Where is the Yellow River Source?

A Controversial Question in Early Chinese Historiography*

Vera Dorofeeva-Lichtmann (Paris)

0 Introduction
Early Han historiography, as represented by the two first imperial histories, the Shiji 史記1 and
the Hanshu 漢書,2 paid special attention to the Yellow River. The “Treatises” on waterways of
the Shiji and the Hanshu – the “Hequ shu” 河渠書 (“Treatise on the [Yellow] River and [its]
Canals”, ch. 29) and the “Gouxu zhi” 溝洫志 (“Treatise on Irrigation Ditches and Trenches”,
ch. 29) – focus on the configuration of its course and canals. The location of its source is the
central issue of the “Memoirs” ([lie]zhuan [列]傳) on the western regions of the Chinese Empire,
the “Dawan liezhuan”大宛列傳 (“Memoir on Dawan,” ch. 123) of the Shiji, the “Zhang Qian,
Li Guangli zhuan” 張騫李廣利傳 (“Memoir on Zhang Qian and Li Guangli,” ch. 61), and the
“Xiyu zhuan” 西域傳 (“Memoir on the Western Region,” ch. 96) of the Hanshu. No other river
or any other landmark received such close attention in the official histories.
The reason for the singling out of the Yellow River and the preoccupation with the “true”
location of its source seems evident – the Yellow River was the main artery of the unified
Empire, serving as its symbol. The necessity to manage the Yellow River for agricultural pro-
duction and avoiding floods was one of the factors that under-pinned political unification.
This paper explores the representations of the Yellow River source discussed in early Han
historiography. The aim is to determine which conceptions of the Yellow River source are
distinguished in the two first official histories, how these conceptions are evaluated and which
eventually became officially recognised.

1 “Appraisals” of the “Dawan liezhuan” of the Shiji and the “Zhang Qian Li Guangli
zhuan” of the Hanshu on the location of the Yellow River source
The question of where the Yellow River source is located is raised in the “appraisal” conclud-
ing the “Dawan liezhuan”. This chapter deals with missions sent to the Central Asia during the
Former Han dynasty.3 Since the “appraisal” is focussed on the Yellow River source problem,
one can see clearly that the quest for its location was the primary goal of these missions.

* I am indebted to John Moffett for corrections of my English. I am also truly thankful to Dorothee
Schaab-Hanke and Martin Hanke for their valuable remarks and suggestions. Any mistakes found in
this paper are my own responsibility.
1 Comp. by Sima Qian 司馬遷, ca 145–87 BC (Beijing: Zhonghua, 1972).
2 Comp. by Ban Gu 班固, AD32–92 (Beijing: Zhonghua, 1975).
3 For surveys of these missions supplied with maps, see Lu Liangzhi 盧良志, Zhongguo dituxue shi 中國
地圖學史 (Beijing: Cehui, 1984), 22–26; Michael Loewe, “The Former Han Dynasty,” in The Cam-
bridge History of China, Vol. 1: The Ch’in and Han Empires (221 B.C. – A.D. 220), ed. Denis Twitchett
and Michael Loewe, (Cambridge, etc.: Cambridge University, 1986), 164–167; Yü Ying-Shih, “Han
foreign relations,” in idem, 407–409. See also a recent study of the Western Region, according to the
Shiji and the Hanshu, by Larisa A. Borovkova, Tsarstva “zapadnogo kraya”: Vostochny Turkestan I Srednyaya
OE 45 (2005/06)
Where is the Yellow River Source? 69

The need to determine the Yellow River source originates from the different locations given for
it in early Chinese texts. Three texts are mentioned in the “appraisal” of the “Dawan liezhuan”
in relation to this issue – the now lost text Yu benji (禹本紀 “Yu’s Basic Records”), the Shangshu
尚書, (meaning specifically the “Yugong” 禹貢 “Yu’s [System of] Tribute” chapter of the Shang-
shu/ Shujing 書經, “The Book of Documents”)4, and, finally, the Shanhai jing 山海經, compiled
about the beginning of the 1st cent. BC.5 The “appraisal” begins with a citation from the Yu
benji, which is the only passage from this text that had survived.6 According to this fragment,
and as reported in the Shanhai jing, the Yellow River emanates from a cosmic mountain –
Kunlun 崑崙 (昆侖)7 – related to the Queen-mother of the West (Xiwangmu 西王母) grant-

Asiya po svedeniyam iz “Shi tszi” I “Khan’ shu” [The kingdoms of the “Western Region”: Eastern Turke-
stan and Central Asia, according to the Shiji and the Hanshu]. (Moscow: Institut vostokovedeniya
RAN – Kraft+, 2001).
4 The “Yugong” (ca. 5th to 3rd cent. BC) is chapter 6 of the guwen edition of the Shangshu (Shangshu zhengyi
尚書正義, SBBY ed.), 6.1a-19b. Cf. The Chinese Classics, Vol. 3, Part 1: Shoo King, trans. James Legge
(Hongkong and London: London Missionary Society’s Printing Office, 1865), 52–67; Chou King, trans.
Séraphin Couvreur (Ho Kien Fou: Imprimérie de la Mission Catholique, 1897, repr. 1971), 61–89; “The
Book of Documents,” trans. Bernhard Karlgren, Bulletin of the Museum of Far Eastern Antiquities 22
(1950), 12–18, “Glosses on the Book of Documents,” idem, 20 (1948): glosses 1352–1396. For the dat-
ing of the “Yugong”, see Qu Wanli 屈萬里, “Lun Yugong zhucheng de shidai” 論禹貢著成的時代,
Zhongyang yanjiuyuan lishi yuyan yanjiusuo jikan 35 (1964), 53–86. For a survey of the textual history of the
Shangshu/Shujing and its scholarship, see Kai Vogelsang, “Inscriptions and Proclamations: On the Au-
thenticity of the ‘gao’ Chapters in the Book of Documents,” Bulletin of the Museum of Far Eastern Antiquities
74 (2002), 138-148.
5 I share the interpretation of the character jing 經 (literally “warp”) in the title of the Shanhai jing ad-
vanced by Yuan Ke 袁珂, Shanhai jing jiaoyi 山海經校譯 (Shanghai: Shanghai guji, 1980), 180–184, who
argues that jing here is not used in the sense of jingdian 經典 (“classic, canon, book”), as in the majority
of text titles comprising this character, but rather means jingli 經歷 (“to go through, to pass, to un-
dergo, to stretch over”). This point of view is apparently supported by Richard E. Strassberg, A Chinese
Bestiary: Strange Creatures from the Guideways Through Mountains and Seas (Berkeley, etc.: University of
California, 2002), who, unfortunately, does not provide much elucidation on his innovative translation
of the title Shanhai jing. I prefer a more neutral translation of jing as “itinerary/itineraries”. For the sys-
tem of itineraries derived from the Shanhai jing, see my article (2003a), “Mapping a ‘Spiritual’ Landscape:
Representing Space in the Shanhai jing,” in Political Frontiers, Ethnic Boundaries, and Human Geographies in
Chinese History, ed. Nicola di Cosmo and Don Wyatt (London: Curzon, 2003), 35–79; idem [2003b],
“Text as a Device for Mapping a Sacred Space: A Case of the Wuzang shanjing (‘Five Treasuries: The
Itineraries of Mountains’),” in Dickhardt, Michael and Vera Dorofeeva-Lichtmann (eds.), Creating and
Representing Sacred Spaces, Göttinger Beiträge zur Asienforschung 2–3 (special issue) (Göttingen: Peust &
Gutschmidt, 2003), 147–210.
6 The “appraisal” apparently singles out this lost text. The Yu benji is cited once and then mentioned two
more times in the “appraisal”. It is not referred to elsewhere in early Chinese texts. One may suggest
that the text disappeared shortly after the Shiji was composed.
7 For a comprehensive account of descriptions of Kunlun Mountain in early Chinese texts, lists of its
identifications, discussion of the possible meanings and origins of its name, perspectives of its interpre-
tation and extensive bibliography, see Manfred W. Frühauf, “Der Kunlun im alten China. Versuch ei-
ner Positionsbestimmung zwischen Geographie und Mythologie,” minima sinica – Zeitschrift zum chinesi-
schen Geist, 1 (2000), 41–67; 2 (2000), 55–94. For a recent essay on Kunlun Mountain from a large histo-
rical and cultural perspective, see Augustin Berque, “Kunlun, la montagne cosmique,” L’Alpe 32, prin-
temps (2006), 78–82.
OE 45 (2005/06)
70 Vera Dorofeeva-Lichtmann

ing immortality.8 The same location of the Yellow River source is found in a series of other
texts – the Mu Tianzi zhuan 穆天子傳 (“Narrative of the Son of Heaven, Mu”, discovered in
AD 279 in the tomb of King Xiang 襄 of the Wei 魏 kingdom, who had died in 296 BC), the
Erya 爾雅 dictionary (ca. 2nd cent. BC), and the Shuijing zhu 水經注 (“Commentary on the
‘Itineraries of Rivers’”) by Li Daoyuan 酈道元 (AD 469?–527).9

