Introduction

As a mode of performance, drama has a history of at least two millennia in China, but it first emerges as a literary art in the second part of the thirteenth century with the rise of miscellaneous comedy (zaju 雜劇) or northern drama (beiqu 北曲) as a new genre.¹ This was a form of musical drama that required one actor, playing the lead male or female character, to sing four suites (one per act) of eight to twenty songs. In some cases the metrical complexities of these songs might have been beyond the power of improvisation of the average actor or actress and required the intervention of an author who conceived of the play as a whole and drafted the text of the songs in the context of each linked scene. While thirty early zaju have been preserved in editions from the fourteenth century, the large majority of the plays from this period has been preserved in manuscript or printed editions from two or more centuries later. These later manuscripts derived first of all from Ming-dynasty (1368–1644) court agencies in charge of dramatic performances and often have been heavily edited to make them suitable for performance in front of the emperor. With a few notable exceptions, later printed versions were based primarily on these palace manuscripts but were once again edited in order to make them suitable reading matter for highly educated literati. The most popular of these printed anthologies was A Selection of Yuan Plays (Yuanqu xuan 元曲選), published in 1615–1616, which provided beautifully illustrated and nicely printed versions of one hundred zaju; in his two prefaces to the anthology the editor, Zang Maoxun 臧懋循 (d. 1621), prided himself on the frequent and extensive changes he had introduced in each of these plays. Zang’s anthology represents a culmination of the printing of zaju anthologies, a canonization of format that provided full stage directions and dialogues for all roles. Educated in drama by reading this anthology, many modern scholars have failed to comprehend that the shape of the earlier fourteenth-century editions was determined by their use as role texts² and therefore do not demonstrate the same fully articulated format as later editions. As a result these scholars describe the fourteenth-century editions as incomplete or defective, and the conclusions they reach in their research on “Yuan drama”—based on what are clearly Ming texts—are often anachronistic. It should be obvious that if we want to study the earliest phase of Chinese dramatic literature and its subsequent development, we must begin with the earliest preserved texts. If we indeed base our research on these fourteenth-century editions, we encounter a world of raw power and vitality, high passion and bloody revenge. If we then compare the plays in these earliest editions with the manuscript and printed versions of the late Ming, we find that these elements have been considerably toned down in these later editions if not covered completely by a thick veneer of conventional Confucianism.³

In this volume we provide translations of seven plays in their fourteenth-century editions. In four of the seven cases we accompany that translation with a rendition of the same play in either a Ming palace manuscript or a late Ming printed edition. While in some cases the differences between versions are seemingly minor, in others they are extensive; but in each of these four cases these differences are substantial enough to speak of the later editions as completely new plays. Our first criterion in selection of these seven Yuan editions has been their intrinsic quality. We have also taken into account to what extent they would make good reading despite their nature as a role text (some texts, not translated here, make for very difficult reading because we do not understand the story they dramatize very well). We have made every effort to provide a representative selection that does justice to the thematic variety within the genre but have not included translations of Yuan editions that we have published earlier.⁴ The seven Yuan dramas and the four Ming rewrites we include are meant to provide a comparative body of texts through which students and scholars of drama can trace the changes in plot, characterization, style, and meaning that a single story can go through in a process of change and adaptation over a period of three hundred years. As we have mentioned elsewhere,⁵ this is a necessary step in understanding the social and cultural world from which each stratum of these texts derives and to whom they were addressed.

The seven plays in fourteenth-century editions and the later Ming rewrites we have selected are as follows:

1.  Yuan edition: The Orphan of Zhao 趙氏孤兒
Ming edition: The Orphan of Zhao Greatly Wreaks Revenge 趙氏孤兒大報仇 (A Selection of Yuan Plays version)

2.  Huo Guang Remonstrates as a Ghost 霍光鬼諫

3.  Yuan edition: Xue Rengui Returns Home Clad in Brocade 薛仁貴衣錦還鄉
Ming edition: Xue Rengui Returns Home in Glory 薛仁貴榮歸故里 (A Selection of Yuan Plays version)

4.  Chen Jiqing Is Enlightened to the Way on a Bamboo-Leaf Boat 陳季卿悟道竹葉舟
Ming edition: Chen Jiqing Mistakenly Boards a Bamboo-Leaf Boat 陳季卿誤竹葉舟 (A Selection of Yuan Plays version)

5.  Tippler Zhao Yuan Encounters the Prior Emperor 好酒趙元遇上皇
Ming edition: Tippler Zhao Yuan Encounters the Prior Emperor (Yu Xiaogu manuscript)

6.  The Affair of the Eastern Window Exposed 東窗事犯

7.  Little Butcher Zhang Immolates the Child to Save the Mother 焚兒救母

The seven texts, which we here designate as the “Yuan printings” or “Yuan texts,” are all four-act music dramas in which each act has as its core a suite of arias (taoshu 套數) sung to one of nine musical modes (gongdiao 宮調) and utilizing a single rhyme throughout one act. Within a single play, each of the four suites utilizes a separate mode, a more or less standard arrangement of arias for that mode, and all songs are assigned to a single performer, either a male or female lead. The majority of these Yuan printings provide only the texts of the arias and minimal stage directions (guanmu 關目) that may include some summary dialogue for the male or female lead; stage directions for other actors in the play—if given at all—tend to be limited to prompts to tell the male or female lead when to act, speak, or sing. Only plays printed from the early fifteenth century onward include fully written-out stage directions, rhymed poems, and dialogue (binbai 賓白) for every actor.

THE THIRTY MISCELLANEOUS COMEDIES PRINTED IN THE YUAN AND THEIR TRANSMISSION

Zaju, performed by a small troupe of actors, is part of a large body of oral performance genres that gained popularity in the Song and Yuan eras. Several of these narrative forms, including the performance ballads in all keys and modes (zhugongdiao 諸宮調), were a form of prosimetric literature (shuochang wenxue 說唱文學 or jiangchang wenxue 講唱文學) that used almost the same song forms as later drama (the qu 曲) and were sung by a single narrator.⁶ Usually only one actor or actress sings onstage in zaju, which may be a borrowing from this type of oral narrative. Twenty-seven of the Yuan printings are clearly recognizable from their stage directions as role scripts for the male or female lead of a performing troupe,⁷ while three editions provide us with only the texts of the songs.

The woodblocks from which these plays were printed appear to have been produced during the middle to late fourteenth century (that is, from the second part of the Yuan dynasty to the first decades of the Ming). We have no way of identifying the printing date of the editions, nor do we have any supporting historical evidence of the existence of such scripts before this time. A few of the thirty texts appear to have been printed in Dadu 大都 (modern Beijing),⁸ the capital of the Yuan, but most were produced in Hangzhou, the Yuan administrative center for the conquered former territory of the Southern Song dynasty. While some texts can be grouped together because of shared features of layout, each of the thirty plays appears to have been printed individually.

These thirty independent editions of Yuan northern drama are now known collectively as Thirty Miscellaneous Comedies Printed in the Yuan (Yuankan zaju sanshi zhong 元刊雜劇三十種), a name given them by the pioneering modern drama scholar Wang Guowei 王國維(1877–1927) in his preface to the 1924 Shanghai reproduction of a recutting and reprinting of the plays by Kyoto University in 1914. Before this, they were known as New and Old Miscellaneous Comedies Printed in the Yuan (Yuankan gujin zaju 元刊古今雜劇), a name bestowed on them by the renowned book collector Huang Pilie 黃丕烈 (1763–1825), who at one time owned these texts. In a postscript he wrote to the collection he remarked that he purchased these texts, along with many other rare editions, from the Dipping In for a First Taste Studio (Shiyintang 試飲堂) of Gu Tingyu 顧聽玉 (ca. 1800) located in Suzhou.⁹ Huang acquired the plays in two different sets. The first was a group of twenty-five plays, for which he compiled the following list of contents, separating the publication information from the title of the play:

Fascicle 1: Five Plays (五種共一冊)

Newly Printed in Old Hangzhou, a Full Text with Plot Prompts 古杭新刊關目的本
Li Taibai Is Banished to Yelang 李太白貶夜郎

Newly Printed with Plot Prompts 新刊關目
Yan Ziling Drops His Hook at Seven-Mile Rapids 嚴子陵垂釣七里灘

Newly Compiled in Dadu 大都新編
King Zhao of Chu: The Most Distant Relative Goes Overboard 楚昭王疏者下船

Newly Printed in Old Hangzhou, a Full Text 古杭新刊的本
Yuchi Gong Wrests the Spear Away Three Times 尉遲恭三奪槊

Newly Printed in Old Hangzhou, a Full Text with Plot Prompts 古杭新刊的本關目
In Wind and Moon in the Courtyard of Purple Clouds風月紫雲亭

Fascicle 2: Six Plays (六種共一冊)

Newly Compiled in Dadu 大都新編
In a Dream Guan and Zhang, a Pair, Rush to Western Shu關張雙赴西蜀夢

Newly Printed with Plot Prompts 新刊關目
A Clever Wench Sports in the Wind and Moonlight 詐妮子調風月

Newly Printed, a Full Text 新刊的本
On Taihua Mountain Chen Tuan Rests with Easy Mind 太華山陳摶高臥

○○○○
The Orphan of Zhao 趙氏孤兒

Newly Printed, a Full Text [Plot Prompts Complete] 新刊的本[關目全]
Xue Rengui Returns Home Clad in Brocade薛仁貴衣錦還鄉

Newly Printed in Old Hangzhou with Plot Prompts 古杭新刊關目
Aiding King Cheng the Duke of Zhou Acts as Regent 輔成王周公攝政

Fascicle 3: Three Plays (三種共一冊)

Newly Printed in Dadu with Plot Prompts, a Complete Edition 大都新栞關目的本
The Affair of the Eastern Window Exposed 東窗事犯

Newly Printed in Old Hangzhou, a Full Text 古杭新刊的本
Huo Guang Remonstrates as a Ghost 霍光鬼諫

Newly Printed with Plot Prompts 新刊關目
Zhang Ding Cleverly Investigates the Moheluo Doll 張鼎智勘磨合羅

Fascicle 4: Four Plays (四種共一冊)

Newly Printed in Old Hangzhou with Plot Prompts, a Full Text 古杭新刊關目的本
Great King Guan and the Single Sword Meeting 關大王單刀會

Newly Compiled with Plot Prompts 新編關目
Duke Wen of Jin Immolates Jie Zi Tui 晉文公火燒介子推

Newly Printed with Plot Prompts 新刊關目
Ma Danyang Thrice Converts Crazy Ren 馬丹陽三度任風子

Newly Printed with Plot Prompts 新刊關目
A Beauty Pining in Her Boudoir: Praying to the Moon 閨怨佳人拜月亭

Fascicle 5: Four Plays (四種共一冊)

Newly Printed with Plot Prompts Complete 新刊關目全
Xiao He Pursues Han Xin 蕭何追韓信

Newly Compiled in Dadu with Plot Prompts 大都新編關目
A Noble’s Grandson: Story of the Undershirt 公孫汗衫記

Newly Printed with Plot Prompts 新刊關目
A Slave to Money Buys a Creditor as His Enemy 看錢奴買冤家債主

Newly Printed with Plot Prompts 新刊關目
The Exalted Emperor of Han Washes His Feet to Anger Ying Bu 漢高皇濯足氣英布

Fascicle 6: Three Plays (三種共一冊)

Newly Printed with Plot Prompts 新刊關目
Tippler Zhao Yuan Encounters the Prior Emperor好酒趙元遇上皇

Newly Printed with Plot Prompts 新刊關目
Chen Jiqing Is Enlightened to the Way on a Bamboo-Leaf Boat陳季卿悟道竹葉舟

Newly Printed with Plot Prompts 新刊關目
Zhuge Liang Burns the Stores at Bowang 諸葛亮博望燒屯

