The organisation of stage stations in Central Asian colonial provinces of the Tibetan Empire according to Pelliot tibétain 1096r

Abstract Based on the first English translation of the Old Tibetan document with the shelf mark Pelliot tibétain 1096 recto, the article analyses the internal organisation of a stage station (sluṅs) in the Central Asian colonial provinces of the Tibetan Empire. It examines officials and offices that constituted a stage station, as well as persons who were using its services. By comparing the information contained in the document with later reports of foreign travellers, the article reconstructs the organisation of a stage station. It also brings to light certain traits that were apparently common to the first historically attested relay system of the Tibetan Empire and the succeeding system introduced by the Mongols during the thirteenth century ce.

The Old Tibetan (OT) document examined in the following article provides us with an exclusive insight into the organisation of a stage station in Central Asian colonial provinces of the Tibetan Empire. Tibetan post services and the transportation system as such have thus far drawn little attention of Western scholars, of whom only Uebach has devoted a study to the relay system of the imperial period. 2 To the best of my knowledge, there exists only one 1 I would like to acknowledge financial support provided by grant BI /- of Deutsche Forschungsgemeinschaft in years -. I wish to thank Diana Lange for helping me with the identification of stage stations on the maps of the Wise collection.
The Tibetan script is transliterated according to the principles put forward in J. Bialek, 'Towards a standardisation of Tibetan transliteration for textual studies', Revue d'Etudes Tibétaines  (), pp. -. Tibetan proper names are hyphened in order to enhance their readability in the text flow. Only the first letter is capitalised. 2 detailed study on later developments of and foreign influences on the Tibetan relay system, namely P. Maurer, 'The Tibetan Governmental Transport and Postal System: Horse Services and Other Taxes from the th to the th Centuries', Buddhism, Law & Society  (), pp. -. Owing to the scarcity of sources, however, Maurer concentrated on the organisation of the relay system as such, leaving aside the functioning of its most basic units-the stage stations. 3 The present study attempts to fill this gap by analysing the only thus far known OT document that sheds light on the internal organisation of a single stage station.
Pelliot tibétain  recto (hereafter: Pt r) is an original OT document, to be specific a summons concerning a dispute over two lost or stolen horses. Like all texts from the Pelliot tibétain collection, Pt r was discovered at the beginning of the twentieth century in Cave  of the Mogao Caves, southeast of Dunhuang, and brought to Paris by Paul Pelliot. 4 The document is composed in Old Literary Tibetan (OLT). 5 Unfortunately, despite numerous attempts we still lack clear criteria on which to date single documents of the period. 6 Accordingly, the date and the place of the composition of Pt r, as well as its 'authorship', remain unknown. Since the text is an original judicial document, it was most probably written in a law court by an authorised person. Two arguments speak in favour of the hypothesis that Pt r originated in Central Asian colonies of the imperial Tibet: () 3 Uebach devoted a part of her article to the internal organisation of a stage station, but her interpretation of Pt r (the backbone of her research) is problematic (see below). In addition, interesting information on the relay system of pre-modern Tibet can be found scattered throughout Lange's meticulous study of the nineteenth-century maps in the Wise collection. See D. Lange, An Atlas of the Himalayas by a th Century Tibetan Lama. A Journey of Discovery (Leiden, ). some of the proper names of persons involved in the case are of non-Tibetan origin (see section entitled Persons below); and () the text explicitly mentions Sá-cǔ (i.e. Dunhuang) as the place of residence of two horse owners. The document is complete, bearing eight seals of persons involved in the case: six seals of guarantors (Qan-hwa-hwa, Den-bun-ɣde, Čan-stag-bzér, Yo-gaṅReɣu-skyes, Gñi-ba Lha-mthon, and Síg-śin-śin), a seal of the defendant Yo-gaṅG.yu-la-skyes, and a seal of a witness who was an anonymous judge from aristocracy (záṅlon zál čhe pa). The legal aspects pertinent to the document have already been comprehensively discussed by Brandon Dotson and so do not need to be restated here. 7 The present article concentrates on the organisation of stage stations (sluṅs) in the period of the Tibetan Empire. Namely, Pt r provides some details on a sluṅs, people related to it, as well as services offered by a sluṅs. Therefore, its primary objective is to present the first annotated translation of the document in a Western language, accompanied by a diplomatic transliteration, and a glossary (see Appendix). In the discussion section, the contents will be scrutinised in order to enhance our understanding of the sluṅs-institution.

