Mad Rulers and Worthy Sons

A Translation and Analysis of the Newly Excavated Zhouxun

by Andrej Fech

This book presents the first study and translation of the ancient text Zhouxun (Instructions of the Zhou). Lost early but mentioned in the history of the Han dynasty, a copy appeared recently and was published as part of the Peking University collection. The work is unique in its claim that ancient Chinese monarchs determined their succession based solely on the abilities of their heirs, disregarding ritually sanctioned criteria, such as primogeniture and the nobility of the maternal lineage. Power transfer here combines hereditary and meritocratic factors in new and different ways. Mad Rulers and Worthy Sons provides a complete translation together with a detailed analysis of the history, structure, and contents of this important document. The book opens a new chapter in our understanding of political values and governmental procedure in early China.

Paperback: $34.95, plus S&H; PDF file download, $15.00



This is an exceptional and brilliant first English translation and analysis of the Zhouxun (Instructions of the Zhou), an excavated manuscript of 211 bamboo slips in 4882 characters that forms part of the Peking University collection and dates from the Western Han. Through this work, readers experience a wealth of narrative materials on the worthiness and political acumen of various local rulers who lived in the second half of the 4th century BCE. The author’s careful structural organization, vibrant explanations, and prolific commentaries truly enhance our deep understanding and profound appreciation of the moral exemplarity of the sovereign and the intellectual history of meritocracy in early China. The book is a must for anyone studying ancient China, traditional thought, and classical models of rulership.

—Robin R. Wang, Professor of Philosophy, Loyola Marymount University

What is the key to long term success of a political system? This hitherto hidden treasure from early Chinese history argues for meritocracy-based transfer of power with popular support. It’s an ideal mechanism for ancient monarchies as well as modern political systems that aspire to political meritocracy with democratic characteristics!

—Daniel A. Bell, Chair of Political Theory, University of Hong Kong


Andrej Fech (b. 1973), Ph.D., is Assistant Professor in the Department of Chinese Language and Literature at Hong Kong Baptist University. His research focuses on early Chinese intellectual history: Daoist thought, excavated manuscripts, and comparative philosophy.


  1. Part One: Discussion == 1. Discovery and Publication — 2. The Protagonists – 3. Textual Organization — 4. Rhetorical Strategies — 5. Philosophical Concepts — 6. Wider Context
  2. Part Two: Translation



CHAPTER 1: Discovery and Publication

The Zhouxun 周馴(訓)or Instructions of the Zhou is part of the Peking University collection (Beijing daxue cang Xi Han zhushu 北京大學藏西漢竹書), a group of bamboo-slip manuscripts dated to the Western Han and donated to the institution in 2009 (Beijing daxue 2011, 43, 49–57).1 In addition to the Zhouxun, the corpus also contains a copy of the Laozi 老子, the Zhao Zheng shu 趙正書 (The Book of Zhao Zheng), Wang Ji 妄稽, Fan Yin 反淫 (Contra Immoderation) and some other texts (Foster 2017, 168).

As the texts were illegally retrieved by a private party, the circumstances of their discovery and curation remain unknown. Hence, the Peking University manuscripts, alongside the Shanghai Museum and Tsinghua University collections, can be characterized as “looted” artefacts. As such, they present scholars with a series of concerns.

Among the most serious ethical problems is the fact that the purchase of looted manuscripts raises the concern of complicity of academics in practices of grave desecration by encouraging a black market in stolen artefacts (Goldin 2013a, 153–160). As for scholarly issues, not being able to study manuscripts in their original archeological environment gives us only a limited picture of their function and purpose.2 Finally, regardless of how genuine a particular looted document might appear to be, we simply have no means of proving that it is actually not a forgery, at least when using conventional practices of manuscript authentication.3 In the case of the Peking University collection, the suspicion of forgery was raised against the Laozi, its most prominent text (Xing 2016). Even though this suspicion has been persuasively refuted (Foster 2017, 167–239; Staack 2017), the mere fact that it could be raised is indicative of the problematic status of looted manuscripts.

Being aware of these issues, I believe that not investigating the already available looted manuscripts would be equally detrimental given the valuable information they contain, even in their fragmented state. The study of the texts from the Shanghai and Tsinghua collections, has, for instance, changed our understanding of early Chinese historiography and philosophy in many ways. Likewise, the study of the Peking university manuscripts, in general, and the Zhouxun, in particular, has the potential to provide new evidence about early China.