The above twenty-five specimens were originally bound in three fascicles; I rebound them in six fascicles.¹⁰

FIGURE 1   Pages 1b–2a of Huang Pilie’s list of twenty-five Yuankan editions

Huang Pilie acquired a second batch of five plays, which he lists in a different format. For the first twenty-five, he separated the title of the play from the advertising blurb, replicated in the translation above by the indented lines (see fig. 1). The second group of five is arranged differently. The list is broken into two sections: the first four titles are listed under the enigmatic rubric “Tenth Row of Yuan Print Editions” (Yuankan Shihang 元刊十行), and the fifth title is listed under the heading “Sixteenth Row of Yuan Print Editions” (Yuankan Shiliu Hang 元刊十六行). These may indicate the place the texts held within his total collection of all Yuan printed editions, of which he had many. Furthermore, instead of separating the advertising blurb from the main title (see fig. 2), he wrote them all in the same line:

Tenth Row of Yuan Print Editions

Fascicle 1: Two Plays (二種一冊)

Newly Printed: Friends in Life and Death: Fan, Zhang, Chicken, and Millet 新刊死生交范張雞黍

Newly Compiled: Clerk Yue Avails Himself of Iron Crutch Li to Return from the Dead 新編岳孔目借鐵拐李還魂

Fascicle 2: Three Plays (三種共一冊)

Newly Compiled, a Complete Edition with Plot Prompts: Zhang Qian Kills the Wife Instead 新編足本關目張千替殺妻

Newly Printed in Old Hangzhou: Little Butcher Zhang Immolates His Child to Save His Mother 古杭新刊小張屠焚兒救母

Newly Printed True Edition: Dispersing Family Wealth Heaven Grants a Son Born in Old Age 新刊的本散家財天賜老生兒

The above were originally bound in one fascicle; I rebound them in two fascicles.¹¹

We have no way of determining how these two lists were constituted. It is clear that Huang acquired these five plays at the same time, bound in two fascicles.¹² It is unclear whether he then rebound them in the same order as in their original binding or if he changed that order. They might have been rebound in thinner volumes for easier reading. Or, since the texts were stored in wooden boxes to be kept on shelves, they may have been rebound to make storage easier. One thing seems clear: the order of the plays is completely random.

FIGURE 2   Huang Pilie’s list of five Yuankan editions

While this is the earliest historical reference to the texts we now possess, we have information on their earlier existence based on their use in the eighteenth century to collate later Ming-dynasty print and manuscript editions. Since Iwaki Hideo published his article on the textual history of the Yuankan zaju sanshi zhong in 1961, it has been accepted that these thirty plays were once in the library of Li Kaixian 李開先 (1502–1568), a mid-Ming bibliophile, dramatist, poet, and critic.¹³ Iwaki reached his conclusion on the basis of the notes of He Huang 何煌 (ca. 1680–1740), who had collated several Yuan texts against later zaju editions, both printed and manuscript. In the four instances in which He Huang used a Yuan edition (not necessarily the ones we have today), he has left the following colophons:

1.  The Single Sword Meeting [collation of a Yuan print edition and an early Ming manuscript]:¹⁴ Newly Printed in Old Hangzhou, a Complete Edition: The Great King Guan and the Single Sword Meeting. “Used a Yuan print edition for collation on the fourteenth day of the eighth month, the yisi year of the Yongzheng reign [September 20, 1725]” 雍正乙巳八月十日用元刻本校.¹⁵

2.  Zhang Ding Cleverly Investigates the Moheluo Doll [Gu mingjia ed.]. “I have finished collating a Yuan print edition formerly held by Li Zhonglu [Kaixian]. I have abandoned Qingchang’s [that is, Zhao Qimei] collation because it is wrong. Zhongzi [that is, He Huang]. The twenty-first day of the eighth month, the yisi year of the Yongzheng reign [September 27, 1725]” 用李中麓所藏元槧本校訖了。清常一校為枉廢也。仲子。雍正乙巳八月二十一日.¹⁶

3.  A Slave to Money Buys a Creditor as His Enemy 看錢奴賣冤家債主 [Xijizi ed.]. “Used a Yuan print edition to collate and correct the text by lamplight on the twenty-sixth day of the eighth month, the yisi year of the Yongzheng reign [October 2, 1725], Zhongzi” 雍正乙巳八月廿六日燈下用元刻校勘 仲子.¹⁷

4.  Friends in Life and Death: Fan, Zhang, Chicken, and Millet 死生交范張鷄黍. “One day after the seventh night of the seventh month of the yiyou year of the Yongzheng reign [August 2, 1729], while collating [this text] to a Yuan print edition, twelve songs were missing [in the Yuan text]. This allowed the [missing songs] to be restored. This might hit the mark” 雍正己酉秋七夕後一日，元槧本校，中缺十二調，容補錄。耐中.¹⁸

Three of these texts were collated within a two-week period; the fourth, Friends in Life and Death, was done nearly four years later. This fourth play was also part of the second batch of five plays noted in Huang Pilie’s catalogue list, and this would seem to indicate that He Huang acquired these two batches of texts at different times and perhaps from different sources.¹⁹ Now, while these four cases show a near concordance between the collated text and the Yuan printing that remains extant, only He Huang’s collations in the Moheluo clearly identify the source of the Yuan printing as being from Li Kaixian’s library. And while the collations in the Moheluo are nearly the same as those preserved in the thirty Yuan editions, there are enough minor discrepancies between the collation interpolations and the Yuan text of the Moheluo to keep us from claiming that the Yuankan text we have today and the one held by Li Kaixian and used by He Huang are the same. As Zhen Weini remarks,

Although there are not many discrepancies, they do in fact exist. Therefore we cannot rely on this text [of the Moheluo] to absolutely affirm that the extant Yuan printings were those collected by Li Kaixian. But, from the fact that the text He Huang used and the extant Yuan printings are very similar, we can at least acknowledge that the extant Yuan printings and those that came from Li Kaixian’s library stem from the same source and that they represent the original features of scripts that were created in the Yuan.

Because the other three texts have no clear evidence from their contents that they were in Li Kaixian’s library, and although what is copied in He Huang’s collation is highly similar to the extant Yuan texts, we can still only confirm that the Yuan texts used by He Huang for collation are preserved in the current Yuankan zaju sanshi zhong, but we cannot use them as evidence to prove any relationship between them and those held by Li Kaixian.²⁰

THE MULTIPLICITY OF YUAN TEXTS

In another case, He Huang has used a manuscript edition held by Li Kaixian to collate the play Wang Can Ascends the Tower (Wang Can deng lou 王粲登樓) (Gu mingjia ed.). In his colophon, he remarks,

On the eighteenth day of the eighth month of the yisi year, the third of the Yongzheng reign [September 24, 1725], I used a manuscript copy of Li Kaixian’s to collate and to correctly rewrite several hundred characters. This [Gu mingjia ed.] is missing twenty-two arias that have been dropped, and there are two arias switched in sequence, and I have corrected this and supplied the missing arias from a manuscript edition. The manuscript edition was missing the complete dialogues. The mistakes and ill-informed nature of the dialogues were unbearable, many times more so than the arias, but there was no source to correct it. I can only hope that there will be a curious and profound scholar who will be able to add the dialogues according to the stage directions. And even more that there is a truly knowledgeable and good person whose power is enough to have a famous actor perform this [that is, my recension]—that would be a wonderful thing. I have written this in anticipation of such a thing. Recorded by He Zhongzi of Xiaoshan. 雍正三年乙巳八月十八日，用李中麓抄本校，改正數百字。此又脫曲廿二，倒曲二，悉據抄本改正補錄。抄本不具全白。白之繆陋不堪，更倍於曲，無從勘正。冀世有好事通人，為之依科添白。更有真知真好之客，力足致名優演唱之，亦一快事。書以俟之。小山何仲子記.²¹

In the fourth act of the Gu mingjia edition, which is obviously truncated, He Huang has inserted eight songs from the manuscript edition, including five after the song Shuixianzi. In a short collation note, he also remarks, “Another edition has these five songs after the song Tianshui ling: Dianqian huan, Qiao pai’er, Gua yugou, Gu meijiu, and Taiping ling.”²² Thus, we know that He had at his disposal two different Yuan editions, one of which is noted as a “manuscript,” each with a different sequence of songs in the fourth act:

Gu mingjia edition	Li Kaixian manuscript	So-called “Another edition”
【雙調 新水令】	[Same]	[Unknown]
[Same]	【駐馬聽】	[Unknown]
[Same]	【雁兒落】	[Unknown]
[Same]	【得勝令】	[Unknown]
[Same]	【甜水令】	[Unknown]
[Same]	【折桂令】	[Unknown]
【喬牌兒】	【喬牌兒】	[Unknown]
【水仙子】	【水仙子】	[Unknown]
【雁兒落】	【川撥棹】	【殿前歡】
【得勝令】	【七弟兄】	【喬牌兒】
【梅花酒】	【掛玉鈎】
【收江南】	【沽美酒】
【鴛鴦煞】	【太平令】

This is an important point, because it demonstrates that there were multiple editions of scripts of the same drama printed or circulating as manuscripts during the Yuan. We are fortunate that we can compare the changes He Huang made to the aria Yingxian ke (迎仙客) in the third act of Wang Can Ascends the Tower with a citation of the same song in the music formulary and rhyme book known as The Rhymes of the Central Plain (Zhongyuan yinyun 中原音韻), written by Zhou Deqing 周德清 (1277–1365) in 1324.²³ In The Rhymes of the Central Plain we find the aria cited as

雕簷紅日低,	Carved eaves, red sun lowering,
畫棟彩雲飛;	Painted rafters, colored clouds flying;
十二玉欄天外依。	Twelve jade balustrades recline beyond heaven.
望中原,	Gaze at the Central Plain,
思故國。	Think on my old state,
感慨傷悲,	Painful recollections, broken by sorrow—
一片鄉心碎。	A strip of homebound heart sunders.24

In Zhou’s work this is presented as a song stripped of its extrametrical syllables (“padding words,” chenzi 襯字) and titled simply “Ascending the Tower” (Deng lou 登樓). Whether this was intentionally done so that it could be read as an independent poem or whether “Ascending the Tower” was supposed to bring to mind the drama Wang Can Ascends the Tower is unclear. Later, when He Huang used a Yuan edition for his collation, he left the so-called padding words intact, which suggests that the text He Huang used was a script:

調簷外紅日低,	Beyond Carved eaves, red sun lowering,
畫棟畔彩雲飛;	Beside Painted rafters, colored clouds flying;
十二欄杆在天外依。	Twelve balustrades recline there beyond heaven.
我這裡望中原,	Here I Gaze at the Central Plain,
思故里。	Think on my old state,
不由我感歎傷悲,	Unconsciously I Painfully sigh, broken by sorrow—
越鬧得一片鄉心碎。	Vexed even more until A strip of homebound heart sunders.25

In these two renditions of the song, both presumably dating from the mid to late Yuan period, we also see two small lexical changes (underlined above): 十二玉欄 → 十二欄杆 and 感慨 → 感歎. These are small changes, but particularly in the second instance it changes the mood from a single subject who is silently struggling with painful recollections to a visible act of communicating those strong feelings externally to a second actor. In the drama of course, this is necessary because Wang Can is talking with Xu Da (see fig. 3). In The Rhymes of the Central Plain, the first reading is preferable because the poem is contextualized within the cliché scene of a lone subject climbing the tower to meditate on the view. These two readings point to the possibility of songs circulating in two different ways: both as independent lyrics and as dramatic verse.

