1. Introduction

Recent discoveries of Shijing 詩經 (Classic of Odes) materials, including the Anda manuscript (probably mid-fourth century bce) received by Anhui University in 2015Footnote ¹ and Shijing slips unearthed from the tomb of the Marquis of Haihun 海昏 (i.e. Liu He 劉賀 [92–59 bce]) in Nanchang 南昌Footnote ² have brought scholarly attention to the previously neglected issue of how stanzas in the Shijing are sequenced.Footnote ³ We have learned that the sequence of stanzas within a poem was not fixed across all editions before the first century bce, especially for poems in the “Guo feng” 國風 (Airs of the states). Scholars have compared the order of stanzas within these newly discovered texts with the order of stanzas in the Mao shi 毛詩 (Mao tradition of the Shijing) and attempted to explain the differences.Footnote ⁴ Some scholars believe that there exists a single most “reasonable” (implicitly “original” in some cases) stanza permutation.Footnote ⁵ However, as most repetitive stanzas and even some non-repetitive stanzasFootnote ⁶ are essentially non-linear, and as there is no intrinsic order to the sequence of many poems in the “Guo feng”, I maintain that it is unproductive even to attempt to find a “superior” stanza order. Rather than comparing various stanza orders in this way, I would like to consider how the fact that there were changing stanza orders both in the Warring States period and after the Western Han might challenge traditional hermeneutics.

As the order of stanzas in the early period appears to have been unsettled, we have to re-evaluate the only early text that has clear references to stanza numbers, the Zuozhuan 左傳 (Zuo Commentary). While these references were previously considered evidence that a stanza order had been uniformly agreed upon before the empire,Footnote ⁷ this opinion has been undermined by the Anda and Haihun manuscripts, which reveal that different stanza permutations existed in the Warring States period (453–221 bce) and even in the early Western Han (202 bce–9 ce). While some scholars use the Shijing-related materials in the Zuozhuan to support the way stanzas are ordered in the Mao shi,Footnote ⁸ this approach is problematic due to the possibility that the Zuozhuan was revised according to the Mao shi.Footnote ⁹ Nevertheless, the references to stanza numbers in the Zuozhuan may still reveal the early steps taken in the process towards stabilizing the order of stanzas, such as fixing non-repetitive stanzas in poems that are only partially repetitive.

Different stanza permutations also existed after the Western Han. The Xiping shijing 熹平石經 (stele inscriptions of Classics during the Xiping [172–178] era) at the end of Eastern Han (25–220) reveals differences in the way stanzas were ordered between the official Lu shi 魯詩 (Lu tradition of the Shijing) and the current Mao shi.Footnote ¹⁰ Within the Mao transmission lineage, standardization has also been a long process. Some poems with repetitive stanzas might not have been fixed even as late as the third century ce. I will show below that the sequence of some stanzas in the Mao shi changed even after Zheng Xuan's 鄭玄 (127–200) annotation of the text, taking several medieval comments on the “Lü yi” 綠衣 (The Green coat; Mao no. 27) and “Zhonggu you tui” 中谷有蓷 (In the valley there are motherworts; Mao no. 69) as examples.

The existence of various stanza permutations challenges a specific kind of traditional interpretation of the Shijing, namely the hermeneutical rule of “orderly progression”.Footnote ¹¹ This approach imposes a progressive sequence on poems with repetitive stanzas and assumes that the current stanza order is unalterable and meaningful. This approach has been adopted in both pre-modern and modern annotations of the text. For example, according to Kong Yingda 孔穎達 (574–648), the three stanzas in the poem “Fuyi” 芣苢 (Mao no. 8), which depicts women picking Chinese plantain, illustrate the process of gathering more and more wild vegetables. In this interpretation, the three stanzas establish a chronological order that can only be deciphered by interpreting the rhyming characters in each stanza in a particular way.Footnote ¹² However, this kind of interpretation is missing from the Han commentaries. Just as establishing the order of stanzas took a long time, the development of this type of approach was gradual. As I will show below, the Mao and Zheng commentaries only applied the rule of orderly progression occasionally, while Kong applied it more frequently, a difference probably attributable to the disappearance of the other three lineages and the establishment of a single orthodox Shijing text.Footnote ¹³ By the time of Qing (1644–1912) and modern commentators, however, this principle had become widely accepted as the standard interpretation. Yao Jiheng 姚際恆 (1647–1715) was perhaps the first influential commentator to borrow critical vocabulary such as zhangfa 章法 (standards of structuring sections) from wenzhang xue 文章學 (the study of essays) to analyse the relationship between stanzas in the Shijing. Influenced by the presumption of wenzhang xue that a well-written composition should be logically arranged,Footnote ¹⁴ Yao regarded the stanza permutation in the current Mao shi as significant and unchangeable. After the May Fourth Movement, the debate about whether repetitive stanzas are essentially the same or progressive became part of a larger discussion on the nature of the Shijing. Once the notion of “Guo feng” as a collection of folk songs took hold, repetitive stanzas came to be viewed as “developing”, rather than merely repetitive. This theory is now the mainstream interpretation and generally accepted by contemporary scholars, even after the discovery of the Anda manuscript. This belief in there being an unalterable stanza order not only influences how rhymes are interpreted but also shapes how lines and verses are annotated. Only by recognizing that the order of stanzas was neither fixed nor self-evident can we revise interpretations that derive from this assumption.

