1 From the “Comprehensive summary” 合論 of his Chuci tingzhi 楚辭聽直 (Xuxiu siku quanshu), 1a.

2 The Conflict of Interpretations: Essays in Hermeneutics, ed. Ihde, Don (London: Continuum, 1989), 320Google Scholar.

3 From Xiong's, Yang 揚雄 (53 bc–ad 18) “Rebutting the ‘Li sao’” 反離騷, Han shu 漢書 (Beijing: Zhonghua shuju, 1962)Google Scholar, 87A.3516.

4 Chuci buzhu 楚辭補注, compiled by Xingzu, Hong 洪興祖 (1090–1155), punctuated by Huawen, Bai 白化文 et al. (Beijing: Zhonghua shuju, 1983, 2002), 4.120–62Google Scholar. In addition to the specific sources quoted below, Jiaxin's, Wang 王家歆 Chuci jiuzhang jishi 楚辭九章集釋 (Taipei: Shangwu yinshuguan, 1980, 2007)Google Scholar has been a helpful reference throughout my readings of the “Nine Avowals”. I render zhang 章, which could also simply mean “piece”, as “avowal” to reflect Wang Yi's 王逸 gloss as ming 明, “to declare, to avow”. Implicitly he is also glossing it as a loan for zhang 彰. In general, zhang means “to make clear, to publicize”, as in the Book of Documents: “The superior men will then become prominent” 俊民用章. See Shang shu zhengyi 尚書正義, in Shisanjing zhushu 十三經注疏 (Taipei: Shangwu yinshuguan, 1965)Google Scholar, 12.23a. Cf. also its use in “Si meiren” 思美人, l. 52: “Ah! But in spite of my wretched abode my reputation is illustrious” 羌居蔽而聞章 (Chuci buzhu, 4.149).

5 My rendering of this title is admittedly tendentious, but reflects one of the possible meanings of the title and helps to focus attention on the actual content of the poem. Some alternatives would be “Unspooling my thoughts” or “Unravelling my longings”. Unfortunately there is no single English word that can convey the dual meaning of si as both the neutral contents of the mind and intense longing for another person.

6 Chuci buzhu 4.127 (1/88) (citations of the “Nine Avowals” will be given in this format, including first juan number and page, then in parentheses the poem number within the “Nine Avowals” and line number within that poem).

7 For this definition of strange loop, see Hofstadter, Douglas R., I Am a Strange Loop (New York: Basic Books, 2007), 102Google Scholar. More generally on the significance of the strange loop or “tangled hierarchy” see Hofstadter, , Gödel, Escher, Bach: An Eternal Golden Braid (New York: Basic Books, 1979)Google Scholar.

8 One of Hofstadter's simplest examples comes in his verbal description of the content of M.C. Escher's lithograph “Print Gallery” (Gödel, Escher, Bach, 715): “What we see is a picture gallery where a young man is standing, looking at a picture of a ship in the harbor of a small town, perhaps a Maltese town, to guess from the architecture, with its little turrets, occasional cupolas, and flat stone roofs, upon one of which sits a boy, relaxing in the heat, while two floors below him a woman – perhaps his mother – gazes out of the window from her apartment which sits directly above a picture gallery where a young man is standing, looking up at a picture of a ship in the harbor of a small town, perhaps a Maltese town – What? We are back on the same level as we began, though all logic dictates that we cannot be.”

9 Iser, , “The recursive loop”, in The Range of Interpretation (New York: Columbia University Press, 2000), 99–133Google Scholar.

10 Another example, suggested by a reviewer for this article, might be the formulaic poetry of the Shijing, as studied by Ching-hsien, Wang in The Bell and the Drum: Shih ching as Formulaic Poetry in an Oral Tradition (Berkeley: University of California Press, 1974)Google Scholar. Certainly the Shijing is full of formulaic repetition and circular structures. But it seems to me generally to lack the reflexive elements (Hofstadter's “level-crossing” and “paradoxical”) which I find more evident in certain Chuci poems.

