Status, Power, and Punishments: “Household Workers” in Late Imperial China

Abstract In the past four decades or so, China scholars have shone a new light on the history of labour in late imperial China, particularly on the role of the household as a unit of production and on the contribution of women to commercial production and family income. Beyond members of the kin group itself, attention is seldom paid to the individuals brought into the Chinese households solely to provide additional manpower. To “break the carapace” of the late imperial Chinese household, this article focuses on the often-omitted “household workers”, that is, on its enslaved (nubi) and hired (gugong) constituents. It approaches the topic from the angle of the vulnerability of these non-kin “workers” to punishments and violence. To evaluate their vulnerability to punishment and gauge the disciplinary powers of the household heads, it examines the relationship between punishments and “household workers” in Ming law. It then explores lineage regulations, before moving closer to the ground by mobilizing a wider variety of day-to-day sources, such as contracts and narrative sources produced in the context of the late Ming and early Qing crisis.

explanation (the topic used to be widely studied by Chinese and Japanese historians). 8 A better explanation might be the lack of even approximate figures and lack of consensus about the numerical importance of servile labour and its evolution. Some historians have estimated that servile labour decreased from the Ming (1368-1644) and finally disappeared during the eighteenth century. 9 Evidence nonetheless points to an increase in the enslaved population during the Ming (up to a few per cent of the whole population in the 1640s, according to my estimates). 10 It also points to a hardly measurable decrease from the eighteenth century onwards (not a complete disappearance). 11 This article intends to "break the carapace" of the late imperial Chinese household a little more by shining light on the often-omitted slaves and other "household workers". I approach the topic from the angle of the vulnerability of non-kin "workers" to punishments and violence. The term "household workers" is not a contextual (emic) category. I use it to refer to two distinct legal groups: nubi (enslaved) and gugong (hirelings). Everyone worked (or was expected to work) in the Chinese household, including women whose domestic and productive labour ideally took place "inside" (nei), while men worked "outside" (wai). 12 Although everyone worked, not every outsider was brought into a household as a worker. Wives (qi), concubines (qie), uxorilocal husbands (zhuixu), and adoptees ( yinan) were undoubtedly incorporated into the kin group for their ability to work. Still, their primary function was to ensure the perpetuation of the agnatic descent line. Nubi and gugong, on the contrary, were brought into the household to provide labour, not to integrate the kin group.
Gender certainly affected the lived experiences of household workers in various ways. As disposable and marketable assets, women may have been more numerous among the enslaved than among hirelings (considering that the latter worked "outside" their own households). The fact that household heads had both a moral duty to give female slaves into marriage when coming of age and, until the eighteenth century, legitimate access to their bodies for sex and reproduction also led to differentiated life and work cycles. 13 Yet, the gender dimension seldom surfaces in the sources used in this article, where nubi and gugong are mainly considered as uniform social and legal groups.
To evaluate their vulnerability to punishments and to gauge the disciplinary powers of the household-heads, I first examine the relationship between punishments 8 Besides numerous articles and book chapters published since the 1920s, two major reference works were published in the last decades of the twentieth century: Wei Qingyuan, Wu Qiyan, and Lu Su, Qingdai nubi zhidu (Beijing, 1982); Jing Junjian, Qingdai shehui de jianmin dengji (Hangzhou, 1993). and "household workers" in Ming law. Then, I explore its transposition into lineage regulations. Finally, I move closer to the ground by mobilizing a wider variety of day-to-day sources, such as contracts and narrative sources produced during the late Ming and early Qing (1644-1911) crisis.
Although I focus on the late Ming to early Qing period (when the enslaved population was at its highest and before the eighteenth-century reconfiguration of status laws), the sources I use cover the whole late imperial period (i.e. the Ming and Qing dynasties). 14 The Great Ming Code (my main source in the first section) was promulgated in the last decade of the fourteenth century, but it was still very much in use in the seventeenth century. Similarly, many of the genealogical records used in the second section were printed during the Qing era but contain regulations written in the Ming. Although the sources do not always allow term-to-term comparisons of the two groups, they open different windows on their relations to punishments.