Map 110

In modern physical maps Kunlun is represented as a mountain range in the north-western

part of the Tibetan Plateau (see Map 1), but at least seven major identifications of Kunlun
with different mountains in the western regions of contemporary China are proposed in
traditional Chinese scholarship, and even more identifications, sometimes well beyond the
western regions, are advanced in sinological literature.11 The Shangshu, specifically the

8 For studies of references to Xiwangmu in ancient Chinese texts and her relation to Kunlun, cf. Riccar-
do Fracasso, “Holy Mothers of Ancient China (A New Approach to the Hsi-wang-mu Problem),”
T’oung Pao 74 (1988), 1–46; Manfred W. Frühauf, Die Königliche Mutter des Westens: Xiwangmu in alten Do-
kumenten Chinas (Bochum: Projekt-Verlag, 1999); Michael Loewe, Ways to Paradise. The Chinese Quest for
Immortality (London: George Allen & Unwin, 1979), 86–126; Suzanne E. Cahill, Transcendence and Divine
Passion – The Queen Mother of the West in Medieval China (Stanford, CA: Stanford University, 1993).
9 It seems noteworthy that this text was composed in the Northern Wei (AD 420–534) that occupied the
basin of the Yellow River, see the discussion of the treatises of waterways focussed on the Yellow River
below.
10 Reproduced from Harold Fullard (ed.), China in Maps (London: George Philip and Son, 1968), 9.
11 See Frühauf 1 (2000), 43–47.
OE 45 (2005/06)
Where is the Yellow River Source? 71

“Yugong” chapter, delineates the Yellow River using a different and rarely mentioned land-
mark – Jishi 積石 Mountain. This location of the Yellow River source, according to my
knowledge, is not given in any other early Chinese text – apart from in citations of the
“Yugong”, for instance, in the early dynastic histories. Jishi occurs, in particular, in the Shan-
hai jing,12 which is not surprising, as this text is the most comprehensive account of land-
marks, especially mountains, surviving from ancient China.13 But this is a rare case. The
identification of Jishi is, however, no less complicated than that of Kunlun. It is sufficient
to note here that two mountains under the name of Jishi are found in modern maps of
China. One Jishi Mountain (the so-called Small Jishi, Xiao Jishi) is situated on the southern
bank of the Yellow River in Gansu 甘肅 Province, not far from its border with Qinghai
青海 Province, about 25 km north-west of Linxia 臨夏 city (see no. 1 added to Map 1).14
The part of the Yellow River to the west of the Small Jishi is called the Jishi Gorge 積
石峽. Another Jishi Mountain (the so-called Big Jishi – Da Jishi, also called A’nyêmaqên 阿
尼瑪卿 Mountain), more precisely a long mountain range, is found much farther up the
Yellow River, in Qinghai province, in the deep nook of the river (see no. 2 added to Map
1). The latter Jishi is apparently shown as the Jishi of the “Yugong” in the historical maps
of China edited by Tan Qixiang.15
Further attempts to determine which of the possible identifications of Kunlun and Jishi are
the right ones, and where precisely they are found on the earth’s surface, does not seem to be
a fruitful approach to these landmarks in the context of the “true” location of the Yellow
River source, since this question, as will be demonstrated below, is really of an ideological
rather than a geographical nature. For this reason, in this study Kunlun and Jishi are regarded
as cosmological referents rather than geographical ones. This characteristic of Kunlun Mountain has
been pointed out by Deborah Porter, who argues that it is first and foremost a “mythic con-
struct”.16 Since Jishi, as a location of the Yellow River source, plays a role similar to that of
Kunlun, it should also be considered from a cosmological perspective. The main function of
such toponyms is that of a marker of a cosmologically significant “position”. Identification of
such toponyms with real landmarks or their location on the earth’s surface necessarily depends
on specific spatial conceptions or political and ideological needs that differ both between writ-
ten sources and historical periods. The results are bound to be controversial, as in the cases of
Kunlun and Jishi. The only sensible way forward is to explore the locations of these landmarks
as given in the ancient texts, so as to determine their cosmological “positions”, and, eventually,
their cosmological and ideological functions.

12 But not as the location of the Yellow River source. This occurrence will be discussed below.
13 The first part of the Shanhai jing, the Shanjing 山經 (“Itineraries of Mountains”), comprises descrip-
tions of 447 mountains.
14 This Jishi became a Buddhist “holy” mountain.
15 Tan Qixiang 譚其驤 (ed.), Zhongguo lishi ditu ji 中國歷史地圖集 The Historical Atlas of China. Vol. 1:
Yuanshi shehui, Xia, Shang, Xizhou, Chunqiu, Zhanguo shiqi 原始社會、夏、商、西周、春秋、戰國時期.
The Primitive Society, Xia, Shang, Western Zhou, the Spring and Autumn, Warring States Period (Beijing: Ditu,
1982), map on 43–44; idem, Vol. 2: 秦、西漢、東漢時期 Qin Dynasty Period, Western Han and Eastern
Han Dynasties Period (idem, 1983), maps on 3–4, 13–14 and 40–41.
16 Deborah Porter, “The Literary Function of K’un-Lun Mountain in the Mu T’ien-tzu Chuan,” Early
China 18 (1993), 73–106, esp. 76–85.
OE 45 (2005/06)
72 Vera Dorofeeva-Lichtmann

This approach is pursued by Porter (1993), who is primarily concerned with the implica-
tions of the north-western cosmological and astronomical “position” of Kunlun, derived
from the Mu Tianzi zhuan and related texts. To compare with, Marina Kravtsova argues
that, according to the Chuci 楚辭 (“The Songs of Chu”) poetic corpus, Kunlun is associa-
ted with the West, and that neither semi-cardinal directions nor the five-fold framework
played a considerable role in the Chu picture of the world, which was built around the
West-East axis instead. She demonstrates the key role of Kunlun in the conception of
space, characteristic of Chu culture, that of the sacred centre located far in the West.17
Therefore, the cosmological “position” of Kunlun may vary in different sources. I shall
focus on the “position” of Kunlun according to the surviving texts cited in the “ap-
praisal” – the “Yugong” and the Shanhai jing.
Now let us take a closer look at the “appraisal” in the Shiji. It reports that the expeditions
to the West did not confirm that the Yellow River emanates from Kunlun. As the result, this
information is declared as completely false, and texts transmitting it, specifically the Yu benji
and the Shanhai jing, are condemned as unreliable. The “appraisal” clearly demarcates such
texts from the “Yugong” chapter of the Shangshu. The “appraisal” is as follows:18
太史公曰：
禹本紀言：
「河出崑崙。崑崙其高二千五百餘里，日月所相避隱為光明也。其上有醴泉瑤池。」
今自張騫使大夏之後也，窮河源，惡睹本紀所謂崑崙者乎？
故言：
九州山川，尚書近之矣！至禹本紀山海經所有怪物，余不敢言之也。
The Lord the Grand Scribe states:
The Yu benji says:
The [Yellow] River emanates from Kunlun Mountain. [As far as] Kunlun [is concerned,] its
height is over 2500 li. [Kunlun is] the place where the sun and the moon avoid and hide away
from each other in order to sparkle and shine [in turns]. On its summit are the Sweet Spring and
the Jade Pool.
[However] now since Zhang Qian has been sent to Da Xia, [and] exhaustively explored the source of
the [Yellow] River, did [he] witness [with his own eyes] the Kunlun referred to in the [Yu] benji?
Therefore [I] say:
[For] the “Nine Provinces”, the [itineraries by land marked by] mountains [and the] river [itinerar-
ies],19 [it is] the Shangshu (= its ch. 6, the “Yugong”) [that] is close [to their correct locations]! As
far as the amazing things and beings contained in the Yu benji and the Shanhai jing [are concerned],
I do not [even] dare to speak about them!20

17 Marina E. Kravtsova, Poeziya Drevnego Kitaya: opyt kul’turologicheskogo analiza [Poetical tradition of an-
cient China: Attempt of culturological review] (St. Petersburg: Tsentr Peterburskoe Vostokovedenie,
Orientalia, 1994), 168–183. Besides, she makes interesting observations on some relations of Kunlun
to the south-western direction in later sources (idem, 190–194).
18 Shiji 123.3179.
19 The compound of these four characters is a compressed reference to the “Yugong”.
20 Cf. J. J. M. de Groot (trans.), Chinesische Urkunden zur Geschichte Asiens, zweiter Teil, Die Westlande Chinas in
der vorchristlichen Zeit (Berlin and Leipzig: Walter de Gruyter, 1926), 44–45, and Burton Watson (trans.),
Records of the Historian; the Shih chi of Ssu-ma Ch’ien (New York and London: Columbia University, 1969),
299. The provided translation is my own and differs slightly from those by de Groot and Watson.
OE 45 (2005/06)
Where is the Yellow River Source? 73

