The Niu–Li Factional Strife: The Origins of a Historiographical Fiction

Abstract

The “Niu–Li Factional Strife,” named after Niu Sengru (779–847) and Li Deyu (787–850), is an enduring theme in Tang history. Based on accounts of personal animosity, a narrative evolved in which Niu and Li have become the ringleaders of two factions that drew in almost all high-profile literati of the ninth century. This article revises traditional and modern narratives of the Strife by first showing that the scattered and contradictory evidence in the earliest sources does not bear out the model of a decades-long struggle between two factions. Second, it demonstrates how “Niu and Li” first arose as an emblem of Tang weakness and a rallying cry for unity within the bureaucracy under the Northern Song two centuries later. Finally, it shows how modern historians picked up the loose ends and remoulded them into a struggle between different classes against the backdrop of factious politics in Republican China.

At present, one third of the court officials form factions.¹

As for factions: Searching for them, one finds no traces,

speaking of them, their existence is doubtful.²

In the winter of 850, the erstwhile powerful minister Li Deyu 李德裕 (b. 787) died,³ demoted and disgraced, on the tropical island of Hainan 海南, about a thousand miles south of Chang'an 長安, the main capital of the Tang 唐 (618–907).⁴ For modern historians, Li's death marks the end of a struggle between two factions (pengdang 朋黨) that had divided the court for almost half a century and involved many individuals that were prominent on the literary and political scene at the time: the “Niu–Li Factional Strife” (Niu Li dangzheng 牛李黨爭), named after Li Deyu and his opponent, Niu Sengru 牛僧孺 (779–847).⁵ The strife held the court under its spell for decades and drew the entire bureaucracy into its maelstrom. In addition, according to the literary historian Fu Xuancong 傅璇琮 (1933–2016), it holds the key to understanding the evolution of Chinese literature: “The intricate situation of mid- and late Tang literature [here: the first half of the ninth century] can only be tackled from the vantage point of the Niu–Li Factional Strife.”⁶

The absence from the sources of any long-lasting political agenda beyond holding on to power behind those alleged factions has made it difficult for historians to establish clear criteria for their demarcation. Therefore, and due to the nature of sources for the Tang, many historians have turned from the political to the literary realm and based claims of membership of certain individuals in a faction on their assumed aesthetic values, stylistic choices, and attitude towards the examinations. In contrast to the Eastern Han 東漢 (25–220) and Song 宋 (960–1279), there is no hard evidence, no “Great Proscription” (danggu 黨錮), and no “blacklist” of those who were presumed members of a faction in contemporary sources.⁷ Historians have mostly relied on circumstantial evidence to establish the existence of factions under the Tang. Andrew Eisenberg and Howard Wechsler have reconstructed the role of factions under the first two Tang emperors, Gaozu 高祖 (Li Yuan 李淵, r. 618–626) and Taizong 太宗 (Li Shimin 李世民, r. 626–649);⁸ Michael Dalby devotes a section in The Cambridge History of China to the Strife but concedes that “the poor quality of the evidence severely constrains any effort to discover a deep significance in the Niu–Li controversy.”⁹ Dalby and Wechsler offer competing definitions of factions, which highlight the differences between the early and the late Tang as well as the difficulties in coming to grips with Tang factions in general. Wechsler states in respect to the early Tang:

[A faction is] a subgroup in a decision-making body working for the advancement of certain policies or people; its members have common objectives that serve to keep them together afterward. Although a faction lacks the permanence of a political party, it is nevertheless organized around long-term interests rather than around a single, specific issue—such as a policy decision—and thus forms an alignment of significant duration within a decision-making body.¹⁰

While Wechsler puts the emphasis on “long-term interests” shared by the members of a faction, which are untraceable in the Niu and Li factions, Dalby places the focus on “networks of personal relations:”

These groups were called ‘factions’ (tang), both at the time and also in subsequent historical accounts, but they were not what we think of as factions or parties in the modern sense. Rather than being closely knit, well defined and well-disciplined pressure groups with a basis in common economic, political or ideological interest, the ninth century Chinese tang were loose associations among politicians, arising from complex networks of personal relations that defy easy characterization.¹¹

Dalby's emphasis on personal relations aptly captures the reality of mid- and late Tang politics. Unlike in the Song or in modern China, factions in the ninth century did not form around shared political goals or philosophical positions; the centrality of personal relations, including pedigree and kinship, outweighed such ideological concerns. The “official clans” (shizu 士族) that competed for positions in the bureaucracy stood at the center of network ties that helped officials to advance their careers.¹² In addition to those ties internal to the clans (e.g., common descent, marriage), the examination system added another layer of personal relations, for instance between examiner and examinee, or between graduates of the same year.¹³ Shared literary preferences for “Ancient-Style Prose” (guwen 古文) over “Parallel Prose” (pianwen 駢文) may have played some role in forging ties in the cultural realm. Yet factions reflected aesthetic choices no more than they did ideological or political ones, because these were too weak to be translated into policies under the Tang. Often, officials used factionalism as a blanket accusation to cast their opponents as “petty men” (xiaoren 小人) who worked for their and their network's personal gain, as opposed to themselves, whose actions as “gentlemen” (junzi 君子) were rooted in the “Way” (Dao 道). In that respect, Tang factions did have some things in common with those of the Northern Song 北宋 (960–1127).¹⁴

Even if factions were endemic to the court and the bureaucracy of the mid- and late Tang, the surviving sources do not bear out the narrative of two stable factions engaged in a constant tug of war over the course of several decades. Narratives of the Niu–Li Strife rely on selective readings of the sources which, at closer inspection, suggest that factional networks in the ninth century were fleeting and overlapping, so that even their protagonists were seldom sure about the divisions and affiliations. This article argues that the Strife is a historiographical fiction that emerged in the Five Dynasties 五代 (907–960) and Northern Song, but only reached maturity in the twentieth century. It first shows that the evidence for personal animosity between Niu and Li scattered over sources close to the events does not bear out the idea of two closely-knit factions, let alone of a continuous struggle between them. Second, it demonstrates how “Niu and Li” evolved into a symbol of Tang weakness centuries after the events, while still lacking features of a coherent narrative. Finally, it shows how Niu and Li resurfaced in late imperial and Republican times, when historians rewrote history as a struggle between different classes or strata of society and used the Strife as a cautionary tale against factious party politics that led to national disunity.

Although there are hundreds of references to factions in the Old Tang History (Jiu Tangshu 舊唐書), the closest and most reliable source for Tang history, none of them indicates that Niu and Li engaged in a decades-long conflict. Even the identity of the protagonists remains unclear: while Niu can only refer to Sengru, some authors suggest that Li does not refer to Deyu, but to Li Fengji 李逢吉 (758–835) or Li Zongmin 李宗閔 (d. 846), two members of Niu's faction, and that “Niu–Li” only refers to one faction. Others argue that Li Zongmin and not Niu Sengru was the head of the faction, and therefore call it the “Strife between the two Li (i.e., Deyu and Zongmin) factions” (Er Li dangzheng 二李黨爭).¹⁵ Finally, “two Li” is sometimes taken to point to one faction headed by Li Zongmin and Li Fengji. In sum, the Strife might implicate two factions or one faction, and neither Niu Sengru nor Li Deyu may have been its main protagonist.¹⁶

As the quotations in the epigraph of this article show, emperors and ministers of the ninth century undoubtedly worried about factions. However, there is little information on whether and, if so, how they conceptualized them. They certainly did not think of factions as durable political units with shared political or, to paraphrase Fu Xuancong, aesthetic goals. They bothered even less with definitions of “factions;”¹⁷ instead, they attached the morally charged label to their opponents to discredit them because it carried rhetorical weight thanks to its classical precedence. Accusing someone of forming a faction offered enough grounds for suspicion, as it meant that a person was not working for the common or dynastic good, but for his personal gain and that of his partisans. As Ari Levine puts it: “By condemning adversaries as treacherous and factious, rhetoricians by default claimed ethical superiority.”¹⁸ Yet, that condemnation had to be employed prudently: “The overtones of moral decay in the word tang [dang] were so strong that accuser, as well as accused, might find himself on the road to exile.”¹⁹ Even Dalby admits that such accusations may have had little basis in reality, or effects on the bureaucracy:

[C]hange-overs and fluctuations among Shang-shu sheng [尚書省] officials were not extraordinarily frequent during these years, when factionalism was at its height. In fact, with the single possible exception of the year 830, the movement in and out of the executive department was far less marked than at times of turbulence such as the An Lu-shan [安祿山] rebellion [755–763] or the reign of Shun-tsung [順宗, Li Song 李誦, r. 805].²⁰