2. Analysing variations in stanza sequences in the “Guo Feng”: Warring States Period to early Western Han

Both the Anda manuscript and Haihun strips provide clear information as to how stanzas were sequenced. The Haihun strips, while badly damaged, contain a record of the sequence of stanzas at the end of each stanza.Footnote ¹⁵ These strips contain six poems in the “Guo feng” that have a different stanza order from the Mao shi.Footnote ¹⁶ The well-preserved Anda manuscript includes 57 poems, all from the “Guo feng”,Footnote ¹⁷ and both the slip numbers at the end of each strip and the markings on the back of the strips indicate how they were supposed to be arranged. Sixteen of these poems have different stanza sequences from the Mao shi, constituting almost 30 per cent in total.Footnote ¹⁸ Many scholars have observed that the majority of the poems that have a different stanza order (13 out of 16) are repetitive (i.e. poems with repetitive stanzas), with some concluding from this that stanza inversion is more likely to occur in repetitive poems.Footnote ¹⁹ However, Table 1 indicates that there is no difference in the proportion of stanza inversions between repetitive and non-repetitive poems in the Anda manuscript.

Table 1. Percentages of stanza inversion in poems with repetitive and non-repetitive stanzas in the Anda manuscript

The limited number of examples of stanza inversions means that Table 1 may not be statistically significant. Nevertheless, the fact that three out of seven non-repetitive poems have a different stanza permutation in the Anda manuscript does challenge the idea that stanza inversion is a characteristic primarily of repetitive poems. Although many poems with different stanza permutations are repetitive, this is likely due to the large number of repetitive poems in the “Guo feng” (138 out of 160). Non-repetitive poems in the “Guo feng” may also have undergone stanza inversion.

Moreover, scholars have noted that in most cases where stanzas in the Anda manuscript have been inverted (11 out of 16), the first stanza remains the same while other stanzas trade places. Eight out of 12 such cases in the Haihun strips are the same, and the first stanza is unchanged.Footnote ²¹ The first stanza of a poem appears to be more stable than other stanzas, even in fully repetitive poems. Some scholars suggest that whether the first stanza was stabilized or not really depended on its first line. For example, Yuasa Kunihiro 湯淺邦弘 states that when the first line appears exclusively in the first stanza, the first stanza “is less likely to undergo inversion”, but when it is present in all stanzas, every stanza is likely to be inverted.Footnote ²² Indeed, the first lines in each stanza of all five poems in the Anda manuscript that have undergone first-stanza inversion (Mao nos 19, 46, 113, 114, and 131) match this description. However, two poems in the Haihun strips (Mao nos 90 and 262) break this trend: the first lines in the first stanzas of these poems are unique, but the first stanzas have still been inverted. How do we explain this difference?

The first lines of each poem are important largely because titles are usually derived from them.Footnote ²³ Compared to the first line, the title of a poem seems to function more effectively as an indication of the stability of the first stanza. Once the title is established, the title helps facilitate memorization of the poem and aids in differentiating one poem from another. Users and transmitters of the Shijing may have easily been prompted by the title to remember how the poem begins. The significance of titles can be seen in the comparison of two fully repetitive poems that have different permutations, “Bao yu” 鴇羽 (Plumes of the bustards; Mao no. 121) and “Fengyu” 風雨 (The wind and rain; Mao no. 90). The first stanza of the former remains unchanged in the Anda manuscript, while the first stanza of the latter (in the Haihun strips) is the second stanza in the Mao shi. In both poems, the first lines appear only in the first stanza, but in “Bao yu” the title only appears in the first stanza. In “Fengyu” the title recurs in every stanza, and thus it could not be used to help stabilize the first stanza.Footnote ²⁴ Therefore, it seems that titles play a more crucial role in fixing the first stanza than the first line.Footnote ²⁵

Although it is unclear whether all the poems in the Anda manuscript had the same titles as those in the Mao shi (the only two extant titles are the same), from other excavated or looted texts that mention titles – such as the “Kongzi shilun” 孔子詩論 (Confucius's discussion of the Shijing) in the Shanghai Museum strips – we know that most titles (in this case 52 out of 59) are the same as those in the Mao shi. That is to say – assuming that the looted “Kongzi shilun” is an authentic ancient text – most titles were fixed by the third century bce, and some of the first stanzas were likely fixed in turn by the titles.

In Table 1, repetitive poems are further categorized by whether they are fully or partially repetitive. As shown above, stanzas in fully repetitive poems are relatively unfixed and the order of the stanzas is prone to change, though not if the title consists of unique words that only appear in the first stanza. Similarly, partially repetitive poems allow for the position of repetitive stanzas to be swapped and exchanged even while non-repetitive stanzas are rather stable. For example, the four-stanza “Juan'er” 卷耳 (Juan'er plant; Mao no. 3) has two non-repetitive stanzas (the first and last stanzas) and two repetitive middle stanzas. In the Anda manuscript, the first and last stanzas of this poem are the same, while the middle two stanzas trade places. Likewise, in Mao no. 126 in the Anda manuscript, only the non-repetitive stanza remains in the same place while the two repetitive stanzas switch positions. In brief, non-repetitive stanzas in partially repetitive poems in the “Guo feng” are always located at the beginning or end of a poem,Footnote ²⁶ and it seems less likely that these stanzas would have been made to trade places.

To summarize, the analysis of early stanza permutations reveals three points. First, it seems no less likely that non-repetitive poems will undergo stanza inversion than repetitive poems. Second, the first stanzas in a poem are relatively stable, particularly when the title appears there exclusively. Finally, non-repetitive stanzas in partially repetitive poems, which are usually located at the beginning or end of a poem, are more likely to be fixed than are repetitive stanzas.

3. Different stanza orders after the early Western Han

Even after the establishment of the empire, poems in the Shijing had stanza orders that varied. The few surviving fragments of the Xiping stele inscription indicate that by the end of the second century ce, the order of stanzas for some poems in the official Lu shi was different from the order in the current Mao shi, which suggests that different scholarly lineages during the Han dynasty may have had their own preferred stanza arrangements. Some 300 years after the three lineages had been established as official traditions, there was still no stanza order that was universal across the different schools.