Almost the same “appraisal” concludes the biography of the generals who led the expeditions
to Central Asia, Zhang Qian and Li Guangli, in the Hanshu:21
贊曰：
禹本紀言：
「河出昆侖，昆侖高二千五百里餘，日月所相避隱為光明也。」
自張騫使大夏之後，窮河源，惡睹所謂昆侖者乎？
故言：
九州山川，尚書近之矣！至禹本紀山經所有，放哉！
The “appraisal” states:
The Yu benji says:
The [Yellow] River emanates from Kunlun Mountain. [As far as] Kunlun [is concerned, its]
height is over 2500 li. [Kunlun is] the place where the sun and the moon avoid and hide away
from each other in order to sparkle and shine [in turns].
[However] since Zhang Qian has been sent to Da Xia, [and] exhaustively explored the source of the
[Yellow] River, did [he] witness [with his own eyes] the Kunlun referred to?
Therefore [I] say:
[For] the “Nine Provinces”, the [itineraries by land marked by] mountains [and the] river [itiner-
aries], [it is] the Shangshu (= its ch. 6, the “Yugong”) [that] is close [to their correct locations]! As
far as what is contained in the Yu benji and the Shanjing [is concerned], [this, on the contrary,] gets
away from the point!22
One can see that the very similar “appraisals” have minor yet interesting differences. One of
them is that Ban Gu mentions only the first (core) part of the Shanhai jing, the Shanjing (“Itiner-
aries of Mountains”). One could suppose that this is due to a copying error, were it not for the
fact that the same “error” can be found in the citation of the “appraisal” from the Shiji in the
Lunheng,23 a philosophical treatise contemporary to the Hanshu, too. In other words, Wang
Chong, the author of the Lunheng, reproduces the “appraisal” from the Shiji with the difference
being that he mentions the Shanjing instead of the Shanhai jing. This means either that the Shan-
jing circulated during this time as a separate text, which seems probable because of its differ-
ence in many respects from the second part, the Haijing 海經 (“Itineraries of Seas”), or that
due to its core role the Shanjing could stand for the entire text.
Another noteworthy difference between the “appraisals” is that Ban Gu omitted the refer-
ence to the guaiwu 怪物 (amazing beings and things).24 The omission results from Ban Gu’s

21 Hanshu 61.2705.
22 Cf. A. F. P. Hulsewé (trans.), with an introduction by Michael Loewe, China in Central Asia (the early
stage: 125 B.C. – A.D. 23). An Annotated Translation of Chapters 61 and 96 of the History of the Former Han
Dynasty (Leiden: Brill, 1979), 237–238.
23 Lunheng 論衡, by Wang Chong 王充, AD 27–97 (SBBY ed.), section 31: “Tan tian” 談天, 11.3b; see
the translation by Alfred Forke, Lun-hêng. Philosophical Essays of Wang Ch’ung. Part I. (Leipzig: Harras-
sowitz, 1907), 254. The “appraisal” of the “Dawan liezhuan” is cited in the Lunheng with some minor
differences that are pointed out in Forke’s translation. For a comparison of the parallel versions in
Shiji, Lunheng and Hanshu, see the table in the appendix.
24 For “amazing being and things” in Chinese culture, see Robert Ford Campany, Strange Writing: Anom-
aly Accounts in Early Medieval China (Albany: Suny, 1996) who refers to them as “anomalies”.
OE 45 (2005/06)
74 Vera Dorofeeva-Lichtmann

intention to completely separate his conception of “terrestrial organisation” (dili 地理)25 from
any allusions to spirits implied by the guaiwu.26
In any case, though slightly different, the “appraisals” of the Shiji and the Hanshu share
completely the evaluation of the Yu benji and the Shanhai jing as unreliable sources of locations,
in marked contrast to the “Yugong”, which is considered as the main reference text for “true
locations”. The location of the Yellow River source serves here, therefore, as a criterion for
evaluating the reliability of terrestrial descriptions.27
In order to fully understand the spatial conception underlying the “appraisals”, it is neces-
sary to thoroughly explore the references to the Yellow River source in the two surviving texts
mentioned here – the “Yugong” and the Shanhai jing – especially since, in contrast to the Yu
benji, they are not cited in the “appraisals”. The primary goal of this exploration is to determine
“positions” of Kunlun and Jishi Mountains, according to these two texts.

2 The Yellow River Source and locations of Kunlun and Jishi in the “Yugong”
and the Shanhai jing
2.1 Jishi and Kunlun in the “Yugong”
The description of the Yellow River in the “Yugong” corresponds to the third river itiner-
ary. Some remarks on the system of itineraries of the “Yugong” are necessary. Two sets of
itineraries are featured in this text – nine itineraries by land marked by mountains and nine river
itineraries. The itineraries are described immediately after the “Nine Provinces”, whose de-
scription constitutes the major part of the “Yugong”. The “Nine Provinces” cover the ba-
sins of the Yellow and Yangzi Rivers and serve as a general framework for representing the
“civilised world” in Chinese culture. Apart from being structurally consistent with the
“Nine Provinces”, the itineraries are marked off by many landmarks mentioned in the de-
scriptions of the “provinces”, and are delineated across them. The itineraries, therefore,

25 This conception is formulated in the “Dili zhi” 地理志 (“Treatise on Terrestrial Organisation”)
chapter of the Hanshu, which does not have a counterpart in the Shiji. Ban Gu’s conception of dili
builds on the “Yugong” as its theoretical foundation and can be defined as “an orderly administrative
territorial division of the Empire established by the ruler and aimed to symbolize world order,” see
Vera Dorofeeva-Lichtmann, “I testi geografici ufficiali dalla dinastia Han alla dinastia Tang,” in “La
scienza in Cina,” ed. Karine Chemla et al., in Enciclopedia Italiana, Storia della Scienza, Vol. II, ed. Sandro
Petruccioli. (Rome: Istituto della Enciclopedia Italiana, 2001), 190–197; idem, “Geographical Treatises
in Chinese Dynastic Histories: ‘No Man’s Land’ Between Sinology and History of Science,” in Science
and Cultural Diversity, Proceedings of the XXIst International Congress of History of Science, Mexico City, 8–14
July 2001, ed. Juan José Saldaña (México City: Sociedad Mexicana de Historia de la Ciencia y de la
Tecnología-Universidad Nacional Autónoma de México, 2005), 2262–2279.
26 I discuss this question in the paper “Ritual Practices for Constructing Terrestrial Space (Warring
States-Early Han)” presented at the International Conference Rituals, Pantheons and Techniques: A His-
tory of Chinese Religion before the Tang (14–21 December 2006, Paris), organised in the framework of the
project on the history of Chinese religions by John Lagerwey). Contributions to the project will be
published by Brill.
27 Bruce Brooks in his contribution of March 26, 1999, to the WSW discussion list labelled “SJ 123”
questions Hulsewé’s suggestion that the “Dawan liezhuan” is based on its Hanshu counterpart. The
differences between the “appraisals”, he argues, show clearly that the “appraisal” of the Hanshu is de-
duced from that of the Shiji and not vice versa, which provides an additional argument against this sug-
gestion.
OE 45 (2005/06)
Where is the Yellow River Source? 75

complement the general nonary framework serving as ways of communication between the
“provinces”.28
Some river itineraries encompass several rivers, some just one. The latter is the case for
the third itinerary, which comprises only the Yellow River. The delineation of the Yellow
River begins from a place called Jishi 積石:29
導河積石，至于龍門，…
[Yu] delineated30 the [Yellow] River [beginning from] Jishi [Mountain], [then] reached Longmen
(Dragon’s Gate) [Gorge] …
This is the second occurrence of Jishi in the “Yugong”. The first appears several passages
earlier, in the description of the western “province” Yong 雍. This province is located in
the basin of the Wei 渭 River and is the last in the sequence of the “Nine Provinces” in
the “Yugong”. Here Jishi is the starting point of tribute transportation from Yong prov-
ince:31
浮于積石，至于龍門西河，會于渭汭。
[The tribute was] floated (at = ) from Jishi [Mountain] towards the Longmen (Dragon’s Gate) [Gorge
and] the Western [part of] the [Yellow] River, [and then was] gathered at (or joined with) the nook of
the Wei [River].
Transportation of tribute concludes descriptions of each “province” in the “Yugong”. Since
the tribute is mostly sent by river, the context is consistent with the description of the Yel-
low River as the third river itinerary. The beginning of this passage is, indeed, very similar
to the beginning of the third river itinerary.
Although Kunlun is not recognised in the “appraisals” of the “Dawan liezhuan” and the
“Zhang Qian Li Guangli zhuan”, as the source of the Yellow River, it appears among the