All narrators of the Strife, from the Song to the present, had to work from a small and scattered source basis. The Niu–Li Strife emerged as an emblem of the discord at the late Tang court that accelerated the downfall of the dynasty and served as a cautionary tale for subsequent regimes in historical writings of the eleventh century. It remained an emblem or a vignette for the late Tang over the following centuries, until it matured into a linear narrative of factional strife (dangzheng) in the early twentieth century. In their efforts to write “new histories” that put collectives such as the nation and society at the center, twentieth-century historians remodeled the political landscape of the late Tang as a struggle between social groups or classes (jieji 階級).²¹

The second part of this article focuses on how Niu and Li emerged as an emblem of factiousness in the Northern Song. Ari Levine has shown how the discursive language of factionalism changed in the eleventh century,²² when historians began to ascribe the decline and fall of the Tang to factions in its bureaucracy in reaction to the factionalism of their own day. Northern Song histories, notably Ouyang Xiu's New Tang History (Xin Tangshu) and Sima Guang's Comprehensive Mirror to Aid in Governance (Zizhi tongjian), served all later narratives as sources despite their lack of coherence with regard to the Strife.²³

Twentieth-century historians created a coherent narrative of the Strife in their attempts to retell history in a linear or continuous mode, not as a cyclical succession of dynasties.²⁴ They used the repeated occurrences of factionalism as building blocks in their narratives, to point to continuities and breaks with the imperial past. They did this partly as a reflection of the endemic factionalism of the early Republican era, partly by projecting modern political vocabulary and concepts such as “(political) parties” (dangpai 黨派, zhengdang 政黨) and class onto premodern China. Specifically, historians gathered scattered evidence from traditional sources and used the framework of factional strife to bring the disjointed traditional accounts into a continuous narrative. Although some of its key terms such as “Niu faction” (Niu dang 牛黨), “Li faction” (Li dang 李黨), and the composite “Niu and Li” appear in the Old History, as a continuous narrative the Strife only crystallized in the twentieth century.

Inscribing the narrative of a protracted strife between two factions into the events and sources of the ninth century contributed to the linear reconfiguration of Chinese history. Some historians even tried to periodize the entirety of that history according to changing constellations of factions, which thereby became a model to analyze Chinese history along developmental lines as conflicts between different classes within the ruling elite. The third part of the article compares the theories of factionalism of two modern historians, Chen Yinke 陳寅恪 (1890–1969) and Wang Tongling 王桐齡 (1878–1953): Chen remodeled Tang political history as a succession of conflicts between different classes. His 1943 Draft Exposition of the Political History of the Tang remains the classic expression of the Strife in the twenty-first century.²⁵ Wang wrote the first linear history of factionalism in China.²⁶

Niu and Li from the Five Dynasties to the Northern Song

The source closest in time for the history of the Strife is the Old Tang History, whose compilers submitted the work to the court of the Latter Jin 後晉 (936–47) in 945. Although it was compiled hastily under a collapsing regime, historians since the Song have praised the work for its coherence and integrity.²⁷ The Old History contains primary documents such as memorials and edicts, often transcribed verbatim, which makes it a reliable source for the history of the Tang. Not surprisingly, historians have often used it, in particular its biographies, to support their narratives of the Strife. This raises the question whether the Old History and its representation of events bear out those narratives.

There are 104 references to pengdang in the Old History,²⁸ spread over the course of the dynasty but with a cluster of twelve mentions—half of them from an exchange at court analyzed below—in chapter 176, which contains biographies of Li Zongmin and other purported members of the Niu faction.²⁹ The “evaluation” (zan 贊) at the end of that chapter refers to the factions of the “Two Li and Three Yang” (er Li san Yang 二李三楊).³⁰ Together, “Niu and Li” appear twice in the Old History;³¹ in both instances, “Li” refers to Zongmin. Among the four appearances of “Niu and Li” in the New History, “Li” refers to Deyu once, in his biography;³² the other three times it refers to Zongmin.³³ Sima Guang is the first who uses “Niu and Li” exclusively in reference to Sengru and Deyu.³⁴ The compound also does not appear in the great anthologies of Tang literature Wenyuan yinghua 文苑英華, Quan Tangwen 全唐文, or Quan Tangshi 全唐詩.

Apart from the concentration of factionalism in chapter 176, the different amount of information on Niu and Li is conspicuous: The compilers of both Histories each gave Li Deyu a lengthy biography,³⁵ while Niu only received treatment in group biographies.³⁶ This might be happenstance: despite the survival of an “entombed epitaph” (muzhiming 墓誌銘) and a “spirit path stele” (shendaobei 神道碑) for Niu,³⁷ the compilers may have had less material to work with, as his surviving body of works is significantly smaller than Li's.³⁸ They may also have judged him less worthy of an individual biography. The New History even removed Niu's biography from its position in the Old History—between the upright officials Linghu Chu 令狐楚 (766–837), Linghu Tao 令狐綯 (802–879), Xiao Mian 蕭俛 (jinshi 791), and Li Shi 李石 (782–845)—into the heart of factionalism, between Li Fengji, Yuan Zhen 元稹 (779–831), Li Zongmin, and Yang Sifu.³⁹

The basic outline of the Niu–Li Strife has remained the same since the early twentieth century, when historians combined events and evaluations scattered across a variety of traditional sources, only few of which are close contemporaries to the Tang, into a seemingly coherent narrative. Although they wrote as modern academic historians who employed “scientific” methods, they rarely escaped the temptation of “allocating praise and blame” (baobian 褒貶) in the tradition of imperial scholar-officials. Thus, throughout the twentieth century and into the twenty-first, historians felt (and feel) the need to take the side of either Niu Sengru or Li Deyu.

The foundations of the Strife were laid in one of the scandals caused by the nepotism ingrained in the Tang examination system, which relied heavily on patronage and recommendation.⁴⁰ In 808, Huangfu Shi 皇甫湜 (777–835), Niu Sengru, and Li Zongmin wrote policy essays for a decree examination that criticized “powerful favorites” (quanxing 權倖).⁴¹ The New History suggests that this innuendo referred to grand councilor Li Jifu 李吉甫 (758–814), Deyu's father. Jifu complained to the emperor about what he saw as slander, whereupon the examiners were demoted and the promotions of Huangfu, Niu, and Li Zongmin rescinded. The event is said to have sparked the Strife by planting hatred between Jifu's son, and Niu Sengru and Li Zongmin.⁴² Only Huang's essay survives, and its target remains vague.⁴³ Some writers suggested early on that the essays did not criticize Li Jifu at all, but the court eunuchs.⁴⁴

In 821, it was Deyu's turn to get involved in an examination scandal, when Duan Wenchang 段文昌 (733–835) and Li Shen 李紳 (772–846) wrote letters to the examiner of that year, Qian Hui 錢徽 (755–829), urging him to let their favorites come out on top. Surprisingly, Qian did not pass a single one of their protégés; instead, a number of relatives of other officials—Li Zongmin's son-in-law, a son of the former grand councilor Pei Du 裴度 (765–839), and a relative of the second examiner, Yang Rushi—made the grade that year. Li Deyu, Li Shen, and Yuan Zhen backed Duan when he contested the result, forcing Muzong to demote Qian, Yang, and Li Zongmin and order a retake, presided over by Bai Juyi 白居易 (772–846) and Wang Qi 王起 (760–847). Unsurprisingly, no graduate of the first round passed the second.⁴⁵

One or both scandals mark the birth of the Strife according to modern narratives, although they expose the nepotism intrinsic to the examination system rather than a factional conflict that lasted over several decades. Fan Zuyu 范祖禹 (1041–1098), principal author of the Tang sections in Zizhi tongjian, remarked in his Tang Mirror (Tangjian 唐鑑): “The factionalism of the Tang started with the policy essays of Niu Sengru and Li Zongmin, and resulted in the sacking of Qian Hui.”⁴⁶ But only in the twentieth century did the two scandals come to serve as starting points in a plot that associated almost every prominent official of the 820s to 850s with one of the two factions. Ninth-century court politics thereby came to resemble a revolving door through which one group rises to the top of the bureaucracy, while the other bides its time in the provinces until the tables turn. Niu Sengru and Li Deyu died in 847 and 850, respectively, but the fallout of their conflict continued into the 850s, when members of the Niu faction still served as grand councilors.⁴⁷ As Han Guopan 韓國磐 (1920–2003) put it: “The majority of bureaucrats of the day were dragged into the whirlpool of this factional conflict.”⁴⁸