The State's Perspective: Enforcing Status and Empowering Masters
The Great Ming Code (Da Ming lü) was promulgated in its final version in 1397. Since the first Ming emperor (Hongwu, r. 1368-1398) had prohibited the alteration of dynastic laws, by the end of the dynasty the Code was out of step with social and economic realities. 15 Although it merely unveils an ideal(ized) vision of the late Ming social order, it nonetheless gives us a sense of the state ideology of the time. It also allows us to outline the contours of always implicit legal statuses and social categories, to seize the state-promoted conceptions of social dynamics of power, and to highlight the role of punishments in the control of "household workers".

Recruitment and Punishments
At the state's level, punishment(s) played various roles in relation to "household workers". First, punishment was, in theory, the only legitimate path towards enslavement, alongside capture in war, despite the well-documented proliferation of private enslavement through (self-)sales and abductions. Relatively silent about enslavement, early Ming official sources suggest that, as a group, nubi were to remain numerically marginal. They also show that the state legally recognized enslavement and intended to maintain a firm monopoly on the production of nubi through punishment. 16 Two centuries after the fall of the dynasty, the author of the Essentials of the Ming Institutions (Ming huiyao) still presented enslavement in Ming times as the result of punishments meted out to the relatives of major criminals: "In the Ming system, the children of war captives (fuhuo) and the persons seized (chaomo) as [relatives of] criminals were awarded as nubi to the families of meritorious officials (gongchen Besides containing tax evasion, prohibiting private enslavement was meant to preserve the agrarian and self-sufficient social order envisioned by the founder of the Ming in which the population was hereditarily bound to their land and occupations and laboured for and by themselves. 18 As expounded by jurist Gao Ju in the early seventeenth century, the people "must fundamentally work hard and to the best of their ability. They must not own nubi. Only meritorious officials can. Commoners who harbour and raise [nubi] transgress their condition". 19 If enslavement was, in theory, a monopoly of the state, nubi ownership was, also in theory, a privilege of the state's most prominent servants. And it was the state's duty to protect the population from enslavement. 20 From a conceptual standpoint, enslavement was thus, first and foremost, a punitive regime for criminals by association, not a labour regime. Unlike nubi, which emerged as a legal category at the dawn of the imperial era, 21 the gugong category was a legal innovation of Ming law. 22 Gugong were not criminals at all. Unlike "raising" (xuyang) nubi, hiring people on a daily, monthly, or annual basis was not objectionable, except in cases of hiring others to perform one's state-imposed labour duties. 23 In the Ming legal context, gugong thus mainly referred to commoners hired to perform work for others on a time-limited and contractual basis in exchange for food and monetary compensation. 24 Punishments had nothing to do with the mobilization of hirelings.
To us, who are accustomed to regarding contractual labour relations mediated through money as the epitome of "free" and autonomous work, the difference between nubi and gugong seems as radical as the opposition between slavery and freedom. The twist, however, is that nubi and gugong were not so radically opposed in Ming law. Despite entirely different recruitment processes and degrees of social autonomy, within the boundaries of their relations with their master/employer, nubi and gugong were treated as cognates. They appeared in the same legal statutes. They were governed by the same legal regime of punishments derived from the parent-child relationship by analogy, placing them in a similar relation of subordination to the authority of the household head.

Nubi and the Analogy with Children
The punishments prescribed by the Great Ming Code were not solely designed to punish the crimes and offences committed by nubi and gugong. the level of power conferred upon masters and employers by the state, as well as the protection afforded to that power.