28 In contrast to the “Nine Provinces”, the systems of itineraries are given much less consideration in
sinological literature. I have briefly discussed the system of nine itineraries by land in relation to the
itineraries of the Shanhai jing. See my article “Conception of Terrestrial Organization in the Shanhai
jing,” Bulletin de l’Ecole Française d’Extrême Orient 82 (1995), 77–78, Fig. 10 on 100, which shows a sche-
matic representation of the itineraries. See also Dorofeeva-Lichtmann (2003b), 153–154. Brashier ex-
plored it recently in the context of interaction with mountain spirits. See Ken E. Brashier, “The Spirit
Lord of Baishi Mountain: Feeding the Deities or Heeding the Yinhang?,” Early China 26–37 (2001–
2002), 177–178. Having drawn a similar conclusion – that the sets of mountains represent itinerar-
ies – we propose slightly differing schematic representations of these itineraries. The system of river
itineraries has become of special interest in relation to the recently discovered description of the
“Nine Provinces” in the Rongcheng shi 容成氏 manuscript (Shanghai Museum Bamboo Manuscripts
series, late 4th cent. BC), since this description is focussed on the river networks. The manuscript is
published in Ma Chengyuan 馬承源, Shanghai bowuguan zang Zhanguo Chu zhushu 上海博物館藏戰國楚
竹書, Vol. 2 (Shanghai: Shanghai guji, 2002), 91–146, photographs of the slips; 247–293, annotated
transcription by Li Ling 李零. I discussed the Rongcheng shi version of the “Nine Provinces” at the 17th
Conference of the Warring States Project (17–18.09.2003, Leiden. I elaborate on some issues of this
discussion and refer to other studies of the Rongcheng shi version of the “Nine Provinces” in my forth-
coming article “The Rong Cheng shi version of the ‘Nine Provinces’: Some Parallels with Transmitted
Texts,” in Proceedings of the 8th International Conference on China (The 8th China Cultural Week), 17–
22.01.2005, Lisbon (Portugal), ed. Ana Maria Amaro and Dora Martins.
29 Shangshu 6.14b. §23 in Karlgren’s division of the “Yugong”.
30 Dao 導 (to delineate) marks the beginning of description of each river route in the “Yugong”.
31 Shangshu 6.14b. §19 in Karlgren’s division of the “Yugong”.
OE 45 (2005/06)
76 Vera Dorofeeva-Lichtmann

landmarks of the “Yugong”. It is also found in the description of Yong “province”, imme-
diately after the delineation of tribute transportation routes. Kunlun occurs in the next and
the last passage of Yong’s description. Yet, here it does not refer to a mountain, but is the
name of one group of the Western “barbarians” (and their territory) that are given here:32
織皮崑崙、析支、渠搜、西戎即敘。
Karlgren translates this not entirely clear passage as follows:
The felt[-wearing people of] Kunlun, Xizhi, Qusou, [these] Western Rong were made orderly.
Whatever interpretation of this passage is preferred, it is evident that, according to the
“Yugong”, Kunlun refers to the same Western section of the described territories as Jishi,
but the text is too imprecise to allow one to determine how these landmarks are located
with respect to each other. It is only clear that they are located not too far-away from each
other.

2.2 Kunlun and Jishi in the Shanhai jing

In the Shanhai jing Kunlun Mountain occurs several times,33 three times as the source of the
Yellow river. It is mentioned twice in its first and core part, the Shanjing 山經 (“Itineraries
of Mountains”) or the Wuzang shanjing 五藏山經 (“Five Treasuries: The Itineraries of
Mountains”), and once in the later layer of this text. The first occurrence is found in the
“Xici sanjing” 西次三經 (“The Third Itinerary of Western [Mountains]”) sub-chapter:34
西南四百里，曰昆侖之丘。… 河水出焉，而南流東注于無達。
400 li to the south-west there is Kunlun Peak. … The [Yellow] River emanates from it and flows
south, in the east pours into the Wuda.
References to rivers in the Shanjing in the majority of cases are done in a uniform and for-
mulaic way – they comprise the initial point of the described river – the mountain it “ema-
nates” (chu 出) from, the cardinally-oriented direction of the beginning of its course (e.g.,
“flows south” nanliu 南流), and its cardinally oriented final point – usually a river or a sea
the river “pours into” (zhu yu 注于). The final direction of the Yellow River course (east)
does not raise any questions, as the east is its dominant direction. But the landmark marking
its final point is rather unclear. Literally Wuda means “Unreachable”, and formally it should
be a river, a sea or a lake, as the Yellow River “pours into” it, yet Guo Pu 郭璞 (AD 276–
324), the author of the earliest surviving commentary on the Shanhai jing, considers it to be a
mountain name. But whatever it is, the Yellow River arrives to the east.
The occurrence of Kunlun in the “Xici sanjing” is of special help and importance for
our study, as it allows one to derive a rather precise cosmological location for Kunlun Mountain.
Indeed, the Shanjing represents the central part of the inhabited world through a system of
twenty-six itineraries marked by mountains. The itineraries are arranged according to the
four cardinal directions and the centre. The north and the south are delineated by three
itineraries each, the west and the east by four each, and the centre contains twelve. For an
attempt at a reconstruction of this system, see a representation of the itineraries of the Shan-

32 Shangshu 6.12b. §19 in Karlgren’s division of the “Yugong”.

33 For a detailed discussion of these occurrences, see Frühauf, 2 (2000), 58–67.
34 Shanhai jing (Shanhai jing jianshu 山海經箋疏, edited by Hao Yixing 郝懿行, SBBY ed.), 2.17a-18b.
OE 45 (2005/06)
Where is the Yellow River Source? 77

jing by Wang Chengzu，redrawn by Richard Strassberg who supplied it with translations of

toponyms and some additions and useful notes, for instance a numbered list of represented
itineraries (Map 2).35 I have discussed this system in detail elsewhere.36
Wang represents itineraries as mountain ranges, which is not entirely accurate, as the
text features single mountains that serve as itinerary markers and are placed at considerable
distances from each other, but his reconstruction gives a good idea of this system.37 Each
mountain has a precise position in its itinerary. The position is given by referring to the
preceding mountain in the sequence – as the distance and the cardinally oriented direction
to be taken from the preceding mountain. Kunlun Mountain is the eighth mountain in the
third western itinerary, which comprises altogether twenty-two mountains (the “Third
Guideway Through the Western Mountains” in Map 2). Its general direction and, conse-
quently, the order of enumeration of mountains is from the centre to the west. I indicated
approximate position of Kunlun on Wang’s reconstruction. One can see clearly that in the
system of itineraries Kunlun Mountain is located in the west and is slightly shifted to the
north.
The next occurrence of Kunlun as the source of the Yellow River is found in the first
itinerary of the “Beishan jing” 北山經 (“Itineraries of Northern Mountains”), the most
western among the northern itineraries, so that here Kunlun is also related to the north-
west. Kunlun is mentioned in the description of Dunhong 敦薨 Mountain (mountain No.
17 in the itinerary comprised of twenty-five mountains):38
又北三百二十里，曰敦薨之山，… 出于昆侖之東北隅，實惟河原。
Then 320 li farther to the north there is Dunhong Mountain… [As far as] the emanating from Kunlun
Mountain’s north-eastern corner [is concerned], this is, in effect, the source of the [Yellow] River.
Finally, the north-western position of Kunlun is pointed out in another chapter of the text, the
Hainei xijing 海內西經 (“Western Itineraries of Inside the Seas”):39
海內昆侖之虛，在西北，帝之下都。… 河水出東北隅，以行其北西南，又入渤海，又出
海外，即西而北，入禹所導積石山。
Inside the Seas there is the great peak of Kunlun Mountain, located in the north-west, is the lower
capital-city of the (Supreme) Deity. … The [Yellow] River emanates from its north-eastern corner. [Pass-
ing] through its northern, western and southern [sides] (= turning around Kunlun from the north, west
and south), then pours into the Bohai (Bohai Lake, not to be confused with the Bohai in the north-east of
China), then emerges from the [Bo]hai, immediately turns to the west, and then in the north enters into
Jishi Mountain, [as] paved by Yu.

35 Wang Chengzu 王成組, Zhongguo dilixueshi: Xian Qin zhi Ming dai 中國地理學史，先秦至明代 (Pe-
king: Shangwu, 1988 [revised edition; first edition: 1982]), 19, map 2, redrawn in Strassberg (2002), 37,
fig. 10. Strassberg, however, mistakenly considers it to be a “modern map of mountains” described in
the Shanjing. I, in contrast, point out the schematic character of Wang’s map and its affinity with Chi-
nese cartographic tradition, see Dorofeeva-Lichtmann (2003a), 38, 43.
36 Dorofeeva-Lichtmann (2003a, 2003b).
37 I discuss Wang’s reconstruction in Dorofeeva-Lichtmann (2003a), 38, and in my article “Mapless
Mapping: Did the Maps of the Shanhai jing Ever Exist?,” in The Warp and the Weft: Graphics and Text in
the Production of Technical Knowledge in China, ed. Francesca Bray, Georges Métailié and Vera Dorofeeva-
Lichtmann (Leiden: Brill, forthcoming).
38 Shanhai jing 3.5b.
39 Shanhai jing 11.2b.
OE 45 (2005/06)
78 Vera Dorofeeva-Lichtmann

