The Patterned Guidelines of Shazhou (Shazhou tujing) and geographical practices in Tang China

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers. L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés. The Patterned Guidelines of Shazhou (Shazhou tujing) and geographical practices in Tang China Alexis Lycas

temporally-identified documents, the easily recognizable subgenre of "patterned guidelines" (tujing 圖經) amounts to almost one quarter of the total number of the transmitted local writings. With regard to the history of geographical knowledge, they are considered as the forebears of "local gazetteers" (difang zhi 地方志), the single most important genre for our knowledge of the local history and geography of the second millennium of the Chinese empire.
By geographical "genre" and "subgenre," I refer to historical actors' categories as displayed by the official classification of historical knowledge. During the Tang, works falling under the geographical (dili 地理) genre of writing could be further divided into subgenres according to the period during which they were compiled and their terminologies and titles: most of the documents that were located in the imperial library were "accounts" (ji 記), "treatises" (zhi 志), "maps" (tu 圖), "traditions" (zhuan 傳), "illustrated treatises" (tuzhi 圖志), and "patterned guidelines" (tujing). 10 Some scholars have suggested that tujing, tuji 圖記 and tuzhi 圖志 were interchangeable suffixes referring to identical works: local treatises with maps. 11 Without overemphasizing the importance of the titles' terminology-several titles were given retrospectively and some titles can indeed be found under different suffixes-I contend that there are differences between these subgenres of geographical writings with regard to their period of production, their content, and their geographical scope.
Temporal discontinuities among subgenres suggest that "accounts" (ji) and "treatises" (zhi) were mostly produced during the period of the Northern and Southern dynasties. Then, between the Sui dynasty (581-618) and the period of the Five Dynasties (907-960), patterned guidelines became the foremost subgenre of geographical writing. Finally, local gazetteers gradually replaced patterned guidelines during the Song, and became the predominant genre throughout late imperial times. 12 Whether "illustrated" (by a tu 圖) or not, the "accounts" and "treatises" written between the Han and Tang dynasties were mainly concerned with ethnographic accounts and contained elements pertaining to cultural history, such as local customs and practices as well as records of oddities. This does not mean that such elements were not found in later patterned guidelines, but rather that they were a prominent component of "accounts" and "treatises" and a minor one of patterned guidelines, which tended to be systematic instructions on how to best administer localities. Furthermore, most ethnographic accounts were produced in specific regions of the territory, mainly the south, and especially the Middle Yangzi area; conversely, patterned guidelines were produced more evenly in localities all across the realm. As for the differences between "illustrated treatises" and "patterned guidelines," the former number significantly fewer than the latter. That was because the "illustrated treatises" were imperial geographies, large scale compilations of regional and local documents-including patterned guidelines-that were meant to describe the empire in its entirety, such as the Illustrated Treatises of the Counties and Prefectures of the Yuanhe Era (Yuanhe junxian tuzhi 元和郡縣圖志). 13 The 90 extant patterned guidelines are made up of approximately 350-400 surviving quotations, providing a representative amount of textual material to reconstruct developments in geographical knowledge during that period. 14 However, most of the transmitted quotations of Han-Tang patterned guidelines reflect the concerns and strategies of Song compilers, rather than those of the original Han-Tang authors. Consider standard entries in Song encyclopedias such as the Taiping yulan 太平御覽 (Encyclopedia of the Taiping Era Read by the Emperor) and the Taiping guangji 太平廣 記 (Vast Records from the Taiping Era), both of which explicitly quote from patterned guidelines: the former only quotes bare topographical descriptions of a place, while the latter selects excerpts of "strange accounts" (zhiguai 志怪). The reception of these quotations by commentators of the second millennium of the empire displays a binary and limited vision of the content of patterned guidelines, and, moreover, completely overlooks their structure. 15 By comparing unaltered extant content from manuscript copies with transmitted quotations, however, it is possible to bridge the gap between what was transmitted and what was omitted. Indeed, examining the Shazhou tujing shows what was deemed worth mentioning in and quoting from a tujing and what was not. It also becomes possible to complete information regarding what was transmitted in space-from the locality to the capital-and what was 10 See, for instance, the bibliographical chapter of the Book of Sui (Sui shu 隋書): Wei (656/1973, Ch. 33, pp. 982-988); and Blitstein (2019, p. 339). 11 Huang (1993, p. 147);Hargett (1996, p. 409). 12 Huang (1993, p. 153). 13 Liu, Zheng, & Liu (2016). 14 Hua (2007a). 15 F. Li (982/1960;978/1981). transmitted in time-from one repository to another. Thus, I use the Shazhou tujing on an "empirical" basis to reconstruct the layout, structure, and content of patterned guidelines. I first examine the layout of P.2005. I then outline its structure and explain how its content is mostly a product of local knowledge. Lastly, I turn to the question of how this utilitarian document facilitated geographical practices, namely movement. But before that, it is necessary to unlock the meaning of "patterned guidelines" (tujing).