There are other cases in which it is Zhou Deqing himself who points out variants to dramatic songs that were circulating at the same time when he wrote his book in 1324. For instance, he transcribes a song from the first act of Ma Zhiyuan’s (?1254–?1320) Lü Dongbin Thrice Drunk at Yueyang Tower (Lü Dongbin sanzui Yueyang lou 呂洞賓三醉岳陽樓) as

據胡床,	I sit upon this folding chair,
對瀟湘。	And face the Xiao and the Xiang.
黃鶴送酒仙人唱,	A yellow crane urges on wine, a transcendent sings,
主人無量醉何妨？	The host has no limits, what problem is there to getting drunk?
若捲簾邀明月,	If you furl the curtain and welcome the bright moon,
勝開宴出紅妝。	It would be so much better than setting out a feast and sending out “those rouged and made up.”26
但得一樽留墨客,	Just get one cup to detain the “ink guest”27
是兩處夢黃粱。	And in two places “dream of yellow millet.”28

FIGURE 3   Original page, Wang Can Ascends the Tower, Guming jia ed., with He Huang’s emendations

Zhou then goes on to remark on variant recensions of the play that were circulating at that time:

This is a lyric from the opening act of Yueyang lou, and the exquisiteness of it lies in the seven characters “A yellow crane urges on wine, a transcendent sings.” Superlative language! Moreover, “wine” is in the rising tone in order to change its pronunciation, and the crux of the line hinges on this. There are those who do not understand the meaning of the words, and take “send” to mean “physically bring over.” They say, “How can a yellow crane bring wine?” And they change the [two words “sending wine”] to “dance as a pair.” Now they really do not know the affair of the yellow crane: an immortal used a pomegranate shell as a place to draw a crane, in order to repay the wine tender. When the guest drinks and clasps it in his hands, then the crane seems to dance in order to press the drinker to drink the wine. There were never two cranes, so how could a pair dance? Moreover, it loses the whole point of drinking wine. “To send” means something like “the beauties of Wu press one to drink wine.” How hard indeed, to cure the vulgar scholar. 此是《岳陽樓》頭摺中詞也。妙在七字『黃鶴送酒仙人唱』。俊語也。況『酒』字上聲以轉其韻,務頭在其上。有不識文義,以送為齎送之義,言『黃鶴豈能送酒乎』？改為『對舞』,殊不知黃鶴事,仙人用榴皮畫鶴一隻,以報酒家,客飲,撫掌則所畫黃鶴舞以送酒。初無雙鶴,豈能對舞?且失飲酒之意。送者,如吳姬壓酒之謂.²⁹

“Immortal cranes dance as a pair” is indeed the change that appears in the two late Ming versions of the play:

我這裡據胡床,	Here I sit on this folding chair
對瀟湘。	Facing the Xiao and the Xiang.
有黃鶴對舞仙童唱,	There are a pair of Yellow Cranes dancing, an immortal lad singing,
主人家寬洪海量？	The host has a capacity as full as the sea!
直吃得捲簾邀皓月,	Just drink until you furl the curtain to welcome the hoary moon
再誰想開宴出紅妝。	No one will think again of setting out a feast and sending out “those rouged and made up.”
但得一樽留墨客,	Just get one cup to detain the “ink guest”
我困了也,	I am sleepy.
我可是兩處夢黃粱。	I truly am dreaming of yellow millet in two places.30

We are fortunate to have Zhou Deqing’s note. Were we to possess the Yuan performance text that has the phrase “A yellow crane urges on wine, a transcendent sings” and compare it with the Ming editions, the natural assumption would be that later editors had made these changes. This is the common tendency in modern studies, which often use the collation of Yuan and Ming texts as a way to find a direct chronological link between a presumed antecedent and its presumably evolved recension. That such an assumption takes place, and for it to hold such a powerful position in the study of drama, seems related to the belief that one author produced each individual play and that all other variants are introduced by others, particularly by actors or editors.

PLAYWRIGHTS, ACTORS, MANUSCRIPT PLAYS, AND PRINTED SCRIPTS

Such modern editorial policy posits a single authorial voice. One of the stronger and more enduring theories of literature in China is that writing is both an expression of a refined ethical interiority—as expressed by the commonly cited phrase “poetry speaks of heart’s direction” (shi yan zhi 詩言志)—and as such a direct reflection of the personality of the writer, as in the equally cliché phrase “see the writing and know the person” (jian qi yan zhi qi ren 見其言知其人). But these are elite values, and when they appear in traditional and modern studies on theater or drama, they are clearly a result of a retrospectively applied concept that derives from the literary and cultural training of the writer.

In point of fact, none of the early Yuan editions carry any signature of authorship. Attributions of these plays to writers were made later on the basis of the first bibliography of drama, The Register of Ghosts (Lugui bu 錄鬼簿), a bio-bibliographical study of early and middle Yuan qu writers written by Zhong Sicheng 鐘嗣成 (ca. 1270–1360) that lists titles of dramas under playwrights’ names. On the basis of that list, and another in A Formulary of Correct Sounds for an Era of Great Peace (Taihe zhengyin pu 太和正音譜), a musical and prosodic study of qu lyrics written by the dramatist and critic Zhu Quan 朱權 (1378–1448), Wang Guowei rearranged the thirty plays and assigned each an author (some plays were classified as anonymous). This certainly seems a logical process, but there is a problem in that the titles of the plays in the Yuan edition and those in the bibliographies often differ. Moreover, there exists no direct and provable tie between any Yuan edition and the later bibliographies. This is why this anecdote about the dancing cranes in The Rhymes of the Central Plain is so important: it does list one play with a direct attribution to an author, and it also contains coevally variant lines of that play that are found in later editions.

Zhou Deqing’s anecdote about contending contemporary versions is equally important because it demonstrates that these plays are not purely authorial creations but a corporate enterprise in which a professional scriptwriter, profoundly knowledgeable about the stage, worked—if not directly with an actor as in the case of Ma Zhiyuan’s collaboration with another playwright and two actors in authoring The Yellow Millet Dream³¹—at least in the environment of cooperation between actors, writers, and troupes to produce scripts for performance. Those that were popular were then copied and spread to other troupes. During such a process changes could be freely introduced into the scripts. This may be why, in the earliest editions, we find no names attached to the scripts—there were too many hands involved in their production. The extant texts that we have from this early period should probably be called an accidental selection of scripts rather than a representative body of what was there at the inception of written zaju. While Li Kaixian, He Liangjun 何良俊 (1506–1573), and Zang Maoxun claim to have had hundreds or even a thousand scripts in their possession, the Chinese penchant for unspecific and exaggerated whole numbers simply to indicate “a large quantity” makes their claims unreliable in terms of a concrete number. Yet there were probably hundreds of scripts circulating during the Yuan and Ming transition.

A problem arises, though, when we consider that what we have are printed texts, not handwritten scripts. A simple hand copy would suffice for the act of passing performance scripts back and forth between acting troupes or adapting other writers’ scripts for a particular performance;³² this simple need cannot account for mass printed versions. The thirty texts we have, which come from different places and different times and demonstrate a variety of formats, reveal a considerable difference in quality of printing. But they all share one feature: they are short enough to be hand copied within a few hours. Commercial printing implies mass circulation and thus an audience of readers or listeners rather than performers. The most likely hypothesis would be that these texts were printed primarily for the benefit of those members of the audience who experienced difficulty understanding the texts of the arias when sung.³³ The fact that the majority of printed texts come from Hangzhou may well reflect the linguistic situation in that city. The twenty-seven role texts that are extant were probably simply performance scripts turned over to print houses. In several of these plays, stage directions³⁴ and plot prompts become fewer and fewer as one moves from the beginning to the end of the play. This begs a question: does the reduction indicate that fewer prompts were necessary at the end or did printers begin to drop them to save space? The latter is a logical process in which text nonessential for a reader would be eliminated. This may explain why there are three texts with no prompts at all—it may be that those prompts originally were there, but the printer, realizing that the majority of people were purchasing them to be able to understand the arias, simply dropped the plot prompts from the very beginning.

Since these role texts were printed for mass circulation, we must acknowledge that they simultaneously represent both a performance and a reading text. Although they were certainly not the closet editions (antou ju 案頭劇) of the later Ming, they were indeed meant to be read by a public. Bao Jianqiang is of the opinion that they were printed for entertainers and master teachers to use in training students.³⁵ Before we reject this notion, we should acknowledge that many entertainers were capable of reading and even producing some of the lyrics (and, one supposes, the musical notations entered by hand) in a script. The Collection of the Green Bower (Qinglou ji 青樓集) lists a number of entertainers who read and had prodigious memories, were superior calligraphers, and who exchanged written poems with literati of the day.³⁶ The women of The Collection of the Green Bower were the most talented of the empire, but it also appeared that itinerant performers and low-class hucksters could at least read the script. In the southern drama A Playboy from a Noble House Opts for the Wrong Career (Huanmen zidi cuoli shen 宦門子弟錯立身), a reworking of an earlier zaju, we find the following passage in one of the incorporated northern songs:

FEMALE sings:

Shanghua shi³⁷

My face haggard and worn, all because of you,

And just which classics do you apply yourself to in your study every day?

YOUNG MALE sings:

Don’t waste time on idle talk,

Bring out those popular musical plays!

FEMALE sings:

Look at the scripts,³⁸

And rehearse them all for me from the very beginning!³⁹

And, when memory failed, the worst of the profession could run their grimy hands over the pocket scripts, as noted in Gao Andao’s 高安道 (ca. 1250–1300) “Exposing an Insipid Troupe” (Sang dan hangyuan 嗓淡行院):

【三】 Third from Coda⁴⁰

妝旦不抹彪,	The one dressed as female lead is not sexily made up,
蠢身軀似水牛,	Her gross body just like a water buffalo;
嗓暴如恰41啞了孤庄狗。	Her voice screeches until it turns as hoarse as a homestead dog.42
帶冠梳硬挺著粗脖項,	Carrying cap and comb, she stiffly sticks up her coarse neck,
恰掌記光舒著黑指頭。	Fishing out a script, she openly spreads out blackened fingertips.
肋額的相迤逗,	Those who put up the advertisements really led us on
寫著道翩躚舞態,	By writing, “A swirling, whirling posture in her dance,
宛轉歌喉。	A lovely, modulated voice in her song.”

These passages, as enlightening as they are about audience, are about reading, not production. They shed no light on who actually produced the texts.

Thus, it remains difficult to make any conjecture about these plays as a group. We must remember that they are the scant remnants of what must have been hundreds of scripts circulating in the Yuan and early Ming periods.⁴³ We have no way of even knowing if the thirty extant plays are a representative group of what was in circulation. We can presume with some confidence, however, that the scripts that have come down to us were all produced within the space of entertainment, and that this space (as a site of a particular habitus) would be replicated within the text itself.

CONVENTIONS OF PRINTING, CONVENTIONS OF FORM

The plays as they are represented in the editions we have now at hand show they were intended for users who had a deep knowledge of the conventions of stagecraft. The printed texts have no divisions that physically separate the acts: one act runs together with another. Scenes, however, are meticulously divided: a line of text will be terminated at the end of one scene within an act, and the new scene will begin on the next line (see fig. 4).