In its most basic expression, the Great Ming Code is an elaborate grid of correspondence between incriminations and punishments. Each incrimination corresponds to a reference punishment. Each punishment could then be modulated in aggravating or extenuating circumstances. As an example, "assaulting" (ou) others "with hands and feet" without causing injuries was liable to ten blows of a chi (light stick), whereas assaulting others "with other tools" (aggravating factor) was liable to thirty blows. 25 In addition, punishments were further modulated according to the relative position of the offender and the victim. Relational asymmetries shifted from one person to the next and derived from status and/or relative position within the family and household hierarchy. Status was the main mitigating factor when the offender and the victim were unrelated (i.e. when they belonged to different households, families, and lineages). Several social groups (like officials and members of the imperial clan) enjoyed legal privileges in the Ming system, but "debased people" ( jianmin, of which nubi were the main component) were systematically discriminated against in relation to commoners (liangmin, or "honourable" people). A commoner who physically assaulted an unrelated nubi was sentenced to the reference punishment for "assaulting others" minus one degree. In contrast, a nubi who struck an unrelated commoner was sentenced to the same reference punishment increased by one degree. 26 When the offender and the victim were relatives or lived in the same household, the criteria of gender, age, rank of birth, generation, and ritual proximity combined to produce context-specific relational asymmetries. Expressed in degrees of mourning, ritual proximity acted as a catalyst of pre-existing hierarchies based on inferiority/juniority (beiyou) and superiority/seniority (zunzhang). 27 Assaulting one's father or mother (seniors of the first degree of mourning) or one's paternal grandparents (seniors of the second degree of mourning) was a far more serious offence (punished with death by decapitation) than assaulting outsiders to the household. Striking one's children ( juniors of the second degree of mourning), on the contrary, was without legal consequences so long as it did not cause death. 28 As the closest, most vertical and asymmetric of all family bonds, the parent-child relationship also served as a major reference point to frame nubi status and to enforce their subordination to the household head ( jiazhang). The nubi-children analogy is ubiquitous throughout Ming normative and moral sources. Its ubiquity does not mean that nubi were socially equal to their master's children. Framing the masternubi relationship in such paternalistic terms nonetheless permitted borrowing from the most asymmetric relation of all. A few moralists, especially in the late Ming period, took this analogy at face value to outline the contours of a genuinely reciprocal, "benevolent", and "humane" relationshipeven sometimes to question the legitimacy of enslaving one's fellow human beings. To many others, however, the fatherly 25 Gao, Ming lü jijie fuli, 20:1a. 26 Ibid., 20:22a. 27 Hsieh, Concubinage and Servitude, pp. 46-47. 28 Gao, Ming lü jijie fuli, 20:38b.
"benevolence" expected from masters was no more than a way to legitimize enslavement in Confucian terms. 29 Principles of reciprocity and benevolence are barely mentioned in dynastic law. The analogy with children is nonetheless constantly and selectively mobilized to narrow the range of identities that nubi could assume vis-à-vis others. It was used to frame a specific regime of punishments that, in all circumstances, placed nubi under the direct control and exclusive authority of the household head and his closest kin. For instance, like other household members, nubi were required to provide asylum to and conceal the crimes of those who "lived together" (tongju) without fear of prosecution. However, they were denied such privilege themselves. In the same manner, whereas junior and inferior members of the household could not be accused of "theft" (dao) in a legal sense (all personal belongings were considered part of the collective property), nubi who stole from household members were prosecuted for theft. 30 Ming law thus conveniently excluded nubi from genuine household membership when access to privileges and property was at stake. At the same time, it assimilated them with children in imposing the specificities of the most asymmetric hierarchical relation. 31 As an example, according to the statute on the desecration of graves, children and nubi were the only ones to face the death penalty for damaging the corpse of a deceased parent or master when "smoking out foxes" (xun huli) on their graveyard. 32 According to Ying Jia (1494-1554), the severity of the law reflected the specificity of the relationship between (grand)children and (grand)parents, which, by extension, applied to nubi. 33 Similar examples of the analogical use of the parent-child relationship are found in the three statutes dealing specifically with the offences and crimes committed by nubi against the household head and his close relatives (statutes on fornication, insults, and assaults). 34 When comparing the punishments meted out to nubi and children, we observe that for serious offences, nubi were treated like children. For striking, causing death, or intentionally killing one's parent, grandparent, or master, all were sentenced to the same forms of the death penalty. However, for lesser crimes, like injuring or killing "by mistake", nubi were sentenced to slightly more severe punishments. Those differences were usually explained by the difference in nature inherent in the master-nubi and parent-child bonds: children were considered as naturally more inclined to be "respectful and cautious", whereas nubi were perceived as being of a different "kind" (lei) and more "neglectful" by nature. 35 A close comparison between the regimes of punishments of children and nubi also shows that the analogy reaches a limit when we move away from the direct master-  nubi relation. With very few exceptions, children were always liable to lesser penalties than nubi when the victim was a more distant relative. For instance, children who injured their relative of the second degree of mourning were sentenced to three years of penal servitude. In contrast, nubi who injured their master's relative of the second degree of mourning were sentenced to death by decapitation. 36 Thus, unlike children whose relative identity changed according to their direct relation to others in the family, nubi's relations to others in the household were always indirect and mediated through the fatherlike figure of their master (leading to generally harsher punishments).