| TERMINOLOGY
What are patterned guidelines? Or, rather, why do I choose to translate tujing as "patterned guidelines"? One of the most common misconceptions of the nature of tujing is evident from its translation as "map-classic," or similar translations that promote the idea of the "map" governing the document. 16 Both terms, "map" and "classic," are problematic renditions of tu 圖 and jing 經, respectively.
Jing is usually translated as "classic" (for canonical texts) or "scripture" (for religious texts). Yuan Ke 袁珂 (1916-2001 suggests that in the context of ancient geographical writings, the jing in the title of the Shanhai jing 山海經 (Itineraries of the Mountains and the Seas) could not mean "classic," but rather "itinerary," "trajectory," or "guideline," since jing bears the idea of a spatial progression (as in jingli 經歷). 17 This interpretation can also be applied to other geographical works that have the jing component in their title. The anonymous 3rd century CE Shuijing 水經 (usually translated as Water Classic) is merely a list of major rivers and their basic courses, and certainly not a "classic." The same can be said of the xingjing 星經 ("itineraries of the stars") genre from the Han period. Beyond the obvious correlation between astronomical and geographical knowledge as exemplified in official dynastic histories and monographs, xingjing and tujing have striking similarities in how they put forth questions of motion and trajectories, whether those concern the stars (xingjing) or earthly routes (tujing). 18 In spite of these elements, translating jing as "itinerary" in the context of tujing seems too restrictive: tujing certainly contained a wealth of topographic itineraries along rivers, across mountains and between postal stations, but they also contained topical entries which do not indicate motion or trajectory. Therefore, I suggest using "guidelines" to translate jing, because this term encompasses the geographical, encyclopedic, and utilitarian aspects that coexist in such documents. 19 Turning to tu 圖, Francesca Bray has thought of tu as "templates for action." 20 As I show in this paper, this framing seems particularly well adapted to the way I understand not only tu but also tujing as a whole. Jing and tu are related to each other and do not function independently in this context. Whether one decides to translate tu as "chart," "map," "illustration," "diagram," "pattern," or "organized principle," it must be understood in conjunction with the other component of the title. Yet I argue that in premodern geographical documents the graphic, non-verbal representation embedded in the meaning of tu was less important than one would expect. 21 No maps for the period under scrutiny here have survived. Additionally, early maps did not prioritise accuracy, tending instead to simplify movement in space into an almost abstract form. 22 Moreover, I suggest, for the following reasons, that the tu component of a tujing might very well mean that it was not a map: (a) as I show below, the content of the Shazhou tujing indicates that it was possible to "navigate" on the ground without maps using textual descriptions alone; (b) as no clear reference to the presence of graphic elements can be found, patterned guidelines probably evolved from an earlier form, that did contain "maps" but 16 Will (1992, p. 7); Huang (1993, p. 137);Bol (2004, p. 7); Hu (2003, p. 163); Mostern (2011, p. 75); F. Lin (2017, p. 10);Felt (2017, p. 379); A. Wang (2018, p. 51);Feng (2018, p. 174), to name a few. For translations that stress the opposite perspective, where the written elements govern the document, see Reiter (1990, p. 316): "gazetteers" with maps, "illustrating" texts; Lien (2009, p. 21): "guides, with maps" ;Milburn, (2015, p. 5): "illustrated guides"; Wen (2017, p. 41): "illustrated geography"; and Sun (2017, p. 20): "illustrated records." 17 Yuan (1980). Vera Dorofeeva-Lichtmann (2003) also translates jing as "itineraries". 18 Many thanks to Daniel Morgan for pointing out these similarities. certainly lost this component over time while keeping the "original" terminology; (c) there are, in other Tang documents, instances showing that patterned guidelines were studied or produced alongside "maps" (ditu 地圖). 23 Furthermore, Wolfgang Behr explains tu as related to "positioning" and "ordering," making it an element that is organized, situated, and located. 24 Thus, I choose to translate tujing 圖經 as "patterned guidelines" rather than "mapclassics" because a "pattern" reflects the carefully ordered aspect of a tujing, and also because these documents were not "classics" but dynamic materials meant to be used empirically. Manuscript copy P.2005 was produced on a 935 cm long paper scroll, divided into 22 folios of similar heights (27.3-28.6 cm). According to Jacques Gernet and Wu Chi-yu, the supple and smooth paper is of superior quality.