Individual scenes that do not involve the lead actor are identified by the term zhe 折, which designates their speech acts or actions; these markers are meant to clarify the order in which subsidiary actors appear onstage and mark the relationship between them. We judge this to be a common practice in early theater, since Ming court manuscripts often do not mark the act divisions either, but they do not use the term zhe for marking scenes of subsidiary actors since these actors’ speaking parts are produced in their entirety in the manuscripts, obviating the need to signify the sequence of their appearance onstage. Ming editors later adopted the term zhe to identify “act” divisions on the printed page. At one level, the term remains the same in usage: as a marker between discrete sections of the drama. That is, in the Yuan editions it marks separate scenes and in the Ming commercial printings it is meant to clearly break the text into reading units that order the narrative (not the performance) and make reading and commentary easier.⁴⁴ As Bao Jianqiang and Hu Chengxuan remark, “The zhe in the Ming printed editions reflects the actions of closet drama of the literati and the zhe in the Yuan texts reflects the performing actions of the entertainer” 明刊本的『折』反映出文人的案頭化行為,而元本的『折』反映了藝人的演出活動.⁴⁵

In the Yuan texts, what later become separate acts are signified as a unit by the use of a single suite of arias, all written to the same rhyme and performed in the same mode. It is usually the case that the designation of mode, which is a common feature of later commercial editions, is missing. This causes little problem, since the suites generally have a set sequence of tunes (taoshu) that are recognizable as belonging to a certain mode by their first aria. There are, in fact, only six exceptions to this general rule in all the thirty plays:

FIGURE 4   Acts 1 (end) and 2 (beginning) of Banished to Yelang

Name of mode is listed:	1. The eighteenth song of the play Newly Compiled: Clerk Yue Avails Himself of Iron Crutch Li to Return from the Dead Xinshui ling is marked Shuangdiao《新編岳孔目借鐵拐李還魂》第十八曲【新水令】標注『雙調』
	2. The ninth song of Newly Compiled in Dadu with Plot Prompts: A Noble Grandson: The Story of the Undershirt is titled Dou anchun and is marked instead as Yuediao《大都新編關目公孫汗衫記》第九曲【斗鴳鶉】標注『越調』
Mode is listed, title of first song is missing:	1. The twenty-fourth song (first song of the third suite) of Newly Printed in Old Hangzhou with Plot Prompts: Aiding King Cheng the Duke of Zhou Acts as Regent is marked as Yuediao mode, but the title of the song, Dou anchun, is missing 《古杭新刊關目輔成王周公攝政》第二十四曲標注『越調』而沒有曲牌【斗鵪鶉】
	2. The thirty-sixth song (first song of the fourth suite) of Newly Printed in Old Hangzhou Complete Edition: Yuchi Gong Wrests the Spear Away Three Times is labeled as Zhenggong mode, but the title of the song, Duanzheng hao, is missing 《古杭新刊的本尉遲恭三奮槊》第三十六曲標注『正宮』而沒有曲牌【端正好】
Modes are marked with unknown mode names:	1. The thirty-sixth song (first song of the fourth suite) of Newly Printed with Plot Prompts: Zhang Ding Cleverly Investigates the Moheluo Doll is marked as Zimu diao Zui chunfeng46《新刊關目張鼎智勘魔合羅》−−第三十七取標注『子母調醉春風』
	2. The twenty-sixth song of the play Newly Compiled: Clerk Yue Avails Himself of Iron Crutch Li to Return from the Dead is titled Gudiao da qingge 《新編岳孔目借鐵拐李還魂》第二十六曲『古調大青哥』

Both performers and professional writers would know the modes and structural variations of each modal suite by heart, and there was no necessity to include well-known names of modes in a performance text as long as the name of the first song of the set was provided; conversely if the name of the mode was provided, the name of the first song could be omitted without any problem. The last two examples listed in the table are interesting because they present the possibility that these notations were introduced as ways to alert the singer to something that lies outside the ordinary: a method of performance in one case and, in the other, a reversion at that moment to an older mode, perhaps, to which the song Da qingge once belonged and to which the performer had to revert in this particular instance.

Any suite of songs may be preceded by “a wedge” (xiezi 楔子) of one or two songs. Because a very limited number of tunes can be used for a wedge, these songs are often not explicitly designated; if they are, the term “a wedge” follows the name of a tune. The songs of a wedge often conclude an independent scene, and so late Ming editions will designate a scene containing wedge songs as “a wedge,” but that is not the practice in Yuan editions. Not all plays have wedges, and there is no play that has more than two. While modern editors have little problem in dealing with wedge songs preceding a suite, they have been more confused by the songs following at the end of a suite as later editions tended to do away with such “appendices.” In Newly Printed with Plot Prompts: Chen Jiqing Is Enlightened to the Way on a Bamboo-Leaf Boat 新刊關目陳季卿悟道竹葉舟 at the end of the third suite (to the Nanlü mode), there are the following seven songs:

Jiejie gao	【節節高】
Yuanhe ling	【元和令】
Shangma jiao	【上馬嬌】
You simen	【游四門】
Sheng hulu	【勝葫蘆】
Houting hua	【後庭花】
Liuye’er	【柳葉兒】47

Among the thirty Yuan plays, this is the only case of a third suite being followed by a set of songs in a different mode and a different rhyme. Rather than describing the world of spirits and the conversion of disciples as preceding arias had done, these intrusive songs shift the focus to a lyrical exposition of the ideal happiness of Chen’s rustic life. While we do not find other examples in late Ming texts, there is at least one modest counterpart in an early fifteenth-century play by Zhu Youdun 朱有燉 (1398–1439), A Leopard Monk Returns to the Laity of His Own Accord (Baozi heshang zi huansu 豹子和尚自還俗). When at the end of the third suite of songs Lu Zhishen, a bandit turned monk, has convinced his mother that he will not return to a life of crime, he promises his mother he will provide her with a comfortable existence in the village of his patron, and the main suite of songs is then followed by two songs, separated by a long section in recited verse, that sing the joys of the countryside.⁴⁸ One is tempted to read these songs in praise of the simple country life as a parody of longer sets of songs on the life of immortals, such as the one found in Zhuye zhou.

It is more normal for the fourth set of songs to be followed by two songs, either Houting hua followed by Liuye’er, or Gu meijiu followed by Taiping ling. This phenomenon is found in the following three plays:

1.  In Newly Printed in Old Hangzhou with Plot Prompts, a Complete Text: Li Taibai Is Banished to Yelang, there are two songs after the fourth suite (in Shuangdiao mode) that are written to the Xianlü mode: Houting hua 後庭花 and Liuye’er 柳葉兒.⁴⁹

2.  In Newly Printed in Dadu with Complete Plot Prompts and Complete Edition: The Affair of the Eastern Window Exposed there are two songs after the fourth suite (in Yuediao mode) that are written to the Xianlü mode: Houting hua and Liuye’er.⁵⁰

3.  In Newly Printed in Old Hangzhou, a Complete Edition: The Great King Guan and the Single Sword Meeting there are two songs at the end of the fourth suite (to the Shuangdiao mode), Gu meijiu沽美酒 and Taiping ling 太平令.⁵¹

In modern editions, these “extra” songs are usually labeled “inserted songs” (chaqu 插曲) or “dispersal scenes” (sanchang 散場), however they are not labeled as such in the Yuan texts. These designations, all supplied by modern annotators, are essentially reintroduced into the text through the practice of reading backward from the idea of the “perfectly formatted” drama found in Zang Maoxun’s A Selection of Yuan Plays and other late Ming editions. What these additional songs at the end of the fourth suite have in common is that they represent drastic breaks with the narrative story line of the dramas and serve as the core of separate scenes. The two songs at the end of Banished to Yelang are sung by Li Bai, but probably only after he has arrived in the dragon’s palace; the two final songs in The Affair of the Eastern Window Exposed are sung by the ghost of Yue Fei, not the male lead, and the two songs in The Single Sword Meeting may well be sung by other players besides the main lead—or by Guan Yu in his divine manifestation.⁵²

There are seven plays among the thirty Yuan printings that conclude with the legend sanchang (散場) or chuchang (出場), but these terms never refer specifically to any songs in the play. Coeval descriptions of theatrical performances strongly suggest that these terms refer to an independent performance routine (such as a dance routine) that might be put on as a bonus for the audience following the completion of the show. Only one of the thirty Yuan printings may have preserved the text of a sanchang song. This is Newly Printed in a Complete Edition with Plot Prompts: In Wind and Moon in the Courtyard of Purple Clouds, in which the main text of the play is followed by an independent “Partridge Heaven”⁵³ song (Zhegu tian 鷓鴣天). We know that the “Partridge Heaven” dance was a popular sanchang routine, and we even have the name of one actress who was famous for her performance of this item.⁵⁴

The endings of these early plays are far less formal in structural terms than all the later editions, and they exhibit a variety of ways to end the fourth act (see fig. 5).

Later texts generally rationalize the endings, making them all conclude with a coda in the fourth suite of songs. But if we examine the endings of the thirty Yuan plays, we find the following:

1.  Seventeen of the thirty plays use the mode Shuangdiao and begin the suite with the song Xinshui ling. Only one of these seventeen plays (number 8) has a suite that is formally concluded by a coda.

2.  Seven of the plays are designated as having a dispersal section and an eighth (number 14, Courtyard of Purple Clouds) has an untitled final passage that we can confidently designate as a dispersal scene.

3.  Three of the plays have a final section designated as a “judgment to send one from the stage” (duanchu 斷出) or “sent from the stage with a sacrifice” (jichu 祭出), and four more plays have stage directions that indicate such actions were carried out (numbers 12, 20, 25, 26). These are all scenes enacted by an “emperor role” (that is, in the context of Yuan plays, either an emperor or a feudal lord) or by the King of the Underworld. Since such characters were prohibited from appearing onstage after the beginning of the Ming dynasty, later editions relegate this function to an official deputy of the emperor.

FIGURE 5   Endings of Yuankan texts

We can conclude from this fragmentary evidence that it was not necessary to end a suite of songs with a coda, and that there was far more flexibility in form before the texts were rationalized for consumption by Ming readers. And, we can conclude that the overall structure of the texts represents a drama that would have the following constituents:

1.  A short wedge that most often occurs at the beginning of the play, before the first suite of songs, but may also occur in other places where it might function as an interlude.

2.  Four full-length suites of songs.

3.  A separate demi-act at the end, in which some form of judgment occurs, and which usually comprises two songs written to a different rhyme than the fourth act. These scenes are related to the action in the play but are clearly distinct from the action of the main plot, although they may be a result of that action.

4.  The texts indicate that there is also a metatheatrical ending, a dispersal scene unaffiliated with the action of the play, which was likely used to send the players from the stage and the audience from their seats.

As Tanaka Issei and Komatsu Ken have demonstrated,⁵⁵ more than half the Yuan texts are concluded with some form of ritual. In three cases, Duke Wen of Jin Immolates Jie Zi Tui, Huo Guang Remonstrates as a Ghost, and Aiding King Cheng the Duke of Zhou Acts as Regent, the final acts are concluded by sacrifices. In the case of Jie Zi Tui and Huo Guang, these are scenes to carry out sacrifices to settle the souls of the two departed. In the first case the stage directions are quite short, but specific:

(收尾)

不爭

你個晉文公烈火把功臣盡,

枉惹萬萬載朝廷論議。

常想趙盾捧車輪,

也不似你個當今帝王狠。

駕云了 祭出

散場

Coda

So now

You, Duke Wen of Jin, brought this meritorious minister to his end with fire so fierce,

And uselessly stirred up court debates for tens upon tens of thousands of years.

And thinking back on the one who bore the wheel for Zhao Dun,

Find no one as evil hearted as the emperors of this age!

After EMPEROR speaks.—Send off the stage with a sacrifice.

Dispersal Scene⁵⁶

In Huo Guang, however, it is more detailed:

駕提天明了 拿二淨上了 駕斷了、安排祭出了

(落梅風)

滅九族誅戮了髫齔,

斬全家抄佑了事產。

可憐見二十年公干,

墓頂上灩灩土未乾。

這的是承明殿霍光鬼諫。

散場

After EMPEROR has indicated that it is dawn—after TWO COMICS are apprehended and enter—After EMPEROR makes his judgment and after sending him off with sacrifice.

(Luomei feng)

Nine branches of the clan exterminated, all the young slain,

The whole family decapitated, all the family wealth seized.