Moreover, even within one single lineage, the Mao shi, the order of stanzas could still be easily changed. The early transmission process of the Mao shi is highly obscure, and the text as it exists today is an accumulative patchwork that contains many layers of commentary.Footnote ²⁷ Since the Mao shi was not established in the court until the end of the Western Han dynasty, it was not as settled as other sanjia recensions in the early stages of its transmission.Footnote ²⁸ Even during the Eastern Han dynasty, when many famous scholars valued and annotated the Mao shi, it was still a marginal Shijing tradition. It was only after Zheng Xuan's exegesis (i.e. Mao shi zhuan jian 毛詩傳箋) at the end of the Eastern Han dynasty that the Mao shi began to become the mainstream tradition of the Shijing. There was, therefore, a period of nearly 400 years when the Mao shi was not fixed and when it could have been easily altered.

Although Zheng contributed substantially to the stabilization of the Shijing, his annotation was fiercely attacked by Wang Su 王肅 (195–256) almost immediately after his death.Footnote ²⁹ This led to 400 years of disagreement between proponents of Wang and Zheng.Footnote ³⁰ This rivalry was only settled in the 600s, after Yan Shigu 顏師古 (581–645) stabilized the characters in the five classics at the emperor's behest in 630 and Kong Yingda presented his exegeses on the five classics – including his Mao shi zhengyi 毛詩正義 (The corrected interpretation of the Mao shi) – in 641. Twelve years later, Kong's work was honoured as the official standard for the imperial examination. Only at this point was the Mao shi truly fixed.

I will provide two examples showing that the stanza order of the Mao shi was altered after Zheng Xuan. One is found in Zheng's exegesis, and the other is from scattered Guoyu 國語 (Discourses of the states) comments. Both suggest that there may have been another version of the Mao shi (or even several versions) that contained different stanza sequences around the third century ce.

“Zhonggu you tui” 中谷有蓷 (In the valley there are motherworts; Mao no. 69) has a typical structure of three fully repetitive stanzas, with the title appearing in each stanza. The first couplets of each stanza in the sequence of the current Mao shi are listed below.

The rhyming words in the current Mao shi are gan 乾 (dry), xiu 脩Footnote ³³ (withered), and shi 濕 (wet), in that order. However, Zheng Xuan expounds on the sequence of this poem differently:

It seems that Zheng saw a version of the Shijing that reversed the first and last stanzas, and he naturally followed this sequence in his summary. When Kong Yingda came to interpret Zheng's annotation, seeing as he was working from the same version of the Mao shi as we have today, he struggled to reconcile the inconsistency:

Kong's rationalization of the sequence is rather unconvincing. Since the analysis of the Anda manuscript above has shown that the order of stanzas in fully repetitive poems could easily be altered, this further suggests that the concept of an authorial arrangement of stanzas may be an illusion. Regardless, the discrepancy between Zheng's analysis and Kong's comment implies that the order of the stanzas of this poem in the Mao shi changed after Zheng's annotation.

“Lü yi” 綠衣 (The green coat; Mao no. 27) is similar in that the order of stanzas in this poem appears to have been changed after Zheng. We can infer this from a reference in the Guoyu, in which a female character recites the third stanza of it (賦《綠衣》之三章).Footnote ³⁶ Wei Zhao 韋昭 (204–273), probably the most influential medieval commentator of the Guoyu, illustrates this recitation as follows:

Wei Zhao was likely referring here to the Mao shi, not one of the other three recensions, since a commentary on the Mao shi under his name is listed in the imperial library catalogue in the Sui shu 隋書 (History of the Sui). However, the couplets he quotes are not in stanza three in the present Mao shi. They appear instead in the last stanza (stanza four). If this annotation is not merely a mistake, then the version of the Mao shi that Wei saw was different from the one that currently exists. Both “Lü yi” and “Zhonggu you tui” illustrate that even after the third century ce the sequence of stanzas was still not stable for some poems with repeated stanzas.

This prompts us to reconsider at least three questions. First, how should we understand the references to specific stanza numbers in the Zuozhuan? Second, is it possible to find the “best” stanza order among all the different permutations? Is there an intrinsic order across stanzas in the Shijing? Third, since the order of stanzas has now been called into question, what do we do with traditional interpretations based on the idea that stanzas must progress teleologically and in an orderly manner, interpretations that presuppose an unalterable sequence of stanzas? How has this hermeneutic principle been established? What interpretations will have to be reconsidered when we abandon it?

4. Rethinking accounts of stanza numbers in the Zuozhuan

How might unstable stanza sequences affect our understanding of early texts? If all the quotations from the Shijing included only a few lines or couplets – as is the case with quotations from many pre-Qin texts – a varied stanza sequence would be of little concern. However, some quotations from the Shijing in the Zuozhuan are quite exceptional, in that they even mention the specific stanza being quoted (e.g. “reciting stanza X of a poem called Y” 賦某詩之某章).Footnote ³⁸ This unique quotation pattern, referred to by a historical figure in the Zuozhuan as “cutting out a stanza when reciting the Shijing” (fu Shi duanzhang 賦詩斷章),Footnote ³⁹ suggests that these reciters are more concerned with the meaning of a specific stanza rather than with the poem as a whole. This approach was later emphasized by Du Yu 杜預 (222–285), one of the most influential pre-Tang commentators on the Zuozhuan.Footnote ⁴⁰ Before the discovery of the Anda manuscript, some scholars treated these references to specific stanza numbers as evidence of the existence of a fixed version of the Shijing in the Spring and Autumn period (770–453 bce), since the references indicated that both speakers and listeners would know which stanza the quotations were referring to.Footnote ⁴¹ However, this assumption has now been shown to be invalid. Given that there were several stanza sequences at the time, how could the references to a stanza number have been comprehensible?