Some writings by Niu and Li may have had their share in the emergence of the Strife narrative. Li polemicized against factions and their members in his Record of Exhausting Sorrow (Qiongchou zhi 窮愁志), a collection of “essays” (lun 論) he wrote after his banishment to the south.⁴⁹ The most blatant assault against Niu in the Record is the “Essay on Travels in the Land of Zhou and Qin” (“Zhou Qin xingji lun” 周秦行紀論), a response to a tale of strange events (chuanqi 傳奇) ascribed to Niu, Travels in Zhou and Qin (Zhou Qin xingji 周秦行紀).⁵⁰ Niu's biography in the Old History states about the essay: “After Deyu was banished to the south, he drew on vulgar prophecies of a calf to denounce Sengru in his Record of Exhausting Sorrow, calling him ‘His Excellency Great Sacrificial Animal’; such was their mutual hatred.”⁵¹ Both the characters for “calf” (du 犢) and for “sacrificial animal” (lao 牢) contain the character for “bovine” (niu 牛), Sengru's surname.⁵² Most scholars reject the authorship of the tale and essay.⁵³ Zhang Ji 張洎 (933–996) already questioned Niu's authorship of Travels in his Record of Chit-Chat by Mr Jia [Huangzhong 賈黃中, 941–996] (Jiashi tanlu 賈氏談錄), claiming that it was the work of a Li-faction partisan, Wei Guan 韋瓘.⁵⁴ Henceforth, the view that Travels is a forgery became an undisputed fact for historians who tackle the history of the Strife without, however, questioning the validity of the Strife narrative itself.

The “Essay on Travels in the Land of Zhou and Qin” references another spurious document of the Strife, the Daily Calendar of Niu and Yang (Niu Yang rili 牛羊日曆/歷) attributed to Liu Ke 劉軻 (772–835?).⁵⁵ Pulleyblank believes that Liu wrote the Calendar primarily to vent his anger about being passed over repeatedly for promotions by Niu and his party. However, even Pulleyblank seems doubtful about the validity of the Strife framework when he states: “The bureaucracy as a whole were [sic] by no means sharply divided into two groups.”⁵⁶ Consequently, the Calendar does not mention Li Deyu or a strife between two factions, but only the machinations of Niu, Li Zongmin, and others. Of course, absence of evidence is not evidence of absence. The reason Liu does not mention Li might be that he wrote the Calendar in the late 830s, before Li reached the height of his power under Emperor Wuzong 武宗 (Li Yan 李炎, r. 840–846). Be that as it may, the Calendar does not verify the existence of a prolonged strife that started with the two examination scandals.

The Weizhou Affair

Another key component of the Strife narrative that demonstrably caused animosity between Niu and Li is the Weizhou 維州 incident of 831. While Li was Military Commissioner of Xichuan 西川節度使 (modern Sichuan), the Tibetan (Tufan 吐蕃) commander Xidamou 悉怛謀 surrendered the city and former Tang prefecture of Weizhou (northwest of Chengdu 成都)⁵⁷ together with its garrison of Tibetan soldiers and their kin. Li accepted the surrender and petitioned the court for confirmation. Most court officials supported the seizure of Weizhou, which the Tang had lost to the Tibetans after the An Lushan Rebellion; however, Emperor Wenzong 文宗 (Li Ang 李昂, r. 827–840) sided with Niu Sengru who, as grand councilor, opposed it because the Tang were engaged in peace negotiations with the Tibetans. Accepting the surrender and return of Weizhou would yield a short-term victory for the Tang but damage its “trust” (xin 信) with the Tibetans and, furthermore, ignore what was “right” (yi 義). Niu also reminded the emperor of the superiority of the Tibetan cavalry, whose “ten thousand horsemen” (wan ji 萬騎) could reach the gates of Chang'an in less than three days. “Even if we gained a hundred Weizhou, of what use would they be to us?”⁵⁸ The emperor ordered Li to hand Xidamou and his people over to the Tibetans, who, unsurprisingly, brutally “executed them,” including children, “leaving no one alive, to deter the barbarians.”⁵⁹

Several sources offer differing accounts of the event, including the biographies of Niu and Li in both standard histories, and Zizhi tongjian.⁶⁰ The one in Li's Old History biography is based on a petition that Li submitted to Emperor Wuzong in 843, that is, more than twenty years afterwards, asking for the posthumous bestowal of the title of “vice commissioner” (fushi 副使) of Weizhou on Xidamou.⁶¹ It emphasizes the strategic importance of Weizhou and states dryly that Niu “blocked the deliberations” (juyi 沮議), referring to Niu's biography for more details.⁶² The latter's narration of the incident is based on the entombed epitaph by Du Mu, which informed subsequent accounts.

大和六年，西戎再遣大臣贄寶玉來朝，禮倍前時，盡罷東嚮守兵，用明臣附。李太尉德裕時殿劍南西川，上言維州降，今若冠生羌三千人，燒十三橋，擣戎腹心，可洗久恥，是韋臯二十年至死恨不能致。事下尚書省百官聚議，皆如劍南奏。公獨曰：「西戎四面各萬里，來責曰何事失信？養馬蔚茹川，上平涼坂，萬騎綴回中，怒氣直辭，不三日至咸陽橋。西南遠數千里，雖百維州，此時安可用？棄誠信，有利無害，匹夫不忍為，況天子以誠信見責於夷狄，且有大患。」上曰「然」，遂罷維州議。⁶³

In the sixth year of Taihe (832~33), the Western Barbarians once more sent a delegation of big-wigs carrying precious jades to come to court, the pomp twofold that of previous occasions, [offering] to completely withdraw their east-facing defense armies and showing clear-sightedness by submitting as subjects. Great Defender Li Deyu, who at the time guarded Jiannan and Xichuan, reported that Weizhou had surrendered and that, if one equipped 3.000 raw Qiang with official caps and burned the thirteen bridges, it would stir the hearts of the barbarians and the old humiliation could be washed away—a feat that Wei Gao (d. 805) had not been able to accomplish in twenty years until his death. The matter was delegated to the officials of the Department of State Affairs for joint deliberation, who all agreed with Li's proposal. Only His Excellency [Niu] said: “The Western Barbarians, whose realm spans ten thousand li in four directions, will hold us responsible in the future: For what purpose would you go back on your word? Grazing their horses at Weiru River and ascending Pingliang Slope (in Gansu), ten thousand riders may descend on Huizhong and, giving straight vent to their anger, reach Xianyang Bridge in under three days. The Southwest is several thousand li away, even if we gained a hundred Weizhou, how could we make use of them at times like these? A mean fellow cannot bear to abandon honesty and trust if it reaped [only] benefit and no harm. How much more so the Son of Heaven, who faces his responsibility toward the barbarians with honesty and trust and, moreover, is beset by grave concerns?” The emperor agreed and, henceforth, dropped deliberations on Weizhou.

According to Du, the Weizhou affair was a matter of weighing “benefits and risks” (lihai 利害) against “honesty and trust” (chengxin 誠信) for Niu, wherein he sided with the latter. The editors of Niu's biographies in the two standard histories and of Zizhi tongjian quote, with variations, Du's account of Niu's words, although it is doubtful that Du was present during the court deliberations. From the tenth to eleventh century, the historians’ assessments of events shifted: the tenth-century Old History assesses that “Niu had been at enmity with Deyu for a long time, [and] even though their deliberations concerned the border for the sake of the body politic, the defamatory gossip of those who leaned on Deyu boiled up because Niu had harmed his merit. Even the emperor considered it/him disingenuous.”⁶⁴ The eleventh-century New History is more concise: “At the time, everyone said that Niu, because he held a longstanding grudge [against Li], obstructed and undermined him. Even the emperor considered it/him disingenuous.”⁶⁵ Both histories are ambiguous about whether the emperor thought Niu, Li, or the whole affair smelled fishy. The eleventh-century Zizhi tongjian states that “Deyu's hatred against Sengru thereupon grew deeper,” adding: “Hangers on of Li Deyu thus said: ‘Sengru was at odds with Deyu and harmed his merit.’ The emperor grew more estranged to him (Niu).”⁶⁶ Until the eleventh century, writers mostly sided with Li on the matter of Weizhou. It was Sima Guang who reversed that verdict:

臣光曰：論者多疑維州之取捨，不能決牛、李之是非。臣以為昔荀吳圍鼓，鼓人或請以城叛，吳弗許，曰：「或以吾城叛，吾所甚惡也，人以城來，吾獨何好焉！吾不可以欲城而邇姦。」使鼓人殺叛者而繕守備。是時唐新與吐蕃脩好而納其維州，以利言之，則維州小而信大；以害言之，則維州緩而關中急。然則為唐計者，宜何先乎？悉怛謀在唐則為向化，在吐蕃不免為叛臣，其受誅也又何矜焉！且德裕所言者利也，僧孺所言者義也，匹夫徇利而忘義猶恥之，況天子乎！譬如鄰人有牛，逸而入於家，或勸其兄歸之，或勸其弟攘之。勸歸者曰：「攘之不義也，且致訟。」勸攘者曰：「彼嘗攘吾羊矣，何義之拘！牛大畜也，鬻之可以富家。」以是觀之，牛、李之是非，端可見矣。⁶⁷

