1. There has never been any question that the Suoche 莎車 [So-ch’e, sometimes written: So-chü] of the Han period is to be identified with modern Yarkand or Suoche (also written Shache). See, for example, Stein (1907), p. 88; CICA, p. 139, n. 361; Bailey (1985), p. 73.

   “For this city Ptolemaios’s text has four variants σοÃτα, σοÃγα, σότα, σάγα. Here the fourth is nearest to the Saka word. The modern, half-Turk name is Yār-kand. The Turkish yăr is ‘cleft’ and ‘rock’ as in yar tuzï ‘rock salt’. The Turkish has kept the meaning of the original name and added the word kand ‘city’. Here yar is from yarmaq ‘to split’.” Bailey (1958), p. 73.

2. Puli 蒲犁 [P’u-li] = Tashkurghan. The identification of Puli with Tashkurghan is, I maintain, certain. The directions in the Hanshu provide accurate directions and distances from both Suoche or Yarkand – 540 li or 225 km to the east – and Shule or Kashgar – 550 li or 229 km north. These distances prove to be very accurate when checked on modern maps. Also, there is no other settlement on these routes and in this generally barren region that it could possibly be. 
          The dependence on food supplies from Yarkand (referred to in the Hanshu) and the size of the population (about 5,000) haven’t changed since the time of the Former Han, thus providing additional confirmation of its identification. See CICA, pp. 101-102. There is only one piece of evidence that does not fit. In the Hanshu’s description of Suoche [Yarkand], it is said that Puli is 740 li (308 km) to the southwest (ibid. p. 142). This must surely be a copyist’s error for the 540 li listed above.
          Heading south from Tashkurgan the traveller in ancient times had two main choices. The first route (and the only one practicable for laden pack animals such as mules and horses – being not too difficult and providing plenty of fodder). It led over the Neza Tash Pass and along the Ak Tash (‘White Stone’) Valley west along the Ak-su (‘White Water’) River which flows to the northeast until one entered the Wakhan Valley proper.

“Our first day’s journey [from Tashkurgan] was to the foot of the Neza Tash pass, sixteen miles [26 km] in a south-westerly direction up the Shindan stream, which flows through the defile of the same name and falls into the Sirikol river. The defile at several places is extremely narrow, and shut in by precipitous rocks and bold steep hills which rise high above it. The fallen stones and stream boulders make the road particularly bad for many miles. Willows and thorn bushes grow plentifully at the head of the defile, and the hills there lose their bold character, and become rounded and sloping. Our camp was in snow, but large patches of grass free from it were found in the vicinity sufficient for our horses, which ate it greedily, preferring it greatly to the chopped straw we carried for mixing with their grain. This grass was similar to what we found in many parts of the Pamirs, and in the Aktash valley, rich and sweet to the smell, resembling English meadow hay, and relished immensely by our animals. Judging from what we saw of it in the end of winter, it is easy to believe in its fattening properties in summer, as related by Marco Polo and other travellers, and also told us by the Wakhis. Neza Tash, meaning spear-stone, is named from a spear-like pointed rock near the place.
          On the second day we crossed the Neza Tash pass (14,920 feet)[4548 metres], leading over a high range running about north-west, and encamped at the mouth of the ravine leading down from it to the Aktash valley, travelling a distance of seventeen miles [27 km] in a general westerly direction. Snow fell in the night time, and our journey for this and the following three days, covering a total distance of seventy-eight miles [126 km], was made mainly through snow. We found plenty of grass in scattered patches and brushwood fuel at this day’s camping place. We were here joined by a party of Sirikolis with yaks and ponies carrying supplies sent by Hussan Shah to accompany us to Wakhan.” Gordon (1876), pp. 123-124.

Notes on the Neza Tash Pass adapted from: Merzalikova (2003):

“Height: 4328 m [14,199 ft], “The Pass leads down Karasu stream to valley Aksu in “Sarez Pamir.” “On the west side it is very easy to ascent. Descent more difficult steep and stony.”