Most accounts of stanza numbers in the Zuozhuan (16 out of 20) are written from a third-person perspective. The protagonist himself does not say “stanza X of poem Y”; instead it is always presented in the narrative as “someone recited stanza X of poem Y”, with the stanza number written down by the narrator (or by later editors). Therefore, the problem of identifying which stanza was recited would have only concerned readers of the Zuozhuan, not the listeners in the scene. Since many “Guo feng” poems had different stanza sequences in the early period, mentioning a stanza number would only have been meaningful if there had been a relatively settled Shijing version that circulated within a certain group of readers alongside the Zuozhuan. This would have ensured that readers could have accurately identified the corresponding stanza without confusion. The Mao shi is the most likely candidate for what this specific version would have been, for two reasons. First, according to the limited records of the history of the text,Footnote ⁴² the Mao shi is probably the version of the Shijing with the closest relationship to the Zuozhuan in the early transmission process. Master Mao 毛公, the first traceable teacher of the Mao shi,Footnote ⁴³ passed the text on to Guan Changqing 貫長卿 (fl. the late second century bce), who also learned the Zuozhuan from another teacher. This is the earliest record of a version of the Shijing and the Zuozhuan intersecting. At the end of the Western Han, the Mao shi was established in the court for the first time, together with the Zuozhuan. In the Eastern Han dynasty, these two texts were always mentioned together and many famous scholars learned both at the same time.Footnote ⁴⁴ The transmission process of the Mao shi and the Zuozhuan were intertwined through the two Han dynasties. Second, because these two classics were frequently circulated together, they inevitably influenced each other.Footnote ⁴⁵ Many researchers have noticed that many parts of the Mao commentary are identical to the corresponding accounts in the Zuozhuan.Footnote ⁴⁶ In addition, the Shijing quotations in the Zuozhuan are, when compared to the text in all the lineages, the most similar to the text of the Mao shi.Footnote ⁴⁷ In other words, not only did the Mao shi benefit from the Zuozhuan when developing its interpretation, but the Shijing-related materials in the Zuozhuan were also revised according to the Mao shi.

Therefore, we should be very cautious about using the Zuozhuan to support the sequence order of the Mao shi, since the transmission and connection between the two is essentially circular. For example, the Anda manuscript, the Haihun strips, the Xiping inscription, and the Mao shi each have a different permutation of “Huang niao” 黃鳥 (Yellow birds; Mao no. 131).Footnote ⁴⁸ Many scholars have concluded that the stanza order in the Mao shi is the most reasonable.Footnote ⁴⁹ Their evidence is one entry from the Zuozhuan which discusses three outstanding gentlemen who were buried with the deceased Lord Mu of Qin (r. 659–621 bce).

The three stanzas in the Mao shi list the three men in the same sequence as in the Zuozhuan, while other permutations do not. Consequently, some scholars argue that the stanza sequence in the Mao shi is fully justified. They believe that Zhonghang 仲行 must be placed in the middle, since zhong 仲 means “the second sibling”.Footnote ⁵¹ However, this theory faces some intractable problems. First, we know nothing about these three individuals other than their names, and we cannot confirm whether Yanxi 奄息, Zhonghang, and Zhenhu 鍼虎 are given names or style names. The Mao commentary claims that they are given names, while the Zheng annotation suggests that Zhonghang is a style name and the others are given names. Given that terms of seniority among siblings usually appear only in style names, inferring the seniority of these three people based on their given names is far-fetched. Moreover, there is no indication that the names in the Zuozhuan are arranged in any particular order, let alone a “correct” order. Even if the three names were intentionally arranged, the account in the Zuozhuan may have been revised according to the Mao shi as mentioned above, making this insufficient proof to show the superiority of the stanza order in the Mao shi. It is impossible to determine the most “reasonable” sequence of the three men, and early readers of the Shijing seem to have been unconcerned about the sequence, as evidenced by there being four different permutations. “Huang niao” is simply a typical example of a fully repetitive poem in which it is equally likely that every stanza could be moved.

Although the Zuozhuan cannot be used to choose the “best” sequence of stanzas, the references to stanza numbers in it can still provide some valuable information. Out of the 20 instances where stanza numbers are mentioned, 16 are in the “Xiao ya” and the “Guo feng”. These cases are sorted in Table 2.

Table 2. References to stanza numbers in the Zuozhuan (“Xiao ya” and “Guo feng” only)

It is not clear whether all the stanza divisions in the Zuozhuan match those in the current Mao shi. Since these two texts are closely interrelated, I have tentatively used the stanza sequence in the Mao shi as a reference. We can see from Table 2 that the last stanzas are especially interesting, with 11 of 16 examples being mentions of the final stanza. Why are the last stanzas especially noteworthy? Why did writers and compilers of Zuozhuan believe that protagonists would remember and recognize the last stanzas in particular? The last stanzas of fully repetitive poems are unlikely to have been settled, as demonstrated by the analysis of the Anda manuscript and the Haihun strips. However, the non-repetitive stanzas in partially repetitive poems and non-repetitive poems with a linear structure are relatively stable.Footnote ⁵⁴ Almost all of the eleven cases in the Zuozhuan have non-repetitive last stanzas.

The importance of the last stanza is also supported by the Shanghai Museum manuscript. “Kongzi shilun” discusses nearly 60 poems, but there are only two mentions of a specific stanza. The first mention is to the fourth stanza of “Guan ju” 關雎 (Fishhawks; Mao no. 1) (“The fourth stanza is an analogy”. 其四章則喻矣),Footnote ⁵⁵ and the other is to the last stanza of “Da tian” 大田 (The big field; Mao no. 212) (“The last stanza of Datian is about understanding [proper] phrasing and having command of ritual.” 大田之卒章，知言而有禮).Footnote ⁵⁶ The non-repetitive “Da tian”, which belongs to the “Xiao ya” section, has a last stanza that is distinct from other stanzas. It seems that by the third century bce, some non-repetitive last stanzas in the “Xiao ya” and “Guo feng” were probably fixed, and they were likely arranged at the end of the poems.