11 Ouyang Wenzhong gong wenji 歐陽文忠公文集 (Sibu congkan), 44.7a/b.

12 Is the recluse fictional, or autobiographical? Here the two amount to the same thing: autobiography is a trope of fiction.

13 Iser, The Range of Interpretation, 128.

14 Shi, Hu, “Du Chuci” 讀楚辭, in Dushu zazhi 1, 1922, 2–3Google Scholar; Hightower, James Robert, “Ch’ü Yüan studies”, in Silver Jubilee Volume of the Zinbun-Kagaku-Kenkyusho (Kyoto: Kyoto University, 1954), 192–223Google Scholar.

15 Chuci buzhu 4.120–21.

16 On “Xi wang ri” 惜往日. See his Chuci duben 楚辭讀本 (1772; rpt. in Chuci wenxian congkan 楚辭文獻叢刊, ed. Linggeng, Huang 黃靈庚 (Beijing: Guojia tushuguan chubanshe, 2014), vol. 48), 302Google Scholar.

17 See his Qu ci jingyi 屈辭精義 (Jiaqing edition; rpt. in Xuxiu siku quanshu, vol. 1302), 4.24a. Tellingly, modern commentator Wang Jiaxin rejects this interpretation in favour of historical reconstruction (Chuci jiuzhang jishi, 302).

18 See Chuci jizhu 楚辭集注 (Shanghai: Shanghai guji chubanshe, 2001)Google Scholar, 4.72; for sceptical modern scholarship, see, e.g., Kanru, Lu 陸侃如 and Yuanjun, Feng 馮沅君, Zhongguo shishi 中國詩史 (1931; rpt. Hong Kong: Guwen shuju, 1961), 1:127Google Scholar; and Wen Yiduo's 聞一多 posthumously published article, “Lun ‘Jiu zhang’” 論九章, Shehuikexue zhanxian 1981.1: 217–23. More recently, however, the trend has been to accept Qu Yuan's authorship of all poems in “Nine Avowals”.

19 See “Soji to Kutsu Gen – hīrō to sakusha to no bunri ni tsuite” 楚辭と屈原––ヒーローと作者との分離について, Nihon Chūgoku gakkai hō 18, 1966, 86–101, and “Soji bungaku ni okeru ‘Shūshi’ no ichi” 楚辭文学における「抽思」の位置, Tōyōgaku 16, 1966, 9–18. The latter article, in particular, first alerted this author to the internal complexity of the “Nine Avowals”.

20 For instance, after a sophisticated discussion of textual linkages in the Chuci, Okamura argues that the “Li sao” and “Lamenting Ying” must have been composed by different authors, simply because of stylistic differences he has observed – as if a single poet could not vary his methods! Thus the most sophisticated critics of the Chuci inevitably come to rest on the most naïve of readings. See “Soji to Kutsu Gen”, 98.

21 Li Rui 李銳 has a valuable discussion of the various potential implications of repeated lines in ancient texts: “‘Chongwen’ fenxifa pingxi” “重文”分析法評析, Qinghua daxue xuebao (Zhexue shehui kexue ban) 23.1, 2008, 127–34.

22 Nagy's, Gregory version of oral-formulaic theory, as laid out in comparative terms in Poetry as Performance: Homer and Beyond (Cambridge: Cambridge University Press, 1996)Google Scholar, is helpful for thinking about how poetry is composed and re-composed within a tradition.

23 As with so many important points in the study of Chinese poetry, this has already been discussed by Qian Zhongshu 錢鍾書 (1910–98): see Guanzhui bian 管錐編 (Beijing: Shenghuo, dushu, xinzhi sanlian shudian, 2001), 940–44Google Scholar.