Historical context
At the turn of the sixth and seventh century CE, by conquering its immediate neighbours, a small polity centred in the Yar-valley (OLT yar luṅs), sometimes referred to as the Yar-lunṡ Kingdom, arose to become an important military and political actor on the Tibetan Plateau. In the s this polity started its expansion beyond the valleys of Central Tibet, subduing Sum-pa, Zán-zúṅand Ɣa-zá (Ch. 吐谷渾 Tuyùhún) over the following thirty years. These conquests mark the emergence of the Tibetan Empire. With varying luck, the Tibetan Empire then continued its expansion through the seventh and eighth centuries, temporarily controlling territories beyond the Tibetan Plateau, including the Central Asian Silk Roads. Its demise started in the s, triggered by an unstable internal political situation and the declining economy that mirrored the worsening international economic situation from the s onward. 8 The expanding Tibetan Empire required an efficient administrative system to controlpolitically and economically-the newly subdued territories and peoples. To this end an extensive relay system had to be established that could support communication between the socio-political centre of the Empire (now located in the valley of the Skyi-cȟu river) and its dependent territories and colonies. Our knowledge of this system is still in its infancy, and is largely based on sporadic mentions of sluṅs 'stage station'-the nodes of the communication network-and messengers, as in the following passages: The council, convened at castle Phrag by councillor Khri-sum-rȷě-[rcan-bzér], issued great tallies of jurisdiction for mṅan and the upper and lower stage stations (sluṅs).
() da čhab srı̄d gčig čı̄ṅ// mȷ̌al () dum čhen po ɣdıltar mȷad pas () dbon záṅdgyes paɣi bkaɣ phrind () sñan pas kyaṅɣdrul dgos te // () phan chun gyı̄pho ña ɣdoṅba yaṅ// lam () rñiṅpar byuṅnas // sṅa lugs bzín () // bod rgya gñis kyı̄bar // caṅkun () yog du rta brȷ̌es la // (ST Treaty W) Now, the politics being one, because a great agreement was reached in this way, it being necessary to travel with good messages from [lit. of] the pleased nephew and uncle, travelling messengers of both sides appeared on old roads as well. Hence, according to earlier customs, let horses be changed at Can-kun-yog between Tibet and China! In this context, Pt r represents an invaluable source of information on the internal organisation of sluṅs that constituted the basic units of the relay system of the Tibetan Empire. Even though due to its concise and highly technical language the text may occasionally be difficult to comprehend and therefore to translate, it delivers unique details on the functioning of a stage station under the Tibetan rule. Nevertheless, it must be emphasised that the document is a legal one and so the organisation of the stage station is not its main concern.
Translation In the first half of the last autumn month of the dragon year, the messenger Gzáms-khon-khri came to the encampment of Par-kog, 9 asking for one horse of Qan-bcan-zigs-chan. Upon it was necessary to send [the horse] back, 10 deputies of the head of the stage station, head of the encampment, among others, said: "Having taken away the horse, [ will have examined the sincerity (dkar) regarding the very Qab-sab-ñan, among others." Upon having said [so] on the eighth day of the first winter month of this year, up to the fourteenth day [of the month] Qab-sab-ñaṅas well as the worker Zán-ɣdo did not come. Then, having summoned Yo-gaṅ1 2 G.yu-la-skyes, the head of the encampment, to the court, [one] inquired [him].