Your servant Guang says: “Many debaters have scrutinized the seizure of Weizhou, unable to make up their minds about who was right and who was wrong, Niu or Li?⁶⁸ I believe when, in former days, Xun Wu besieged Gu, some of the people of Gu offered to defect with their city, but Wu would not allow it; [instead] he said: ‘If someone was to betray us with one of our cities, it would be utterly despicable to us! Then how on earth can we approve of someone else coming to us with their city? We cannot condone treason for the sake of obtaining a city.’ This caused the people of Gu to kill the traitors and strengthen their fortifications. At that time, the Tang had renewed their friendship with the Tibetans when they received Weizhou. Speaking of its benefit, Weizhou was of small significance while trust [between the Tang and Tibetans] was of great significance; speaking of its harm, Weizhou was secure while the capital region was in danger. Therefore, measuring [benefit and harm] for the Tang, which one should have come first? [Although] Xidamou was a submissive subject to the Tang, he was a treacherous official to the Tibetans. That he should suffer execution—what pity lies therein? Moreover, what Deyu spoke about was benefit, what Sengru spoke about was what was right. That a petty man covets benefit and forgets about what is right is shameful enough, how much more so for the Son of Heaven! Let's compare it to that case of a neighbor's oxen that leisurely wanders into someone's home: Some counsel his elder brother to return it, some counsel his younger brother to steal it. The one who counsels returning it says: ‘Stealing it is not right, furthermore it might cause litigations.’ The one who counsels stealing it says: ‘The neighbor often stole our sheep, why should we be deterred by what is right? An ox is a large piece of livestock, selling it can enrich the family.’ From this point of view, the right and wrong of Niu and Li can clearly be determined.”

Sima Guang's assessment of the Weizhou affair must be read in connection with his opposition to the resumption of hostilities between the Song and Tangut/Xi Xia 西夏 in 1064 under Emperor Yingzong 英宗 (Zhao Shu 趙曙, r. 1063–1067) and especially to the irredentism of Shenzong 神宗 (Zhao Xu 趙頊, r. 1067–1085).⁶⁹ This connection was already pointed out by Hu Guang 胡廣 (1369–1418) in the Ming⁷⁰ and Wang Fuzhi 王夫之 (1619–1692) in the Qing.⁷¹ In the twentieth century, it was taken up by Cen Zhongmian 岑仲勉 (1885–1961) to show that Li Deyu had been “without a faction” (wudang 無黨) in his refutation of Chen Yinke's view on Tang factions, which will be discussed below.

Some “essays on factions” (pengdang lun) from the Northern Song discuss Niu and Li. Yet they seem less occupied with the course of events than with the personal qualities of their protagonists. Wang Yucheng 王禹稱 (954–1001), for instance, writes:

愚讀唐史，見元和、長慶之後至大和、開成間，贊皇{李德裕}、竒章{牛僧孺}、李凉公{李逢吉}輩互為朋黨。[⋯⋯]且竒章全德而不免竄逐，贊皇忌刻、逢吉傾巧而終至大位，又誰咎哉、又誰咎哉！⁷²

[When] I read the historical records of the Tang, I see how, between the Yuanhe and Changqing up to the Taihe and Kaicheng eras, the likes of the Dukes of Zanhuang (Li Deyu), Qizhang (Niu Sengru), and Li (Fengji), Duke of Liang, formed factions [against] each other. […] Moreover, Qizhang was of perfect virtue but could not avoid being banished, Zanhuang was consumed with envy and Fengji full of craftiness, but both reached high office in the end. Alas, whose fault was this!? Whose fault was this!?

It is Sima Guang who is more occupied with the historical causalities. He voices strong opinions on the Strife, both in the Comprehensive Mirror and in his “Essay on Factions.” At the outset of the latter, he approvingly quotes an “Essay on the Destruction of the Tang” (“Huai Tang lun” 壞唐論) by Huang Jiefu 黃介夫 (dates unknown), which, unfortunately, is no longer extant.

黃介夫作《壞唐論》五篇，以為壞唐者，非巢、溫與閹豎，乃李宗閔、李德裕朋黨之弊也。是誠得其本矣。雖然，介夫知其一，未知其二。[⋯⋯]夫宗閔、德裕雖為朋黨，由文宗實使之。文宗嘗曰：「去河北賊易，去朝中朋黨難！」殊不知群臣為朋黨，誰之過也？由是觀之，壞唐者，文宗之不明，宗閔、德裕不足專罪也。⁷³

Huang Jiefu wrote an “Essay on the Destruction of the Tang” in five chapters, in which he argues that the destruction of the Tang was not brought about by [Huang] Chao, [Zhu] Wen, or the eunuchs, but by the evil of Li Zongmin's and Li Deyu's factions. This truly gets to the root of the matter. However, Jiefu knew only one part of it and not yet the other. […] Now, although Zongmin and Deyu formed factions, it had really been caused by Wenzong. Wenzong once said: “To rid Hebei of rebels is easy, to rid the court of factions is difficult!” If scarcely he knew that ministers form factions, then whose fault was it? From that vantage point, what destroyed the Tang was Wenzong's imprudence, Zongmin and Deyu are not to be blamed exclusively.

Sima Guang's response to Huang Jiefu's Essay on the Destruction of the Tang became a model for later historians who identified factions, albeit not limited to those of Niu and Li, as root cause behind Tang decay and collapse. Yet Song theories of factionalism stayed, for the most part, within the frameworks of classical hermeneutics and historical analogism identified by Ari Levine, in that they explained factionalism by a dichotomy between morally superior gentlemen and petty men.⁷⁴ Within that dichotomy, which persists in much of twentieth- and twenty-first-century scholarship, either Niu or Li had to be a superior man while the other was a petty fellow. However, the essay also shows that some Song authors already felt uneasy about this one-sided distribution of praise and blame: even though Sima does not develop a theory of institutional failure or dysfunction, he suggests that factionalism was unavoidable under the weak emperors of the late Tang.

Another example, and perhaps the oldest one that survives which specifically examines Niu and Li (Deyu), which stands out in that it blames neither unilaterally but tries to reach a nuanced judgement of each one's character, is from the Historical Essays (Lidai lun 歷代論, lit. Essays on Successive Dynasties) by Su Zhe 蘇轍 (1039–1112), a younger contemporary of Sima Guang. It contains a long discourse on the Strife simply titled “Niu and Li” 牛李.

唐自憲宗以來，士大夫黨附牛、李，好惡不本於義，而從人以喜慍，雖一時公卿將相，未有傑然自立者也。牛黨出於僧孺，李黨出於德裕。二人雖黨人之首，然其實則當世之偉人也。蓋僧孺以德量高，而德裕以才氣勝。德與才不同，雖古人鮮能兼之者，使二人各任其所長，而不為黨，則唐末之賢相也。⁷⁵

From the time of Xianzong under the Tang, the scholar-officials attached themselves to the factions of Niu and Li. Good and bad did not root in what was right but followed the man according to likes and dislikes. Even though they might become one-time chancellors and generals, not one of them relied singularly on his own abilities. The Niu faction sprang from Sengru, the Li faction from Deyu; although the two men were heads of factions, they still were, in fact, great men of the age. Presumably, Sengru measured higher in virtue while Deyu prevailed in talent. Virtue and talent differ, even among the ancients there rarely was anyone able to combine both; if those two men had relied each on their strength and not formed a faction, they would have made worthy chancellors of the late Tang.

For Su Zhe, Niu and Li are no longer superior or petty men but carriers of personality traits or moral qualities—virtue on Niu's side and talent on Li's—whose unequal distribution prevents both great men (weiren 偉人) from becoming worthy chancellors (xianxiang 賢相). On the surface, Su juxtaposes Niu and Li and their views on policy issues to highlight their strengths and weaknesses; underneath, this juxtaposition reflects Su's opposition to the new policies of Wang Anshi 王安石 (1021–1086),⁷⁶ who “prevailed in talent” while his opponent, Sima Guang, “measured higher in virtue.”