As mentioned above, the north-western cosmological and astronomical position of Kunlun

is much emphasised by Deborah Porter in her study of the Mu Tianzi zhuan, building all her
argumentation on it.40 The north-western position of Kunlun with respect to the “Nine
Provinces” is also derived by John Major from the “Treatise on Topography” (“Dixing
xun” 墬形訓) of the philosophical compilation Huainanzi 淮南子 (finished not later than
139 BC), and he also provides helpful cosmological schemes.41 The consistency of this data
is a bit spoiled by the location of Kunlun “to the south of the western sea” (xihai zhi nan 西
海之南) in one of late chapters of the Shanhai jing, the “Dahuang xijing” 大荒西經 (“West-
ern Itinerary of the Great Wilderness”), which implies a south-western position, but land-
marks mentioned here in relation to Kunlun (e.g. the Chi 赤 and the Hei 黑 Rivers, liter-
ally Red and Black, respectively) are the same as in the passages where a north-western
position is given.42 One can suggest that the western sea, according to the Shanhai jing, may
be shifted to the north, and, indeed, on Wang’s map a part of the western sea is shown in
the north-west, placing the location of Kunlun below it and, at the same time, in the north-
west. In any case, the north-west is the dominant cosmological position of Kunlun, accord-
ing to the Shanhai jing.
The passage from the “Hainei xijing”, originating from the later layer of the Shanhai jing
(Haijing part, Hainei 海內 “Inside the Seas” group of chapters) is apparently an attempt to
make a coherent fusion of the information on the Yellow River course from earlier sources,
in particular from the two passages from the Shanjing discussed above. According to the
former passage (“Xici sanjing”), the Yellow River first heads to the south from Kunlun, and
according to the latter (“Beishan jing”), the Yellow River emanates from its north-eastern corner.
In the “Hainei xijing” the Yellow River emanates from the north-eastern corner of Kunlun
and arrives to the south of it, having encircled it from the north, west and south. In other
words, here a link between the north-eastern corner of Kunlun, as the place where the Yel-
low River comes out of the earth, and the initial southern direction of its flow is established.
Then a compromise between the Shanhai jing and the “Yugong” versions of the Yellow
River source is proposed, and their “reconciliation” is achieved in a very straight-forward
way, through paraphrasing the delineation of the Yellow River from Jishi in the “Yugong”.
The Yellow River emanating from Kunlun becomes an extension of its delineation in the
“Yugong”, and, therefore, is no longer an alternative conception. The proposed extension
of the Yellow River, in effect, matches rather well its delineation in the “Yugong”, where it
is actually said that Yu “delineated the [Yellow] River [beginning from] Jishi” (dao He Jishi 導
河積石), but Jishi is never explicitly referred to as its source (Heyuan 河源).43

40 Porter (1993), esp. 80 ff.

41 John S. Major, “The Five Phases, Magic Squares and Schematic Cosmography,” in Explorations in
Early Chinese Cosmology, ed. Henry Rosemont (Chico: Scholar, 1984), 133–137.
42 E.g. Chi and Hei are mentioned in the discussed occurrence of Kunlun in the “Xici sanjing”, and Chi
in the “Hainei xijing”.
43 However, in the “summary” of Yu’s regulations of the “Nine Provinces” and the system of itineraries
found in the “Yugong” (Shangshu 6.16b, §30 according to Karlgren’s division of the text) there is a
phrase concerning the sources of all the river itineraries delineated by Yu (jiuchuan diyuan 九川滌源
“The nine river itineraries had their sources cleaned”), which implies that the initial points of the de-
scribed river itineraries are their sources.
OE 45 (2005/06)
Where is the Yellow River Source? 79

Map 2

Let us take a closer look at the Kunlun Æ Jishi extension of the Yellow River. According to
the “Hainei xijing”, the Yellow River, having emanated from Kunlun and arrived to the south
of it, then flows to the west where it reached Jishi. This implies that Jishi is located farther to
the west than Kunlun. The relative locations of Kunlun and Jishi derived from the “Hainei
xijing” are consistent with those given in the Shanjing. The advantage of the system of locations
in the Shanjing, especially within a single itinerary, is their high level of precision, which makes
them absolutely incontestable and independent of any interpretation of the content of the text.
Specifically, Jishi Mountain belongs to the same itinerary as Kunlun – the third itinerary of
western mountains whose general direction and order of enumeration of mountains is from
the centre to the west. Jishi is the thirteenth mountain in this itinerary, whereas Kunlun is the
eighth, and, therefore, Jishi is located considerably farther to the west than Kunlun (see its
approximate location in Map 2). According to the description here, Jishi is an important point
on the Yellow River:44

44 Shanhai jing 2.20b.

OE 45 (2005/06)
80 Vera Dorofeeva-Lichtmann

又西三百里，曰積石之山，其下有石門，河水冒以西流。
Then 320 li farther to the north there is Jishi Mountain. There is a stone gate under it (at its southern
slope). The (Yellow) River bursts through it (= through the gate) and flows (via the gate) to the west.
The reference to the “stone gate” (opening) at the southern slope of Jishi is coherent with the
phrase from the passage from the “Hainei xijing” that the Yellow River “in the north enters into
Jishi Mountain”. Since the Yellow River, according to this passage, having emerged from
Kunlun heads south and then west, it necessarily reaches Jishi in the north. The next phrase in
the Jishi description is much more difficult to understand and the proposed translation is tenta-
tive, but the western direction of the Yellow River course is mentioned very clearly.45
The Yellow River is mentioned in descriptions of two other mountains of the third itiner-
ary of western mountains: the first Chongwu 崇吾 and the third Buzhou 不周. Congwu
Mountain “is located to the south of the [Yellow] River” (zai He zhi nan 在河之南 ), and this is
the first among the indications of its location provided in the text. This means that the starting
point of the third itinerary of western mountains is to the south of the Yellow River. The Yel-
low River in the description of the famous Buzhou Mountain butted by Gonggong 共工46 is
less useful as a location reference, as it is said that it flows under the Youze 泑澤 Marsh that
is found relatively far to the east of Buzhou:47
…東望泑澤，河水所潛也，…
…To the east of it (= of Buzhou) one observes from distance (special form of ritual) the Youze
Marsh, this is the place where the He River hides as an undercurrent…
Although an underground river can hardly be located, this reference is coherent with the loca-
tion of Chongwu to the south of the Yellow River, as Chongwu is found to the east of Buzhou.
All these indications allow one to derive the following configuration of the beginning of
the Yellow River course: it emanates from the north-eastern corner of Kunlun, flowing around
it to the north, west and south, then heads further south and pours into the Bohai Lake.
Emerging from here, it turns to the west and to the north in order to arrive at Jishi. Its course
after Jishi is less clear, but in any case, Jishi is the most western point of the Yellow River course men-
tioned in the Shanhai jing. Kunlun, in its turn, according to this text is located not far to the
north-west of the “Nine Provinces”,48 somewhere to the south of the Wei River correspond-
ing to the location of the Kunlun “barbarians” inside Yong “province” of the “Yugong”.49

45 Wang Niansun 王念孫 and Hao Yixing suggest that the character “west” is a mistake for “south”,
but this suggestion is based on cartographic representations of Jishi I shall discuss later and contra-
dicts the Yellow River course derived from the Shanhai jing. See Yuan Ke (1980), 51.
46 For Gong Gong, see William G. Boltz, “Kung Kung and the Flood: Reverse Euhemerism in the Yao
Tien” T’oung Pao 67, 3–5 (1981), 141–153; Anne Birrell, “The Four Flood Myth Traditions of Classi-
cal China,” T’oung Pao 83, 1–3 (1997), 229–235; Marc Edward Lewis, The Flood Myths of Early China
(New York: SUNY, 2006), 55–60.
47 Shanhai jing 2.13b.
48 Such a location is shown, for instance, in schematic representations of Kunlun and the “Nine Prov-
inces” by Major (1984), 133–137.
49 Some parallels to such locations of Jishi and Kunlun are found in Wu Yue chunqiu, where in chapter 6
the deeds of Yu, the ancestor of the Yue kingdom, are described. Here another rare occurrence of
Jishi Mountain is found. It marks the extreme western point of Yu’s round tour (zhouxing 周行) of the
world “Inside the Seas” (hainei), and is one in a group of landmarks that mark the eastern, western,
southern and northern limits of Yu’s travels. Kunlun is mentioned immediately after, which allows
OE 45 (2005/06)
Where is the Yellow River Source? 81