How pitiful: Twenty years of public service,

The moist, rich earth on his grave not yet dry—

This was

“In the Palace of Receiving Sages’ Wisdom, Huo Guang Remonstrates as a Ghost.”⁵⁷

The conclusion of Acts as Regent is more a celebratory ritual carried out after the Duke of Zhou has stepped aside as regent and restored authority to the young emperor. At the conclusion, the Younger of Tang, King Cheng’s younger brother, enters to present “auspicious rice sprouts” (rice plants from different areas that look exactly the same). Since the word “sprout” and “harmony” are perfect homophones, this symbolized the union of different areas, particularly the Zhou homeland and the newly conquered area in the east. When the sprouts were presented, King Cheng ordered his brother to escort the Duke of Zhou to the east. The stage directions again are quite spare: (唐輸獻加[嘉]禾上了 祭出):

After THE YOUNGER OF TANG enters with auspicious sprouts—send off with a sacrifice.⁵⁸

Several other plays, including In a Dream Guan and Zhang, a Pair, Rush to Western Shu, The Orphan of Zhao, and Burning the Child to Save the Mother, all show rituals that were held at the conclusion of the plays. They are, as such, a part of a broader structure of denouement that restores characters to their proper ethical and social status as a way of confirming social harmony. Whether by marriage, reunion of parents and children, addressing the deeds of malefactors, or offering sacrifices to dead souls, these are specifically social acts. In this sense, the sacrifices carried out onstage are less about religious belief and more about using ritual to establish proper family, social, and state relationships.

LATER INCARNATIONS OF THE TEXTS

Since the evidence is strong that these thirty Yuan texts were part of a corporate process of text production, it is no accident that these plays have no author’s names attached to them. The scholarly assumption has been that this is an accident that can be rectified by restoring the names through late fourteenth century catalogues. But this would hold true only if there were a single authorial edition that had never undergone any changes. It is much more likely that the stories on which the scripts are based were freely circulating, and it is also likely that once the scripts were in public circulation, they were changed at will by actors and writers throughout the Yuan; consequently, what we have today, even in the earliest editions, may in fact be considerably different from the first iteration of the drama.

As we include four Ming-dynasty rewrites, it is important to understand the relation between the Yuan editions and the later texts of zaju. Among these later texts of zaju, the thirty Yuan editions have their closest connection with a body of materials collected by Zhao Qimei 趙琦美 (1563–1624) in his Studio of the Transformed Bookworm (Maiwangguan 脈望館) called “Zaju Old and New Copied and Collated” (Chaojiao gujin zaju 抄校古今雜劇). His collection of 242 volumes includes both manuscript versions and printed editions of dramas.⁵⁹

The manuscripts in Zhao Qimei’s collection involve two kinds of texts: first, court editions (from the eunuch agency in charge of theatrical entertainments inside the palace, the Office of Bell and Drum), and second the Yu Xiaogu collection. The process of copying between these two kinds of texts and the resulting format of the page in Zhao’s collations differ significantly. For the palace texts from the Office of Bell and Drum, Zhao’s copies are quite consistent. The format is regular, the characters are bright and clear, and there are normally ten lines of seventeen characters to each half page. The song titles are not in parentheses but are set in a single line two spaces below the normal upper level of text. “Entry” poems are set one space below. The size and boldness of the characters are uniform for both prose and arias, an empty space is left above and below stage directions, and there are few padding words in the arias. When an important person such as an emperor’s name is encountered, the text stops and moves to a new line, elevating the new line two spaces above the text (see fig. 6).⁶⁰ The regularity of his copying suggests that Zhao was scrupulous about maintaining the format of the originals he transcribed, an important asset for understanding the original shape of the texts.

Distribution of Editions in the Maiwangquan Collection

Manuscript editions (172)	Print editions (70)
95 manuscripts drawn from the court	Xijizi, Zaju Selections (ca. 1589) 息機子《雜劇選》
60 have collation notes by Zhao Qimei, giving date of collation	15 plays 8 by known Yuan authors 5 by unknown Yuan authors 2 by known Ming authors
13 clearly noted as palace editions (neifu ben 內府本) but undated
(The above two categories have collation notes and lists of props and costumes [chuanguan 穿關]	Xu family, Zaju by Famous Writers New and Old appended to them)
22 have no clear record that they are palace editions but have lists of props and costumes, leading both Sun Kaidi and Zheng Zhenduo to assume they are palace editions; also no collation notes	55 plays 22 by known Yuan authors 12 by unknown Yuan authors 15 by known Ming authors 6 by unknown Ming authors
(14 of these have the note “random stories” [zazhuan 雜傳] appended and are works by unknown Yuan or known Ming playwrights)
33 that have been collated against manuscript collections in the collection of Yu Xiaogu
23 specify a date of collation; 10 do not
14 are noted as “put together for performance in the Court Entertainment Bureau” (Jiaofang Bianyan 教坊編演)
45 demonstrate no provenance but seem to be palace editions

FIGURE 6   Page 25 of Liu Xuande Goes Alone to the Xiangyang Meeting, from a palace edition showing raised characters for the term “the Sage,” referring to Emperor Xiandi of the Han

This holds true, as well, for Zhao Qimei’s transcriptions from the Yu Xiaogu editions.⁶¹ There is little consistency among the individual plays in Yu’s texts in terms of number of lines and number of characters in each line, but Zhao’s transcriptions again reflect the presumed format of the original: he did not separate the plays into acts but into scenes; he copied the arias in heavier characters to differentiate them from the prose, which was written in smaller and lighter characters; the titles of the songs were placed in parentheses; and he left the spacing between discrete scenes and stage directions—all conventions that we are accustomed to seeing in the Yuankan texts (see fig. 7). The Japanese scholar Komatsu Ken has convincingly argued that the manuscripts deriving from the Yu Xiaogu collection are based on plays as performed by the Court Entertainment Bureau in the 1460s, whereas the texts deriving from the Office of Bell and Drum would appear to reflect performances inside the imperial palace during the Jiajing period (1522–1566).

FIGURE 7   Page 3 of Tippler Zhao Yuan Encounters the Prior Emperor, a manuscript collated against a Yu Xiaogu edition

Both the texts emanating from the Court Entertainment Bureau and the far more numerous scripts from the Office of Bell and Drum are fully written out and provide the complete dialogues for every character. Late Ming evidence indicates that play scripts were submitted to court censors before performance, and this most likely also applied in earlier years of the dynasty. This practice probably also goes a long way to explain the standardization of the texts, which is already observable in the Yu Xiaogu texts but which is even more observable in the palace texts emanating from the Office of Bell and Drum. The manuscripts deriving from the Office of Bell and Drum usually conclude with an encomium to the emperor who was watching the play. For instance in the Three Kingdom’s play Prior Emperor Liu: A Meeting at Xiangyang (Liu Xianzhu Xiangyang hui 劉先主襄陽會), Liu Bei’s judgment that wraps up the action of the play concludes with

For a thousand antiquities [my men’s] names will be spread and carried on,

As they support the altars of state for hundreds upon hundreds of generations.

We pray that

OUR EMPEROR

Will live ten thousand upon ten thousands of years!

It would have been impossible for Liu Bei, at this point, to utter these words to Xiandi of the Han within the context of the play, and they are clearly meant as a eulogy for the Ming emperor.⁶² In contrast, seven of the Yu Xiaogu texts are followed by the phrase “This is the end of the zaju text” (Zaju juan zhong ye 雜劇卷終也) or by a variant of that phrase, followed by a rhymed passage that may have been either sung or chanted. Unlike the conclusion to palace plays performed before the imperial presence, the Yu Xiaogu texts concluded by “the end of the zaju text” have much more generalized summations, often self-referential to the actors or acting troupes and praising the contributions to imperial rule of the civil and military officials in front of whom the Court Entertainment Bureau often performed. While the phrase “the end of the zaju text” occurs in only seven of the dramas, we have one example of a full structure in the play The Story of Sima Xiangru and His Inscription on the Bridge (Sima Xiangru tiqiao ji 司馬相如題橋記), where the following occurs immediately after the coda of the last act:

(眾云)雜劇卷終也。(外云)道甚。(眾答云)瀛洲開宴列嘉賓,祝讚吾皇萬萬春。武將提刀扶社稷,文官把筆佐絲綸。

THE ENSEMBLE speaks: This is the end of the zaju text.

EXTRA speaks: What are you going to say?

THE ENSEMBLE responds:

Set out a feast on Fairy Isle and array the fine guests,

We pray to celebrate Our August One’s myriads of springs;

The military officials raise the swords to support the altars of state,

The civil officials hold their brushes to help the silken words⁶³ spun by the king.⁶⁴

The invocatory line (“end of the zaju text”) has been lined out in a few texts with an ink stroke (by Zhao Qimei?) (see fig. 8),⁶⁵ since it must have been puzzling to later editors.

And, while we cannot argue ex nihilo that it was expunged entirely in some texts of the Yu Xiaogu collection, there is good circumstantial evidence to suggest that this closing scene was a standard part of the performance as represented in the Yu Xiaogu materials—a set of actions, songs, or recited passages that were done after the fourth act, after the play itself was finished. In some dramas we find text that resembles the last couplet of the response just cited (military officials … civil officials) and that exists as an independent couplet unrelated to the main action. While not noted in the text itself, it is probable that these lines were uttered by the dramatic company as a whole in response to a prompt, as in the preceding example, by the “extra,” a person who was offstage and not part of the troupe onstage.⁶⁶ The linguistic and stagecraft artifacts found in the Yu Xiaogu collection are represented in this volume by the rewrite of Tippler Zhao Yuan Encounters the Prior Emperor. The Yu Xiaogu texts can be dated with some reliability to the 1480s, and, according to evidence adduced by Komatsu Ken, there is a probability that some were performed as early as 1464, more than fifty years before the palace editions of the Jiajing reign. This allows us not only to sharpen our understanding of the various strata of Ming court editions but also to begin to sort out the transmission of texts and their possible influence on later collections. For instance, differences between printed texts of the same play that show a great discrepancy in late sixteenth and early seventeenth century editions may be due not only to editorial intervention (by Zang Maoxun, for instance) but also to the different Ming manuscripts on which they were based.

The one thing that all the Ming manuscripts have in common is that they do away with the “imperial role” (jia 駕), due to proscriptions in the Ming from portraying emperors onstage:

“Detailed Explanations” from The Great Ming Legal Code

Now, for entertainers who act out northern comedies or southern plays: they are not allowed to costume as emperors or kings, empresses or consorts, loyal ministers, or brave warriors of historical dynasties nor former sages, former worthies, or images of gods; those who violate this will be caned a hundred times. Those who allow or order such costuming in the houses of officials in charge of the people shall share the same punishment. Those who costume as immortal Daoists, as well as righteous husbands, chaste wives, filial sons, and compliant grandsons, or any who compel people to goodness do not lie within this proscription. 凡樂人搬做雜劇戲文,不許妝扮歷代帝王、后妃、 忠臣、烈士、先聖、先賢、神像,違者杖一百;官民之家,容令妝扮者與同罪。其神仙道扮及義夫、節婦、孝子、順孫、勸人為善者,不在禁限。(大明律講解卷二十六刑律雜犯).⁶⁷

FIGURE 8   Page 24 of Zhuang Zhou’s Dream of the Butterfly, showing the lined-out phrase “this zaju play is finished” 雜劇卷終

The importance of this proscription is brought into relief when we consider that fourteen of the thirty Yuan print editions list an emperor role (jia 駕) in the stage directions, referring either to an emperor or to a king of the preimperial Zhou (there are fifteen if we count He Huang’s manuscript transcription of Wang Can Ascends the Tower). It is also possible that the role of Liu Bei in Dream of Western Shu was played by an emperor role, but since no stage directions are extant, we cannot tell. Conversely, while the emperor in person disappeared from the stage in the Ming, palace editions strongly emphasize imperial power as the ultimate arbiter. A fine example of that is provided by the rewrite of The Orphan of Zhao. Whereas the Yuan printing strongly suggests that the grown-up Orphan will kill his father’s murderer in a private act of bloody revenge, a fifth act that is found in the Yuanqu xuan version and most likely derives from a palace script has him ask permission to kill the murderer, turning private revenge into state punishment. Palace editions also tend to tone down explicit social criticism and gory violence. Reflecting their new performance venue, where ensembles of the finest actors in all role types were clamoring for expanded roles, many palace scripts cut down the number of arias assigned to the male or female lead and expand the part of other actors by assigning them poems and other recitation pieces.