"The messenger Gzáms-khon-khri, having asked for one horse of a man from Sá-cǔ 13 , came to the encampment of Par-kog. Thereupon, as for this horse, both the messenger and the groom Qab-sab-ñaṅprepared to mount the stallion. There were not many messengers. After some messenger-horsemen had come and the horse of the man from Sá-cǔ was bound, 14 I said to the messenger(s) and the groom: I identify rmos with CT smos, v < smo 'to say, to speak'; cf. Nangchen m h y "with ɲe to clarify one's kinship relations (e.g. before getting married)" (R. Bielmeier et al., Comparative Dictionary of Tibetan Dialects (CDTD). Volume : Verbs [Berlin, ], p. ).

12
Takeuchi interpreted Yo-gaṅas a name of a people that remains unidentified thus far. See T. Takeuchi, Old Tibetan Contracts from Central Asia (Tokyo ), p. .

13
Modern Dunhuang; Tib. Sá-cú < Ch. Shazhoū 沙州. 14 OTDO has bya bsdas but the reading of the last two syllables of l.  is uncertain. The first one looks more like čya, whereas the second one begins with a sign that can hardly be identified with any letter of the Tibetan alphabet. Its last letter could be either s or m. Since no such a word as * bsdas seems to be attested in written sources, I read the syllable as bsdam. The meaning and function of bya remain unexplained. 15 A certain Jeɣu-hiṅis mentioned in Pt /Pt : B as surety (see Takeuchi, Old Tibetan Contracts, pp. ff.). It is not certain whether this is the same person as Jeɣu-hin-yir of Pt r.

16
Compare Yolmo [ jıldo] 'courtyard' (CDTD: , s.v. g.yul ɣthag 'threshing floor'). 17 For this reconstruction compare sluṅs phon gyi g.yul thog in l. r. 18 The translation of rnam čhig la as 'once' is purely contextual. 19 The phrase glo ba čhuṅis known from only a few OT documents: At that time a minion of the stage station appeared. thum čhu ma, having appeared afterwards, said "[I] am coming from Jˇu-cǎṅto Lug-luṅto help". 20 Having clarified [the circumstances], 21 [one] decided: "Concerning the lost horses of Hin-che, among others, the head of the encampment, among others, truly feared [its] stealing". 22 [Thus] it was said.
The head of the encampment, upon being inquired, said, "Upon this horse had come to the encampment, I ordered the groom Qab-sab-ñaṅthat [he] must (śig = IMP) bind the horse again. [Qab-sab-ñaṅsaid:] 'Once, both horses were in the courtyard of the head of the stage station. Thereupon, Jeɣu-hin-yir, riding a one [and] [it]. Therefore, [one] decided that the head of the encampment must provide (lit. give) guarantors, summon Qab-sab-ñaṅand Zán-ɣdo, and plead on the full moon day of the first winter month. 23 Sealed for the guarantors of [Yo-gan] G.yu-la-skyes with the guarantor seals of Qan-hwa-hwa, Den-bun-ɣde, Čan-stag-bzér, Yo-gaṅReɣu-skyes, Gñi-ba Lha-mthon, and Síg-śin-śin, among others, with the personal seal of the person concerned (i.e. Yo-gaṅG.yu-la-skyes), and with the witness seal of an aristocrat-judge.

Tibetan Text
The text has been transliterated by the author on the basis of scans made available on Gallica. 24 The document consists of  lines of text immediately followed by eight seals in red gnod par glo ba čhuṅ(Liɣi yul luṅbstan pa, D , spriṅyig, nė r) 'one is dejected about the harm' (Emmerick, Tibetan texts, p. ) In addition, Or./: r preserves the phrase glo čhuṅ, which might be a mere abbreviation of glo ba čhuṅ. We observe that, with one exception (Or./: ), glo ba čhuṅrequires terminative of either a verb stem or a nominalised v. In the former case the verb stem appears to be v: sloṅs su and rku su (< * rkus su; in Pt r). It seems that de Jong treated glo ba čhuṅas a near-synonym of CT sems čhuṅ'a timid mind' (J: b). The latter is attested in modern dialects in the meaning 'caution' (CDTD: ). de Jong's interpretation is supported by another passage from Pt r: rku su yaṅglo ba čhuṅna (l. ) vs brkusu yaṅdog[s] śes (l. ). Both clauses concern Yo-ganĠ .yu-la-skyes, the head of the encampment. In the second passage glo ba čhuṅhas been replaced by dog[s] 'to fear'. On these grounds I propose translating glo ba čhuṅas 'to be desponded, disheartened'. For glo ba 'breast' and its metaphorical meanings in OLT, see J. Bialek, 'Stretching the body, stretching the mind. The OT noun ring revisited', Zeitschrift der Deutschen Morgenländischen Gesellschaft ,  (), p. , fn. .