The Zhouxun was published in the first volume of the two-volume collection containing some Peking University manuscripts in September 2015. In addition to the high-resolution photographs of the slips’ frontside, drawings of their rear side are also provided, elucidating the position of the verso lines. Furthermore, the volume contains a transcription of the text, a commentary through the editors and parallel passages from other texts. Transcription is presented based on the original structure of the Zhouxun, which appears to have included 14 chapters (Beijing daxue 2015, 121–122; hereafter cited as Beida Mss). The second volume of that collection also contains an article by Han Wei, the most comprehensive analysis of the various features of the manuscript at hand to date.


The Zhouxun features 211 bamboo slips measuring 30.2–5 cm in length and 0.8–1 cm in width, which were bound with three threads. When fully written, most slips contain twenty-four characters.4 These parameters distinguish it from most other manuscripts from this collection.5

As all other excavated materials, the manuscript of the Zhouxun contains no punctuation. In fact, except for the characters themselves, the only visible markings on the surface of the slips are: dots (•), signifying the beginning of the chapters (and some passages within the last chapter), which are placed on the top of the slips; repetition marks ( ) placed after a character on the right edge of the slip to indicate that the given character should be read twice at this juncture and verso lines on the back of the slips used to establish the order of their arrangement.6

The published transcription, however, contains several additional elements. Most importantly, the editors have added punctuation, based on their understanding of the text. As will be shown below, their interpretation of the text structure and, consequently, the use of punctuation can be contested in a number of cases. Furthermore, the number of bamboo slips is indicated after the respective last character written thereupon, as in, for instance, bu li xiao ren 不立孝仁126. Accordingly, ren 仁is the last character on slip 126. To be able to identify each individual graph, the combination of two numerals will be used in this book: the slip number slash (/) the number indicating the character’s position on the slip. For instance, because slip 126 is inscribed with twenty-four graphs, the character ren appearing at its end has the individual number of 126/24. In addition, the published version contains the following symbols:

1)()for phonetic loan characters or graphic variants;

2)〈〉for writing errors;

3)【】for lacunae which can be reconstructed with a high degree of certainty based on the internal logic of a passage or other considerations. For instance, the beginning of chapter 4 is transcribed in the published version as:【•維歲四月更旦之日,𪚔(共)大子朝,周昭文公自身貳(敕)之,用兹念也。曰:】. This shows that, although the passage is now missing, there is a good reason to assume that originally it appeared at that particular juncture in the text.

4) 囗 for lacunae where the number of characters can be estimated;

5) … … for lacunae where the character number cannot be estimated.

One is able to read the transcription of the manuscript only after having familiarized oneself with the symbols introduced above.

The Scope

The recovered version comprises 211 bamboo slips brushed with 4882 characters and eighty-nine repetition marks.7 At the same time, the manuscript concludes with the count: “in rough approximation, 6000 (characters)” (dafan liu-qian 大凡六千) (slip 211/14–17). Consequently, it seems that about 1118 characters, i.e., almost 19 percent of the original content, have been lost. This number roughly corresponds to forty-six slips (each accommodating twenty-four graphs), making the original scope of the present copy about 257 slips (for a slightly different estimate, see Han 2015, 249–250). However, the loss rate of 19 percent appears unusually high in the context of other Beida manuscripts, many of which have been recovered with only minor, if any, losses (Han 2011, 67). . Indeed, as I will show in the next paragraphs, an approximation of the original scope of the Zhouxun shows that this high loss rate did not reflect the real situation and, most likely, resulted from inaccurate word count.

The Zhouxun most likely comprised fourteen chapters. For the first thirteen, the number of missing slips has been already established with a high degree of certainty, based on such considerations as verso lines, plot development, etc. It is only for the concluding chapter 14, the longest unit in the text, that the corresponding statistics are more difficult to estimate.

While the number of missing slips in a chapter never exceeds three, eight chapters have either no lacunae at all or lack only one slip. This chart can also help us establish the total number of slips in the initial arrangement of each chapter. The mean percentage of missing slips for the first thirteen chapters would be then slightly under 9 percent. The corresponding number of characters for this text selection was consequently around 4558 (4150 extant graphs plus the extrapolated number of 408).