From there one travelled past the Little Pamir Lake roughly 80 km to Langar:

“…twenty-five miles [40 km] from the lake. A deserted village and traces of cultivation were observed here, and numerous yaks and cattle were seen grazing on the opposite side of the valley. A stream of considerable size also joins at Langar, flowing from the south-east, and a road goes by it to Kunjut, over the Kura pass.” Gordon (1876), p. 129.
          When one reached Sarhad (approximately another 40 km west from Langar), one could cross south over the relatively easy Baroghil Pass (3,798 metres; 12,460 feet) towards Mastuj which leads on either to the Chitral Valley, providing relatively easy access to both the region around modern Jalalabad in Afghanistan, or through Swat to the region of ancient Gandhara near Peshawar.
          If, on the other hand, a caravan headed due west from Sarhad through the Wakhan Valley, it travelled through Badakshān to Bactra (Balkh) and beyond.
          The other alternative was the quickest way into northern India and usually open all year but was extremely dangerous and only suited for travellers on foot. From Tashkurgan one travelled just over 70 km south to the junction of the Minteke River. Heading west up this valley one reached the Kilik and Minteke Passes which both led into upper Hunza from which one could travel over the infamous “hanging passages” to Gilgit and on, either to Kashmir, or to the Gandharan plains.

3. Wulei 無雷 [Wu-lei] = Sarhad. This Wulei must not be confused with the Wulei 烏壘 (‘Black Fort’) = Yengisar – see note 20.7 below. The character wu 無 in ancient times was pronounced with an initial m sound and was used to represent Sanskrit mo or ma, as in the mantric term, nama – ‘I humbly trust (or adore)’. The EMC forms given in Karlgren, Pulleyblank and GR all also indicate this initial m sound.
          The section on Puli in the Hanshu specifies that Wulei is 540 li (225 km) west of Puli (CICA, pp. 101-102). If this route is measured out carefully it brings one precisely to the strategically placed settlement of Sarhad, where the route again forks. One branch heads south over the Baroghil Pass, the other heads west along the Wakhan Valley to Badakshān and the heartland of the Kushan empire in what is now northern Afghanistan.
          Unfortunately, we again have a case of an error in the Hanshu, for in its entry on Wulei it says, “to the south it is a distance of 540 li to P’u-li.” Ibid. p. 103. This contradicts the statement a couple of pages previous that, “to the west [from Puli] it is a distance of 540 li to Wu-lei.” Fortunately though, the text of the Hou Hanshu confirms that Wulei must be to the west of Puli, as it states: “Going west from the kingdom of Suoju (Yarkand), and passing through the countries of Puli (Tashkurghan) and Wulei, you arrive among the Da Yuezhi (Kushans).”
          Yu (1998), p. 60, follows Matsuda’s hypothesis that: “Wulei was situated in the present Little Pamir, which was between the valley of the River Aksu which flows northeast to the upper reaches of the River Murg-āb, and the valley of the River Aksu which flows west to the upper reaches of the Āb-i Panja.” However, as he confuses Shuangmi with Nandou, which he assumes refers to the lower reaches of the Gilgit River, he doesn’t realise Wulei was actually situated somewhat further west, near modern Sarhad, at the head of the Baroghil Pass leading into the upper Yarkhun Valley and on to Mastuj and Chitral – or ancient Nandou. (See the note 13.6 on Shuangmi above).

4. Li Chong – Protector General 16-23 CE. See CICA, p. 64, n. 196, where the dates were taken from Huang Wen-pi (1948), p. 179 f. After an abortive attack on Yanqi or Karashahr in 16 Dunhuang.” Translated and adapted from Chavannes (1907), p. 198, n. 1.

6. Chanyu 單于 [sometimes incorrectly written: Shan-yü] = ‘Khan’ or ‘King.’ This is the same title used by the Xiongnu sovereign, which is possibly related to the title Khan – see note 1.55 above.

7.       “Hostages in Chinese history may be somewhat arbitrarily classified into the following groups:

1. “Exchanged hostages” – to guarantee a friendly relationship between two states or two other groups.