Another helpful indication of the Yellow River course is the references to rivers emanating
from mountains in itineraries located in its neighbourhood. Rivers emanating from the third
through the ninth mountains of the fourth itinerary of western mountains located to the north
of the third itinerary, roughly parallel to it, flow east and pour into the Yellow River. This is coherent
with the location of the first mountain of the third itinerary of western mountains to the south
of the Yellow River. And now if one looks at the rivers emanating from the fourth through
the seventh, and the ninth mountains of the first itinerary of northern mountains which goes
from the centre to the north and is located to the east of the fourth itinerary of western moun-
tains, these rivers flow west and pour into the Yellow River. This allows us to delineate this part of
the Yellow River course from north to south, and to the north from the beginning of the third
itinerary of the western mountains. This implies that the Yellow River arrives to this part of
the Yellow River from the north, having encompassed the source of the Wei 渭 River, the
western tribute of the Yellow River, and the northern tributes of Wei – the Jing 涇 and the
Luo 洛 Rivers.50
The beginning of the Yellow River course derived here is not that evident, and it certainly
created some difficulties to Wang Chengzu. For instance, he does not mention the turn to the
west from Kunlun in his delineation. A good indication of the fact that his reconstruction has
some inconsistencies is his delineations of the Wei, the Jing and the Luo Rivers (upper left
corner of Map 2). It seems that for some reason Wang could not find a proper place for these
rivers. The proposed delineations are unrelated to the Yellow River, contradicting references
to these rivers in the text. The corrected beginning of the Yellow River course incorporated
into Wang’s map is shown in Map 3.
In sum, the references to Jishi and Kunlun in the “Yugong” allow us to locate these two
landmarks somewhere not far from the western border of the western “province” Yong, but
there are no indications in the “Yugong” as to how Jishi and Kunlun are located with respect to
each other. In addition, Kunlun is not a mountain name here, but rather the name of “barbari-
ans”. The elaborate system of locations applied in the Shanhai jing, in contrast, enables one to
derive fairly precise respective “positions” of Jishi and Kunlun mountains. According to the
Shanhai jing, Jishi is located considerably farther to the west than Kunlun and is the most western
“station” of the Yellow River course. In this respect the Shanhai jing does not contradict the
“Yugong” – here Jishi is also the most western point of the Yellow River course. The difference
between these two texts is, therefore, the role of Kunlun – in the “Yugong”, Kunlun is a secon-
dary landmark (most likely the name of “barbarians”) unrelated to the Yellow River; in the
Shanhai jing Kunlun is one of the core mountains and the Yellow River source, while Jishi is
secondary to it.

one to suggest that it is located inside these limits, in Wu Yue chunqiu 吳越春秋, attributed to Zhao Ye
趙曄, mid of 1st cent. AD (SBBY ed.), 6.3a.
50 This relation between the Yellow River, the Wei, the Jing and the Luo Rivers can be derived from
references to them in the fourth itinerary of the western mountains.
OE 45 (2005/06)
82 Vera Dorofeeva-Lichtmann

Map 3

3 The Yellow River source in the “Treatises” on waterways of the Shiji and the Hanshu
The “appraisals” of the “Dawan liezhuan” and the “Zhang Qian Li Guangli zhuan” first ques-
tion Kunlun Mountain as the source of the Yellow River and the reliability of sources that
locate it there, making a clear choice in favour of the “Yugong” as the true account of loca-
tions.

OE 45 (2005/06)
Where is the Yellow River Source? 83

The same choice is made in the other two chapters of the Shiji and the Hanshu that de-
scribe the entire course of the Yellow River – the “Hequ shu”51 of the Shiji, and the “Gouxu
zhi”52 of the Hanshu. But, in contrast to the “appraisals” of the “Dawan liezhuan” and the
“Zhang Qian Li Guangli zhuan”, the treatises on waterways do not even mention Kunlun and
texts referring to it, exclusively relying on the “Yugong”.
The “Hequ shu” is mostly concerned with the Yellow River and related canals. The treatise
describes the construction of various canals as means of irrigation and transportation, changes
in the course of the Yellow River and breaches of its dikes. The description is done in a
chronological sequence and is preceded by an introduction that evokes overcoming the Flood
by the mythical emperor Yu and his regulations of terrestrial space. The introduction begins
with a citation that formally refers to the “Xiashu” 夏書 (“The Book of Xia”),53 the part of
the Shangshu devoted to the Xia dynasty and comprising the “Yugong” as its first entry. In
effect, though, it is a selected citation of the preface to the complete reproduction of the
“Yugong” in the “Xia ben ji” 夏本紀 (“Main Records of the Xia Dynasty”) chapter of the
Shiji.54 The preface relies on the “Yugong”, specifically on its beginning and the “summary”
of Yu’s regulations of the “Nine Provinces” and the system of itineraries,55 yet also mentions
many details of Yu’s actions that are nowhere in evidence in the Shangshu. The reference to the
“Xiashu” reproduces with some differences certain passages of the preface, including the pas-
sages that paraphrase the “Yugong”.56
Sima Qian then stresses the great threat of the Yellow River to the Central Empire (Zhongguo
中國) as a problem of central importance to be solved and, therefore, the reason for Yu’s regula-
tion of its course. This is followed by a citation of the delineation of the Yellow River course
according to the “Yugong”, though with some minor differences, one of which highlights Jishi
as the initial point, done by adding the preposition zi 自 (“from”) in front of Jishi:57
故道河自積石歷龍門，…
For this reason [Yu] made a way for the [Yellow] River from Jishi [Mountain] all through the Longmen
[Gorge] …58

51 Shiji 29.1405–1415; cf. Les Mémoires Historiques de Se-ma Ts’ien, Vol. 3, 2ème partie, trans. Édouard Cha-
vannes (Paris: Leroux, 1899), 520–537; Records of the Grand Historian of China translated from the Shih chi of
Ssu-ma Ch’ien, trans. Burton Watson, Vol. II (New York and London: Columbia University, 1961),
70–71; trans. Watson (1969), 230–238; Syma Tsyan’ – Istoricheskie Zapiski (“Shi tszi”) [Sima Qian – His-
torical Records (Shiji))], trans. Rudolf V. Vyatkin, Vol. 4 (Moscow: Nauka, 1986), 194–200).
52 Hanshu 29.1675–1699.
53 The first characters of the introduction are “Xiashu yue 夏書曰” (The Book of Xia says:).
54 The “Yugong” is reproduced here with minor differences. For the preface, see Shiji 2.51; cf. Les Mé-
moires Historiques de Se-ma Ts’ien, Vol. 1, trans. Edouard Chavannes (Paris: Leroux, 1895), 101, Syma
Tsyan’ – Istoricheskie Zapiski (“Shi tszi”) [Sima Qian – Historical Records (Shiji)], Vol. 1, trans. Rudolf V.
Vyatkin and Vsevolod S. Taskin (Moscow: Nauka, 1972), 152; The Grand Scribe’s Records, Vol. I. The
Basic Annals of Pre-Han China by Ssu-ma Ch’ien, ed. William H. Nienhauser (Bloomington & Indianapo-
lis: Indiana University, 1994), 22–24.
55 Shangshu 6.1b and 6.16b, respectively. §§ 1 and 30 in Karlgren’s division of the “Yugong”.
56 Phrases from one through twelve following the Xiashu yue.
57 Shiji 29.1405.
58 The character li 歷 (“to go through, to pass, to undergo, to stretch over, to get to”) used here accen-
tuates the considerable distance between Jishi Mountain, whichever identification of it is accepted,
OE 45 (2005/06)
84 Vera Dorofeeva-Lichtmann

The citation is cut into two parts by explanatory notes by Sima Qian on the special difficulties
of managing the Yellow River. In conclusion, Sima Qian points out the utmost importance of
its regulation. This ensured orderly functioning of the entire system of waterways – the “nine
rivers” (jiuchuan 九川) and the “nine marshes” (jiuze 九澤 ) – eventually bringing harmony to
the “civilised world” (“All the Xia” zhuxia 諸夏) to the benefit of the three dynasties.
The title of the “Gouxu zhi” does not contain the name of the Yellow or any other river,
and instead refers to systems of irrigation in general. There is, however, little difference in the
content of the two treatises. The “Gouxu zhi” first reproduces with minor variations the
“Hequ shu”59, and then adds to it data on irrigation canals constructed after the compilation of
the Shiji until the reign of Wang Mang (AD 9–23).
In sum, the “Treatises” on waterways of the Shiji and the Hanshu proceed from the de-
lineation of the Yellow River, as described in the “Yugong” and beginning from Jishi. The
“Yugong” plays here the role of the absolute and unquestionable authority on the subject, and
its delineation of the Yellow River serves as the “theoretical foundation” of the “Treatises”.

4 The imperial conception of the Yellow River source, according to the “Xiyu
zhuan” of the Hanshu
The “appraisals” together with the “Treatises” on waterways give the impression that the im-
perial conception of the Yellow River source is Jishi, but things are not as simple as they may
seem at first glance.
The introductions of the “Treatises” obediently reproduce the delineation of the Yellow
River from Jishi in the “Yugong”, but do not say that Jishi is the source of the Yellow River. The
“appraisals” consider the references to Kunlun as the source of the Yellow River as completely
false for the reason that it was not confirmed by the expedition of Zhang Qian, who “exhaus-
tively explored the source of the [Yellow] River” (qiong Heyuan 窮河源). But having stressed this
thorough exploration of the Yellow River source, the “appraisals” rather surprisingly are not
precise about where the true source is located. Instead, a general statement about the reliability
of the “Yugong” is provided. Such vagueness is not accidental. The reason is that the expedi-
tions to the West, as twice reported in the main text of the of the “Dawan liezhuan”, discov-
ered, in effect, a source of the Yellow River, but one differing both from Kunlun and Jishi:60
于窴之西，則水皆西流，注西海；其東水東流，注鹽澤。鹽澤潛行地下，其南則河源出
焉。多玉石，河注中國。
[As far as the territories] to the west of Yutian [are concerned], all [their] rivers flow west, pour into
the Western Sea. Rivers to the east of it (= of Yutian) flow east, pour into the Salty Marsh (= Lake
Lobnor). The Salty Marsh [further] clandestinely goes under the earth. [And as far as the territories] to
the south of it (= of Yutian) [are concerned], the [Yellow] River source emanates from here. [There
is] much jade, the [Yellow] River [then] enters (literally pours into) the Central State.61

and Longmen Gorge. Longmen Gorge is situated on the part of the Yellow River where, having
made a large loop around the Ordos in the north, it descends from the north to the south above the
place where the Fen 汾 River, its north-eastern tribute, pours into it (Shenxi province, Hancheng
county).
59 Hanshu 29.1675–1684.
60 Shiji 123.3160.
61 Shiji 123.3173.
OE 45 (2005/06)
Where is the Yellow River Source? 85