THE THIRTY PLAYS IN MODERN SCHOLARSHIP AND CRITICISM

Despite their rediscovery and subsequent reprinting nearly a hundred years ago, the thirty independent editions of Yuan northern drama have had little effect on the study of zaju as a genre of literature.⁶⁸ Their primary influence has been in providing earlier recensions of stories used in drama as well as data for examining the structure, tonal patterns, and lexicon of early northern theater. This has been due, in our estimation, to the fact that the study of drama itself is a modern phenomenon that has assembled the corpus of early zaju retrospectively under the influence of reading practices that were a legacy of literati culture that continued into the modern age. This has created a situation, nearly unknown in other cultures, in which the early texts of drama (or portions of dramas) are unnaturally divided into two major categories: performance scripts and literary editions. This is certainly due to the fact that in China—as distinct from cultures in which narrative and drama stand as foundational genres—lyric poetry and historical or philosophical prose stand at the fount of cultural production and have ineluctably shaped the reading practices of literati.

Moreover, the distinctly modern discipline of drama studies does not start in China until the cusp of the twentieth century. Prior to this time the majority of formal writing by traditional critics on zaju and other forms of drama was limited primarily to structure (tonal patterns, rhymes), music, performability, audience, biography, and bibliography. Wang Guowei is usually singled out as the progenitor of modern drama studies in China, and his book Song Yuan xiqu kao 宋元戲曲考 has been canonized in the secondary literature, becoming the default starting point for students of drama.⁶⁹ As is well-known, Wang Guowei and those who follow him treat zaju as literature rather than performance text. As Ren Zhongmin pointed out in a lecture transcribed in 1983,

Wang Guowei was a literatus through and through; he was not a performer. His appreciation of The Story of the Western Wing, The Story of the Lute, and all Yuan drama is limited to the texts of the northern and southern songs; he did not emphasize the prose dialogue, the stagecraft, or the structure. His eyes were always on the literary compositions, therefore [in writing his book] he had to choose the term xiqu [the narrower term, literally, “the songs of plays”] and reject the idea of xiju [the broader term, literally, “plays”]. Because his knowledge of drama is so one-sided, we do not need even bother with his notions of “real drama” or “strong and complete” drama. What is unfortunate is that those who followed him blindly, like the historians of drama Zhou Yibai and Zhang Geng, had no choice but to call their works histories of xiqu, not daring to call them histories of xiju.⁷⁰ 王國維是個徹底的文人,不是藝人。他欣賞《西廂》、《琵琶》和一切元劇，乃以其南北曲的文章為限，至于賓白、排場、結構等等，亦非他所重了。他的著眼僅在詞章，故他取文體名時，就不得不用『戲曲』，而放棄『戲劇』，他對戲劇的認識如此偏差，就不必于所謂真戲劇、健全的戲劇上找他的麻煩吧。可嘆的是盲從他的一些人，如周怡白和張庚等戲劇史家，都不敢稱『戲劇史』而稱『戲曲史』.

This is not the place to debate the value of Wang’s contribution, which has created a minor cottage industry of text,⁷¹ but to pursue the line of questioning that Ren opened up about Wang’s reading preferences. In this lecture, Ren goes on to say that as a literatus trained in classical literature, Wang subscribed to the hegemonic status of textual learning, which for Ren is encapsuled in the phrase “Everything else is worthless, only reading is the highest achievement” 萬般皆下品，唯有讀書高. Furthermore, according to Ren, Wang’s book was highly popular because it resonated with an overarching scholarly preference for textual learning; and by relocating the authority to judge drama from the performers themselves (and their patrons) to the scholar and the norms of their literary tradition, it simply trumped the question of performers’ status as interpreters of culture. That is, the values of an elite meritocracy were interposed between the object of study—drama as a whole—and cultural discourse, effectively effacing both performer and nonelite writer. This was not something new. In the early fifteenth century, Zhu Quan 朱權 (1378–1448) included this short passage in his A Formulary of Correct Sounds for an Era of Great Peace:

Zaju: those performed by actors are called “prostitute-entertainer plays” and therefore called “of the stage.” Zi’ang, Master Zhao⁷² said,

The zaju performed by playboys from good families is called “the work of the professional entertainer”; that performed by prostitute-entertainers is called “layman acting.” Now good families value a sense of shame, so there are very few who actually perform. Now they have become very few and, paradoxically, those performed by the professional prostitute-entertainer are now called “professionals.” This has missed it by a mile. Someone asked me, “What is the reason for this?” And I responded by saying, “Zaju comes from great Confucian scholars and men of deep learning, composed by men of poetry and scholarship. They are all from good families. If they were not created by our kind, would the prostitute-entertainer have anything to perform? I have traced this to its root in order to bring its principles to light and therefore take them to be lay amateurs.”

Guan Hanqing said, “It is not they who are skilled in their own profession, it is the work of our group. They do no more than perform the labor of slaves, providing laughter and working diligently to please and thereby serve our likes. What these disciples act out is the ‘wind and moon’⁷³ of our whole group.” I have seized on this because, although it was said in jest, it conforms to principle.

Among the sons of good families are those who are thoroughly versed in the rules of rhymes and metrics; moreover they are born into an era of burgeoning peace, and delight under a rule of harmony. They want to revert to antiquity in order to affect the present in order to adorn this great peace. That which they now act out was called “play of happy streets” in the Sui, “music of the Pear Garden” in the Tang, “play of the garden of pleasure” in the Song, and “music of ascendant peace in the Yuan.”⁷⁴

雜劇，俳優所扮者，為之『娼戲』，故曰『勾欄』。子昂趙先生曰:『良家子弟所扮雜劇，謂之行家生活，娼優家所扮者，謂之戾家把戲。良人貴其恥，故扮者寡，今少矣，反以娼優扮者謂之行家，失之遠也。』或問:『其何故哉？』則應之曰:『雜劇出於鴻儒碩士，騷人墨客所作，皆良人也。若非我輩所作，娼優豈能扮乎？推其本而明其理，故以為戾家也。』關漢卿曰:『非是他當行本事，我家生活。他不過為奴隸之役，供笑獻勸，以奉我輩耳。子弟所扮，是我一家風月。』雖是戲言，亦合于理，故取之。良家之子有通于音律者，又生當太平之盛，樂雍熙之治，欲返古感今，以飾太平。所扮者，隋謂之『康衢戲』，唐謂之『梨園樂』，宋謂之『華林戲』， 元謂之『昇平樂』.

This short passage is perhaps the earliest extended discussion of the break between performers and literary specialists. There seems to be little doubt that, while writers trained for the examinations did not create the performance genre of zaju, they were instrumental in changing it from simple skits to a more sophisticated dramatic genre. The metrical requirements of composition, the cultural knowledge needed to sustain a complex and multilayered subjectivity presented through the lead singer, and the linguistic and stylistic ability to write poetic lyrics probably lay outside the reach of all but the most educated entertainers. This calculus is further complicated in this passage by the implied relationship between learning and status. When Zhao Mengfu comments that the writing is produced by great Confucian scholars, we do not think he meant that eminent Confucians actually wrote the plays but that the “sons of good families” were raised in a family tradition (jiafeng 家風) in which classical learning was part of everyone’s life when growing up. This knowledge was put to use in drama; but that very act was further complicated by having to adapt this classical knowledge into a universal colloquial language and the specific cultural world that stamped fiction and drama. Of course not every entertainer was incapable of this—as witnessed by the coauthors of Ma Zhiyuan, Hua Li Lang, and Hongzi Li Er⁷⁵—but for the most part entertainers simply did not have the training and therefore did not possess the stylistic gifts or ethical weight that this training necessitated.⁷⁶

In a sense, then, Wang Guowei was simply the last in a long line of elite writers that stretched back to the fourteenth century who all chose to emphasize writing and text over performance. Wang held a quite equivocal view about the quality of zaju, recognizing its beauty but also (in terms of literati culture) its flaws. He was less sanguine about what Zhu Quan had called “Confucian scholars and men of deep learning” and actually reified the divide between the social and cultural status of early text producers and those he thought of as “real dramatists” from the elite world of the literati. In his Luqu yutan, published in 1909, he wrote,

Bai Pu was the only scholar-official to write zaju. None of the other writers, like Guan Hanqing, Wang Shifu, Ma Zhiyuan, or Zheng Guangzu, held an important position, but they resided somewhere between the scholar and the actor. Therefore, there is something about their writing that stands out by itself, but so does the shallowness of their learning and the vulgarity of their minds. The reason that xiqu never was part of even the most insignificant literature was probably due to this. When it came to the Ming, then many literati began to dabble in xiqu. In the early period there was Gao Ming ([given name] Dongjia) and in the later, Tang Xianzu ([given name] Linchuan)—both were widely learned and elegant gentlemen. It was only when it reached Kong Shangren ([given name] Jizhong) and Hong Sheng ([given name] Fangsi) of our dynasty [that is, the Manchu Qing dynasty] that they began completely to sweep away several hundred years of rankness, but the vitality [of drama] was about to be exhausted.⁷⁷ 士大夫之作雜劇者，唯白蘭谷樸耳。此外雜劇大家，如關、王、馬、鄭等，皆名位不著，在士人與倡優之間。故其文字誠有獨絕千古者，然學問之弇陋，與胸襟之卑鄙，亦獨絕千古，戲曲之所以不得與於文學之末者，未始不由於此。至明，而士大夫亦多染指戲曲。前之東嘉、後之臨川，皆博雅君子也;至國朝孔季重、洪昉思出，始一掃數百年之蕪穢，然生氣亦畧盡矣.

At the time that Wang wrote this, neither the trove of 242 texts from the Maiwangguan nor the Yuankan zaju sanshi zhong had come to light,⁷⁸ and it was inevitable that he would turn to Zang Maoxun’s 1615–1616 anthology as his basic source material. He explained his choice thusly:

The world often criticizes Zang Jinshu ([proper name] Maoxun) for making so many unnoted changes in his printing of the Selection of Yuan Plays. From my perspective the early Ming manuscript of Zheng Tingyu’s zaju, King Zhao of Chu: The Most Distant Relative Goes Overboard, which was in the Jiahui Tang library of the Ding family in Qiantang,⁷⁹ is full of errors and execrably inferior, and is a long way from [the quality] of the edition found in a Selection of Yuan Plays. Now many Yuan plays underwent unseen changes at the hands of entertainers and had long lost their authenticity. The printing by Jinshu derived from editions of the Office of Imperial Theater obtained from Liu Yuanbo, and as he mentioned in his preface, “were not the same as current commercial editions.”⁸⁰ Later people discussed [Zang’s texts] by holding up examples from commercial editions and selections from the Songs from the Music Bureau in an Era of Peace and Joy (Yongxi yuefu), so it is logical that there would be so many conflicts. 世多病臧晉叔（懋循）刻元曲選，多所改竄;以余所見錢塘丁氏嘉惠堂所藏明初鈔本鄭廷玉《楚昭王疏者下船》雜劇，謬誤拙劣，不及元曲選本遠甚。蓋元劇多遭伶人改竄，久失其真。晉叔所刊，出于黃州劉延伯所得御戲監本，其序已云，『與今坊本不同』。後人執坊本及《雍熙樂府》所選者而議之，宜其多所抵牾矣.