The text was written in black ink and the characters were in general very well drawn. The copy suffered from humidity over time and traces of grease are now apparent. The bottom right part of the first folio is now damaged. 26 It seems likely that it was the work of a copyist, either a local clerk under the authority of the highest ranked official at the prefectural level, or a professional copyist from a workshop in the capital, Chang'an. 27 There are several scribal mistakes, which were not always corrected, indicating that it was probably not an officially commissioned manuscript copy. In such cases, additional characters were added in a smaller size, below a character, or in the margin, indicating that the text was reread and revised by someone with additional knowledge, as discussed below.
The layout of the document has two striking features. The first is the presence, in all manuscript copies of the Shazhou tujing, of a clear indentation (see Figure 1). The second is the presence of a very wide upper margin. How can these phenomena be explained? The practice of indentation was common in official documents such as requests and official letters. 28 These indented headers may also have looked familiar to a medieval user of dictionaries. However, they seem to be unprecedented in manuscripts containing written accounts of events. Thus, the presence of such strong indentation would classify the Shazhou tujing as a technical document-although it gives ample room to anecdotes and storytelling in general.
Since the manuscript is incomplete and since no indication is given, it is difficult to determine what was on the upper margin, or if there was anything on it in the first place. There could be several material explanations for such a wide upper margin being left blank: (a) the copyist needed more space precisely because he was copying from another manuscript, and thus placed his paper scroll below the original; (b) there were supplementary headers (although the first lines of Chapter 1 as displayed in S.2593v seem to contradict this, see section 4 below); (c) blanks were left on purpose in order to add other elements at a later stage, such as visuals or commentaries; (d) in an economic context with a shortage of quality paper, the use of "more" paper than necessary could suggest that it was an important copy-such as an official document intended to be sent to the court, although it was probably never sent given the mistakes it contained-or that the person who had it copied had sufficient means to overlook the reproduction cost. 29 At any rate, these blanks are unique and were presumably intentional. Since this technique of systematically indenting a narrative is unprecedented in earlier manuscripts, the Shazhou tujing is the earliest extant example.
This aspect alone challenges preconceptions regarding the organizing principles of premodern geographical documents, the beginnings of the standardization of geographical knowledge, and the practical dimensions of its usage.
This indentation is particularly obvious in the following entry, which displays three different levels (see also above, Figure 1, 6th to 10th column from the right): The example of the Yiqiu canal shows that the standard structure is defined by headers and subheaders that display names and provide initial mentions of distances worth knowing. The header is followed by the canal's location of origin and the hydraulic constructions linked to the entry, which are displayed with their own measurements. The entry ends with an etymological explanation relating to the name of the place under consideration-such etymologies were common in toponymic practices. The layout is a visual aide and the characters are equally important "signs" and "visual expressions of knowledge." 31 Together, they help the reader to identify faster and better the information sought; they function in the same manner as the styles and the navigation pane in a modern Word document. The Shazhou tujing is a written account, but it is encapsulated in a technical document that has a strong and easily recognizable structure. It is this very structure that helps the intended audience to navigate through the text. As Lin Kaishi 林開世 argues for later documents, it derives its efficacy and authority from its form (xingshi 形式). 32 The practical consequence for users of tujing is the facilitation of the actual navigation in the physical space. These lines from the very beginning of Chapter 1 provide a general table of contents, which displays how the usual type of information that could be found in a tujing was classified. Before going any further, it is necessary to present what is in the manuscript copies that we have: Chapter 1, on Shazhou 沙州: S.2593v , held at the British Library. Only six lines of text survive, from the beginning of the manuscript. It presents the administrative toponyms, the characteristics of the soil-that is, whether or not it is suitable for agriculture-and the types of trees, with an assessment of which trees grow in Shazhou in comparison with the traditional types of trees that are supposed to be found throughout the realm.