According to Table 2, other stanzas of partially repetitive poems, such as the fourth stanzas of “Guan ju” and “Zai chi” 載馳 (Gallop; Mao no. 54), are also mentioned sometimes, though much less frequently than the last stanza. “Guan ju” and “Zai chi” may be among the most popular and influential poems in the “Guo feng”.Footnote ⁵⁷ If the records in the Zuozhuan are to be trusted, “Zai chi” is one of the few poems in the entire Shijing that has a known author and which was composed in response to a famous contemporary event. It was being quoted by other politicians in official meetings only a few decades after its composition. As for “Guan ju”, it is not only the first poem in the Shijing,Footnote ⁵⁸ but it is also the most well-known song in the corpus. “Guan ju” is often referenced for all of the “Guo feng”, and it is the only poem from the Shijing that Confucius comments on not once but twice in the Analects.Footnote ⁵⁹ This poem was evidently of extraordinary importance. Significant and popular songs often receive greater attention, and the stanzas within them are likely to have been fixed earlier than stanzas in less famous ones. Overall, different poems and stanzas were stabilized at different times. Several well-known poems probably had a rather fixed stanza order at least by the Warring States period, and non-repetitive stanzas in partially repetitive poems (normally the first or last stanzas) are also more likely to have been settled in a specific position by this time. This observation aligns with my previous analysis of partially repetitive poems in the Anda manuscript.

5. Is there an intrinsic stanza order?

Many scholars have compared various stanza permutations with the Mao shi, with the aim of identifying the most “reasonable” order. However, is there an intrinsic order across stanzas? Can we possibly find it?

In the first section of this paper, I proposed that an analysis of the Anda manuscript suggests that non-repetitive poems in the “Guo feng” may have been no less likely than repetitive poems to have undergone stanza inversion. This conclusion may seem, at first glance, counterintuitive. Some non-repetitive poems do have an apparently linear structure. For example, the three stanzas in the “Cai pin” 采蘋 (Gathering the waterplants; Mao no. 15) describe the process of picking waterplants, placing them in vessels, and finally depositing them in the ancestral shrine. Since these three actions happen in chronological order, it would be hard to alter the sequence of the stanzas. However, not many non-repetitive poems in the “Guo feng” have this kind of unalterable stanza order. Some stanzas in the same poem are fundamentally about the same thing, but the perspective changes. Take the non-repetitive “Pao you kuye” 匏有苦葉 (The gourds have bitter leaves; Mao no. 34) for example. All four stanzas in this poem describe the scene of a young lady waiting for a gentleman on the riverbank. Some stanzas are from the point of view of the lady and directly express her thoughts, while other stanzas describe the surroundings, but there is no underlying order across the stanzas as a whole. For these non-repetitive poems, the order of the stanzas can be easily changed without affecting the content, just as with repetitive poems.

While one may be tempted to think of non-repetitive stanzas as intrinsically fixed in their order, it is often impossible to determine any particular sequential logic between them. Therefore, the attempt to search for the most “logical” permutation might be unproductive. For example, both Yuasa Kunihiro 湯淺邦弘 and Kang Tingshan 康廷山 suppose that the Anda version of “Sitie” 駟驖 (Iron-Black Horses; Mao no. 127) is better than the Mao shi version (in which the last two stanzas are reversed), with Kang arguing that the order of the three stanzas in the Anda manuscript is more logical, for it depicts the hunting event starting with its inception (stanza one), then moves to the preparation for the hunt (stanza two), and finally shows the hunt itself (stanza three).Footnote ⁶⁰ Yuasa similarly believes that the spiritual composure of the rider in stanza two “enables success in the subsequent hunt (stanza three)” and further states that “the Anda preserves the original state of the text, whereas the Mao shi seems to have been miswritten”.Footnote ⁶¹ Nevertheless, all three stanzas are essentially about the same topic – a hunt – and the Mao shi version may be equally valid. The Mao shi version is as follows:

Zheng Xuan interprets the last stanza as a flashback depicting the way the hunt was carried out, thus stating that it should precede the second stanza. However, there is a lack of substantial textual evidence for his argument. Kong Yingda, on the other hand, contends that all three stanzas portray the scene of hunting, including the pleasure enjoyed in the northern park (the park where the ruler goes hunting) in the last stanza.Footnote ⁶³ Kong's understanding seems to be supported by the content of this poem, as each stanza focuses on a slightly different aspect of the hunt. The first stanza mentions the participants, the second focuses on the ruler, and the third portrays the relatively peaceful time between active chases when the hounds rest in the carriages. As these periods of rest could occur at any point during the hunt, it is difficult to establish a definite order to the stanzas. Readers should not therefore impose an order on the stanzas of this poem.

The “intrinsic order” of repetitive poems, if indeed it even exists, is even more challenging to ascertain. A few repetitive poems (fewer than 10%) seem to have an intrinsic order across repeated stanzas, and this order is driven by a logic that is chronological, spatial, or numerical. For example, the first two repetitive stanzas in Mao no. 10 describe two scenes, one where the protagonist of the poem has not seen the gentleman, and one where the protagonist catches sight of the gentleman. In the three stanzas of Mao no. 98, the person waiting for the protagonist first stands between the gate and a screen wall, then he stands inside the courtyard, and finally he stands in the hall. Mao no. 53 describes a set of horses, and the number of horses increases across the three stanzas from four to five, and then from five to six. However, even among these poems where the order seems fairly immutable (e.g. the number of horses going from four to six and then from six to five would be very awkward), it is still possible to reverse the sequence of the three stanzas without altering the poem's overall meaning (e.g. a pattern of six, five, and four sounds equally reasonable).

