of the Caucasus; he locates the Oxus as rising from the extreme western end of the Caucasus. In Sogdiana, there is a region called Vandabanda "between the Caucasus and the Imaum" (6.12.4). The mountainous land of the Komedes lies north of that region, and they also must live west of it, since their territory adjoins that of the upper Koas. East of Vandabanda, in the land of the Sakas, the Byltai live "by the Imaum range" (6.13.3), evidently between the east-west and north-south extensions of the Imaum.

Ptolemy's Vandabanda has been placed in Badakhshan, or even in Farghana, primarily because the names sounded alike. However, Ptolemy's report suggests another site. It cannot be disputed that Vandabanda lay somewhere in the region of the upper course of one of the tributaries of the Oxus, which in this instance means the Panj. In reality, Vandabanda had nothing to do with Sogdiana, but the mere fact that Ptolemy attributed it to Sogdiana suggests that the source he used placed Vandabanda east of the south-north course of the Panj, approximately in Wakhan. With Ptolemy's map in mind, the location of Vandabanda as between the Caucasus and Imaum is incomprehensible. If, however, the Imaum is considered to be the Himalayas, and the Caucasus means the eastern Hindu Kush, then one may conclude that Ptolemy's original sources referred to the area between these two ranges, i.e., the region along the Gilgit and tributary valleys. In this interpretation of Ptolemy's map, Vandabanda extended to Wakhan and Gilgit. As for the name of the region, Markwart was correct: the spelling of Ovandabanda is a dittograph that originated in the merging of two words that in the original text were two variants of a single name: Ovanda and Banda, whose nominal meaning is "fortified border; border fortress." The famous mountain fortresses in Wakhan (and perhaps also those farther to the south, in Gilgit and Chitral [? ] ), which were clearly designed to guard the borders, already existed at the time. The original name probably denoted the area immediately adjoining those fortresses and then expanded to cover the entire region.

The reports from Classical authors can be refined and supplemented by Chinese sources, in particular the Qian Han Shu. Jibin (Kaspiria) beyond the Hanging Passage undoubtedly corresponds to Kashmir. However, there is no exact information on the site of Jibin, and it is possible that the source sometimes refers to a more extensive territory than modern Kashmir. Northeast of Jibin and subject to it lay the territory of Nandou. To the east lay Wucha; the road from Wucha to libin must have run through Nandou. In the north, Nandou bordered on Xiuxun; in the west, on Wulei and the Great Yuezhi; and in the south on Chuo Qiang (i.e., the Tibetans of Baltistan). Thus, some scholars have placed Nandou in Gilgit and in the same area have located the Hanging Passage, by which they meant mountain passes of the Hindu Kush that communicated with the valleys of the Hunza and Yasin. This location implies that there was a close relationship between Nandou and Jibin and that to the south, Nandou was the neighbor of Chuo Qiang.

Other evidence from the Qian Han Shu does not agree with this. In preface, however, one should keep in mind that it is not appropriate to take the description of the location of the neighboring areas of a country in the ancient sources literally, and attempt then to transfer the description directly to the modern map. These descriptions appear to be based only on travelers' information, and if, for example, the source says that a region is to the west of a certain country, it means that from some point within the borders of the given country the road to the center of the neighboring region led west.

Nandou's domain was west of Wucha, i.e., Sarikol, indicating that the road from Sarikol, turning west, passed through the territory of Nandou. Thus, Nandou can only be Wakhan. West of Nandou lay the land of the Great Yuezhi. Nandou's immediate neighbors most probably were Shuangmi and Xiumi, two territories that had been subjugated by the Great Yuezhi in the preKushan period. As is now accepted, Shuangmi corresponds to Chitral, which in reality lies west of Gilgit. Xiumi has been convincingly identified with Wakhan. While this at first appears to eliminate the possibility of including Wakhan in Nandou, later Chinese sources from a period when Xiumi actually did correspond to Wakhan give the capital city of Xiumi as Saijiazhen (= Ishkashim). Hudud al-`alam also calls Sikashim the capital of Wakhan. Therefore, Xiumi (or Xumi) was a domain that included Wakhan and Ishkashim and was centered in the latter. In this ancient period, Xiumi might have been bordered by the region of Ishkashim. Between Ishkashim and Wakhan stands the ancient fortress of Yamchun, which was in existence in the Kushan period perhaps marking the border between Xiumi as a part of the Great Yuezhi and Nandou.

It is more difficult to interpret the report that Xiuxun was Nandou's neighbor to the north. Nevertheless, this appears to be correct. The Xiuxun/Komedes, who had settled far up the Panj River, might have come as far south as Shughnan and Ishkashim and penetrated into the domains of the Great Yuezhi. In this area, they would have bordered on Nandou. In addition, the road to the center of Xiuxun in the Alai valley actually did lead directly north from that point.

Finally, the domain of Wulei is described as the western neighbor of Nandou, although according to the modern map it is to the north. Wulei is convincingly placed in the Alichur and Aqsu-Murghab valleys. On the west, it also bordered the Great Yuezhi; a group of fortresses, ruins of which still stand in the valleys of the Gunt and Shahdara, may mark the boundary. Ptolemy placed the Byltai roughly in those areas where the Wulei lived. According to Ptolemy's information discussed above, the Byltai occupied the area west of the north-south extension of the Imaum. Since, according to Maes Titianus' itinerary, that branch passed through the "Halting Place for Merchants who trade with Sera" (Ptolemy 1.12.8; 6.13.1) situated in the region of modern Irkeshtam, one may conclude that the Imaum mentioned in connection with the Byltai corresponds to the Sarikol range. The domain of Wulei (Puli) was a Tibetan nomadic tribe, although Ptolemy situates his Byltai in the country of the Sakas and gives their name in a form attesting to a passage through the Saka area (the typical Scythian particle -ta).