As Wang points out in this remark, Zang Maoxun took his selection of one hundred from texts he had stored in his own library and from two hundred plays from the Imperial Theater Office. As Zang’s prefaces to the Selection of Yuan Plays show, he also picked his own texts based primarily on the criterion of how well they would educate southern writers to the wonders of zaju. In essence, then, Wang selected a text of the play to canonize what had previously been selected by a late Ming literatus as a representative work from a group of plays that had in turn been selected (or created) by the eunuch agency in charge of providing scripts for imperial performance.

Moreover, Zang Maoxun’s prefaces are written in the context of dramatic debates of the sixteenth and seventeenth centuries. In his second preface to the Selection of Yuan Plays, in particular, he sets forth his own criteria for drama (summarized here):

1.  That qu, including both zaju and xiwen, are in a direct line of descent within the orthodox genre of poetry and therefore are subject to the same standards of critical ethical judgment

2.  That qu, nevertheless, is a broader form, drawing from a wide range of sources and therefore more difficult to write

3.  That it must not only present a seamless combination of these sources but also meld the refined (ya 雅) and vulgar (su 俗) into a realm of “sentimental diction” (qingci 情辭) that is “stable and balanced”

4.  That it must create an organic space onstage, created from the appropriate register and dialect of the characters’ language and from a representation of events that is of “true color”—a recognizably mimetic and accurate reproduction of any event’s “basic appearance.”⁸¹ This space is separated from the outside not by physical barriers but by a virtual landscape that language generates from the center and which remains as a centripetal force.

5.  That it must recognize the difference in regional forms of language but that it must be clear about tonal distribution, appropriate dialect and register in terms of rhyme and tone, and that it must abide by accepted rules of composition

Finally, the perfect presentation of drama relies, according to Zang, on the skilled writer and the skilled performer, each working in their respective spheres. As in Zhu Quan’s essay, Zang’s preface is careful to keep the two spheres of writer and performer separate. Zang also clears space for his own emendations to the text, particularly his rewritings of the fourth act, by claiming that zaju authors were “worn out” by the fourth act.⁸² One of the ways of looking at Zang’s prefaces is that they are in conversation with the writings of Wang Jide 王驥德 (d. 1623), He Liangjun, and Wang Shizhen 王世貞 (1526–1590) at a critical moment in the sixteenth century when zaju is in decline and southern drama is on the ascendancy. Zang finds his greatest affinity with He Liangjun and clearly accepts the idea that qu as song was derived from a poetic legacy that could be traced back to the Book of Odes.⁸³ Thus, it had to undergo editing to bring the songs into the ethical realm that such a tradition denoted.⁸⁴ Consequently, there are clear agenda in Zang’s prefaces that center on the relationship between genre, culture, and prosody, as well as a personal anxiety about establishing his own point of view as authoritative in a moment of contending and powerful voices.

Our point here is that the canon of zaju and the modern study of the history of drama are both retrospective creations based on later, even modern, conceptions of what they should be. As opposed, say, to the tradition of shi poetry, which has a long history of development in terms of canon formation (a canon that has changed over the years) and a parallel development in poetic theory and criticism, drama is discovered as a literary genre only as it emerges into elite literati consciousness in the early Ming. And when it emerges into their consciousness, it is fraught with issues of class status, ethical status, as well as where it should be placed within an already well-defined constellation of culturally orthodox literary genres. Moreover, reading practices that had been shaped by years of education had inculcated aesthetic values that included complicated homologies between genre, linguistic usage, class, interiority, and ethics. So, in addition to the actual social class differences between performers and authors, a respectable canon had to be formed that would also adequately represent these more abstract and culturally determined values of elite culture. As opposed to what we might call the natural and symbiotic development of a canon and its accompanying critical apparatus, the corpus of zaju (and to some extent early southern drama as well) that was selected to appear in print retrospectively created an artificial and imaginary canon into which earlier textual antecedents, discovered after the fact well into the twentieth century, did not fit.

The relatively little attention paid to the Yuankan texts may also stem from the fact that they have been introduced into the field primarily in three ways: (1) as “original” zaju texts (原本) that preserve the “original features of zaju” (元雜劇本來面目),⁸⁵ (2) through textual scholars who are interested purely in a history of editions or in setting a text, and (3) as a data set for linguists working both in syntax and phonology. In the case of setting texts, the emphasis has been on providing “corrections” to variant forms of graphic representation, or replacing homophonic “borrowings” with the “real” characters. In nearly all cases emphasis has been laid on the poor quality of the received text of the Yuan printings: “There are many incorrect characters, the simplifying of written characters does not conform to normative conventions, there are critical instances of omissions, the characters are unclearly written, and the printing blocks are worn-out—none of which makes it easy for the reader” 錯別字多，簡寫不規範，脫漏情況嚴重，白字漫漶不清，版子又斷爛，頗不便于閱讀.⁸⁶ As Zheng Qian wrote in the preface of his pioneering critical edition of the thirty Yuan plays,

These texts are aria scripts published by commercial presses in the Yuan. The cutters were extremely careless and inferior; incorrectly written characters, missing characters, homophonic borrowings, simplified and vulgar characters are found all over the pages. Sometimes it is cut in such a way that it makes no recognizable character. If we discuss the format and style of the lines, we find the prose sections and arias are indiscriminately mixed, and the song titles are often incorrectly carved or missing altogether. And in addition [the texts] have another deficiency, which is that the prose dialogue is incomplete, either retaining only simple prose lines of the male or female lead or lacking any dialogue altogether. In this case, if there is no other edition available, the plot details often cannot be clearly understood. Since it has these two major deficiencies, even seasoned readers of Yuan drama experience the effort needed to read these texts, let alone beginners. And this is precisely the reason that these texts are not widely circulated, although they are fine and important.⁸⁷ 這部書是元代書坊所印的『小唱本』，刻工非常草率拙劣，錯字、掉字、同音假借字、簡體俗字、滿紙都是，有時簡直刻得不成字形。講到行款格式，則賓白與曲文常是混在一起，分不出來，曲調牌名曄常有誤刻或漏刻。此外還有一種毛病，就是賓白不全，只有正末或正旦的簡單說白，或竟全無賓白。於是，別無他本諸劇的情節常是弄不清楚。有了這兩個缺點，對于元劇修養有素的人，對此書也有時頗感吃力，更不必說初學。此書雖好雖重要而不太通行，就是這個緣故.

Zheng goes on to remark that it took him nearly thirty years to create his own edition, due in part to the need to learn the conventions of Yuan printings,⁸⁸ to master the formal features of prosody, and to have access first to the collotype edition from Kyoto and then to the Ming woodblock and manuscript editions.⁸⁹ His ultimate aim was to provide a general reader for early drama (一般通用的讀本).

Despite Zheng’s desire to make a reading edition, his work is still consulted today mostly as a philological text. The same holds true for the other collated editions of Yuankan texts. The dominant classroom and scholarly text remains the Yuanqu xuan. The reasons why are well expressed by the late Xu Shuofang in his article “Zang Maoxun and His Yuanqu xuan,” in which he addresses the problems inherent in giving up the eminently readable Selection of Yuan Plays for the earlier editions. He points out that while the Yuan editions may be thought by some to “be closer to the original text” (接近于原本),⁹⁰ their lack of dialogue makes them difficult to use. Xu considers the fact that Wang Guowei (and, by extension, others) legitimize the Yuankan as “authentic” by accepting Zang Maoxun’s proposal that “the dialogue was created by actors while they were performing. So, it is often coarse and repetitive.” Xu counters by pointing out the necessity of the dialogue to move the plot along, which cannot be accomplished by the arias alone. He cites several examples of compelling prose passages that he attributes to the literate authors of the writing societies (shuhui 書會) and notes that elite readers are always biased toward the poetry of the arias, seldom taking serious note of the dialogue as an integral part of the dramas. “To attribute such fine dialogue,” he says, “to momentary inspiration of the actor, in which anyone can add or subtract at will, if not a momentary slip of the tongue [on Zang’s part] then is to let the point of view of orthodox literature become a bane” 把。。。中的科白歸之于演員的即興之作，人人得以自由增減，不是一時失言就是正統文學觀點在做祟.⁹¹ On the contrary, Xu argues, because they contain the dialogue, Zang Maoxun’s texts are closer to the original dramas. The Yuankan texts, “with incomplete dialogue, difficult to understand arias, even having undergone meticulous putting in order … still are hard for a general readership to accept as literary texts” 賓白不全，曲文難懂，即使經過細心整理，元刊本也難以作為文學作品為一般讀者所接受.⁹²

In sum, the reading practices of a Jiangnan literatus (Zang Maoxun) and his community of readers in the seventeenth century are imposed retrospectively on a set of texts that have their own, hidden, evolutionary nature. Not only do the standards of this group become the basis for selecting and anthologizing earlier texts, they are enacted as a practice that is free from the traditional prescriptions for editing, collecting, and printing that hold for other forms of literary text. And, when drama studies become a discipline under the pressure of modernization in the twentieth century, these texts are canonized as the representative body. Moreover, later scholars who mature intellectually reading the same texts of orthodox literature maintain a standard mode of reading that subsumes dramatic texts to the more standardized and culturally comfortable realm of traditional poetics. Literati, as Xu points out, will always prefer the arias because they are poetry.

The constant stress on the difficulty of reading the Yuan editions is due to the fact that their conventions are not those of the seventeenth century or those of that particular group of Suzhou literati who were creating the corpus of drama and xiaoshuo as reading material in the early seventeenth century. Only Zheng Qian, of all modern scholars, has a notion that they can be readable as literature—and he openly acknowledges the difficulty of preparing oneself for that task. But his positive example only highlights the larger failure to see these texts historically, and to accept the fact that they represent a reading practice of an audience for whom these texts were perfectly understandable. For example, “wrong” characters often turn out to be homophones (baizi 白字), simple phonetic prompts to the correct pronunciation of a word or term. In that context they are perfectly understandable. Difficult orthography turns out to be conventional;⁹³ missing dialogue would be about well-known stories or plots; and different arrangements or length point to a differing sensibility or strictures onstage.⁹⁴ That the texts have remained underutilized as literature—due to this complex combination of social class, ethical status, easy acceptance of literati recensions as “authentic” texts, and completeness of other editions—should not deter us from seeing them at the fount of a written legacy. They are difficult to read, but we should attempt to read them in their own context, appreciate the literary quality they have, try to understand the literary quality they had in the eyes of their readers (and listeners), and see how they contribute to a legacy of textual materials. Too often the thirty Yuan plays are completely separated from later Ming editions as an anomaly, but as we hope to have shown, they form a natural unit not only with Li Kaixian’s Revised Plays of the Yuan Masters (Gaiding Yuanxian chuanqi 改定元賢傳奇) but also with the Yu Xiaogu manuscripts in the collection of the Studio of the Transformed Bookworm and other early manuscripts from the Ming Court Entertainment Bureau (Jiaofang Si 教坊司). In context, they allow us not only to understand textual development but also to comprehend how literary taste changes over the ages before it is codified by the tradition of high learning.

NOTES

    1.  By “comedy,” we mean a play in the colloquial tongue that generally portrays ordinary people or popular representations of historical figures, has loosely connected scenes, and ends in a moment of happy reunion.

    2.  That is, scripts that were produced for either the single female or male lead to use.

    3.  See Idema 1996 and West 2004.

    4.  Shi Junbao 石君寶, Wind and Moon in the Courtyard of Purple Clouds (Fengyue Ziyun ting 風月紫雲庭), in Idema and West 1982, 236–77; Guan Hanqing 關漢卿, A Beauty Pining in Her Boudoir: The Pavilion of Praying to the Moon (Guiyuan jiaren baiyueting閨怨佳人拜月亭), in West and Idema 2010b, 77–103; Zhuge Liang Burns the Stores at Bowang (Zhuge Liang Bowang shao tun 諸葛亮博望燒屯), in Idema and West 2012, 197–235; Guan Hanqing, The Great King Guan and the Single Sword Meeting (Guan dawang dandao hui 關大王單刀會) and In a Dream Guan and Zhang, a Pair, Rush to Western Shu (Guan Zhang shuangfu Xi Shu meng 關張雙赴西蜀夢), in Idema and West 2012, 236–95, 296–315.