The analogy between nubi and children was thus a malleable and utilitarian legal fiction. It extended to nubi the almost absolute authority that parents exercised over their children so that nubi were permanently relegated to the lowest levels of the household hierarchy (in addition to being relegated to the lowest levels of society as "debased people"). The punishment matrix so created was also a powerful institutional machinery that empowered masters and protected their paternalistic authority against the transgressions and challenges of the enslaved. By threat or by actual referral to the judicial authorities, masters could at all times reassert their authority over enslaved people who, like children, could not appeal to the courts against them. 37

Disciplining Nubi and Gugong
The pending question is that of the extent of power vested in masters/employers by the state. Although not clearly outlined, its contours can be inferred from the content of the three above-mentioned statutes on crimes committed by nubi against the household head and his close relatives, which also extended to gugong.
As seen above, the statute on assaults ("Slaves striking the household head", no. 337) was ruthless against nubi and gugong. 38 Its ruthlessness was nonetheless perceived by jurists as central to the defence of hierarchies, as it not only "addresse[d] affrays between masters and nubi" but also "[kept] the world running by distinguishing between statuses and by correcting denominations". 39 In other words, its centrality lay in the defence of the authority of the household head since "correcting denominations" (zhengming) meant, in Confucian terms, ensuring the proper correspondence between "the names" and "the substance" of social roles. 40 After enumerating the punishments prescribed against disobedient and renegade nubi and gugong, statute no. 337 addresses situations where masters and employers (and their relatives of the first two degrees of mourning, who shared their disciplinary power) struck and killed their subordinates. With regard to nubi, the law only considered cases of homicide. It distinguished between accidental death in the process of legitimate disciplining, the intentional killing of "guilty" nubi, and the arbitrary killing of "innocent" nubi. The intentional killing of "guilty" (youzui) nubi was liable to a moderate sentence of 100 strokes. The issue was not that masters could beat their 36 Gao, Ming lü jijie fuli, 20:25a. nubi, but that in the process they "arrogate[d] the right to beat and kill" (shanzi sizi ousha), which was a monopoly of the state. 41 The "arbitrary" (feili) killing of innocent nubi was liable to a more severe sentence of one year of penal servitude (plus the emancipation of the family of the deceased enslaved). 42 "Accidental" or "unexpected" (xiehou) death resulting from "legitimate disciplining" (yifa juefa) of "disobedient" (weifan jiaoling) nubi, however, was not liable to prosecution. According to Gao Ju, "legitimate disciplining" meant beating on the buttocks. 43 In other words, although the law had a whole arsenal of provisions tailored to address crimes and offences committed by nubi (from insulting to killing one's master), the state granted masters almost unlimited disciplinary powers against disobedient and unruly behaviours, so long as they did not cause death.