…而漢使窮河源，河源出于寘，其山多玉石，采來，天子案古圖書，名河所出山曰崑崙云。
…And the Han ambassador exhaustively explored the source of the [Yellow] River,62 [according to
the results of his exploration,] the [Yellow] River emanates from Yutian, its mountains have a lot of
jade, [that the Han ambassador collected] and sent back [to the Han court] (translation of these two
characters, according to commentary). The Son of Heaven relied on the old “graphic representations”
and books [that say] that the mountain the famous [Yellow] River emanates from is called Kunlun.
The first passage is completely missing in the “Zhang Qian Li Guangli zhuan”, the second
is reproduced, but the reference to Yutian is omitted – the entire phrase “Heyuan chu Yutian
河源出于寘” (The [Yellow] River emanates from Yutian) is just erased.63 Having dropped
these references from the biography of Zhang Qian and Li Guangli, Ban Gu provides a
much more elaborate and slightly different version of the discovered Yellow River source in
another chapter on the Western Region, the “Xiyu zhuan”, one that does not have a coun-
terpart in the Shiji. This chapter begins with a detailed description of the Yellow River
course from its sources to Jishi.
According to Ban Gu, the Yellow River has two sources – Congling 蔥嶺 Mountain (liter-
ally, “Onion Mountain”, usually identified with the Pamirs)64 and Yutian 于闐. According to
Yan Shigu’s 顏師古 (AD 581–645) commentary, tian 闐 is the same as tian 窴 in the passages
just discussed of the Dawan liezhuan.65 Yutian is the ancient name of Hotan or Khotan.66
Yutian is located by (below) the Nanshan 南山 Mountains (the “Southern Mountains”).67 The
Yutian branch of the Yellow River flows north and pours into the Congling branch. After the
two branches unite, they pour into Lake Puchanghai 蒲昌海 (literally the “Lake of Cane that
flourishes wildly”) or Yanze Marsh (literally the “Salty Marsh”, Lake Lobnor). Then the Yellow
River flows under the earth, and reappears to the south of Jishi Mountain, where it becomes the
[Yellow] River of the Central State (zhong guo), that is the core territory of the Empire:68
其河有兩源：一出蔥嶺山，一出于闐。于闐在南山下，其河北流，與蔥嶺河合，東注蒲昌
海。蒲昌海，一名鹽澤者也，去玉門，陽關三百餘里，廣袤三百里。其水亭居，冬夏不增
減，皆以為潛行地下，南出於積石，為中國河云。
The [Yellow] River has two sources, one emanates from Congling Mountain, one emanates from
Yutian. Yutian is situated below Nanshan Mountain, its [branch of the Yellow] River flows north, and
unites with the Congling [branch of the Yellow] River, [then] pours east into Lake Puchanghai. Lake
Puchanghai, another name is Yanze Marsh, is situated more than 300 li away from Yumen and Yang-

62 This phrase is repeated in the “appraisals”.

63 Hanshu 61.2696. This is an additional argument in favour of the conclusion that the “Zhang Qian Li
Guangli zhuan” relies on the “Dawan liezhuan” and not vice versa.
64 The Pamirs are located to the west and east of the border between Tajikistan and Xinjiang-Uygur
Autonomous Region.
65 Hanshu 96A.3872, n. 5.
66 Hotan or Khotan, modern Chinese name Hetian 和田, is located in the south-west of the Xinjiang
Uygur Autonomous Region. Formerly Hotan oasis was one of the important “stations” on the Silk
Road.
67 It is clear that the Nanshan Mountains mentioned here cannot be the Nanshan Mountains found in
modern maps of China (the latter are located too far to the east of Hotan). The mountains that would
fit a location near Yutian are, actually, the modern Kunlun chain. If we locate the Nanshan in the
place of modern Kunlun, that is, along the southern boundary of the Tarim basin, they would make a
proper counterpart to the Tianshan Mountains that frame the Tarim in the north and whose other
name is Beishan Mountains, literally the “Northern Mountains”, see Map 1.
68 Hanshu 96A.3871.
OE 45 (2005/06)
86 Vera Dorofeeva-Lichtmann

guan, [its] width and length (from north to south) are [both] 300 li. Its water stops and stays without
moving [here], in winter and summer does not increase or decrease, [instead] all [the water] clandes-
tinely go under the earth, and in the south reappears from Jishi, becoming the [Yellow] River of the
Central State.
For a rare cartographic representation of this idea of the sources of the Yellow River help-
ing to visualize it, see Map 4.69
In sum, the new geographical information accumulated during the expeditions to the
West required a revision of both the ancient textual traditions locating the Yellow River
source. The reliability of the “Yugong”, however, was not to be even partially questioned,
as this would have contradicted the high ideological status appointed to this text in the early
Han historiography.
Sima Qian apparently did not manage to find a coherent solution to the Jishi problem –
the discrepancy between the delineation of the Yellow River from Jishi in the “Yugong”, and
the discovered source (Yutian 于窴). As a result, he twice mentions the discovered source in
the main text of the “Dawan liezhuan”, but masks it in the “appraisal” of this chapter.
The situation for Ban Gu was even more complex and delicate. On the one hand, he for-
mulated the conception of “terrestrial organisation” based on the “Yugong”, and thus recog-
nised this text as the highest authority among terrestrial descriptions. On the other, more in-
formation on the Western Region had become available by his time. In addition, he could not
openly revise Sima Qian, even though his position on the Yellow River source was contradic-
tory. Nevertheless, Ban Gu found an elegant compromise between the new geographical in-
formation available on the Western Region and the authority of the “Yugong”, and still for-
mally respected Sima Qian’s position. He reproduced the main points of Sima Qian’s chapter
on the expeditions to the West expressed in the “appraisal”, having first eliminated references
to the discovered source inconsistent with it, and then elaborated on this issue in a new com-
plementary chapter on the Western Region. Here Ban Gu did not invent much especially new
with respect to Sima Qian. All the main constituent elements of his conception of the Yellow
River sources were expressed by Sima Qian in the “Dawan liezhuan” – the Yellow River starts
in the West and first pours into Lake Lobnor (Yanze), then its waters flow underground and
reappear in the core territory of the Empire.70 But Ban Gu transformed this vague passage
placed somewhere in the middle of the “Dawan liezhuan” into a detailed and clear description
that opens a new special chapter on the Western Region.
Let us take a look at this conception with respect to the real topography of the Western
Region (Map 1). The expeditions to the West discovered Lake Lobnor and the rivers that flow
towards it from the west and the south-west. Ban Gu proposed that these rivers should be
considered as the two sources of the Yellow River that – together with the lake – form a unit
separated from the main body of the Yellow River by an underground part. This conception
perfectly corresponded to the needs of the imperial ideology – it clearly distinguished between

69 Han xiyu zhuguo tu 漢西域諸國圖 (“Map of all the countries of the Han Western Region”, AD 1260–
1264) reproduced from Yan Ping et al., China in Ancient and Modern Maps (London: Philip Wilson,
1998), 79. For convenience, I have added to the map pinyin transcriptions of toponyms mentioned in
the “Xiyu zhuan”.
70 See especially the passage in Shiji 123.3160 cited above.
OE 45 (2005/06)
Where is the Yellow River Source? 87

the “inside” and the “outside” sections of the Yellow River, that is, the sections located in the
core territory of the Empire (Zhongguo) and in the Western Region.

Map 4
The division of the Yellow River into “inside” and “outside” sections is carefully mirrored in
the textual structure of the Hanshu. These two sections are described not only in different
chapters of the Hanshu, but in hierarchically unequal parts of the history. The “inside” section
of the Yellow River is described in the “Treatise” on waterways (“Gouxu zhi”), that is, in the
“Treatises” (zhi 志) part of the dynastic history, which deals with selected items of imperial
ideology that were considered of special importance.71 The “outside” section of the Yellow
River is described in the “Memoirs” (zhuan 傳) part, which is comprised of biographies of
distinguished persons and descriptions of peripheral (“barbarian”) lands, the subjects of quali-
tatively lesser significance for the Empire.
This separate representation of the Yellow River allowed Ban Gu to find a proper “con-
ceptual” place for Jishi – as the Yellow River source in the “inside” part,72 and, at the same
time, to incorporate into the description of the Yellow River the new geographical information
about the Western Region. The “position” of Jishi as a quasi-source of the Yellow River is an

71 A “Treatise” in a dynastic history is a systematic description of a certain subject of imperial impor-

tance, see B.J. Mansvelt Beck, The Treatises of Later Han: Their Author, Sources, Contents and Place in Chi-
nese Historiography (Leiden: Brill, 1990), 36–55; Balazs distinguishes four categories of these subjects –
I. Rites and Customs; II. “Sciences”; III. Government (State) Institutions; and IV. Bibliography, see
Étienne Balazs, La bureaucratie céleste: Recherches sur l’économie et la société de la Chine traditionelle (Paris: Gal-
limard, 1968), 57; cited (in English translation) by Mansvelt Beck (1990), 54–55. The book Mansvelt
Beck referred to focuses on the “Treatises” compiled by Sima Biao 司馬彪, ca. 240 – ca. 306, later
added to the Houhan shu 後漢書, edited by Fan Ye 范曄, 398–446 (Beijing: Zhonghua, 1973).
72 For instance, in Map 4 it is noted that Jishi is the Yellow River source, apparently of the core territo-
ries of the Empire.
OE 45 (2005/06)
88 Vera Dorofeeva-Lichtmann

innovation of Ban Gu’s, as compared to Sima Qian, who, apart from citing the “Yugong”,
avoids any reference to this landmark.