2. “Unilateral hostages” – to guarantee allegiance and loyalty.

a. “External hostages” might be taken by one of two belligerent parties from another during negotiation for an armistice or surrender. In more peaceful times, hostages might be taken by a powerful state from a weak state, by a suzerain from its vassal states or dependent tribes, or by a lord from a group of individuals at the time when they were offering their allegiance.
b. “Internal hostages” might be taken by a ruler from his military of civil officials, especially from those who were stationed along boundaries or sent out on an expedition.

   In all instances the hostage was normally a member of the sender’s family, in the majority of cases, his son. Hostages from several families might be required on a single occasion….
          Taking hostages was a standard practice of the Han dynasty for controlling small barbarian states. Princes sent for this purpose were known as chih-tzu 質子, “hostage sons,” or shih-tzu 侍子, “attending princes.” The latter term was used because such hostages were often made attendants at the court or guards at the imperial palaces. They were lodged in the capital and treated kindly. On the other hand, they were subject to Chinese law and punishments. Thus, a hostage prince from Lou-lan 樓蘭 is known to have been castrated in the reign of Wu-ti.
          Interestingly enough, the Hsiung-nu in Han times also required hostages from their satellites as a surety for allegiance. Certain small states in the Western Regions found themselves hemmed in between the Chinese and the Hsiung-nu and were obliged to send hostages to both powers. Since a hostage prince would usually prove friendly to the court where he had lived, he was in a good position to receive its support in a bid for power after his return. Bitter wars were at times waged between princes who had returned from the Han and Hsiung-nu, and this constituted an important phase of the struggle for control over the Western Regions between the two powers. When the Hsiung-nu had been defeated and weakened, they also send hostages to China, notably, in 53 B.C. and in 20 B.C.
          A detailed study of external hostages may reveal the degree and direction of outward expansion of a dynasty. For example, at the beginning of the Latter Han era in 45 A.D., eighteen states of the Western Regions sent in hostages with tribute requesting a tu-hu 都護 or protector general be stationed in that area. The Emperor, however, did not feel that China was ready to control that region and ordered the hostages to be returned.” Yang (1961), pp. 43-46.

8. Guisai 媯塞 [Kuei-sai]. Chavannes (1907), p. 200, n. 1, notes that the character gui was used for Wei, the Iranian name for the Oxus River, and the character sai represented Saka (or Sai), so the term might have implied that Guisai was a principality on the Upper Oxus governed by a Saka prince but subject to Yarkand – but he considered it very doubtful. For further discussion of this hypothesis see CICA, p. 164, n. 514.

9. Wulei 烏壘 = Yengisar. This Wulei (‘Black Fort’) must not be confused with the Wulei 無雷= Sarhad – see note 20.3 above.
          The Hanshu records that Wulei was the seat of the Protector General at that time and records that it was 350 li (145 km) east of Kucha, and 300 li (125 km) west of Weili (Korla). This places it beyond any doubt at the oasis of Yengisar. (CICA, pp. 164, 177).
          Because Stein (1928), p. 794, was not aware of the true length of the Han li (estimating it at about 1/5 of a mile, or about 322 metres, instead of the true length of 415.8 metres), he was unable to properly choose between the three oases of Bugur, Yangisar, and Chadir, as the site of Wulei. He favoured the Bugur oasis as the probable site because it is the largest oasis between Korla and Kucha (ibid. p. 796). However, he also gave Yangisar (Yanghi-hissar) serious consideration, and noted its strategic importance:

“The importance of Yanghi-hissār is increased by the fact that a route leads from it across the high range northward of the Yulduz plateau at the head of the Kara-shahr valley. It was stated to be the first practicable route east of Kucha to the plateau, and to be much used by Mongols taking supplies from the oasis to their grazing grounds. The pass crossing the watershed was said to retain snow all through the year ; but these hardy Mongol customers, I was told, find it practicable even during the winter months.” (Ibid. p. 791)

10. Ligui 驪歸 [Li-kuei]. Unidentified town – possibly near Khotan – see note 4.3 above.

11. Weishi 位侍 [Wei-shih] is possibly not a personal name. The characters are not normally used for people’s names and can be translated as: ‘Attendant to the Throne’ or, perhaps, ‘Next-in-line to the Throne.’ See also: Chavannes (1907), p. 201.