As to gugong, the law neither considered homicides nor distinguished between innocent and guilty gugong. The principal line of demarcation was that which separated legitimate discipline from damages to the body. Like masters, employers could not be prosecuted for killing gugong "accidentally" or "unexpectedly" in the process of legitimate discipline. Nor could they be punished for simply "beating" gugong. However, they faced severe punishment if they beat and caused "fractures or worse" (zheshang yishang). For instance, death resulting from a beating was liable to three years of penal servitude, whereas "deliberate" killing was liable to death by strangulation. 44 The different levels of protection granted by the state to gugong and nubi show that, despite being legally akin and subjected to analogous regimes of punishments, there was a major legal difference between the two categories. Unlike nubi, gugong were not permanently "debased" and cast out of the society of "honourable" people. Their absence from the statutes addressing crimes and offences between unrelated "mean" and "honourable" people shows that the socially demeaning bearing on nubi applied to gugong only within the boundaries of their employer's household and for the duration of their employment. 45 Inside their employer's household, gugong were considered "inferiors and juniors". Like nubi, they could be prosecuted for "theft" and were required to conceal crimes committed by "seniors and superiors" without benefiting from the reciprocal privilege. In some cases, they were liable to slightly lesser sentences than nubi. But when subordination to the authority of the household head had to be emphasized (as in cases of crimes with strong ritual significance like the desecration of their employer's buried body, premeditated murder, and illicit sexual relations), they received the same punishments as nubi and children. 46 As underlined by Zhang Kai (1398-1460), inside the household, "gugong do not compare to nubi, […] yet, both are mean dependents" ( jianli zhi tu). 47 41 Feng, Da Ming lü jishuo fuli, 8:23b. According to the same author, "guilty" meant committing a crime punishable under the law. 42 Gao, Ming lü jijie fuli, 20:28a. 43 Ibid., 20:28b. 44 Ibid., 20:26a. 45 Chevaleyre, "Serving and Working for Others", p. 179. 46 Gao, Ming lü jijie fuli, 18:61b-63b, 19:5b, 25:11b. To summarize, modalities of entry into a "labour" relation, duration, and conditions of remuneration were not prominent factors in determining the level of power that "employers" were entitled to exercise over "workers". The key factor that made gugong akin to enslaved people was that they worked for others, not themselves. To benefit from others' "benevolence and solidarity" (enyi, in the form of "wages"), they were called upon to "serve and be commanded" (yishi). 48 Legal commentators of the Ming and Qing era all recognized that to receive "a [money] price" (guzhi) in exchange for performing "service" (yi) was socially different from being enslaved and serving for a lifetime (zhongshen). 49 However, to "serve" (which in practice meant to obey orders) was the starting point of a process of entrenched hierarchical differentiation that required complete obedience and subordination. To ensure that social hierarchies, roles, and order were preserved, the state placed serving people (hirelings as much as enslaved) under the indisputable and exclusive authority of the household heads, granted them extended disciplinary powers, and protected their paternalistic power with an arsenal of highly asymmetric legal punishments.

The Lineage Perspective: Paternalism and Discipline
In addition to "universal" dynastic law, the corporate lineages that flourished in the Ming also drafted their own regulations "to govern their houses" (zhijia). 50 Ordering one's house was not simply a matter of family organization. It was a political duty that contributed to maintaining the broader social order. But it was also a vital means for patrilineal descent groups to ensure their self-governance, protect their assets, and curtail state interferences with their affairs.
Included in registers ( jiapu, zupu) that lineages printed at great cost and kept secret, these regulations touched upon many aspects of lineage activity (such as the management of collective assets, the education of the younger generations, the allocation of lineage subsidies, the organization of rituals and solidarity, hierarchies, and the observance of proper behaviours). 51 As tools of labour and social management, these regulations also shine a light on household-level control mechanisms, on the perceptions that lineage authorities had of their power, and on the anxieties that disciplinary methods were meant to address.

Lineage Regulations
Chinese lineage registers are usually composed of a wide variety of documents (genealogic tables, prefaces, deeds, maps, etc.), some of which had a significant normative value. "Admonitions" ( jiaxun) and "regulations" (zugui, zonggui, jiafa, etc.), in particular, were designed to prescribe and correct individual behaviours and thereby preserve harmony. 52 If the two genres conflate somewhat in content and purpose, admonitions generally consist of didactic texts written by illustrious ancestors, whereas regulations usually take the form of structured lists of rules and prohibitions (sometimes modelled after the penal code) to be used in the administration of lineage justice.
Not all Chinese families could afford to structure themselves in elaborate ways. Compiling documents, writing one's lineage history, and engraving woodblocks were resource-consuming tasks. Printing registers was thus a marker of social standing and an expression of adherence to state-sponsored Confucian values. The standards and norms that these texts unveil are thus mainly those of the propertied Confucian elite.