20
Due to the unknown meaning and function of thum čhu ma the interpretation of the whole passage remains tentative. Ju-cȟaṅand Lug-lun, presumably toponyms, are otherwise not attested. 21 dbyaṅs seems to have been a technical term frequently used in judicial texts in conjunction with bčad 'decided' or zál če 'sentence'. On the other hand, in Pt  it co-occurs with the verb bslab/bslabs (v slob 'to learn; to teach') in two forms: dbyaṅand sbyaṅ(s). On this rather meagre evidence I relate dbyaṅs to CT sbyoṅ'to exercise, to practise; to study' and ɣbyoṅ'to be skilled'; all derived from √bjaŋ.

Discussion
The term sluṅs occurs seven times in the document, sometimes as a simple lexeme, sometimes forming part of a compound (e.g., sluṅs chugs, sluṅs phon). However, its explanation requires examination of at least one more technical term: chugs. To elucidate their meanings, I will first examine persons mentioned in the document who were closely related to the sluṅs and then look at the organisation and services of the latter.

Persons
The document mentions several persons related to the sluṅs. The exact nature of the offices they held is not completely clear, but we learn that the institution was hierarchically organised with a sluṅs phon 'head of the sluṅs' at its head. The following discussion particularises the functions of the persons involved in the events reported in Pt r.
In ll. - 'deputies (sna) of the head (dphon) of the sluṅs' are mentioned, one of whom is chugs phon, 'head of the chugs'. The phrase sluṅs gyi dphon can be identified with sluṅs phon recurring in ll.  & . 25 The compound dphon sna suggests that a sluṅs had a superior called dphon (specifically, * sluṅs dpon), who had at least a few deputies (sna), one of whom was called chugs phon (< * chugs dpon, lit. 'head of the chugs'). From this a hierarchy emerges: a sluṅs phon supervised a chugs phon. The sluṅs phon remains anonymous in Pt r and, we may assume, was not conceived of as in any way involved in the case. 26 However, it is possible that the sluṅs phon would have been brought to justice in case the chugs phon would not have appeared in the court. In Uebach's words, "[t]he most important task of the chief of the post-station (sluṅs phon), apart from checking the insignia of emissaries, was to check whether the seals of the missives the emissaries carried were intact or had been tampered with. If the missive showed signs of having been opened, the emissary was sent back to the previous station for an investigation. There was a potential death penalty if the emissary was found guilty" (Uebach, 'Notes on the Postal System', p. f.) This information is based on a passage from Pt  (ll. r-; for its discussion see A. Macdonald, Ariane, 'Une lecture des Pelliot Tibétain , , ,  et  '. in The head of the chugs in the sluṅs concerned was Yo-gaṅG.yu-la-skyes (ll. -). The head of the chugs was subject to the head of the sluṅs. He was summoned to the court after the groom Qab-sab-ñaṅand the worker Li Zán-ɣdo had not arrived. Hence, we can infer that the head of the chugs was directly responsible for the groom; he was in the capacity of giving orders to the groom (ll. -). He also had to take responsibility for groom's misdeeds. In his accusation the horse owner Hin-ce was asking whether looking after horses was not the duty of the head of the chugs (ll. -).