The picture finally derived from first century BCE to first century CE sources is that the country of Kaspiria/Jibin generally corresponds to Kashmir. The adjoining area of Vanda(Banda)/ Nandou encompasses Gilgit with Hunza and Yasin and also evidently Wakhan and the adjoining parts of the Pamir plateau. The agreement of the various sources on the neighboring areas in this region is remarkable. Despite the fact that Ptolemy took these reports from new itineraries and fit them roughly into Eratosthenes' old cartographic scheme, his description clearly corresponds to the analogous Chinese data: Kaspiria/ Jibin lies south of the region of Vanda(Banda)/ Nandou; the Komedes/Xiuxun are to the north and west; and, finally, the Byltai/Wulei in fact are neighbors to the north.

Important evidence regarding the population of the region in question is provided by materials from the archaeological work carried out in the eastern Pamirs under the direction of Bernshtam and Litvinskii, namely, the location and investigation of burial mounds and other monuments of the seventh-first centuries BCE in the valleys of the Aqsu-Murghab, Alichur, and Pamir rivers in the vicinity of Lake Rangkul. Most date from the fifth-third centuries BCE; the most recent (second-first centuries BCE) are concentrated in the eastern and southern sections of the area in which they appeared earlier. The material culture resembles that of the Sakas, `but this is decidedly not true of the anthropological data. Like the modern inhabitants of the western Pamirs, the Sakas are primarily of the anthropological type characteristic of the Central Asian region between the Amu Darya and Syr Darya.

The population of the eastern Pamirs of the earlier period belonged to a completely different type, that of the eastern Mediterranean area. Therefore, anthropologists are in agreement that genetically these people are not related to the Saka tribes who inhabited the areas farther to the north: their ties lie to the southwest, perhaps in the Hindu Kush and northern India, and in all likelihood they are the descendants of a more ancient local population. In fact, although the modern population of Wakhan belongs to a different type, there is reason to believe that, at least in the last centuries BCE, a people of the same physical type and culture as the modern population of the eastern Pamirs lived there. As yet, there are no paleoanthropological materials on the given period from areas farther to the south, but their modern population-the Kashmirians and Burashki of Hunza and Yasin in particular (the latter are direct descendants of the ancient Kaspians)-are representatives of the Indo-Afghan type, a modern variant of that predominant in the ancient population of the eastern Pamirs. It is clear that the anthropological data coincide with the conclusions about the Kaspians (Kaspirs) presented in this article.

In summary, in the sixth-third centuries BCE, the Kaspians occupied the entire Hindu KushHimalaya mountainous region sketched here. At that time, the population of the eastern Pamirs consisted of only the extreme northeastern segment of a vast body of Kaspian tribes. It is possible that somewhere in the Murghab valley these tribes might have come into contact with the Amyrgian Sakas. Amyrgian migration in the second century BCE only brushed against the western and southern sections of the region inhabited by the Kaspians, while the eastern Pamirs remained untouched. In that area, Kaspian territory shrank for another reason: pressure from the Tibetans. The Byltai, one of the Tibetan tribes which had settled in the northwest foothills of the Kun-lun in the second-first century BCE, penetrated to the Pamir plateau, perhaps from the northeast, and drove the Pamir Kaspirs to the extreme east and south of their former territory. As early as the first century BCE, this Tibetan tribe evidently occupied part of the Gez Darya valley south of the Juandu area of settlement. At that time, generally only Kashmir was still called Kaspiria. The land of the Kaspians farther to the north became part of Vanda (Banda); the memory of the fact that it had once been a single entity with Kaspiria was preserved in the concept of it as a Kaspian dependency. Together with Wakhan, it evidently included those parts of the Pamir foothills that were still in the hands of the local Kaspians.

Further displacement of the Pamir nomads was related to the expansion of the domains of the Great Yuezhi. As Bernshtam noted, the presence of fortresses in the western Pamirs attests both to the existence of agricultural oases and to an active defense against the nomads of the eastern Pamirs. If those fortresses guarded the borders of the domain of the Great Yuezhi in the preKushan period, the shift of boundaries to the east as far as the Congling (Sarikol) range (in the mid-first century CE, when Vima Kadphises conquered Kaspiria [Jibin] and probably the associated area of Vanda [Banda]) reflected a farther advance on the nomads of the eastern Pamirs. Ultimately, in the fourth-sixth centuries Wakhan and the neighboring regions probably were inhabited by settled farmers who came from the Khotan oasis.

This reconstruction of the history of individual Saka tribes is, of course, a working hypothesis to be corroborated or rejected upon further investigation. In this respect, much depends on new data, particularly the study of the toponymy of areas once inhabited by the Sakas and Kaspians. The sequence of key moments in their history remains to be clarified by detailed archaeological investigation of Kashmir and Gilgit, Badakhshan, and regions of Kashghar and Khotan.