As their authors often underscored, lineage regulations had a function like that of dynastic law: "The realm has legal statutes ( falü). Families have admonitions and regulations (xungui). Both are the same. How can the realm be governed when statutes are not enforced? How can households be ordered when regulations are not enforced?" 53 Lineage justice was thus not a mere substitute for imperial justice, but an internal and lower-level body of dispute resolution designed to maintain order and prevent, as much as possible, the involvement of the imperial administration. As stressed in the previous section, the state protected the authority of the lineage-and household heads and invested them with powers to control and discipline their kin and non-kin subordinates. On occasion, lineage regulations were also granted binding power when stamped by the administration or endorsed by the emperor. 54 Like dynastic normative sources, lineage admonitions and regulations seldom provide details about concrete and potentially unflattering daily practices. "Household workers" are far from being present in every lineage register and are far less present than members of the kin group, particularly children whose education was their principal concern. 55 A first observation can nonetheless be made from the relative frequency of "household workers" in those texts. Gugong are seldom mentioned and mainly surface in texts dealing with activities that required additional labour input and generated costs to be paid by the lineage (in food and money allocations), like agriculture, construction works, and ritual activities. 56 Thus, it seems that despite the ambiguous status of gugong as temporary and partial insiders to the household, their presence and management were not a significant concern for the lineages.
Nubi are slightly more present in lineage registers. They are the subject of more detailed and varied prescriptions. Like the state, lineages strictly restricted the acquisition of nubi to only one supply source. To them, nubi were not criminals condemned by justice but outsiders who voluntarily sold themselves or their offspring "because of cold and hunger" ( jihan). 57 In keeping with the tenets of dynastic law and the legal ban on trafficking, 58 they prohibited practices such as buying people of "unclear origin" and from traffickers; 59 or profiteering from indebtedness to enslave others. 60 If lineage authorities wished above all to avoid (shameful and costly) prosecutions, their prescriptions illustrate the gap between state norms and social practices of enslavement.
Lineage authorities similarly regarded the use of nubi as a privilege and made the management and allocation of enslaved manpower a prerogative of their appointed managers, not of individual members. Huo Tao (1487-1540)'s influential Family Admonitions (Jiaxun), for instance, allocated the patriarch and the estate managers a handful of nubi for such tasks as guarding the main gates, receiving visitors, measuring the grains, and so forth. Aside from that, only members who enjoyed academic prestige (like officials returning home on sick leave) could receive nubi from the lineage for their personal service. 61 Huo Tao barely mentions the possession of nubi by individual members, suggesting that acquiring and owning nubi was a monopoly of the corporate lineage. Huo nonetheless alludes to privately owned nubi, suggesting that the practice was at least tolerated. 62 More generally, the example of Huo Tao's Admonitions unveils a picture of the lineage order that is similar in many respects to the Ming founder's ideal agrarian society. Lineage members were expected to have honourable occupations and to work for and by themselves, not to rely on the work of others. 63 Enslaving outsiders for the collective use of the lineage was unquestionably legitimate and even necessary, but it had to remain limited in numbers and to specific tasks.

Paternalistic Control
The most salient feature of lineage admonitions and regulations, however, is their display of a pervasive tension between paternalistic principles and anxieties caused by the presence of nubi. As seen from their tables of contents, sections dealing with the "use" and "control" of nubi are usually paired with sections on "commiseration" and "magnanimity". 64 No similar example can be found for gugong, but both categories are sometimes caught together in the nets of paternalistic discourses. We can read in Pang Shangpeng's (1524-1580) Family Admonitions (Pangshi jiaxun) the following: With regard to hirelings and enslaved: Except for those who are cunning, ignorant, and lazy and who should be expelled, for those who are fit for use 57 Huo, "Taiyuan Huoshi zhongfang zupu jiazhen", p. 90. (kanyong), [one shall ensure that] they are timely provided with food and beverages, that they are neither hungry nor thirsty, and that their work (lao) and rest (yi) are balanced. Tao Yuanming said: "They are also children of human beings. You shall treat them well." If you expect [them] to provide their best efforts (sili), you shall first gain their compliance (huanxin). As to those who are loyal (zhong) and diligent (qin) and that can be relied on, you must extend special generosity (zhouxu) in order to show stimulation. 65 An influential text (reprinted in various registers until the Republican era), 66 Pang's Admonitions show that gugong were indeed perceived as akin to nubi, although the majority of lineage regulations seldom elaborate on their management as they do about nubi. If some authors seemed to genuinely acknowledge the contribution of "household workers" (first of all nubi) to the survival and wealth of their lineage and sometimes even express authentic commiseration, 67 many other authors saw treating subordinates well merely as a utilitarian means to instil gratitude and obedience.