The early fourteenth-century Wenxian tongkao cites entries from the catalogues Junzhai dushu zhi 郡齋讀書志 and Zhizhai shulu jieti 直齋書錄解題 of a now lost Yuanhe Record of Faction(s) (Yuanhe pengdang lu 元和朋黨錄), which recounts the events of the “Niu–Li faction(s)” from 808 to 859.⁷⁷ The two catalogues diverge on the dating of the Record, and the reference to the policy essays of Niu and Li (i.e., Li Zongmin), Niu Li duice 牛李對策, suggests that it relates the events of one faction rather than a strife between two factions.

元和朋黨錄一卷。鼂氏曰：唐馬永易記牛李朋黨始末，自牛僧孺試賢良迄令狐綯去位。陳氏曰：（元和錄三卷）池州石埭縣尉維揚馬永易明叟撰。自元和三年牛李對策以至大中十三年令狐綯罷相唐朋黨本末具矣。[……]崇觀政和閒人也。又有馬永卿大年者，從劉元城游，大觀三年進士。當是其羣從。館閣書{目}以永易{錫}為唐人，大誤也。⁷⁸

Record of the Yuanhe Faction(s) in one juan. Mr. Chao states: “Ma Yongyi⁷⁹ of the Tang records the events of the faction of Niu and Li, from Niu Sengru taking the palace decree exam to Linghu Tao leaving office.” Mr. Chen states: “(The Yuanhe Record in three juan) written by Ma Yongyi, (courtesy name) Mingsou, of Weiyang in Shili county, Chi prefecture. The entirety of the factions of the Tang, from the policy essays of Niu and Li in the third year of Yuanhe (808) to Linghu Tao's removal from the state council in the thirteenth year of the Dazhong reign (859), are all recorded herein. […] [Ma] was a man of the Chongning, Daguan, and Zhenghe eras (1102–18). There was also a Ma Yongqing, (courtesy name) Danian, who was an associate of Liu Yuancheng (dates unknown), passed the Advanced Scholar's degree in the third year of Daguan (1109), and must have been his (Yongyi's) paternal kin. The Catalogue of the Imperial library and Archives writes that Yongyi was a man of the Tang, what a grave error!

If Ma Yongyi was a man of the Tang, as Chao Gongwu states, then his Record of the Yuanhe Faction(s) would be a prime candidate, if not for being the source of the Strife narrative, then for being the earliest text that tries to bring all the events that later formed the basis of the narrative together in a linear mode. If Ma, however, was a man of the early twelfth century, which seems more likely, his Record postdates even the Comprehensive Mirror. In that case, Record of the Yuanhe Faction(s) was in all likelihood an early example of a genre of historical writings that emerged in the Song: the “topical narratives” (jishi benmo 紀事本末), to which we will return below, and which bring together events that were disconnected in the purely chronological histories of the “annalistic” (biannian 編年) style such as the Comprehensive Mirror.⁸⁰

The Evolution of the Strife after the Northern Song

At least three responses to Sima's appraisal of the Weizhou affair survive from the Southern Song: one by Hong Mai 洪邁 (1123–1202), who argues against critics who thought Li was more capable than Niu and praises Sima for setting the record straight.⁸¹ The second comes from Zhu Xi 朱熹 (1130–1200), who faults both Li and Niu: Li had been wrong in accepting the surrender of Weizhou, so returning the city was acceptable. However, handing over Xidamou and his soldiers to be slaughtered by the Tibetans was excessive. Niu had argued for what was right considering the peace agreement, but his heart betrayed personal motives when he undermined Li's position at court. Li's plans might have been deceitful and in contradiction with the peace accord, yet his heart had been correct. The ultimate fault, however, lay with the Tang court, which should not have signed a peace treaty with barbarians to begin with.⁸² Finally, Hu Yin 胡寅 (1098–1156), in his Limited Views from Reading History (Dushi guanjian 讀史管見), argues from a “fervently revanchist”⁸³ position that Weizhou had been Tang territory from the start and castigates Niu for “harming a great strategy for the sake of small trust” (以小信妨大計也) while Li had “planned affairs of state with great righteousness” (以大義謀國事).⁸⁴ The responses by Zhu and Hu must be seen in light of the humiliation that the Song had suffered at the hands of the Jurchen Jin 金 (1115–1234) and the opposition against the treaties the Song court was forced to sign with the Jurchen.

The next step in the evolution of the Niu–Li narrative, and perhaps the most important one until the twentieth century as it connected all events into a linear structure for the first time, is Yuan Shu's 袁樞 (1131–1205) Topical Narratives from the Comprehensive Mirror of 1174. It takes entries from Sima Guang's Comprehensive Mirror and, as the title suggests, rearranges them verbatim according to topic, without adding extra content. Chapter thirty-five contains an account of the “Calamities of Factions” (“pengdang zhi huo” 朋黨之禍) of the late Tang,⁸⁵ running from the examination scandal of 821 (not that of 808) to Li Deyu's death in 850. The summary of the scandal is identical in the Mirror and Topical Narratives and reads as follows: “From that time onward, Deyu and Zongmin each split into factions. Their mutual attacks lasted for forty years.”⁸⁶

Here, it is Li Zongmin, not Niu Sengru, who was entangled with Deyu in a feud that lasted for decades. Niu does not appear in the scandal of 821, but only enters the picture two years later, in 823, in a context seemingly unrelated to Deyu. This, however, only serves as backdrop for Niu's ascent to grand councilor (tong pingzhang shi 同平章事), which brings Deyu back into the picture:

時僧孺與李德裕皆有入相之望。德裕出為浙西觀察使，八年不遷，以為李逢吉排己引僧孺為相，由是牛、李之怨愈深。⁸⁷

At the time, Sengru and Li Deyu both had prospects of becoming chancellors. Deyu was dispatched to become Inspection Commissioner of Zhexi and was not promoted (to court) for eight years. He thought Li Fengji handicapped him and supported Sengru to become chancellor, therefore, Li's and Niu's mutual hatred grew deeper.

From there, the pendulum swings back to focus on the Li faction, but not the one of Li Deyu or Li Zongmin, but the one of Li Fengji and Li Shen. The conflict between Niu Sengru and Li Deyu resurfaces several paragraphs further down. There, Yuan explains that Deyu and Sengru rose to become grand councilors under Wenzong thanks to the sponsorship of Pei Du for Deyu and Li Zongmin for Sengru. That insertion closes with the Weizhou incident and its aftermath and, from there on to the end, focuses on the conflict between the two Li factions (er Li zhi dang 二李之黨) of Deyu and Zongmin. The conflict between Sengru and Deyu seems like an afterthought or addendum. Yuan's narrative of factionalism ends with the dictatorship of Li Deyu under Wuzong, his fall from grace under Xuanzong 宣宗 (Li Yi 李怡, r. 846–859), and the deaths of Li Shen, Li Zongmin and, finally, Li Deyu. Rather than painting a picture of the bureaucracy divided into two factions of Niu and Li (or one Niu–Li faction and the rest), as most modern narratives do, Yuan presents a more complex—and thereby perhaps more realistic—image, in which not one but three Li factions (of Fengji, Deyu, and Zongmin) exist at the same time. Twentieth-century historians, in their drive to put the writing of history on a more “scientific” basis by using terms and explanatory models largely taken from the West, reduced that complexity by bifurcating Tang factions into two distinct classes or groups. Their search for simple monocausal explanations and laws of development represented a regression from earlier modes of writing history.

After the Song, interest in the Weizhou affair waned and then resurfaced in the late Ming and early Qing. Wang Shizhen 王世貞 (1526–1590) discusses it in an “Essay on Niu Sengru” (“Niu Sengru lun” 牛僧孺論),⁸⁸ and Wang Fuzhi 王夫之 in Discourses on Reading the Comprehensive Mirror (Du Tongjian lun).⁸⁹ However, their discussions bear little relation to the narrative of a factional strife and they do not draw a connection to the earlier examination scandals. Instead, they see Weizhou as evidence that Li and Niu held a personal grudge against each other, and as an argument against reconciliation with barbarians. Wang Fuzhi discusses Niu Sengru, Li Zongmin, Li Deyu, and their factional alignments in a section preceding his discussion of Weizhou, but its content and, indeed, title: “The sudden formation of factions” (“Pengdang shuhu lihe” 朋黨倏忽離合), belie the narrative of a structured factional strife that emerged later. Their associations, according to Wang, had arisen on the spur of the moment, and it was impossible to determine conclusively who was aligned with whom or who was right and who was wrong. The decline of the Tang and Song had followed the same principle, in that weak rulers “sat like corpses” (shiju 尸居) and sighed at the chaos that emerged around them when officials split into constantly shifting alliances, creating a power vacuum. Wang develops a domino theory of factionalism: a self-righteous minister under a weak emperor tolerates when his partisans attack his opponents. Thereby, the first domino is flipped, and the ripple effects spread through the whole bureaucracy.⁹⁰ In Wang's view, it would have required a strong and determined ruler to break the tide and create unity among the officials, but mid- to late Tang emperors clearly did not fill that bill.