Pt r documents three distinct offices or positions that seem to have been directly involved in taking care of horses: rta rȷi, khuṅs po, and bu gñer. Because rta rȷi apparently had some kind of superiority over khuṅs po, I think it more proper to translate the former as 'groom' and the latter as 'worker' (see below). I understand groom as denoting a person responsible for the management of horses in all aspects, whereas worker would have been responsible for feeding, cleaning, etc. To judge from the etymology of bu gñer (< * bu gñer ba), the term denoted a minion helping in the sluṅs.
A sluṅs had a groom-sluṅs gyi rta rȷi. In the sluṅs under discussion it was Li Qab-sab-ñan( ll.  & ). 27 The latter was responsible for horses kept in the sluṅs; he had to bind (skri) them and look that they did not run away (l. ). Therefore, when the horses got lost he was the first suspect (l. ). His immediate superior was the head of the chugs (chugs phon), to whose orders the groom had to obey (ll. -).
Li Zén/Zán-ɣdo is once called khuṅs po (l. ) 28 but his role in the events is enigmatic. In l.  we read that the groom Qab-sab-ñaṅhanded a light brown horse over (gthad) to Zán-ɣdo and the horse got lost. From then on Zán-ɣdo, together with Qab-sab-ñan, was accused of losing the horse. They were summoned to the court but did not appear (l. ). Consequently, the head of the chugs, Yo-gaṅG.yu-la-skyes, was summoned and Études Tibétaines dédiées à la mèmoire de Marcelle Lalou [Paris, ], pp. -, and R. A. Stein, 'Tibetica Antiqua . L'usage de métaphores pour des distinctions honorifiques à l'époque des rois tibétains', Bulletin de l'École française d'Extrême-Orient  (), pp. -), but the latter document does not put sluṅs phon in charge of messengers; it does not even mention sluṅs phon. Hence, Uebach's conclusions, even though possible, are premature regarding the textual sources at our disposal. 27 Dotson considered the syllable li in Li Qab-sab-ñaṅand Li Zén-ɣdo (see below) to be a family name (Dotson, 'Introducing Early Tibetan Law', p. ). The latter is a typical transcription of a Chinese name, and so here Li can be identified with the Chinese family name 李. The given name Zén-ɣdo is also attested in Pt :  (cf. Takeuchi, Old Tibetan Contracts, p. ). The name Qab-sab-ñaṅis more problematic. The given name consists of three syllables. Neither Qab nor Sab-ñaṅare found separately, but the name Sam-ñaṅrecurs in documents analysed by Takeuchi and was reconstructed by the latter as a Chinese given name (ibid., p. ). Sam-ñaṅcould have resulted from the assimilation of the original -b to the following nasal: -b > -m / _ñ-. However, in all cases Sam-ñaṅfollows a Chinese family name, but in Pt r it comes after the syllables Li Qab. The problem remains unsolved. . Accordingly, etymologically the most plausible explanation of khuṅs po would be 'a male person (-po) affiliated to an office (khuṅs)'. Since the office in question was a sluṅs and the khuṅs po's duties included taking care of horses, I propose translating the term simply as 'worker', understood as denoting an employee who does manual or non-executive work. obligated to bring the groom and Zán-ɣdo to the court. It follows that Zán-ɣdo was likewise employed at the sluṅs and subject to Yo-gaṅG.yu-la-skyes. Moreover, because he received the horse from the groom Qab-sab-ñan, he must have also been subject to the latter. If khuṅs po denoted an official, he was ranked below rta rȷi.
A third person, apparently helping with horses, was sluṅs gyi bu gñer (l. ) 'minion of the sluṅs'. Nothing is known of this official apart from his relation to the sluṅs and the fact that he occurred to help (l. ). It is also not clear why is he mentioned in the case; the passage (ll. -) seems out of context.