Most of the repetitive stanzas seem to lack this kind of unalterable structure. Many contemporary readers perceive these poems as having a progressive order mainly because they rely on existing commentaries, not the poem itself, to determine the “best” order. For example, the stanza sequence of “Chou mou” 綢繆 (Tightly bound; Mao no. 118) in the Mao shi has been generally considered to be superior to the one in the Anda manuscript. This assessment is based on the inherited interpretation of the rhyming words in the first couplets of each stanza. The couplets are listed below as they appear in the Mao shi version.

These couplets contain two groups of rhyming words: one group refers to different plants (薪-芻-楚) while the other refers to different positions of the Three Stars (天-隅-戶). The second group has attracted much more attention since a hierarchical system cannot be found among the plants. Li Hui 李輝 and Cao Jianguo 曹建國 both believe that the Mao shi's sequence is better than the Anda version's sequence of “zai tian 在天 -zai hu 在戶 -zai yu 在隅”. Cao follows the Mao commentary and asserts that the stanza order of the Anda manuscript “goes against common sense and must have been a transcription error”.Footnote ⁶⁵ Similarly, Li follows Zhu Xi's 朱熹 (1130–1200) annotation and claims that “the stanza order of the Mao shi version is better since it is arranged chronologically”.Footnote ⁶⁶ Since Zhu's annotation of these words shares the same logic as the Mao commentary, I list the comments from the Mao commentary on the rhyming words below:

The Mao explanation presents an unalterable sequence of stars rising from the east, the south-east, and the south.Footnote ⁶⁸ However, there is no clear indication of direction or position in the poem itself. Although zai hu probably refers to the south, the yu mentioned in the second stanza is ambiguous since there are multiple corners in each direction.Footnote ⁶⁹ Furthermore, the comment on zai tian is especially questionable as there is no evidence in the early texts to support the explanation that this is “in the east”. Zai tian means only “in the sky”,Footnote ⁷⁰ so the Three Stars mentioned in this stanza could conceivably be visible from every direction. In other words, Mao's interpretation of the motion of stars across the stanzas can only be valid if we accept his unsupported explanation of each rhyming word – and this explanation is based entirely on the presumption that there is a stable order to the stanzas. As in this example, previous comments cannot serve as reliable evidence for determining which order of stanzas is most reasonable. In short, readers should avoid assuming that there is an intrinsic order to the stanzas without checking for substantial evidence within the poem itself. As most of the repetitive stanzas are essentially non-linear, seeking a “better” stanza order, as exemplified by “Chou mou”, is futile.

6. The history of the hermeneutical tradition of “Orderly Progression” – and a reconsideration

As shown above, most non-repetitive and repetitive stanzas of the poems in the “Guo feng” do not have a logical order, despite the belief of many traditional commentators that the stanza order of such poems was gradually progressive. However, this belief was not dominant in the early stage of Shijing hermeneutics, and it took considerable time to become well established. An examination of the formation of this tradition can help undermine its tacit acceptance that dominates scholarship today.

The earlier analysis of “Chou mou” in this paper revealed that while signs of this hermeneutic approach are visible in the Mao commentary, the approach was only used to interpret a few poems. Another noteworthy case is the Mao comment on “Tao yao” 桃夭 (The peach tree is young and beautiful; Mao no. 6). The first couplets of each stanza and the corresponding Mao comment are as follows:

The lines in bold lend support to a progressive interpretation where the flower-like appearance comes first, followed by female virtues and, finally, perfect prosperity. Still, there is little evidence in the poem to support this interpretation. It is unclear why it is the fruit of a peach tree, not anything else, that is analogous to a woman's virtues, or why the description of fruit precedes the reference to leaves, which obviously goes against the natural process of growth. The word zhenzhen is also problematic, as it is unclear why it means the “greatest” prosperity (zhi sheng 至盛). As in the case of “Chou mou”, this annotation of the rhyming words seems to be influenced by a pre-existing interpretation. It is notable that the bold lines are distinct from the preceding annotations of specific words. These lines focus on the metaphorical and moral meaning of the couplets, rather than on explaining individual characters or phrases. Such abstract comments are rare among the many layers of materials in the Mao commentary and may have been added later than the plain annotations that proceed them. For the majority of the Mao commentary, there is no progressive interpretation of the order of stanzas, even when the poem itself seems to proceed in a seemingly apparent way. Take for example the first couplets of each stanza and the relevant Mao comment on “Shu li” 黍離 (That glutinous millet hanging down; Mao no. 65):

It seems natural to take the sprouts, ears, and grain of panicled millet in these three stanzas as marking three moments in the growth of a plant. If so, these stanzas are listed in chronological order. However, it seems unlikely that the writer of the Mao commentary, based on the comment, would have been enthusiastic about this interpretation. The Mao commentary suggests that each stanza be read independently, and it is unclear whether the commentator believes that the poet saw different parts (or stages) of panicled millet all at once or that the poet was viewing the millet at different times. The silence about this in the Mao comment indicates that the commentator refused to build a link from one stanza to another. In addition to this, the Mao commentary rarely differentiates between rhyming words but tends to group them into the same semantic category. The typical way the Mao commentary annotates rhyming words is by commenting “rhyming word A is like rhyming word B” (A, 猶 B也).Footnote ⁷³ In other words, these comments do not often reflect the thematic rule of orderly progression.