24 Chuci buzhu 4.121 (1/1–4).

25 Glosses on xi 惜 here vary among “begrudge”, “pity”, and “regret”, a range hard to cover with a single English word. I take the sense to be that the speaker feels conflicted between sympathy for his lord, and frustration at his own situation. For song 誦, see Shijing 191/10: “I, Jiafu, have made this song/To lay bare the king's disorders” (following Legge, James, She King (Hong Kong: London Missionary Society, 1871), 314Google Scholar; and also Guoyu jijie 國語集解 (Beijing: Zhonghua shuju, 2002)Google Scholar, 1.11: “… The bards with blind pupils offered their song …” 矇誦, with song glossed as “words of remonstrance and advice” 箴諫之語. Thus a song is on one hand simply a recitation, like a fu, but it tends to be a politically motivated one.

26 Huang Linggeng 黃靈庚 points out that suo 所 (for which there is a variant fei 非) can be read conditionally. See Chuci zhangju shuzheng 楚辭章句疏證 (Beijing: Zhonghua shuju, 2007)Google Scholar, 5.1266–67.

27 Chuci buzhu 4.134 (3/17–18).

28 Huang Linggeng points out that this line is parallel to line 28 of “Rueful Remonstrance”: “But I have gone astray and cannot find the gate to his favour” 迷不知寵之門. So miao 眇 has a sense close to mi 迷, “confused”. See Chuci zhangju shuzheng, 1408–9.

29 Cheng Yu-yu 鄭毓瑜 has written recently about the way that the poet's sorrow is physicalized in the Chuci. See Yinpi lianlei: wenxue yanjiu de guanjianci 引譬連類：文學研究的關鍵詞 (Taipei: Lianjing, 2012), 88–99Google Scholar.

30 Reconstructions are based on Baxter, William H., A Handbook of Old Chinese Phonology (Berlin: Mouton de Gruyter, 1992)CrossRefGoogle Scholar, as modified in Schuessler, Axel, ABC Etymological Dictionary of Old Chinese (Honolulu: University of Hawai‘i Press, 2007)Google Scholar.

31 David R. Knechtges discusses the use of descriptive binomes in the Han fu, and the problem of translation, in the preface to Wen xuan, or Selections of Refined Literature, Volume Two: Rhapsodies on Sacrifices, Hunting, Travel, Sightseeing, Palaces and Halls, Rivers and Seas (Princeton: Princeton University Press, 1987), 2–13Google Scholar.

32 In this light I have found useful the revisionist study Huaixing, Shen 沈懷興, Lianmianzi lilun wenti yanjiu 連綿字理論問題研究 (Beijing: Shangwu yinshuguan, 2013)Google Scholar. Shen points out difficulties with the dogma of recent Chinese linguistics that lianmianzi are disyllabic simple-morpheme words. In many cases this cannot easily be demonstrated. Shen even criticizes the conventional twentieth-century interpretation of “butterfly” (p. 258). Whichever view one takes on particular words, the point is that these issues cannot be settled by the adoption of an a priori solution at a theoretical level that ignores the variation of individual contexts.

33 “A Note on Ode 220”, i Studia Serica Bernhard Karlgren Dedicata (Copenhagen: Ejnar Munksgaard, 1959), 190–98Google Scholar, rpt. in Selected Works of George A. Kennedy, ed. Li, Tien-yi (New Haven: Far Eastern Publications, 1964), 463–76Google Scholar, at 475.

34 Chuci buzhu 4.159 (9/77–78).

35 Chuci buzhu 4.127 (1/77–80).

36 While Wang Yi glosses yu 紆 as “crooked” (qu 曲), Hong Xingzu more precisely identifies it as “entangled” (ying 縈). See Chuci buzhu 4.127.

37 Chuci buzhu 4.141–2 (5/7–14).

38 Zhu Xi glosses xiao 校 as he 覈 (Chuci jizhu 4.86).

39 A variant omits chu 初. Huang Linggeng argues that ben di 本迪 should be read as bei you 伓（倍）由, “to betray the source” (Chuci zhangju shuzheng, 1497–1500). This is the most parsimonious emendation that makes sense of both text and Wang Yi's commentary. The graph bei 伓 appears frequently in Mawangdui and Guodian manuscripts.