Chapter 3, on Shazhou 沙州: Two manuscript copies, P.2005 (beginning missing, 513 lines) and P.2695 (beginning missing, 79 lines), both held at the Bibliothèque nationale de France. 35 P.2005 is an almost complete manuscript. These 513 surviving lines are important as they represent one third of the total part of the Shazhou tujing that focuses on the prefectural seat, and thus the only testimony of the unaltered layout, structure, and content of a tujing. Although a small part of the beginning is missing (the bottom half of the first folio, a total of 120 characters), it starts off with the section on rivers. More specifically, it must have started with the course of river Ganquan 甘泉, since that is the main waterway flowing in and out of Shazhou, and since the second header, "Canals," starts with canals located on the Ganquan. The first 12 lines begin with depictions of surrounding flora and fauna. A total of four rivers (shui 水) are described, followed by seven canals (qu 渠), one moat (haoqian 壕塹), three marshes (ze 澤), two dikes (yan 堰), one ancient dam (gudi 古堤), one ceremonial hall (dian 殿), several salt pans (xianlu 鹹鹵) and three salt ponds (xian chishui 鹹池水), one oasis (po 泊), 19 abandoned postal stations (yi 驛), a prefectural school (zhouxue 州學), a county school (xianxue 縣學), a medical school (yixue 醫學), two altars (tan 壇), four shrines for deities (shen 神), one "oddity" (yiguai 異怪), two temples (miao 廟), one tomb (zhong 塚), three halls for conducting royal affairs (tang 堂), one trench ( fortresses (gucheng 古城), a description of "Zhang Zhi's ink pond" (Zhang Zhi mochi 張芝墨池), a list of 19 entries that the prefecture does not possess, 20 propitious omens (xiangrui 祥瑞), and finally one folk song (geyao 歌謠).
Chapter 5, on Shouchang county 壽昌縣: P.5034, held at the Bibliothèque nationale de France, 181 lines survive. 36 Although the manuscript is badly preserved with many missing characters, it remains possible to identify the following entries: two (or more) Buddhist monasteries (si 寺), one stupa (kan 龕), one county school (xianxue 縣學), one altar (tan 壇), four mountains (shan 山), two marshes (ze 澤), two springs (quan 泉), one lake (haishui 海水), two canals (qu 渠), two gills (jian 澗), two ancient passes ( guan 關), two(?) ancient fortified walls ( gucheng 古城), the Shicheng garrison, one monastery (si 寺), six roads (daolu 道路), four abandoned fortresses ( gucheng 古城), one watchtower (ting 亭), and one Zoroastrian temple (xianshe 祆舍). Two elements pertaining to the structural features of P.2005 make me think that patterned guidelines such as the ones found in Dunhuang belonged to a genre of geographical documents that was highly codified in Tang times, and probably earlier. After the entry on "Zhang Zhi's ink pond"-which could be an interpolation ordered by the prefect Liu Wukui 李無虧 (fl. second half of the 7th century)-the reader is presented with a list of 19 elements such as "military colonies" or "local worthies" that should be accounted for in a regular tujing. 39 The list is followed by a statement: "None of the types of entries described on the right exist in the current prefecture." 40 Although they do not supplement any content, these entries are listed nonetheless. This might be because these elements once existed in Dunhuang, and were thus part of the locality's history. But I think it is more likely that these are entries a reader would expect to find in any Tang-dynasty tujing. The same hypothesis is applicable to the case of postal stations, which will be examined below: they are mentioned, despite having been abandoned between the time they were 36 Shazhou tujing 沙州圖經 [Ch. 5], P.5034, Dunhuang Manuscripts, Pelliot Collection, BnF. 37 Will (1992, pp. 18-22); Hargett (1996, pp. 414-415); Bol (2001, p. 45); Dennis (2015, p. 30). See also Ikeda On's hypotheses (Ikeda, 1975, p. 38). 38 For a study and translation of S.5448 (Dunhuang lu), see Giles (1914). 39 These 19  erected and when they were recorded in the Shazhou tujing. These lists of elements are undoubtedly structural features of patterned guidelines, at least during the Tang, and they also speak of and define what patterned guidelines were: structured documents whose content was constantly evolving.