The Modern Re-Invention of the Niu–Li Factional Strife

The dramatic changes to Chinese society in the nineteenth and twentieth centuries also impacted on the discourse of factionalism. The fall of the last imperial dynasty, the Qing 清, in 1912 and the introduction of a new political system required the use of a new language, which was translated, first and foremost via Japanese, from European languages.⁹¹ In cases where a word or expression from premodern sources was deemed fit to translate a new concept, such as in the case of “religion” (zongjiao/shūkyō 宗教), the expression was usually gutted of its premodern meaning. In contrast, the modern words for political parties, zhengdang and dangpai, are true neologisms that do not appear before the last years of the nineteenth century;⁹² “factional strife” (dangzheng) likewise is a newly invented term that does not appear before the late Qing. All three—zhengdang, dangpai, and dangzheng—only started to gain currency after the Hundred-Days Reform of 1898, when the traditional language of factionalism and the modern language of party politics briefly overlapped in accusations against the reformers, before the former gave way to the latter in the new century.⁹³ The expression “Niu Li dangzheng”, so prevalent in scholarship on the Tang today, does not even seem to appear before the mid-twentieth century. The sources analyzed thus far refer to the “(two) factions of Niu and Li” (Niu Li [er] dang 牛李[二]黨) or “of the two Li” (Er Li dang 二李黨). Chen Yinke, who may be the inventor of the expression, uses “Niu Li dangzheng” but once;⁹⁴ more often, he refers to the “parties/factions of Niu and/or Li” (Niu Li [dang]pai).⁹⁵ Wang Tongling, in his History of Factionalism, although he chose dangzheng for the title of his book, simply refers to the “factions of Niu and Li” (Niu Li dang).⁹⁶

Conceptual historians rightly warn against any attempts to project the meaning of modern words such as minzhu 民主 (“democracy”) or guojia 國家 (“nation-state”) backwards into occurrences of such compounds in premodern texts, or to establish the modern meaning of words by reference to that of their constituent characters in Classical Chinese or Literary Sinitic (wenyan wen 文言文).⁹⁷ The meaning of a word or phrase can only be established synchronously by reference to its usage within a given language community, not diachronically by reference to an imputed “original” meaning. The occurrence of minzhu or guojia in premodern texts does not indicate, let alone prove, the existence of a concept of democracy or the nation-state in premodern China. However, readings of the premodern past by modern historians constitute a complex case of “translingual practice,”⁹⁸ as modern historians often break down the boundaries between premodern and modern language usage by, among other things, quoting extensively and without translation from premodern sources, and couching modern academic discourse in Literary Chinese grammar. For instance, when Chen Yinke uses dang 黨, pai 派, dangpai, etc., he does not distinguish between premodern “factions” and modern “parties.” Even when a modern historian such as Wang Tongling attempts to draw a distinction between the two terms, his language slides from the source language into the modern discourse of party politics and back. Rather than viewing such instances of “translingual practice” as slips of the pen or terminological inaccuracies, we may more fruitfully read them as conceptual work (“Begriffsarbeit”) through which modern historians try to integrate the knowledge structures of premodern sources into the emerging academic discourse of modern China.

To complicate matters further, “Song historiography” (Zhao Song shijia 趙宋史家) or “New Song Learning” (Xin Songxue 新宋學)⁹⁹ cast a long shadow on the twentieth century, which often replaced “Confucian” frameworks with Western models of social forces or classes. The prefaces that Chen Yinke wrote for his Draft Exposition of the Political History of the Tang give a glimpse into the influence that Yuan Shu's Topical Narratives and other works had on twentieth-century Tang history. First published by the Commercial Press 商務印書館 in Chongqing 重慶 in 1943, today's book is a reconstructed version that Shao Xunzheng 邵循正 (1909–1972) collated from an earlier draft, to which Chen added a new preface, dating to 1942. The copy Chen mailed from Hong Kong to Shanghai for publication in 1941 was lost in the turmoil of the Second Sino-Japanese War. In 1988, Shanghai Ancient Books Publishing House 上海古籍出版社 printed the facsimile of a handwritten Tangdai zhengzhi shilüe gao 唐代政治史略稿; in the preface, Jiang Tianshu 蔣天樞 (1903–1988), a former student of Chen's, claims that this is the “clean copy” (qingxie gao 清寫稿) that the Publishing House had received from the Shanghai banker Wang Jianshi 王兼士 (1905–1995).¹⁰⁰ The middle parts of the two prefaces differ markedly. The preface of the published 1943 version reads:

吾國舊史多屬於政治史類，而資治通鑑一書，尤為空前傑作。今草兹稿，可謂不自量之至！然區區之意，僅欲令初學之讀通鑑者得此參考，或可有所啟發，原不敢謂有唐一代政治史之綱要，悉在此三篇中也。¹⁰¹

Our country's old histories mostly belong to the genre of political history, and the Comprehensive Mirror to Aid in Government is its unprecedented masterpiece. Scribbling the draft before you today can be called the utmost of not knowing one's limitations! However, in my modest intention, I only wish that beginning readers of the Comprehensive Mirror use this as a reference work and perhaps gain a few insights from it. In fact, I do not dare to claim that the essentials of the political history of the Tang are all contained in these three chapters.

Compare this to the 1941 preface in the clean copy:

[⋯⋯]僅欲於袁機仲書中增補一二條目，以便初學，而仍恐其多所疏誤，故付之刊佈，以求並世學者之指正，本不敢侈言著作也。[⋯⋯]¹⁰²

[…] I only wish to supplement one or two clauses to the book of Yuan Jizhong 袁機仲 (i.e., Yuan Shu) to help the beginning student, and I still fear that there are too many omissions and errors, so in sending this to the publishers, I seek corrections from my peers and, in fact, do not dare to pretend this to be a [finished] work.

Bian Senghui recalls how, in a class Chen taught in 1935, his teacher criticized people of the present (jinren 今人), who liked to read Yuan's Topical Narratives because they believed it would conform to Western scientific methods (Xiyang kexue fangfa 西洋科學方法), and neglected its source, the Comprehensive Mirror. That, to Chen, was a mistake, because the sole basis for Yuan's decisions to copy and paste events from the Mirror into one topical narrative were “the questions that his (subjective) mind brought forth” (xinzhong suo chansheng de wenti 心中所產生的問題); moreover, Yuan frequently (though not always) omitted connections between events that occurred in more than one narrative.¹⁰³ Chen's reference to “people of the present” may be a covert criticism of Liang Qichao 梁啟超 (1873–1929), or of historians who wrote “new histories” (xinshi 新史) in the style Liang had envisioned: In 1902, Liang praised Yuan Shu's contribution to the evolution of China's “old histories” (jiushi 舊史), as Topical Narratives was the only work that resembled the genre to which most Western histories (Xishi 西史) belonged. However, Liang also lamented that Yuan had not sought out the causal relations (yuanyin jieguo 原因結果) between events, and Narratives could therefore only be seen as an expedient means (fangbian famen 方便法門) for accessing historical information, and as a “vassal” (fuyong 附庸) to the Mirror.¹⁰⁴ Whatever caused Chen's change of mind in altering the preface to his Political History of the Tang, or whatever he meant by “supplementing” (zengbu 增補) Yuan's book, his reading of Tang factionalism remained indebted to the framework laid down in the Song.

Chen Yinke: Factionalism as Class Struggle

The most influential twentieth-century theory of the Strife, which also serves as a bridge between premodern and modern historiography, is Chapter 2 of Chen's Draft Exposition of the Political History of the Tang, titled “Political Revolutions and Faction/Party Divisions.”¹⁰⁵ This and the first chapter of the book, “The Clans of the Ruling Class and their Rise and Fall,”¹⁰⁶ although they treat different topics, also cut the Tang in half chronologically, as the first chapter gives pride of place to the first half up to the An Lushan Rebellion, while the second privileges the second half. This is testimony of Chen's conviction that the reigns of Empress-regnant Wu 武后 (Wu Zetian 武則天, r. 684–705), and Xuanzong 玄宗 (Li Longji 李隆基, r. 712–756) constitute a watershed. He views the sixth-century court as dominated by a Guanlong Clique (Guanlong jituan 關隴集團),¹⁰⁷ which included the imperial clan and followed the “Guanzhong First Policy” (Guanzhong benwei zhengce 關中本位政策) established by Yuwen Tai 宇文泰 (505–556) under the Western Wei 西魏 (535–557) and Northern Zhou 北周 (557–581), and continued by the Sui 隋 (581–618) and Tang. In spite of the title of the book, Chen places a strong focus on ethnocultural (minzu wenhua 民族文化) rather than political history in the first chapter.