40 Chuci buzhu 4.134 (3/19–24).

41 Niansun, Wang 王念孫 points out that these are both binomes (alliterative and rhyming, respectively) in Dushu zazhi 讀書雜誌, ed. Weijun, Xu 徐煒君 et al. (punctuated ed., Shanghai: Shanghai guji chubanshe, 2015), 2652Google Scholar.

42 Chuci buzhu 4.135; “Lamenting Ying” 3/47–8.

43 Chuci buzhu 4.137 (4/1–4).

44 This line is exactly the same as one quote above from the same poem – Chuci buzhu 4.134 (3/24).

45 Note that Wang Yi implicitly comments on the trope of entanglement by glossing the first line of the poem (which would seem to lack it) as “It means that grief and anger are knotted and gnarled, anxious about trouble and abuse” 哀憤結縎，慮煩冤也. Interestingly, Wang Yi also employs the compound jiegu 結縎 in one of his own poems in the Chuci (see Chuci buzhu 17.317).

46 Chuci buzhu 4.158 (9/57–60).

47 Jie 解 “removed” can also mean “untie, unweave”.

48 Chuci buzhu 4.159 (9/65–8).

49 Wang Yi identifies it as his “minute and subtle thoughts” (xiwei zhi si 細微之思).

50 Chuci buzhu 4.161–2 (9/109–10).

51 This is a point that Okamura emphasizes in arguing that these pieces are all relatively old (“Soji to Kutsu Gen”, 96), but of course an envoi section would be as easy for an imitator to copy as any other feature of the “Li sao”, so the point is not decisive.

52 Okamura, “Soji to Kutsu Gen”, Wen Yiduo, “Lun ‘Jiu zhang’”, and Lu and Feng, Zhongguo shishi, 127. Likewise Hawkes, David writes of it as “clearly inspired by Li sao, from which several of its lines are borrowed”. See Songs of the South: An Ancient Chinese Anthology of Poems by Qu Yuan and Other Poets (London: Penguin, 1985, 2011), 179Google Scholar.

53 Wang Yi's gloss is piao 飄, merely a great gale, but Zhu Xi makes clear that the sense is that of a whirlwind (xuanzhuan zhi feng 旋轉之風). See Chuci jizhu 4.97.

54 The “Li sao” includes these lines: “The whirlwind gathers itself up and then disperses – Leads the clouds and coronas and drives them forth” 飄風屯其相離兮，帥雲霓而來御 (Chuci buzhu 1.29 [l. 203–4]), and Wang Yi glosses piao 飄 as huifeng 回風. Qu Yuan meets a whirlwind while riding through the sky upon a phoenix, but it is not clear from the poem itself whether the storm is a bad omen or perhaps propels him higher upwards. By contrast, “Riding on the whirlwind I will roam faraway” 乘回風而遠遊 in “Seven Remonstrances” (Chuci buzhu 13.249) is clearly positive, and another reference to huifeng in “Lesser Controller of Destinies” 少司命in “Nine Songs” (Chuci buzhu 2.72) probably so.

55 Chuci buzhu 4.155 (9/1–2).

56 Chuci buzhu 4.157 (9/33–6).

57 These images also appear close together in the “Li sao”, at Chuci buzhu 1.28 (l. 197) and 1.29 (l. 205) respectively.

58 Chuci buzhu 4.146 (1–6).

59 Cf. “Unravelled Yearnings”, lines 11–12.

60 Following Jiang Liangfu, I understand line 5 here parallel to “Li sao”, line 41, with jianjian 蹇蹇 a loan for jianjian 謇謇.

61 Chuci buzhu 4.124 (1/37–40).

62 Following Zhu Xi, I amend 中情 to 善惡 to fit the rhyme and based on the parallel line in “Li sao”, Chuci buzhu 1.36.

63 Chuci buzhu 4.152 (7/63–6).

64 See Hollander, John, Melodious Guile: Fictive Pattern in Poetic Language (New Haven: Yale University Press, 1988), ixGoogle Scholar.