    5.  West and Idema 2010b, ix–xxxv.

    6.  For a general overview of prosimetric narrative, see Idema 2010b, 343–67; for a basic introduction to the features of northern drama, see West and Idema 2010b, ix–xxxv.

    7.  On the use of role types in drama rather than characters, see West and Idema 2010b, x, xvii–xix.

    8.  While some texts claim to have been compiled in Dadu, only one text claims to have been printed in Dadu.

    9.  Huang Pilie 1958, 23b; preface dated Dec. 18, 1804.

  10.  Ibid., 1a–4a.

  11.  Ibid. (sec. 2), 1a.

  12.  See n. 19.

  13.  Iwaki 1961, 80–87; see also Miao 2004, 15–17. On Li Kaixian, see Idema 2005–2006; Tan 2009; and Zhen 2008.

  14.  For a succinct description of the extant editions of zaju, see West and Idema 2010b, xix–xxxi; on Ming palace editions, see Du 2008; Zheng Li 2007; and Komatsu 2001.

  15.  Guan 1958, 25b; see also Cai 1989, 2:637.

  16.  Meng 1958, 32b; see also Cai 1989, 2:790.

  17.  Zheng Tingyu 1958b, 31b; see also Cai 1989, 2:774.

  18.  Gong 1958, 32b; the arias to which he alludes are found in act 1, 9a–11b, and act 2, 15b–17b; see also Cai 1989, 2:796.

  19.  Zhen 2008, 65.

  20.  Ibid. Li Kaixian, with the aid of a number of assistants, produced his own anthology of Yuan drama titled Gaiding Yuanxian chuanqi 改定元賢傳奇 (Revised Plays by Yuan Authors). To judge from the few surviving plays of this project, it appears that they worked on the basis of Yuan printings or manuscripts in his collection. While Li and his collaborators retained the full suites of songs and made only minimal changes to the arias, they provided full dialogues for all characters in the play on the basis of a careful reading of the arias, the preserved stage directions, and the cue lines. While we know that sixteen plays were printed, only six have been preserved. Later Ming editions of plays with these titles appear to be based on these editions. See Idema 2005–2006 and Tan 2009.

  21.  Zheng Guangzu 1958, 21b, and Cai 1989, 2:793.

  22.  Ibid.

  23.  Zhou Deqing 2006. On the rhyme categories in Zhongyuan yinyun, see Stimson 1966.

  24.  Zhou Deqing 2006, 299.

  25.  Zheng Guangzu 1958, 20a.

  26.  That is, female entertainers.

  27.  A poet.

  28.  From the story of a person who dreamt of an entire lifetime during the time it took to cook a pot of millet. See chap. 4, n. 19.

  29.  Zhou Deqing 2006, 298.

  30.  Ma 1958, 4a.

  31.  Zhong 1982, 117. See also 204n606, which includes Jia Zhongming’s 價仲明 (1343–1422) note that the authors, Ma Zhiyuan 馬致遠, Hongzi Li Er 紅字李二, Hua Li Lang 花李郎, and Li Shizhong 李時中, all belonged to the Yuanzhen Writing Society 元貞書會. Hongzi Li Er (literally, “Red-Writing Li the Second”) and Hua Li Lang (literally, “Esquire Li the Flowered”) both probably sported tattoos.

  32.  Compare He Huang’s use of the autograph text that Li Kaixian seems to have copied from a Yuan printed version; see n. 16.

  33.  In modern theaters the text of arias is projected over the stage for the benefit of the audience.

  34.  Stage directions can be extremely detailed in the way they instruct the lead actor or lead actress about how to perform his or her role, at times demanding them to exhibit contradictory emotions. For specific examples, see Idema 1993.

  35.  Bao and Hu 2010, 45.

  36.  See Xia 1996, passim; Idema and West 1982, 145–69. We are referring here to the famous exchanges between Guan Hanqing, Hu Zhiyu 胡祗遹 (1227–1293), Feng Zizhen 馮子振 (1257–1327), and Pearl Screen Beauty, Zhu Lianxiu 珠簾秀. See Idema and West 1982, 166–67, and Xia 1996, 82–90.

  37.  This tune does not occur in the repertoire of southern drama, and the song appears to have come directly from the play’s northern antecedent. See West 2008, 89–94.

  38.  Zhangji 掌記 were hand-copied scripts small enough to be put into one’s clothing and carried around. Extant evidence suggests that they contained the text of an act of a play and were annotated with musical notations for quick reference. See Idema and West 1982, 93; Tseng 2009, 353–57.

  39.  Translation revised from Idema and West 1982, 207.

  40.  Following the notes in Hu 2008, 312–16.

  41.  Following Hu Ji’s suggestion that 恰 may be a mistake for 搯.

  42.  As Hu indicates, this line has room for reinterpretation given that it may refer to the “venal official” role type.

  43.  Li Kaixian remarked in his preface to the Revised Plays of the Yuan Masters that he had more than a thousand scripts in his possession. See Li Kaixian 2004, 1808, translated in Idema 2005–2006, 49.

  44.  We should note, however, that in his 1332 work, Rhymes of the Central Plain, Zhou Deqing uses the term zhe 摺 at least once to mark act divisions.

  45.  Bao and Hu 2010, 44.

  46.  Zheng Qian’s remarks are pertinent here. He has rectified this suite by inserting a more normal first song, Fen die’er, based on the Guming jia edition. But he also notes that there are other instances in which the song Zui chunfeng begins a suite, and therefore keeps the original arrangement open as an option. He takes the term zimu diao as denoting a particular singing method (changfa 唱法) and expresses uncertainty about whether or not it is linked to the first song. If it were a “singing method,” then it could be understood perhaps as “sung in contrasting light and heavy manner.” See Zheng Qian 1962, 240–41.

  47.  Fan Kang 1958, 7a–b.

  48.  See West and Idema 2010b, 346–47.

  49.  Wang Bocheng 1958, 9b.

  50.  Jin 1958, 8b.

  51.  Guan 1958, 26a.

  52.  Huang Tianji thinks Zhou Cang may sing these last songs. See Huang Tianji 2001, 107.

  53.  See Idema 2005–2006, 61–64; Shi 1958, 7a; Idema and West 1982, 276.

  54.  The actress Wei Daodao 魏道道; see Xia 1996, 136.

  55.  See particularly Komatsu 2001, 38–43.

  56.  Di 1958, 8b.

  57.  Yang Zi 1958, 7a.

  58.  Zheng Guangzu 1958, 8b.

  59.  Information drawn from Komatsu 2001, 124–42; Zheng Li 2007, 98–99; Du 2007, 48–52.

  60.  Du 2007, 49.

  61.  Zhao’s family had long been acquainted with the Yu family. Zhao’s father, Zhao Yongxian 趙用賢 (jinshi 1571), had, along with Yu’s father, Yu Shenxing 于慎行 (1545–1607), criticized the high Ming official Zhang Juzheng 張居正 (1525–1582) for his failure to carry out proper mourning rites. As a high official himself, Yu Shenxing had many opportunities to be in and out of the court and to have access to court plays, particularly those from the Court Entertainment Bureau, which were often performed for officials rather than the emperor.

  62.  See Idema and West 2012, 181, 195–96.

  63.  Imperial edicts.

  64.  Anon 1958a, 41b.

  65.  Komatsu 2001, 136–37.

  66.  This function of the extra has to be distinguished from the extra as an actor performing secondary male roles. This particular extra sometimes made intrusive comments from offstage in direct questioning of a particular actor or set of actors.

  67.  Cited in Wang Liqi 1981, 11. This proscription was first issued on the twenty-fifth day of the third month of the twenty-fifth year of Hongwu, April 21, 1389, and was restated in several other edicts during the rest of the dynasty. See ibid., 11–16.

  68.  A very recent article by Bao Jianqiang and Hu Chengxuan has finally begun to challenge the use of Ming editions as a basis for examining early text. In their work, they write, “Since the Yuan texts are performing texts that are stamped by the performance of artists and stagecraft, those who edit the texts should value and respect the basic quality of the texts, and move along the path of popular performance and should not measure them or shape them according to the standard or frame of literati editions” 既然元本是藝人性和劇場性的民間通俗演唱腳本，那麼，校勘元本應尊重其版本性質，走民間文藝一路，不以文人版本的標准、框架來衡量、斧鑿它. See Bao and Hu 2010, 44–49.

  69.  On Wang Guowei’s early opinions on drama and his subsequent shift in attitude under the influence of Japanese scholars, see Huang Shizhong 2009; He Yuming 2007; and Llamas 2010.

  70.  Ren and Ji 1983, 34–35.

  71.  In this respect, see He Jinli 2006.

  72.  The great Yuan literatus and painter Zhao Mengfu 趙孟頫 (1254–1322).

  73.  This highly ambiguous term here means “literary elegance,” although it might be tinged with its other connotation, “romantic affair.”

  74.  Zhu 2010, 39–42. Another version of this passage is found in Fei 2002, 44–45.

  75.  See n. 31.

  76.  See Zhu 2010, 42–44.

  77.  Wang Guowei 1964, 274; see also Luqu yutan in Wang Guowei 1983, 4b.

  78.  Ibid., 279 and 8b.

  79.  Gao Sixi 1976, 32–35.

  80.  Zang 1998, 1b.

  81.  This does not necessarily mean an empirical mimesis; a philosophically or conceptually hypothetical reality can also have its “basic color.”

  82.  Zang and many of his contemporaries believed that the Yuan dynasty used zaju in its official examinations and the examination candidates were exhausted by the time they reached the fourth act.

  83.  Zang 1998, 1b. The Shijing (Book of Odes) is a collection of over three hundred songs. The collection was one of the Five Classics, and tradition credited Confucius with the selection and editing of the texts that were included.

  84.  This would, of course, also place Zang in the tradition of Confucius, who supposedly edited the folk songs of the Shijing for court use.

  85.  Sun 1953, 153; Wang Guowei’s preface to the Shanghai reprint of the Kyoto edition of the thirty plays, cited in Zheng Qian 1962, i.

  86.  Xu Qinjun 1980, iii. Xu’s statement is a bit harsher than either Zheng Qian 1962 or Ning 1988.

  87.  Zheng Qian 1962, ii.

  88.  Ibid., 3: “My own knowledge of novels and short stories in the colloquial and drama had grown; moreover, I had seen several kinds of commercial Yuan woodblocks in photoprint, like the Yuefu xinsheng and Sanguo zhi pinghua, and I could recognize all the simplified and alternative-form characters in these works, and all that was left were a few malformed characters that could not be recognized” 我對於戲曲小說，所知已較前為多，並已見過若干種元代坊刻書的影印本，如《樂府新聲》、《三國志平話》之類，此書的簡體別體字已經都能認出，只剩下一些不成形體根本無從辨識的字.

  89.  In reference to the printing of the Maiwangguan collection.

  90.  Xu Shuofang 1993, 14.

  91.  Ibid, 15.

  92.  Ibid, 17.

  93.  See, for example, Ning 1988, 13–17.

  94.  The oft-cited example of The Orphan of Zhao is the primary example of this. The Yuankan text has only four acts, while the Selection of Yuan Plays version has five, leading most scholars to claim that the Yuan play is somehow incomplete, as Xu Shuofang remarks, “like an arrow nocked, its power lies in the necessity of firing it; it’s hard to leave off halfway” 如箭在弦，勢在必發，難以半途廢. See Xu Shuofang 1993, 16.