12. Gumo 姑墨 [Ku-mo] = modern Aksu – the name of both the town and the river. In Buddhist Sanskrit it was known as Bharuka. For details of the etymology and history of the name see Bailey (1985), pp. 71-73, 75. There can be no question that Gumo refers to the present-day oasis of Aksu – literally ‘White Water’:

“At last Mr. Chavannes has corrected, according to the information of Mr. Grenard, a traditional identification which has falsified all the ancient itineraries to the west of Kucha. Proceeding from the current Chinese opinion which identifies Wensu with modern Aksu, we were obliged to place the country of Gumo (Qoum) further to the east next to Yaka-aryk. Mr. Grenard and Mr. Chavannes established that Gumo is in reality Aksu, and that Wensu corresponds to Uch-Turfan. This important correction appears absolutely justified by the facts.” Pelliot (1906), p. 371. [Translated from the French].

“The 262-kilometre (162 mile) journey from Kucha to Aksu takes between five and six hours. The area is frequently visited by light dust storms generating an eery, creeping ‘fog’ around the base of the sand mounds and the occasional ruins of Han-Dynasty (206 BC-AD 220) beacon towers.
          Neolithic artefacts from 5000 BC have been discovered in the Aksu area. By the first century BC news had reached the Chinese imperial court of the Kingdom of Baluka, one of the 36 kingdoms of the Western Regions. The kingdom, aided by the Xiongnu, held out against the Chinese army under General Ban Chao for a time, only to have him march upon the capital city in AD 78 and execute 700 inhabitants.
          Xuan Zang wrote of the kingdom in 629: ‘With regard to the soil, climate, character of the people, customs, literature, these are the same as the country of Guici [Kucha]. The language differs however a little. [The kingdom] produces a fine sort of cotton and hair-cloth, which are highly valued by neighbouring countries’.” Bonavia (1988), p. 160.

See also: Chavannes (1905), p. 553, n. 1; Stein (1921), 1297, and note 21.7 below.

13. shou jiehou 守節侯 [shou chieh-hou] = Probationary Military Prince. “shǒu... (1) HAN–SUNG: Probationary, prefix to a title during the appointee’s first year in service, only after which he was normally entitled to substantive (chen, chen) status and full salary.” Hucker No. 5355. See also Hucker Nos. 755, 756 ff., and 2205.

14. guoxiang 國相 [kuo-hsiang] = Counselor-delegate. “HAN–N-S DIV: Counselor-delegate, the central government’s representative in a Princedom (wang-kuo) or a Marquisate (hou-kuo), equivalent to a Commandery governor (chün t'ai-shou) and a District Magistrate (hsien-ling), respectively; rank 2,000 and 1,000 bushels, respectively; interchangeable with hsiang (Administrator). Apparently co-existed with Administrators (nei-shih) from 140s B.C. to 8 B.C., when the latter post was discontinued. Thereafter the Counselor-delegate was the unchallenged manager of a Prince’s or a Marquis’s domain....” Hucker No. 3514.

15. fuguo hou 輔國侯 [fu-kuo hou] = Bulwark Prince. fu: “Ety., the side-props that prevent a chariot from turning over; hence, lit., to help, to support: Bulwark. (1) Used throughout history as a broad categorical reference to officials and subofficial functionaries in service under the head of an agency....” This is very similar to the title fu-guo gong [fu-kuo kung] which Hucker translates as Bulwark Duke. The usual translation of hou is ‘marquis,’ the next noble title in rank after a wang ‘prince’ and a gong or ‘duke.’ Hucker Nos. 2035, 2075, 2205. However, it is also used in a more general way, especially when referring to foreign leaders or ‘feudal princes.’ See GR No. 3925, 2.a.

16. hsiang 相 = “Ety., an eye beside (behind? peering from behind?) a tree: lit., to assist.... (3) HAN–MING: Administrator: in Han and early post-Han times the senior official in a Princedom (wang-kuo), Marquisate (hou-kuo), or other semifeudal domain....” Hucker No. 2303. 