These were the officials working in the sluṅs. Apart from them the document mentions other persons as well. On several occasions an owner of a horse is spoken of: qan bcan zigs chan gyi rta gčig (l. ) 'one horse of Qan-bcan-zigs-chan' śa ču pa ɣi rta gčhig (ll.  and ) 'one horse of the man from Sá-cǔ' śa ču paɣi rta gñi ga (l. ) 'both horses of the men/man from Sá-cǔ' rta bdag hiṅce (l. ) 'horse owner Hin-ce' hiṅche la scogs paɣi rta (l. ) 'the horses of Hin-che, among others' We have two proper names: Qan-bcan-zigs-chan and Hin-ce. In addition, from l.  we infer that one horse was claimed by a certain Jeɣu-hin-yir. In the next line the same person is said to have fled away with two horses. The circumstances are not completely clear, but it seems that Qan-bcan-zigs-chan and Hin-ce kept their horses in the sluṅs and Jeɣu-hin-yir used the opportunity to steal the horses. Once the text speaks of 'one horse of the man from Sá-cǔ', once of 'both horses of the men/man from Sá-cu'. The most plausible explanation is that both Qan-bcan-zigs-chan and Hin-ce were from Sá-cǔ and each kept one horse in the sluṅs. However, contrary to Hin-ce, Qan-bcan-zigs-chan does not seem to have been involved in the case.
Jeɣu-hin-yir seems to be the thief; he came to the chugs claiming that his horses were there (l. ) but he fled riding on one horse and leading the second one along (l. ).
Pt r mentions yet another person: messenger ( pho ña) Gzáms-khon-khri (ll. - and ), who came to the chugs, asking for the horse of Qan-bcan-zigs-chan. He was apparently sent by Qan-bcan-zigs-chan to bring the latter's horse back. In this context we may remark that persons who attended the sluṅs and changed their horses there were referred to as pho ña (see l. ). 29 The document ends with the (poorly preserved) seals of eight persons involved in the case whose names and positions are given as:

Guarantors: Qan-hwa-hwa
Den-bun-ɣde Čan-stag-bzér Yo-gaṅReɣu-skyes 30 29 Pt r only mentions messengers in connection with the sluṅs. This however does not mean that nobody else was entitled to use the services of the sluṅs, as asserted by Uebach (Uebach, 'Notes on the Postal System', p. ). Merchants or Buddhist pilgrims are two other groups that must have visited sluṅs on their long journeys. We know from later sources that many of the stage stations were located close to market places as shown on the maps of the Wise collection (Lange, An Atlas of the Himalayas, p. ) and, for example, Skra-bdun (Tradün) stage station was even located within Skra-bdun monastery (ibid., pp. -).

30
Possibly a relative of Yo-gaṅG.yu-la-skyes. In this case, Yo-gaṅwould have been a family name (but compare fn.  above).
Gñi-ba Lha-mthonṠ́i g-śin-śinḊ efendant: Yo-gaṅG.yu-la-skyes (chugs phon) Witness: an anonymous aristocrat-judge Internal organisation of the slunṡ The internal organisation of the institution as depicted in Pt r can be partly reconstructed on the basis of the offices that formed it. In the preceding section I discussed the following officials: sluṅs phon 'head of the sluṅs' sluṅs gyi dphon sna 'deputies of the head of the sluṅs' chugs phon 'head of the chugs' rta rȷi 'groom' khuṅs po 'worker' bu gñer 'minion' pho ña 'messenger' The institution itself consisted of several distinct compartments. Its most general name was sluṅs. It was managed by the head of the sluṅs. Within the sluṅs there was a sluṅs chugs (l. ), lit. 'chugs of sluṅs', also simply referred to as chugs, in which horses of messengers were put (bzág). The chugs concerned in Pt r is called 'chugs of Par-kog' (l. ). This suggests that a sluṅs could have several chugs and each of them bore its own name. A chugs was overlooked by the head of the chugs. A chugs had a chugs khor (< * chugs ɣkhor), lit. 'chugs-pen', where horses stayed overnight (ll. -). Grooms and workers took care of horses that were staying in the chugs. The field of responsibility of minions is difficult to establish. The head of the sluṅs had his private courtyard (g.yul thog) in the sluṅs (ll.  and ), which was used to separate horses of special guests from plain horses. The picture of the sluṅs that emerges from Pt r reveals its complex and hierarchical organisation. Regarding the meanings of particular terms that recur in the text, the above analysis allows for the following interpretations: • sluṅs 'stage station', managed by a sluṅs phon 'head of the stage station'; it included one or more 'encampments' (chugs); • chugs 'encampment, camp site, base' denoted a site within or in the direct proximity of a stage station where horses were held; it was managed by a chugs phon 'head of the encampment'. chugs most probably also encompassed accommodation sites for humans, like a special building ( * chugs khaṅ) or tents; 31 31 The word chugs was derived by conversion from v of the verb ɣȷug (for analogous derivatives in OLT, see J. Bialek, 'Old Tibetan verb morphology and semantics: An attempt at a reconstruction', Himalayan Linguistics ,  (), pp. f.). Its etymological meaning can be reconstructed as * 'sth. that is settled, established'. In the meaning 'encampment, camp site' chugs entered into CT lexicon in compounds like chugs khaṅor chugs sa 'caravansary, or merely a level, open place near a village, where traveller's (sic) may encamp, or where public business is transacted' (J: a). chugs sa is also attested in modern Balti with the meaning 'place where one can stay, especially for the raja and his • chugs khor 'pen' (lit. 'encampment-pen') denoted an enclosure in which horses were kept overnight.