5 Conclusions
The following conceptions of the Yellow River source are discussed in early Han historiography:
— Jishi is the source of the Yellow River (the “Yugong”);
— Kunlun Mountain is the source of the Yellow River (the Shanhai jing), and Jishi Mountain is
the most western point of the Yellow River course; in the “Hainei xijing” chapter a com-
promise is proposed – the Kunlun Æ Jishi extension of the Yellow River is presented as
complementing its delineation in the “Yugong”;
— The conception of two sources in the Western Region pouring into Lake Puchanghai or
Yanze Marsh (Lobnor) and separated from the main Yellow River course reappearing at
Jishi by an underground part (the “Xiyu zhuang” chapter of the Hanshu), developing on a
vague outline of this idea in the “Dawan liezhuan”.
The conception of the Yellow River source(s) in the Western Region somewhat resembles the
Kunlun Æ Jishi extension of the Yellow River described in the “Hainei xijing”. Here also the
Yellow River first falls into a lake (Bohai 渤海) and then emerges from the lake and arrives at
Jishi, and since there is no underground part, the Kunlun Æ Jishi extension seems even more
consistent. The reappearance of the Yellow River from underground at Jishi somewhat re-
sembles its “bursting” (mao 冒) through the “stone gate” (shimen 石門) of Jishi described in
the “Xici sanjing”.
Yet, there are some principal differences between the two conceptions that make them ab-
solutely incompatible. Firstly, according to the Shanhai jing, Jishi is the most western point of
the Yellow River course, located far-away in the west, considerably farther than Kunlun. Ac-
cording to the “Xiyu zhuang” chapter of the Hanshu, Jishi is the demarcation point between
the main territories of the Empire, corresponding to the “Nine Provinces”, and the western
“barbarous” periphery,73 and, therefore, is located just at the boundary of the “civilised
world”.
Secondly, both Sima Qian and Ban Gu strived to erase any connection between Kunlun
Mountain and the Yellow River because, for reasons that require a special study, this did not
match the imperial conception of terrestrial space. This imperial conception faded, however,
with the beginning of the period of political disunion after the collapse of the Han Empire in
AD 220. This decline in particular imperial interest in the Yellow River can be seen from the
fact that the “Treatises” on waterways disappear from the dynastic histories compiled after the
Han dynasty.74

73 I am thankful to Christian Lamouroux for having called my attention to this role of Jishi.
74 They are revived after a gap of about a thousand years under the title “Hequ zhi” in the dynastic
histories of the Song and Jin Dynasties, Jinshi 金史 and Songshi 宋史, both compiled almost simulta-
neously under the Yuan Dynasty (AD 1271–1368). In other words, the “Treatises” on waterways re-
appear in the dynastic histories describing the period of time and compiled during the period of time
when the Yellow River basin again became of special political importance due to its invasion by “bar-
barians”. Since then treatises on waterways under the same name are found in all subsequent dynastic
histories.
OE 45 (2005/06)
Where is the Yellow River Source? 89

Map 5

The insistence on the orthodoxy of the officially recognised sources on the Yellow River also be-
came less strict after the time of the Han Empire. In particular, the negative evaluation of the
Shanhai jing in the Shiji and the Hanshu is completely reversed in the Houhan shu, which described
the Later Han dynasty (AD 25–220). Apart from the “Treatises” composed by Sima Biao, the
Houhan shu was compiled by Fan Ye.75 Similarly to the Shiji and the Hanshu, the Shanhai jing is
mentioned in the Houhan shu again in relation to the Yellow River, in the biography of the official
responsible for repairing its dykes, Wang Jing.76 The Shanhai jing, the “Hequ shu” and the Yu-
gong tu 禹貢圖 (“Maps of the ‘Yugong’”, no such maps dating from the Han dynasty have sur-
vived) are reported to have been given to Wang Jing by the emperor, serving as the major refer-
ence texts for repairing the dykes.77 Here the Shanhai jing is even placed in the primary position
among these sources. The royal recommendation to rely on it first indicates the official recogni-
tion of its locations, especially the location of the Yellow River source at Kunlun Mountain.

75 The different parts of the Houhan shu were then combined by Liu Zhao 劉昭 (fl. AD 502–520).
76 Houhan shu 76.2465 (“Wang Jing liezhuan” 王景列傳).
77 This occurred in AD 69.
OE 45 (2005/06)
90 Vera Dorofeeva-Lichtmann

A good illustration of this recognition is found in representations of China and the sur-
rounding world in traditional Chinese cartography, the earliest surviving “world” maps dating
from the Southern Song Dynasty (1127–1279 AD).78 The most current type of these maps
delineate the Yellow River as coming from Kunlun Mountain, and show Jishi a little bit farther
to the east along the Yellow River course (see for instance Map 5).79 Kunlun and Jishi are
found in the upper left part of the map.80 Yet, these maps provide a location of Jishi cardinally
different from that derived from the examination of the Shanhai jing, where Jishi is located to
the west of Kunlun. Having placed Jishi to the east of Kunlun, these maps, in effect, also a-
dopt Ban Gu’s idea of Jishi as a demarcation point between the Western Region and the core
territory of the Empire.81

6 Appendix82
Shiji 123 太 史 公 曰 禹 本 紀 言 河 出 崑 崙 崑 崙 其 高 二 千 五 百 餘 里
Lunheng 31 太 史 公 曰 禹 本 紀 言 河 山 崑 崙 其 高 三 千 五 百 餘 里
Hanshu 61 贊 曰 禹 本 紀 言 河 出 昆 侖 昆 侖 高 二 千 五 百 里 餘

Shiji 123 日 月 所 相 避 隱 為 光 明 也 其 上 有 醴 泉 瑤 池
Lunheng 31 日 月 所 於 辟 隱 為 光 明 也 其 上 有 玉 泉 華 池
Hanshu 61 日 月 所 相 避 隱 為 光 明 也

Shiji 123 今 自 張 騫 使 大 夏 之 後 也 窮 河 源 惡 睹 本 紀 所 謂 崑 崙 者 乎
Lunheng 31 今 自 張 騫 使 大 夏 之 後 窮 河 源 惡 睹 本 紀 所 謂 崑 崙 者 乎
Hanshu 61 自 張 騫 使 大 夏 之 後 窮 河 源 惡 睹 所 謂 昆 侖 者 乎

Shiji 123 故 言 九 州 山 川 尚 書 近 之 矣
Lunheng 31 故 言 九 州 山 川 尚 書 近 之 矣
Hanshu 61 故 言 九 州 山 川 尚 書 近 之 矣

78 For examples of such maps, see Cao Wanru 曹婉如 et al. (eds.), Zhongguo gudai dituji 中國古代地圖集
An Atlas of Ancient Maps in China, 3 vol. (Beijing: Wenwu, 1990, 1994, 1997), a good collection is also
provided in Yan Ping et al. (1998). For discussion of the representations of the Yellow River source
in Chinese cartography, see Vera Dorofeeva-Lichtmann, “Contradicting Representations of the Yel-
low River Source in Song Cartography,” Vestnik MGU, Vostokovedenie, Seriya 13 [Journal of the Mos-
cow State University, Oriental Studies, Series 13] (2006), 49–56.
79 禹貢所載隨山濬川之圖 Yugong suo zai suishan junchuan zhi tu (“Map of [Yu’s] moving along the moun-
tains [as orientation marks in order to blaze paths through highlands (shown in the map by lines)] and
deepening rivers, as registered in the ‘Yugong’”, shortly before AD 1209) reproduced from Yan Ping
et al. (1998), 65. This map, though like many others formally referring to the “Yugong”, in effect, in-
corporates data from other texts, in particular, the delineation of the Yellow River source from
Kunlun.
80 For convenience I added pinyin transcriptions of Kunlun and Jishi in the map.
81 Wang Chengzu in his reconstructed map of the Shanjing (see Map 2) seems to copy the configuration
of the Yellow River course from these maps rather than trying to reconstruct it from the Shanhai jing
(cf. Map 3).
82 I am indebted to Martin Hanke who provided this table for me.
OE 45 (2005/06)