Chen explains the political history of the Tang as a succession of power struggles: first between the “Guanlong clique of mixed Han-barbarian stock” (Guanlong Hu Han hunhe jituan 關隴胡漢混合集團)¹⁰⁸ and an older group or aristocracy of “bureaucratic clans from Shandong” (Shangdong shizu 山東士族) centered on Shandong, Hebei 河北, and Henan 河南; then, after the decline of the former in the wake of Wu Zetian's reign, between the latter and newly-risen classes (xinxing jieji 新興階級) of examination degree holders. Different races (zhongzu 種族) and classes aligned with different factions/parties. As Axel Schneider has noted, Chen relies heavily on the biographies in the two standard histories but treats their subjects as representatives of social or ethnic groups rather than autonomous agents.¹⁰⁹ Accordingly, Chen treats the Niu and Li factions as representatives of different classes within the elite.

According to Chen, Wu Zetian and her successors curtailed the influence of the Guanlong clique by relying on the literary examinations, in particular the “Advanced Scholar” (jinshi 進士) degree, to fill offices. That flushed fresh blood into the bureaucracy, which differed from the aristocracy not only in terms of social class but also, and more importantly, in terms of culture: while the aristocracy maintained strong family traditions of learning in the classics and ritual, the nouveaux lettrés concentrated on the Selections of Refined Literature (Wenxuan 文選) and other more recent texts to prepare for the Advanced Scholar examinations, which required the composition of poetry:

即唐代士大夫中其主張經學為正宗、薄進士為浮冶者，大抵出於北朝以來山東士族之舊家也。其由進士出身而以浮華放浪著稱者，多為高宗、武后以來君主所提拔之新興統治階級也。¹¹⁰

Thus, those among the Tang scholar-officials who advocated the study of the classics as orthodoxy and disdained the Advanced Scholar [degree holders] as ostentatious wordsmiths by and large descended from the old families of the bureaucratic clans from Shandong of the Northern Dynasties. The majority of those who graduated as jinshi and became famous through their ostentatiousness and frivolousness were newly-risen ruling classes selected by the rulers since Gaozong (Li Zhi 李治, r. 649–683) and Empress Wu.

Chen concedes that it would be impossible to separate those two classes entirely, because there were degree holders among the aristocracy as well as families of newly-risen bureaucrats who held on to power over successive generations. Yet their differences rose to the surface in the Strife of the Niu and Li Factions—the latter named after Li Deyu, who descended from the Li clan of Zhao prefecture 趙郡李 in Hebei, and hence belonged to the Shandong aristocracy.¹¹¹

A key figure in Chen's twentieth-century re-narration of the Strife is Zheng Tan 鄭覃, a member of the Xingyang-Zheng clan 滎陽鄭氏 and purported partisan of the Li faction. For Chen, Zheng's biography in the Old Tang History, from which he quotes extensively, provides evidence that the majority of members of the Li faction belonged to the aristocracy.¹¹² He became chancellor under Emperor Wenzong in 835, after the “Sweet Dew Incident” (Ganlu zhi bian 甘露之變) that crushed literati opposition against the eunuchs.¹¹³ As chancellor, he tried to abolish the Advanced Scholar examination because “although Tan understood the meaning of the classics, he lacked the ability for refined compositions and despised the Advanced Scholars for their ostentatiousness.”¹¹⁴ Chen makes him a key figure in his narrative of the Strife. Not only is Zheng one of the foundations of Chen's theory of the Strife, but he becomes an oracle of a self-fulfilling prophecy, constantly suspecting others of factionalism. In 838, when officials were discussing extraordinary rewards of officials that were not commensurate with their merit, he caused a flurry:

鄭覃曰：「陛下須防朋黨。」嗣復曰：「鄭覃疑臣朋黨，乞陛下放臣歸去。」因拜乞罷免。李珏曰：「比來朋黨，近亦稍弭。」覃曰：「近有小朋黨生。」帝曰：「此輩凋喪向盡。」覃曰：「楊漢公、張又新、李續之即今尚在。」珏曰：「今有邊事論奏。」覃曰：「論邊事安危，臣不如珏；嫉惡則珏不如臣。」嗣復曰：「臣聞左右佩劍，彼此相笑。臣今不知鄭覃指誰為朋黨。」因當香案前奏曰：「臣待罪宰相，不能申夔、龍之道，唯以朋黨見譏，必乞陛下罷臣鼎職。」上慰勉之。文宗方以政事委嗣復，惡覃言切。¹¹⁵

Zheng Tan said: “Your Majesty must be on Your guard against factions.” [Yang] Sifu replied: “Since Zheng Tan suspects me of factionalism, I beg Your Majesty to dismiss and allow me to return home.” He kneeled and begged for his dismissal. Li Jue¹¹⁶ interfered: “The factions of the [Yuanhe period] have waned in recent times.” Tan said: “As of late, small factions have emerged.” The emperor said: “That lot has almost passed away completely.” Tan said: “Yang Hangong,¹¹⁷ Zhang Youxin, and Li Xuzhi are still there today.”¹¹⁸ Jue said: “Today, we have petitions on border affairs to discuss.” Tan said: “In discussing border security, I am no match to Jue; in despising wrongdoing, Jue is no match to me.” Sifu said: “I have heard that ‘the imperial entourage mock each other while they carry swords under their cloaks.’¹¹⁹ I am ignorant of whom Zheng Tan just now referred to as forming a faction.” Thereupon, he stepped in front of the incense stand and pleaded: “I brought shame on [the office of] grand councilor. Unable to explain the ways of Kui and Long,¹²⁰ I only face ridicule of [forming] a faction. I must beg Your Majesty to release me of my supporting duties.” His majesty consoled and encouraged him. Wenzong from then on entrusted Sifu with government matters and loathed Tan for his sharp words.

Chen makes this passage from the Old History's biography of Yang Sifu the centerpiece of his theory of the Niu–Li Strife as a struggle between different classes or strata within Tang officialdom. Yet, despite the association of Yang with Niu and Zheng Tan with Li Deyu, pengdang here does not refer to Niu and Li but to the Yuanhe period of Emperor Xianzong, twenty years in the past. Zheng's accusation is not aimed at any specific faction in the present but serves, in the words of Ari Levine, to claim “ethical superiority”¹²¹ over those he wishes to accuse of cronyism. Here, in the tenth-century Old History, we may see the first attempt at drawing a direct connection between the Yuanhe period and factions of succeeding decades. It is only later that this connection became modelled on the reform and anti-reform factions of the Northern Song.

One major problem of Chen's theory is his reticence in giving explicit definitions of terms such as party/faction, class, revolution, etc., which he uses inconsistently and whose meaning therefore remains fuzzy. He postulates a cyclical or helical pattern of progressive struggles between various factions and cliques belonging to different social, ethnic, or cultural groups, each time culminating in a revolution. For his inconsistent use of Marxist vocabulary, Chen was severely criticized during the Anti-Rightist Campaign of the late 1950s.¹²² The fiercest attack, however, came from a fellow Tang historian: Cen Zhongmian, in his History of the Sui and Tang, pointed out weaknesses in Chen's thesis that the Niu and Li factions were divisible into old and new classes based on their attitude towards the examinations, classical learning, and literature.¹²³ He showed that people in both groups had been descendants of old clans and holders of examination degrees at the same time. The Advanced Scholar, which had only thirty graduates each year, was coveted by both aristocrats and social climbers alike. Cen even claimed that Li Deyu had been factionless, and that Sima Guang had praised the cautious Niu over the daring Li in his assessment of the Weizhou affair to express his opposition to the military policies of Wang Anshi under Shenzong—a point premodern historians had raised already.¹²⁴ Following the tradition of Confucian historians to mete out praise and blame, Cen laid down the accusation of factionalism at the feet of Niu and Li Zongmin.