Conclusions
Information on the internal organisation of a sluṅs provided by Pt r is scanty. Nevertheless, in this respect, the document discussed in this article is our best source for the period of the Tibetan Empire. Even though travel literature, native as well as foreign, is exceptionally abundant for the post-imperial period, thus far no detailed descriptions of the internal organisation of stage stations have surfaced. 32 We find occasional pieces of information strewn I think that this interpretation is based on a misunderstanding. Neither Thomas nor Takeuchi have quoted any passage that would unanimously show chugs as referring to a group of humans. Thomas also presented an alternative interpretation: 'camping arrangements' (ibid). I think that in military contexts chugs denoted a base or a camp site too, whereas members of a group stationed there were called chugs pa (Or./: r). Uebach followed Takeuchi in interpreting chugs as 'a small military unit of four watchmen' (Uebach, 'Notes on the Postal System', p. ) and was therefore compelled to conclude that the sluṅs of Pt r had a military watch. That this interpretation is flawed is most clearly seen in ll. - where a messenger comes to a chugs looking for a horse or in ll.  and  where a horse is put in a chugs. Apart from that, Uebach does not seem to be aware of the semantic shift she had to make in order to reconcile the textual data with Takeuchi's interpretation; a group of people, 'watchmen', is taken in her analysis to be identical with the location at which this group served, 'watch'. Mon. morin = rta, whereas the second syllable, jam ('road, route, way or pass', F. D. Lessing, Mongolian-English Dictionary [Berkeley, ], p. ; concerning its etymology, Kotwicz wrote: "Aussi dans la phase initiale ʓˇam ∼ yam pouvait désigner les distances entre les points fixes où l'on relayait les chevaux, plus tard, ces points même, et, finalement, la ligne de communication tout entière, y compris les points de relais et les intervalles intermédiaires", W. Kotwicz, 'Contribution aux études altaïques', Rocznik Orientalistyczny  (), p. ), was independently borrowed into Tibetan as ɣȷ̌am from which ɣȷ̌am mo (also spelled ɣȷ̌aɣ mo, Laufer, 'Loan-words', p. ) 'Poststation' (Sch: b) was derived. Laufer remarked that "under the Mongols, Tibet was divided into twenty-seven ȷ̌am ('departments'), a chief officer (ȷ̌am dpon) being appointed in each" (ibid.; Petech likewise mentioned  (Petech, 'Tibetan relations', p. ), but Maurer spoke of  stage stations, Maurer, 'The Tibetan Governmental Transport', p. ). I assume that the form rta zam resulted from folk etymologisation in which the original -ɣȷ̌am was replaced by a better-connoted throughout the literature, as, for instance, the following remarks in the journey report of Montgomerie: These Tarjums are from  to  miles apart; at each, shelter is to be had, and efficient arrangements are organised for forwarding officials and messengers. The Tarjums generally consist of a house, or houses, made with sun-dried bricks. The larger Tarjums are capable of holding  to  men at a time, but some of the smaller can only hold a dozen people; in the latter case, further accommodation is provided by tents.