As exemplified by Pang's Admonitions, "magnanimity" barely amounted to ensuring that nubi did not suffer from cold and hunger or that discipline and indulgence, rest and labour, were properly balanced. 68 As frequently noted, enslaved people "listen to hunger and cold before listening to orders" (xian ting jihan, hou ting shihuan). 69 Whereas dynastic law mobilized the parent-child analogy to create the conditions of a stringent and unchallengeable relationship of subordination, lineages resorted to paternalistic rhetoric in a pragmatic way: to defuse insubordination, contain resistance, and promote an ideal vision of a pacific relation of domination.
Of course, nubi were not all "satisfied with their lot" (anfen), obedient, and sensitive to paternalistic discourses. Discipline was thus the necessary corollary to commiseration, benevolence, and reciprocity. That provisions on discipline and control always outnumber discourses on commiseration demonstrates that control and discipline were the primary concerns of household heads and lineage authorities. As mentioned in a lineage register printed in the late nineteenth century: Enslaved people (tongpu) are the same to their masters (zhuren) as the minister (chen) is to his sovereign ( jun). Although the situations (shi) differ, the relative positions (fen) are analogous. With regard to the enslaved, a master must treat them with benevolence (en) but also needs means of control (yu). 70 Discipline, however, did not mean tyranny. Household heads were frequently reminded to "cherish" (aixi) the "enslaved they bought and those who came to provide labour and services" (suomai nubi ji laitou gongyi), to refrain from "abuses" (lingnüe), 71 to closely monitor work, and to explain assignments clearly to avoid mistakes and unnecessary punishments. As underlined by Huo Wanjie in the late fifteenth century, "commanding [nubi]" (shi) required "technique" (fang). But when it proved insufficient, "beating and scolding" (dama) became necessary. 72 Punishments, Obedience, and Anxieties Lineage punishments were generally much less violent than those prescribed by the Code. This applied not only to nubi, but also to everyone in the lineage. Lighter and more incremental (even in cases of serious offences like cursing or assaulting one's parents or master), 73 the lineage punishments amounted to warnings, injunctions to reform oneself, partial or total suspension of lineage subsidies, fines, beating with various instruments (first among which "planks" [ze]), expulsion from the lineage, and, in last resort, referral to the administration for processing under the law.
For instance, article 12 of the Family Regulations (Jiagui) of the Zhang lineage from Qinghe (Anhui, printed 1752) provides a rare example of the tension between paternalism and anxieties, as well as of the disciplinary methods used by the lineages. Entitled "Employing nubi" (bipu yishi), the article comprises twenty-four short rules. 74 Only the first one elaborates on the reciprocal and benevolent nature of the master-nubi relationship. It states: "Although [nubi] are said to be inferior and debased (beijian), they are children of human beings ( jie renzi ye). Since they serve the household (gongyi wujia), they must be raised with benevolence (enyang). Only then can we benefit from their manpower (de renli)." 75 The following three rules expound on the risks related to the presence of nubi in the lineage, the first among which is theft and sexual relations with lineage women. The remaining twenty rules follow the model of the Code: they enumerate offences and prescribe corresponding punishments. The crimes covered by this text range from disobedience to conspiracy against one's master. They include theft, illicit sexual relations, arguments, insults, and assaults against lineage members, but also avoiding service, laziness, and lack of dedication when performing a task.
"intelligence with one's master's enemies" (sijiao zhuchou) were liable to the heaviest penalties (sixty blows of the zhang and one hundred blows of the zhang supplemented with expulsion, respectively). Only one case required outside intervention: "plotting rebellion against one's master" (mouni qinzhu), which was to be "referred to the local magistrate so that [the culprit can] be sentenced to death". Thus, except in cases of very serious crimes and homicides (which were not covered but were undoubtedly referred to the authorities), lineage regulations were far more indulgent with nubi than dynastic law (most of the crimes enumerated being liable to heavier sentences under the law).