65 Qu Fu, Chuci xinzhu 楚辭新注 (Guanzhong congshu), “Fan li” 凡例, 2a.

66 This point has not always been appreciated by students of oral-formulaic theory, though in Chinese studies it has been stated clearly by C.H. Wang, e.g., “The protean character of the basic meaning of a Shih Ching word also illuminates the individual compositional art in formulaic vogue when the tradition was ripe”. In other words, the technical sense of “formulaic” is entirely different from the pejorative sense of the word in colloquial English. See The Bell and the Drum: Shih Ching as Formulaic Poetry in an Oral Tradition (Berkeley: Univ. of California Press, 1974), 96Google Scholar.

67 Hofstadter, Gödel, Escher, Bach, 10. The essential insight derives from Kurt Gödel's (1906–78) 1931 paper “On formally undecidable propositions of Principia Mathematica and related systems”. Gödel's proof relies on a remarkable device, showing that mathematical propositions can be reconceptualized through a self-referential enumeration, then using this property to identify certain undecidable ones.

68 Tseng, “An allegory on allegory: reading ‘Ju song’ as Qu Yuan’s Ars Poetica”, Dong Hwa Journal of Humanistic Studies 1, 1999, 98.

69 Tseng, “An allegory on allegory”, 98.

70 E.g. Okamura Shigeru 剛村繁 has argued that “Chou si” 抽思 is a composite of two pieces in “Soji bungaku ni okeru ‘Shūshi’ no ichi” 楚辭文学における「抽思」の位置, Tōyōgaku 16, 1966, 9–18. Okamura's exquisitely precise analysis was a direct inspiration to my arguments here, even though my conclusions differ radically from Okamura's.

71 See “Qu Yuan yanjiu” 屈原研究, originally published in 1942, rpt. in Guo Moruo quanji: lishi bian 郭沫若全集：歷史編 (Beijing: Renmin chubanshe, 1982), 4: 41Google Scholar.

72 As Kaicheng, Jin 金開誠 et al. argue in Qu Yuan ji jiaozhu 屈原集校注 (Beijing: Zhonghua shuju, 1996), 521Google Scholar, citing Yuo Guoen 遊國恩.

73 See Fang yan jiaojian 方言校箋 (Beijing: Zhonghua shuju, 1993)Google Scholar, 13.8a.

74 Dong rong 動容 is a rhyming binome that describes how the autumn wind lays waste to green leaves and other growing things. The second character is also written rong 溶 or 搈. See Bingzheng's, Tang 湯炳正 illuminating discussion in Chuci leigao 楚辭類稿 (Chengdu: Ba Shu shushe, 1988), 332Google Scholar. The sense of the line as a whole is similar to that of the famous opening of the “Nine Suasions”.

75 The unusual sense of yao 搖 as “rapidly” here is attested in Fang yan 方言, as shown by Wang Niansun in Dushu zazhi, 2653. A variant has 遙赴 for 搖起.

76 I take this line to be a concise summation of Chinese political theory. Understanding the troubles of the people or state as a whole must proceed in tandem with self-cultivation and the attainment of individual self-control on the part of the ruler. Peace and contentment in the society at large can be attained only when the rulers are themselves at peace, self-composed, having achieved harmony between their own desires and the world around them.

77 Here I decline to adopt Hong Xingzu's variant text 何獨樂斯之謇謇兮，願蓀美之可光. Though the variant improves the rhyme, this is not dispositive since it is common in early poetry for nasal endings to cross-rhyme. For more on Lord Iris see the discussion below.

78 Wang Yi identifies these as the Three Kings and Five Hegemons.

79 Peng Xian is identified by Wang Yi as a Shang courtier from antiquity, but in fact both Peng and Xian occur frequently in mythological terms and as the names of shamans. Peng is also mentioned in the Analects and other texts as a model of longevity. Modern scholars have distinguished them as two separate sage-heroes of antiquity, as in Liangfu, Jiang 姜亮夫, Qu Yuan fu jiaozhu 重定屈原賦校注, rev. ed. (Tianjin: Tianjin guji chubanshe, 1987), 29–30Google Scholar.