In contrast, Zheng Xuan's annotations show a greater application of such a rule. For example, he connects the three stanzas in the “Zhonggu you tui” and interprets them as describing the damage inflicted on the motherworts increasing over time. We cannot find such an interpretation of this poem in the Mao commentary. Similarly, Zheng interprets the repetitive stanzas in the “Guan ju” as being in chronological order.

In the Mao commentary there is no implication of a progressive order across these stanzas, but Zheng obviously understands these stanzas to be in chronological order. He believes that there is a transition from “not yet gathered” plants to “already gathered” plants across the poem, although there is hardly any support for his argument that the three stanzas describe events that happened at different times. It seems that in Zheng Xuan's time this hermeneutical principle of orderly progression was more widely accepted by commentators, which resulted in more poems being interpreted using this principle.

We can see the theory of orderly progression gaining significant traction 400 years later when Kong Yingda explained almost twice as many poems using this rule. Table 3 compares how many repetitive poems were interpreted by Mao, Zheng, and Kong using this hermeneutical rule.

Table 3. Numbers of repetitive poems with orderly progressive annotations by Mao, Zheng, and Kong

Although the majority of repetitive poems were not understood as having a progressive order, there is a notable increase in the application of this principle in Kong's annotations. Kong sometimes tries to explain the structure of a poem and the function of each stanza in his exegesis of poem titles and prefaces, a popular annotation practice since the Six Dynasties.Footnote ⁷⁷ In doing so, he inevitably imposes a linear progression on repetitive stanzas. For example, he interprets the order of the three stanzas in the fully repetitive “Fuyi” as follows:

This chronological order sounds far-fetched, and it does not appear in either Mao or Zheng's annotations; it seems to have been fabricated by Kong. It is just as possible that all the activities in bold happened simultaneously. The three stanzas are also essentially non-linear, and if the order of the stanzas was inverted, the understanding of this poem would not be affected.

What factors contributed to Kong's interest in interpreting poems in a progressive way? Aside from the rise of a new type of interpretation, yishu 義疏 (meaning and explanation), it is also possible that the disappearance of the other three lineages influenced Kong's choice. According to Lu Deming 陸德明 (c. 550–630), by the very beginning of the Tang the Qi shi had long been extinct, the Lu shi had no influence in northern China, and although the Han shi 韓詩 (Han tradition of the Shijing) still survived, no scholar was transmitting it.Footnote ⁷⁹ With few alternative permutations available, the Mao shi was the only available Shijing text, and it is natural that Kong adopted the Mao shi stanza order as the only reasonable sequence. Moreover, as the person responsible for standardizing the Mao shi, it would have been essential for Kong to further legitimize the text, including its stanza order.

Despite this, Kong only interpreted around 20 per cent of repetitive poems in a progressive way. In terms of his general argument about repetitive stanzas, he does not appear to believe strongly in the progressive order. On the contrary, he claims that many stanzas deal with essentially the same topic. For example:

Kong focuses more on common themes rather than highlighting differences between stanzas. In his comment above, for example, even though three stanzas in the “Cai pin” seem to describe the different stages of preparing a sacrifice, Kong still regards the stanzas as being about the same thing. Thus while he advances orderly progression more than previous commentators, such progression had not yet become a commonly accepted hermeneutical rule.

An important pre-modern commentator who contributed more substantially to this interpretive tradition is Yao Jiheng, a scholar from the early Qing period. In his Shijing tonglun 詩經通論 (General discussions of Shijing), he clearly proposes orderly progression as a rule for interpreting repetitive stanzas (shi li 詩例). He argues that the rhyming words in repetitive stanzas must vary in either order or degree and strongly opposes Kong's general statement about how multi-stanzas “restate the profound emotions”.Footnote ⁸¹ Consequently, in almost all of his exegeses of repetitive poems, he regards the stanza order as unalterable and perfectly reasonable. Look at some examples from his explanations in the first section, “Zhou nan” 周南, for instance:

In his explanations of the structure of the poems, Yao repeatedly emphasizes li 例 (rules) and zhangfa. These terms are not his innovations and were already established in the study of essays (wenzhang xue 文章學) that greatly flourished in the Ming and Qing dynasties.Footnote ⁸³ Although the tradition of wenzhang xue has a long history dating back to the Six Dynasties, for most of its history it largely focused on formal essays rather than on poems.Footnote ⁸⁴ In the late Ming, however, the methodology of wenzhang xue began to be applied to other texts, including novels and poetry.Footnote ⁸⁵ According to this methodology, every sentence and paragraph in a well-written article should be logically connected to previous sentences and paragraphs, and the structure of both cannot be altered. Yao brought this basic assumption to the field of poetic criticism and borrowed some terms from wenzhang xue in his examination of the sequence of stanzas in the Shijing. In other words, the establishment of orderly progression as the dominant explanation for poems in the Shijing was greatly facilitated by applying another hermeneutical system, one first refined through its use on a different genre of texts.

After the May Fourth movement, the question of how to understand repetitive stanzas resurfaced and was closely intertwined with a larger debate, namely the nature of the Shijing. The two most representative scholars in this debate are Gu Jiegang 顧頡剛 (1893–1980) and Wei Jiangong 魏建功 (1901–80). Gu believed that most repetitive stanzas are fundamentally the same, since only one of them would have been taken from the original folk song and other similar stanzas would have been added later by court musicians trying to stretch the song out.Footnote ⁸⁶ Wei wrote a very influential article, “The most important method of folk song's expression: repetition” (歌謠表現法之最要緊者: 重奏復沓), to refute Gu's argument, in which he takes several fully repetitive poems in the Shijing as examples and states that the rhyming words across the stanzas must exhibit meaningful differences. He further argues that repetitive stanzas are the most significant feature of folk songs, and concludes that poems in the “Guo feng” are essentially folk songs.Footnote ⁸⁷ The central issue in this debate is whether or not the Shijing should be considered a collection of unmediated folk songs,Footnote ⁸⁸ but this question is distinct from whether or not repetitive stanzas have an intrinsic order. Even if there is a progressive sequence across stanzas, this does not necessarily indicate that the Shijing was produced by non-elites, and vice versa. Despite this, the belief that the Shijing is a collection of folk songs has become widely accepted since the 1920s, with Wei's opinion being the more commonly accepted. The theory of orderly progression had been regarded almost as an unalterable truth – at least until the discovery of the Anda manuscript.