| LOCAL ACTORS AND THE PRODUCTION OF GEOGRAPHICAL KNOWLEDGE
By moving from the structure to the content, it is apparent that P.2005 is organized around thematic lines or itineraries (roads, rivers), and points or markers (temples, schools). Within the structural framework of the entries listed in the previous section, P.2005 in fact revolves around historical actors who are often mentioned in conjunction with relevant sources, and more importantly, with regard to geographical practices and to how they impact the evolution of the locality. If this is such a core aspect of the Shazhou tujing, how are these actors and practices conveyed? When describing the deeds of historical actors, written accounts (local records, standard histories) and oral history (  epigraphic testimonies, which were expressions of the power of the locality. What is striking here is that beyond their mere mention, they are the central actors in the tujing. I consider that this is one of the specific ways of creating a sense of belonging to a place. By recording how exclusively local actors engineered a locality through their actions, authors of patterned guidelines produced knowledge that is embodied in and made possible by the very medium they use. This performative aspect is reinforced by the fact that actors and authors can be the same person, as the next example will show.
A prominent local actor, prefect Li Wukui was an important figure whose local deeds went unnoticed in the official historiography. However, he is mentioned nine times in P.2005, and is portrayed as a paragon of official benevolence. As Sun Yingying argues, he was probably the author of one or more versions of the Shazhou tujing. 43 That his deeds are presented in an overly favourable manner is not my primary concern here. What matters in terms of geographical information is that the various administrative practices he was responsible of-building a weir to irrigate fields, relocating three postal stations for the benefit of travellers, and reporting four propitious omens, including one he attributes the discovery to a commoner-have agency over the place, through Li Wukui's actions.
The Shazhou tujing is, I argue, a fruitful entry point to reassess important questions pertaining to the production of bureaucratic knowledge in a premodern imperial context. Specifically, it sheds light on the "points of intersection between the central government and local societies," and on the transmission of "local information to central government officials." 44 In other words, it shows how the local informs the imperial in terms of knowledge formation.
The manuscript copies of the Shazhou tujing provide practical solutions for understanding, at a specific local level (a frontier prefecture) and at a particular moment in history (the Tang) trative, and social change stem from actors at the local level. 46 At first glance, the information provided to the reader appears to be a history of the local construction of canals, moats, weirs, and dams. But it actually is a history of how those works should matter to someone in charge of acting upon the locality, in terms of management, usage, and maintenance. When actors are mentioned, it is usually for their actions: how they solve problems relating to the water supply, food preservation, transportation, and so on. This type of practical information was useful for future officials, who would be inspired by the deeds of Chen Xuangui, Li Wukui, and their likes, and also for literati searching for other material. 47 43 After a careful examination of the historical references contained in the folk songs that appear at the end of Chapter 3, Sun Yingying attributes the authorship of the songs to Li Wukui, and therefore considers Li Wukui as a potential compiler of the Shazhou tujing: Sun (2017, pp. 227-229). 44 Dennis (2015, p. 3). 45 Lamouroux (2002); De Weerdt (2016). 46 On the prominent role played by local officials during the Tang, see Twitchett (1957, p. 36;1960, p. 182 Generally speaking, the document conveys an implicit opposition between exceptional individuals (nobility, kings, and officials) and unidentified commoners. 48 In fact, commoners do appear here and there in the Shazhou tujing, as they played quite an important part in the construction of the locality. Some did it implicitly, others more explicitly. There are four instances of commoners (baixing 百姓) explicitly discussed, all in the section on omens: in 666, Yan Hongshuang 嚴洪爽 obtained a propitious stone in a temple, and in 671 Wang Huichang 王會昌 managed to capture a white sparrow; Yin Sijian 陰嗣鑒 observed a multicoloured bird and Yin Shouzhong 陰守忠 saw a white wolf, both in 691. 49 Because these propitious events allegedly had an influence at the imperial level, their names were recorded. But even when commoners remained anonymous, they played an active role in the construction of the place, as cause, consequence, or both: the sustainability of their agricultural practices forced civil and military officials to engage in hydraulic works; they were displaced in times of conflicts and such relocations modified borders; they also participated in the legitimation of a ruler, Wu Zetian, through auspicious omens and the composition of a song glorifying her actions. 