80 Identical with line 17.

81 Congrong 從容 most commonly means “relaxed and leisurely”, but also has a second meaning of “behaviour, appearance, deportment”. See Niansun, Wang, Guangya shuzheng 廣雅疏證 (Nanjing: Fenghuang chubanshe, 2000)Google Scholar, 6A.34b–35a.

82 The location of Beigu has generally been treated as an unsolvable mystery, but Jao Tsung-i 饒宗頤 suggested that it refers to Pugu 蒲姑, a.k.a. Bogu 薄姑 (located in modern Boxing 博興 county, Shandong), in the state of Qi. This is very plausible, particularly as Qu Yuan is said to have travelled on embassies to Qi, and the phonology is passable – Beigu is OC *bâkkâ, Bogu is OC *pəkkâ. See Tsung-i, Jao, Chuci dili kao 楚辭地理考, rpt. in Rao Zongyi ershi shiji xueshu wenji 饒宗頤二十世紀學術文集 (Taipei: Xin wenfeng chuban, 2003), 16: 103–5Google Scholar.

83 Chuci buzhu, 4.141.

84 Okamura speculates that it represents a divinity to whom a shaman could offer complaint, in distinction to the gods with whom one consorts in the “Nine Songs”. See “Soji bungaku ni okeru ‘Shūshi’ no ichi”, 13.

85 Jao Tsung-i argues that the bird represents not Qu Yuan but King Huai (Chuci dili kao, 106–7). This seems possible, but only because the symbolic language of the Chuci is inherently ambiguous, with both courtier and sovereign being represented as a beautiful maiden. Given the ambiguity it is simpler to understand it as referring again to Qu Yuan, the main speaker throughout.

86 See Linggeng, Huang 黃靈庚, Chuci yu jianbo wenxian 楚辭與簡帛文獻 (Beijing: Renmin wenxue, 2011), 85–6Google Scholar.

87 Wenlan, Fan 范文瀾, Wenxin diaolong zhu 文心雕龍注 (Taipei: Xuehai chubanshe, 1991)Google Scholar, 10.176. Hoshikawa Kiyotaka 星川淸孝 argued along these lines that the ci in Chuci refers not so much to the language of Chu, or to the literary form as in the compound cifu, but rather to the genre of the prayer or invocation (zhuci). See Hoshikawa, , Soji no kenkyū 楚辭の研究 (Tokyo: Yōtokusha, 1961), 24–35Google Scholar.

88 Jui-ching, Lu, Fengjian shuqing yu shenhua yishi: Chuci wenxin lun 諷諫抒情與神話儀式：楚辭文心論 (Taipei: Liren shuju, 2002), 3–62Google Scholar.

89 Chuci buzhu 4.161 (9/107–8). In fact this piece and “Regretting Past Days” were already singled out for scepticism in the Song dynasty. Wei Liaoweng 魏了翁 (1178–1237) questioned whether Qu Yuan could have praised Wu Zixu 伍子胥, as both these poems seems to do. See Jingwai zachao 經外雜鈔 (Siku quanshu), 2.16b–17a. In fact, though, it turns out that “Crossing the River” also seems to praise Wu Zixu. For an authoritative discussion concluding that this issue is orthogonal to the question of authorship, see Zhi, Li 力之, “Wang Yi shi ‘She Jiang’ de ‘Wuzi’ wei ‘Wuzi Xu’ wu wu bian” 王逸釋 〈涉江〉 的「伍子」為「伍子胥」無誤辨, in Chuci yu zhonggu wenxian kaoshuo 楚辭與中古文獻考說 (Chengdu: Ba Shu shushe, 2005), 185–96Google Scholar.

90 Ren shi 任石 should be understood as a variant of huai shi 懷石, as in Sima Qian's description of Qu Yuan's suicide (Shiji 84.2490). Cf. Guo Pu 郭璞 (276–324), “Rhapsody on the Yangtze River” 江賦: “I grieve for Lingjun who shouldered a stone [to drown himself]” 悲靈君之任石 (Wen xuan 12.572).