17. These events are echoed in the Khotanese Buddhist “history,” the Prophecy of the Li Country, composed in 746 CE and preserved in the Tibetan Tanjur. We find the story here clearly recognisable but transformed into a pious Buddhist legend where, amongst other changes, the pig becomes a deer and the captured Kashgari king is allowed to escape death and return to Śu-lig (Kashgar) on his promise to return to Khotan after becoming a Buddhist Arhat and ‘pious friend’ of his previous captor, the king of Khotan. 
          The name of the “high official” is given here as Dumo – which is presumably a transcription of Dharma, the king’s name in the legend. This King Vijaya Dharma is said to be the youngest of three sons of King Vijaya Jaya who married the Chinese princess who first brought silkworms to Khotan. If all this can be accepted it places the introduction of silk to Khotan sometime not too long before the middle of the 1st century CE. For more details on the introduction of silk to Khotan see: Appendix A: The Introduction of Silk Cultivation to Khotan in the 1st Century CE, and note 4.1 above.
          The legend is best read in R. E. Emmerick’s translation in his Tibetan Texts Concerning Khotan, London, Oxford University Press, 1967, pp. 35-47. See also: Thomas (1935), pp. 110-11 and 110 n. 9; Hill (1988), pp. 179-190.

18. Aurel Stein (1921), pp. 1230 ff.; (1928), 724; 777 ff., locates Weili 尉梨 [Wei-li] at Kara-kum (actually marked Weili, some 40 km south of Korla, on modern maps), and Weixu at Korla itself. The Hanshu places Weili 100 li (42 km) south of Yanqi (Karashahr), and Weixu the same distance away, but no direction is given. Karakum is twice as far from Karashahr as that.
          By modern road, it is about 47 km southwest from Yanqi (Karashahr) to Korla. The small difference of 5 km from the measurement in the Hanshu can be easily explained by slight changes in the route, or slightly different positions for the town centres during the Han era.
          In addition, Korla has long been the biggest centre in the region after Karashahr itself, having abundant water and extensive farmlands, as well as controlling the main routes to the south and west of Karashahr. Weili is given a population of 9,600 compared to only 4,900 for Weixu in the Hanshu, which also mentions that it adjoins Shanshan and Qiemo (Charchan) to the south.
          I think the evidence makes the positioning of Weili in the region of modern Korla certain. See also the discussion in CICA: 177, n. 585 where there is a mention of extensive ruins to the northeast of the present city of Korla.

“Korla . . . offers nothing of historical interest apart from the site of the Iron Gate Pass (Tiemenguan), seven kilometres (four miles) to the north. This Silk Road gateway, wedged between the mountains and the river, guarded the only ancient route connecting northern and southern Xinjiang. All that remains from the destruction wrought by the Cultural Revolution is a pile of bricks. There are plans to rebuild the huge iron gate for which the pass is named.” Bonavia (1988), pp. 151, 154.

19. jixu 罽絮 [chi-hsü] = felt padding. It is not entirely clear from the dictionaries I have at hand whether this should read ‘felt (or cotton or silk) carpets’ or ‘felt (or cotton or silk) padding.’
          I suspect, that at this early period the material was most likely felt. Sending silk to China would have been a bit like “shipping coals to Newcastle,” and cotton would probably have had to have been imported at great expense from India. Also, the word ji indicates some rather rough material – of fur or hair – see GR No. 878.
          Felt is an extremely important material for nomadic and semi-nomadic peoples and numerous examples of this material have been uncovered in recent years along with the famous desiccated “mummies” of the Taklamakan:

“Nomads use felt not just for its convenience of manufacture. More important, it can be made so dense as to be nearly impervious to wind and water, yet it is far lighter than other waterproof materials like wood and metal. The herders spread great sheets of felt over light frameworks to produce their famous round tents, or yurts..., and they use it for flooring (as rugs), bedding, luggage, saddle gear, hats, cloaks, and other clothing. They use it to make dishes for solid foods and padded cases for the china cups used for drink. Along with horse riding, felt has made the Eurasian nomadic way of life possible.” Barber (1999), p. 37.