50 In technical documents, practices are usually described because they explain how to make things work at the local level. In the political and social context of the first millennium of the Chinese empire, this is usually linked to the empirical functioning of an administrative unit. In P.2005, two related categories and subcategories of technical information emerge: matters of livelihood (agriculture, irrigation) and protection (civil, military, ritual). In each case, the reader, whether an official or a member of the local elite, is provided with verified and reliable information that will help him avoid (past) mistakes, and thus waste as little time as possible when it is time to make decisions. For instance, descriptions of commoners building dams and sluices to irrigate the surrounding cultivated fields empirically show how irrigation leads to agriculture. These are followed by descriptions of seasonal variations and their impact on the river and the soil. 51 Salt production and extraction is facilitated by comparisons between ponds with regard to salt quality. 52 As for ritual matters, ancestral temples, altars, shrines, schools, accounts of the strange, and records of propitious omens all have a specific function. P.2005 provides guidelines to assist in performing rituals at the appropriate time and location. Ritual-or architecture-related peculiarities are mentioned because one needs to know about them: the placement of relics (such as tablets, zhu 主) are indicated both to facilitate access to them and to use them efficiently; temples are listed as repositories of local worthies, whose actions are recorded in their post-mortem inscriptions. 53 Lastly, the 20 omens listed towards the end of P.2005 are all precisely located. They are not simply mentioned as having happened at some point in time. They are written down in conjunction with a location: by the side of a mountain, by the banks of a river, inside a ceremonial hall, and so on. Whether concerned with sources, past events, or local actors, each element that appears in patterned guidelines confirms that these documents were produced in order to fill a gap about knowledge on localities.

| ON THE GROUND: GEOGRAPHICAL PRACTICES
Having illustrated how the reader could situate himself and move within the document, I now show how readers could proceed empirically with it. Obviously, the question of movement is important, but it is not the only one addressed by 48 Min 民 appears quite ironically as ren 人, because min was part of the personal name of Li Shimin, the founder of the Tang  a tujing. Not all users of patterned guidelines consulted them for empirical reasons. Some literati read them to gain knowledge about unfamiliar places before discovering them physically (see Han Yu's example above), or to supplement their empirical experience after a visit. Others used them for their encyclopedic value without any intention of travelling. My focus here is on the usage of patterned guidelines as "utilitarian" documents. As such, they include information on mobility and anything else that newly appointed officials or literate travellers could use to gain rapid and practical knowledge about unfamiliar places, in a reliable and standardized format that such users could expect in any jurisdiction across the empire. As such, I focus on how motion was conveyed through measures and associated markers in P.2005.
Measurements are social and referred to because of their commensurability, meaning that they can be represented and assessed by groups-authors, readers, and users of geographical documents-according to shared standards.
This applies to P.2005 in the case of distances between two points, as well as for the dimensions of various buildings (temples, halls) or landmarks (mountains, lakes). The consequence was that literate people could work their way through Shazhou equally well whether they were from the capital or the southernmost part of the realm.
In order to understand how a Tang  These entries describe the timeframe and reasons for the abandonment of stations. Located on the "southern" route linking Shazhou to Guazhou, these three stations were abandoned in 683, following the relocation of the route along a safer itinerary. They were replaced by stations 13-16. Yet, it was not long before that second "southern" route was also abandoned: in 691, Li Wukui requested to replace it with a "northern" route (stations 1-7) that was less winding and less prone to banditry ( Figure 2). However, the main information regarding postal stations concerns distances and their practical value. Stations are mentioned for their commensurability, which is reinforced by their relational value. These measures make sense to the user as long as they can be followed from point to point, along an itinerary that can easily be traced and surveyed empirically. Furthermore, the 19 postal stations are organized in four geographical and temporal sequences, each comprised of three to seven stations (see Figure 2). Indeed, it is much easier to follow an itinerary with short and regular intervals, such as forts, postal stations, and mountains, rather than a straight line with only points at the beginning and end, and the total distance between the two. Finally, the distances presented in P.2005 are relative; they must be understood in relation to the other distances within one of the four identified sequences of stations. 55 All these mentions of distances imply movement, or the physical motion of a person or document across the topography and between places. 56 Yet what aides to movement were available in Tang China? The first solutions that might come to mind are the map and the compass. These two instruments were already attested then and they were probably used on other occasions, but it is unlikely that they were employed in this context. Maps were not I suggest that the essential information needed to travel between places was given in a different kind of instrument, namely patterned guidelines. They displayed cardinal directions and distances that were both relative and short, which, together with natural or man-made points and markers, facilitated the visual identification of places.