[…] Each Tarjum is in [the] charge of an official, called Tarjumpá, who is obliged to have horses, yaks, and coolies in attendance whenever notice is received of the approach of a Lhasa official. From ten to fifteen horses, and as many men, are always in attendance night and day. Horses and beasts of burden (yaks in the higher ground, donkeys in the lower) […] are supplied by the nomadic tribes, whose camps are pitched near the halting houses. 33 Montgomerie's observations thus concur with the information retrieved from Pt r. A stage station (sluṅs) was a complex institution consisting of several compartments that were called encampments (chugs). Each of these provided accommodation for a distinct group of travellers or messengers (either in houses or in tents). 34 In addition, each encampment possessed its own pen (chugs khor) where horses (or other pack-animals) were kept separately, most probably so that they did not get mixed up and could be returned to their owners after the tax service has been fulfilled. It is conceivable that the messenger Gzáms-khon-khri, who came to the encampment looking for the horse of Qan-bcan-zigs-chan (Pt r: -), was to bring back the horse to its owner. 35 We can speculate that each chugs was dependent on tax services of one particular community of tax-payers, either a group of households, a village, or a nomadic camp: rta zams of later times were supplied with horses and cattle by the nearby living nomads as part of their tax obligations. 36 Montgomerie reports that, depending on the topography of the area, either yaks or donkeys were kept. This agrees with the information from the Old Tibetan Annals that one distinguished between stage stations located in the upper and in the lower parts of the country: mṅan zam 'bridge': * rta ɣȷ̌am > * rta ȷ̌am > rta zam. Das noted two pronunciations: tazam and tajam (S. C. Das, A Tibetan-English dictionary with Sanskrit synonyms [Delhi, ], p. b); the latter still reflecting the original * rta ɣȷ̌am and suggesting that the folk etymology rta zam was a local development and had not spread over the whole Tibetan speaking area (n.b., Das' etymology reading rta zam as 'horse bridge' (ibid. and S. C. Journey, p. ) is obviously mistaken, as already noticed by Laufer, 'Loan-words', p. ). It is feasible that the change * rta ɣȷ̌am > rta zam first occurred around stage stations located in a vicinity of a bridge or a river ferry. Three such stage stations can be identified on the basis of the maps from the Wise collections: Čhu-śul (Chushul; no. The vowel -u-in tarjum and tazum is an English transcription of the Tibetan short vowel -a-in a closed syllable (cf. Eng. sum [sʌm]). Likewise, the word internal -rj-doubtlessly mirrors the English pronunciation of the compound * rta ɣȷ̌am. Apart from the simple compound rta zam, one also encounters formations like rta zam ɣjaɣ mo 'Relaispost, Poststation' (Corff, Auf kaiserlichen Befehl, vol. , p. , .). The latter was most probably coined to disambiguate the meaning of rta zam after the origin of zam (< ɣȷ̌am) had already fallen into oblivion. 33 Montgomerie and Pundit, 'Report of a Route-Survey', pp. f.

34
The list of rta zams provided in ibid., pp. f. also contains a short description of each place, indicating what kind of accommodation was available there. 35 This again suggests that the stage station of which the encampment Par-kog formed part was located not far away from Sá-cu. 36 Maurer, 'The Tibetan Governmental Transport', pp. f. daṅ/ sluṅs stod smad gyıthaṅkhram čhen po btab / (ITJ : -) '[The council] issued great tallies of jurisdiction for mṅans and the upper and lower stage stations'. 37 We find common traces in the organisation of the imperial sluṅs system and the postimperial rta zam system re-established by the Mongols in the thirteenth century. Our knowledge remains very limited, but it is conceivable that the Mongols did not create the system, but rather reformed the existing one that must have survived the disintegration of the Empire, if not for the sake of information circulating then at least to support regional trade.