In summary, the development of the hermeneutical principle regarding the order of stanzas has been a long process stretching from the end of the Eastern Han to the present day. At the time of Mao and Zheng, this theory was not widely accepted. Kong's annotations applied it to more poems than earlier commentators, probably due to the scarcity of extant Shijing versions and the rise of yishu hermeneutics at the time. This rule really only flourished in the late Ming and early Qing, when scholars like Yao Jiheng came to believe firmly that repetitive stanzas must have had an intrinsic order, similar to paragraphs in an essay. The principle was further confirmed in post-May Fourth scholarship, where the notion of the Shijing as a collection of folk songs helped solidify this belief.

The first half of this long process more or less corresponds with efforts to standardize the order of stanzas in the Shijing. When the order of stanzas was largely settled and there was only one orthodox stanza order in circulation, commentators were more likely to justify that order. Conversely, when a certain order of stanzas is accepted by readers, commentators are less likely to change it. Also, changes in the way poems were composed and appreciated from the pre-Qin period to the Six Dynasties affected the way people understood the order of stanzas in the Shijing. During this time poems came to be expected to embody a coherent and sequential progression of thoughts, which is in accordance with the principle of orderly progression. However, poems in the “Guo feng” had been formed and transmitted in a fundamentally different cultural context, and they should not be evaluated by later poetic norms and rules.

By accepting the fact that many “Guo feng” poems lack a set sequence and refraining from trying to find – or, more precisely, trying to impose – a logical order across stanzas, traditional interpretations of changing characters and entire stanzas can be re-examined. How the order of stanzas is interpreted has never been about just one poem, but is instead part of how interpretation itself is done. Therefore, a reconsideration of this hermeneutical rule offers an opportunity to rethink the entire way in which Shijing interpretation has been done. Here, I briefly discuss two cases where the rule of orderly progression has affected not only the explanation of rhyming words, but also how other parts of the poem and even other poems have been approached.

First, once commentators decided on the sequence and logic that extends across stanzas, they became prone to interpret other phrases in a poem in a way that supports their overall logic. For example, Zheng's annotation of the word zai 載 in the “Sitie” is deeply influenced by his understanding of the timeline of this poem. He regards the last stanza (stanza 3), which talks about the preparation for the hunt that took place in the second stanza, as a flashback. Therefore, his annotation of the last stanza is as follows.

Here, Zheng explains the meaning of the word zai in this stanza as “to start”. Given the context, this appears awkward, and so he provides an additional explanation, specifying that zai refers to the starting of the dogs. It seems that Zheng's only reason for annotating zai as such is to provide support for his understanding of the chronological order of the poem as a whole. If zai means “at the beginning”, then the third stanza must have happened before the previous stanzas, which would validate Zheng's understanding of the third stanza being a prerequisite for the actions that take place in the previous stanzas. On the other hand, Zhu Xi believes that zai means “to carry”, and he interprets the last stanza as being the scene that occurs after the hunt is finished. His interpretation of the last stanza is as follows:

By providing a different exegesis of one single word, Zhu's interpretation, based on a different logic of the way the stanzas are arranged, drastically changes the meaning of the last stanza. Of course, zai can mean either “to start” or “to carry”, and there is nothing incorrect about choosing one explanation over the other. However, it is risky to build one's understanding of the whole stanza or the whole poem on the explanation of one word, as has been done by both Zheng and Zhu. As readers, we should approach such examples with caution, recognizing that even annotations that seem objective might also have been influenced by pre-existing notions.

Second, in some cases, commentators may summarize a pattern derived from one poem and apply it to other poems. For example, in both Mao no. 4 and no. 12, two fully repetitive poems, the rhyming word in the last stanza is cheng zhi 成之 (to accomplish it). Judging by the Mao comment, it seems that this pattern has been extended to some rhyming words in other last repetitive stanzas, which are also interpreted as cheng. The Mao comments on some words in the repetitive stanzas in three poems (Mao no. 2, no. 32, and no. 85) are shown in Table 4.Footnote ⁹¹

Table 4. Mao comments on the repetitive stanzas in Mao no. 2, no. 32, and no. 85

All three comments on the words in bold in stanza two are tenuous at best. The definition of momo and yao as “to accomplish” are not obvious understandings of these words. Although the Mao commentary could have given explanations more closely related to how these words are commonly used, the commentary persistently imposes this meaning on them. The most plausible explanation for this is that the commentator already had a preconceived notion of what should have been expressed in the last repetitive stanzas, likely influenced by the cheng zhi in the last stanzas of Mao no. 4 and no. 12. In other words, the Mao interpretation of these three poems is not solely based on the text of the poems but is also affected by other poems in the Shijing. The Mao comments on different poems within this large corpus are all interrelated, not isolated. This holds true for all other commentators; no commentator annotates just one single poem, and the way in which they approach various poems in the large corpus is systematic.

In conclusion, the belief in there being an intrinsic order across stanzas can be seen in various aspects of the interpretation tradition, from the exegesis of individual words and phrases to the overall comprehension of the nature of the Shijing. It is crucial to scrutinize each annotation and uncover the underlying ideology, rather than accepting it as being self-evident. Reconsidering the theory of orderly progression is a step towards re-evaluating the tradition of Shijing interpretation.