Historians have largely abandoned Chen's schematic view of the stratification of mid- and late Tang factions, and his juxtaposition of aristocrats versus new elites based on whether they entered officialdom via hereditary privilege or the civil service examinations. The main thrust of critique came from social history: research into the composition of the elites based on paleographic and epigraphic evidence has shown that officials are not as schematically divisible in terms of social background as Chen believed. Although great clan status was transitory, there was no discernable increase in number of non-great clan elites within the bureaucracy in the second half of the Tang. Instead, the great clans showed an astounding ability to highjack the examination system for their purpose and maintained their status as ruling elite until the very end of the dynasty.¹²⁵

Wang Tongling: Factionalism as Cyclical Pattern

With that, let us turn to Wang Tongling, whose influence on modern Chinese historiography seems obscure when compared to Chen's, but who nevertheless left a substantial œuvre of linear histories of China and East Asia that went through several editions in the 1920s and 1930s.¹²⁶ As a professor at Beijing Normal University 北京師範大學, Wang participated in the training of school teachers and the formation of the history curriculum in Republican China.¹²⁷ Having graduated from Tokyo University in 1912, he also published in and translated from Japanese.¹²⁸ The overlap with Chen's research, especially in the fields of ethnic and factional history, is remarkable, as is the fact that Chen never refers to Wang's work or vice versa.¹²⁹ Although his language seems archaic, even in comparison with Chen's,¹³⁰ and his framework, just like Chen's, went out of fashion once Marxist historiography became the new orthodoxy after 1949, Wang, like Chen, serves as a bridge between traditional and modern approaches to factionalism. In spite—or because—of that, his impact on the Republican academic and public discourse about history remains unstudied.

In 1922, Wang published A History of Factionalism in China through the Ages, in which he elaborates on the “Conflict of the Niu and Li factions after the mid-Tang” (“You Tang zhongye yihou Niu Li dang zhi qingzha” 有唐中葉以後牛李黨之傾軋).¹³¹ While the younger Chen Yinke understood factionalism as conflicts between classes, whose causes he sought in the social and ethnic composition of the Tang elite, Wang sought the root of instability of China's contemporary political system in the recurrent appearance of factionalism that ran through the entire premodern history of China in a cyclical pattern. This pattern always threw China back into the maelstrom of factionalism. Wang identifies ten “sources of chaos” (luanyuan 亂源) in imperial China, of which factions (pengdang) are but one; the other nine are befuddled, violent rulers (renzhu hunbao 人主昏暴), consorts arrogating power (nühou shanquan 女后擅權), female favorites (nüchong 女寵), eunuchs (huanguan 宦官), imperial in-laws (waiqi 外戚), powerful ministers (quanchen 權臣), warlords (fanzhen 藩鎮), popular unrest (luanmin 亂民), and foreign countries (waiguo 外國). It reads like a list of maladies that befell almost every dynastic regime toward the end of its lifecycle from a Confucian viewpoint of history. Wang postulates a break between China's monarchical past and its Republican present: the Republican form of government rendered almost all the above causes obsolete by electing a new president every few years. Only warlords and parties (dangpai substituting for pengdang) remained.¹³²

Wang divides his history into ten chapters that follow a linear chronology and discuss the forms that factionalism took in each period. He starts with the pre-imperial philosophical schools (Xian Qin xuepai 先秦學派, chapter 1),¹³³ which he calls factions, and continues with the Han (chapter 2), the Jin 晉 and Northern and Southern Dynasties 南北朝 (265–589, chapter 3), the Tang (chapter 4), Northern (chapters 5–6) and Southern Song (chapters 7–8), the Ming (chapter 9), and the Qing (chapter 10). The factionalism of each period differs from those of the preceding and succeeding periods. The Niu and Li factions, for instance, differed for Wang from those of the Eastern Han in that the latter were formed of “scholars and gentlemen” (shi junzi 士君子) who worried about state affairs (guoshi 國事) in opposition against the eunuchs, while the former were formed of scholar-officials (shi dafu 士大夫) who opposed each other in their pursuit of fame (gongming 功名)¹³⁴—which, pace Chen, may be tantamount to claiming that they belonged to the same class.

What strikes the reader in Wang's discussion of factionalism is how easily he moves back and forth between factionalism in the imperial past and the present, which distinguishes him from the strictly medieval narrative of Chen Yinke. What he shares with Chen is the constant conflation of modern vocabulary of parties and factional strife with the premodern language of factionalism. It is important to note, however, that he does not envision the succession of factional struggles as a history of social progress or evolution, but as the eternal return of the same.

謂中國政黨不足恃乎？前清民國之興亡，實政黨左右之也。謂中國政黨為足恃乎？何以歐美各國以有政黨興，吾國以有政黨衰也？噫！吾知之矣。歐美各國政黨，皆以國家為前提，有利於國家者，則犧牲黨見以殉之。我國政黨，以黨綱為前提，有利於己黨者，則寧犧牲國家以殉之。此其所以異也。[⋯⋯]然則中國政黨曷為有弊而無利？曰：中國自古為專制政體。專制政體之下，無政黨發生之餘地。其有類乎政黨者，則東漢末年之鉤黨，有唐中葉以後之牛李黨，唐末之清流黨，北宋之元祐黨、熙豐黨，南宋之偽學黨，明末之東林黨、閹黨，皆敵黨加以黨之名，自己並不承認為政黨也。¹³⁵

Can we say political parties in China cannot be relied upon? The fall of the former Qing and rise of the Republic were, in fact, brought about with the support of parties. Can we say that political parties in China can be relied upon? Then why is it that in Europe and America, each country rose with the existence of political parties and our country decays with them? Ah, I know! In Europe and America, each country's political parties make the nation their premise. Whatever is beneficial to the nation, they sacrifice their party views to and are willing to die for it. The political parties of our country make the party program their premise. If there is something profitable for their own party, they would rather offer up the nation as a human sacrifice. That is wherein they differ. […] Then why do political parties in China cause harm and not benefit? It is because China had an autocratic government since ancient times. Under an autocratic government, there is no room for the development of political parties. Those that had something in common with political parties were the proscribed factions of the Eastern Han, the Niu and Li factions after the mid-Tang, the pure stream factions at the end of the Tang, the Yuanyou and Xifeng factions of the Northern Song, the False Learning factions¹³⁶ of the Southern Song, and the Donglin and eunuch factions at the end of the Ming. In all these cases, the factions of their enemies attached the name “faction” to them, they did not at all admit that they were political parties themselves.

Wang spent nearly half of his life—thirty-three years—under imperial rule and even took part in the imperial examinations. His views on factions remain indebted to the moralizing Confucian discourse of premodern China. That discourse saw factions arising when the government was not run by high-minded gentlemen whose loyal pursuit of public interest rooted in the Way (Dao 道) and, who by their very nature (see Ouyang Xiu), could not form factions, but by petty men who pursued their own, small-minded and self-serving interests. At the same time, Wang tried to move beyond that discourse by identifying patterns behind the repeated rise and fall of factions that were rooted in geography, social psychology, and the conservative character of the Chinese. Not only did he try to explain the poor state of China's current political parties by the historical experience of factionalism in the past, he also made predictions for China's future, which he saw in a rather dim light because, in his view, his compatriots were held back by that experience from developing a modern and enlightened party system.

Conclusion

Over the past century, historians have reframed the historical narrative of China in a developmental mode to serve the needs of a modern nation. They identified the mid-eighth to the mid-eleventh centuries as a transitional period from medieval to modern society, which had to be accommodated in any linear model. To achieve that goal, they had to fall back on the new vocabulary that became available to historical research in the nineteenth and early twentieth centuries, and which included concepts such as class and party. They re-narrated the history of the mid- and late Tang by dividing the official elite into two distinct groups or classes under the labels of “parties” or “factions of Niu and/or Li” and the “Niu–Li Factional Strife.” They based their narratives on inconclusive evidence scattered over near contemporary sources, building on foundations that were laid under the Song. In many ways, modern historical accounts of the Tang still reflect Song readings of the Tang. Yet, there are also major differences between modern accounts and those of their Song precursors. In their retellings, modern historians often conflated the language of their medieval sources with the modern discourse on party politics and class struggle. While traditional sources dealt with individual moral exemplars rather than with social groups or historical patterns, modern historians tried to rewrite the past in developmental terms, to show how China's ethnic and class structure had evolved over the centuries, how the factionalism of the imperial past could be overcome in a modern party system, or how China's present was held back by its past. The presence of factions at the court and in the bureaucracy of the mid- and late Tang based on clan affiliations and personal networks spawned by the examination system is hardly deniable. However, pressing them into two or more distinct groups or classes, which helped historians in the past to reconfigure history in a developmental mode, today constitutes an impediment rather than a benefit to understanding the complex social and political dynamics of medieval China. Research into the social composition of the Tang elites based on the epigraphic record, which is much more abundant and readily available now than it was at the time of Chen and Wang, has helped tracing out a more finely grained picture of the complex socioeconomic and sociocultural situation under the Tang. Nonetheless, rather than judging the theories of Chen, Wang, and others as failures, we should see them as part of the never-ending process of rewriting history. By doing so, they contributed to a narrative that, for better or worse, endures to this day.