Now, if the initial scope of the Zhouxun really was around 6000 characters, then chapter 14––the existence of yet another textual unit is very unlikely as the content analysis will show––must have originally contained no less than 1642 characters or around sixty-eight slips. In this case, the chapter would be unique in regard to both its length and the degree of its damage (48 percent). But if we apply the mean ratio of 9 percent to its remnants, we get 804 characters (732 extant plus 72 extrapolated). Consequently, the overall number of graphs in the text must have been around 5362 (4558 plus 804), that is, about 638 characters (26 slips) fewer than recorded in the overall count.

While discrepancies between the word count and the actual scope were a widespread phenomenon in the manuscript culture of early China (Zhang X. 2006, 175), the large-scale divagation in the present case requires explanation. Chen Jian (2015), for instance, conjectures that the word count was so high because it also referred to some textual units outside of the work. Another possible explanation could lie in the fact that the concluding word count was evidently written down in a different calligraphic style than the text proper (Han 2015, 250).

Therefore, it cannot be ruled out that the scribe tallying the word count was either unfamiliar with the real scope of the work or based their assessment on the slips that contained more than twenty-four characters. The overall word count, moreover, is formulated as a “rough” approximation, indicating a potential inclination to round up the total number of characters to six thousand.

Palaeographic and Linguistic Characteristics

The calligraphic style of the Zhouxun is the clerical script of the type “Silkworm’s Head and Swallow’s Tail” (cantou yanwei 蠶頭燕尾), which was found to be closest to the calligraphy of the Dingzhou Bajiaolang 定州八角廊 slips (Yan 2011a, 71). Because the latter were discovered in a grave whose occupant passed away in 55 BCE, it stands to reason that the available copy of the Zhouxun was produced some time prior to that, probably during the last years of Han Wudi (r. 140–87 BCE).8

Numerous characters in the Zhouxun deviate in some way from their (current) standard form. Based on the transcription of the work, the number of these variants is approximately 472. Consequently, to be recognizable to modern Chinese readers, about 10 percent of the text requires interpretation of different degree of sophistication. This is a very low percentage compared to other early manuscripts, especially those from the pre-imperial period. Moreover, the editors identified only twelve writing errors among the variants.

This implies that the present manuscript was produced at a time when written language had already acquired a high degree of stability. That this process was not yet completed can be seen in those (around dozen) cases where one and the same character was written differently.9 Sometimes, different variants of the same character can even be found in adjacent sentences (see, for instance, du  (44/14) and du 獨 (45/12) in ch. 3). The fact that they were not corrected to adhere to the same standard suggests that divergent graphic representations were considered legitimate. A small number of characters in the Zhouxun appear to function as phonetic loans for other characters (for example, ŋ 皇 (71/1) for hwaŋh 況, tsiauk爵 (102/10) for tsiauk 雀, grên 閒 (158/22) for krâns 諫, boŋ 逢 (163/13) for phuŋ豐). Their appearance points to the possible role of orality in the transmission of the Zhouxun. While all these phenomena were very common among early manuscripts, the uniformity of the graphic representation of language in the present text speaks to a rather late date of its production.

1 Some passages in this book are based on my previous work on the Zhouxun (Fech 2018 and 2020).

2 For the arrangement and function of texts in early Chinese graves, see, for instance, Lai 2015.

3 As Kern 2019, 46 has pointed out, forgers might even use blank bamboo slips which are abundant in ancient graves, to make their creation pass the carbon-14 dating test. The only solution to that problem would be testing the ink characters are written in.

4 Beida Mss. 121. There are only nine exceptions in the whole manuscript, which are slips 78, 104, 110, 133, 134, 145, 147, 189 and 194. Among them, slips 110, 133 and 147, containing twenty-seven characters each, conclude chapters 8, 10 and 11 respectively. Therefore, it stands to reason that, in this case, the copyists tried to avoid the situation when a bamboo slip is inscribed only with three characters, which would have happened had they followed the regular pattern of inscription.

5 For the Laozi, see Han 2011, 67; for the Zhao Zheng shu, see Zhao H. 2011, 64; for the Wang Ji, see He J. 2011, 75.

6 For a discussion of the function of verso lines, see Staack 2015.

7 The editors only provide an approximate scope of less than 5000 characters (Han 2011, 72). The number 4882 includes neither the title of the work, nor the four characters contained in the overall count, nor the seventeen characters situated on six bamboo snippets (slips 212–217), which cannot be confirmed to belong to the Zhouxun.

8 Su Jian-Zhou 2017, 233 calls the reign period Taishi 太始 (96–93 BCE) as the most likely date of origin for the Peking University version.

9 On the gradual character of language standardization in early China, see Galambos 2004, 189–194.