The image of Guangxian station (Figure 3), still standing high in a desert environment, shows that the aforementioned criteria would suffice to identify the next postal station located tens of kilometres from it.
The positions of postal stations provide information on a relational regime of distances that is set according to three criteria: (a) cardinal directions, (b) the establishment of the prefectural seat as a reference point and origin, and (c) the distances from one station to the next. The application of all these three criteria must be feasible from a practical point of view.
Such a utilitarian display of geographical information was not unprecedented. Distances were commonly described in this manner, at the local level.
There are examples, from as early as the 3rd century BCE, of local archives in the Middle Yangzi area (modern Hunan and Hubei) indicating point-to-point distances between localities. 59 However, the combination of the layout, structure, and content of an almost complete manuscript copy such as P.2005 shows that patterned guidelines were most likely already codified by the Tang period, and probably earlier.
Furthermore, medieval tujing were instruments for geographical practices that created knowledge from and about localities. The documents' layout and structure made them tools, and their content, through descriptions of actors and practices, helped to create a sense of place.

| CONCLUSION: LOCAL ARCHIVES FOR ACTION
In this article, we have examined the value of Tang-tujing as practical guidelines, by focusing on: (a) the utilitarian value and empirical implications of the layout and structure of P.2005, mainly in terms of motion, and (b) how these documents were produced by local actors to supplement knowledge about a locality. They certainly had F I G U R E 3 Guangxian postal station (number 8). From Z. Li (1998, unpaginated plate) [Color figure can be viewed at wileyonlinelibrary.com] 57 Huang (1993, p. 137); Lien (2009, p. 22). 58 Medieval Arab travelers, navigators, or pilgrims did not use maps; they possessed other means of orientation, that were considered more useful than maps. See Ducène (2017, p. 69). 59 See, for instance, the so-called "Daoli shu" 道里書, Collection of bamboo slips (Ch. 4), Peking University collection of Qin documents, Beijing, China, which displays point-to-point distances between river and land routes: the slips are presented in the form of "X place and X place are at an X li distance." other functions. As part of a wider spectrum of documents that were produced in various localities of the Chinese ecumene between the Han and the Tang, their use depended on regional and political contexts. They were also important repositories of local information. Yet they were not supposed to be kept forever, as they had to be updated at regular intervals, like modern travelogues and guidebooks. 60 Patterned guidelines were cumulative documents, evolving archives that were meant to be used empirically, but also to be consulted by authors and compilers of empire-wide gazetteers and geographies-or of any other official work-who would find relevant information in them, while specific parts of their content (such as that relating to schools, temples, and city walls) were then reused, either in empire-wide geographical repositories that tended to be general gazetteers, or in thematic parts of general encyclopedias. 61 How did patterned guidelines become local repositories of practical knowledge? The accumulation of temporal layers of ordinary practices constructs a place, and the layout and structure of the document facilitate it. The Shazhou tujing thus appears as the reproducible original material of all subsequent Dunhuang local writings and gazetteers: of the other seven local documents found in the Dunhuang library cave, four are actually updated versions of the earlier Shazhou tujing. 62 This makes a tujing a collaborative document that could be and had to be updated regularly. As a subgenre of geographical writing, patterned guidelines were composite documents. They were intended to supplement information on given localities, as they recorded all information about a place one ought to know, even when some of that information-see the empty entries and the abandoned stations above-could not be found in the surveyed location.
This paper was intended as an aide to navigate through a document whose goal was ultimately to help its readers navigate in space. The manuscript copies of the Shazhou tujing explain the actual layout of patterned guidelines and provide a fuller image of their content. Moreover, P.2005 exemplifies the utilitarian value of the geographical information a Tang user could find in a tujing, and thus relates, to a certain extent, to the concept of "common sense geography" recently developed in the context of the ancient Mediterranean. 63 Finally, they confirm that tujing were not only the precursors of local gazetteers, but also evolving repositories of knowledge for both local and empire-wide geographies and geographers, and, indeed, "templates for action."