Such "leniency" was typical neither of the Zhang lineage nor of the treatment of nubi in general. However, one specificity of the Zhang lineage regulations is that, unlike other lineages that prescribed physical punishments for crimes committed by their constituents, the article on nubi is the only one to systematically prescribe physical punishments. 77 This text thus encapsulates most of the concerns shared by owners of nubi and exemplifies the prevalence of punishment over benevolence.
The range of behaviours targeted by lineage regulations first shows that discipline and punishments were less designed to manage nubi labour per se (in terms of regulating work or increasing productivity) than to ensure complete obedience and subordination. Lineages undoubtedly expected to extract labour in exchange for their "benevolence", as suggested by references to getting their "strength" (renli) and by prescriptions against "laziness". 78 However, "dedication" (qin) and "hard work" (lao) were required not only from nubi, they were also required from every lineage constituent whose work contributed to safeguarding and expanding the collective heritage. 79 Like everyone else, nubi were enjoined to work hard, yet their presence in the lineage was seldom framed in the language of work. Their function in the lineage was, first and foremost, to "serve" ( yi). 80 To "serve" essentially meant to obey in all circumstances with diligence and loyalty. This opened the way to labour extraction with little limitation, but discipline was primarily designed to contain the dangers posed by the presence of nubi to lineage order. Perceived as a permanent "source of trouble" (sheng tudu), 81 nubi were often depicted as deceitful, prone to theft, violence, resistance, and laziness, and inclined to abuse their master's influence for profit (which undermined the reputation of the whole family and could bring charges against its members). 82 Yet, the primary anxiety expressed in lineage registers regarding nubi was bloodline pollution. 77 For crimes committed by other members of the lineage, the Zhang Household Regulations usually formulate prohibitions, all introduced by the negative adverb "don't…" (wu) and calling to lineage solidarity, respect of moral prescriptions, hierarchies, and self-preservation. Even for younger and inferior lineage members, infringement only led to reprimands and demanded repentance (hui). 78 Huo, "Taiyuan Huoshi zhongfang zupu jiazhen", p. 89. 79 Pang, Pangshi jiaxun, p. 11; Huo, "Taiyuan Huoshi zhongfang bashi zu Huaiting weng jiazhen fulu", p. 107. Taga, Sōfu no kenkyū, pp. 789, 612, 789, 822; Bray, Technology and Gender, pp. 243-247. 80 Chevaleyre, "Serving and Working for Others", p. 181. 81 Pang, Pangshi jiaxun, p. 12. 82 Wang Yan'gan, Jingjiang Wangshi zongpu (1935), in Taga, Sōfu no kenkyū, p. 672. Lan, Lanshi xuxiu zupu, p. 106; Huo, "Taiyuan Huoshi zhongfang zupu jiazhen", p. 89.
Incorporating outsiders into the lineage (as wives, concubines, adoptees, etc.) was a constant source of anxiety. 83 Lineage regulations were, in this regard, often stricter than the dynastic law. Whereas the latter did not prohibit marriages between enslaved females and ordinary men, 84 lineages condemned unions between lineage members and nubi of any sex in the strongest terms. 85 Such alliances were an offence to one's ancestors (shang ru zuzong), a burden on one's descendants (xialei zisun), and a shame to the whole clan (hezu xiuyu shijian), to the point that infringement of the rule was sometimes liable to elimination from the genealogical tables. 86 This fear of "pollution" is also exemplified by various prohibitions against male nubi coming into contact with the women of the household. 87 Lineages required male and female nubi to strictly abide by gender separation. For instance, in the fifteenth century's Huo lineage from Nanhai (Guangdong province), female nubi were forbidden to "go outside" (chuwai) upon reaching fourteen sui old (circa thirteen years old), while male nubi were forbidden to "go inside" ( jinnei) upon reaching sixteen. 88 From the lineage perspective, nubi (and, to a lesser extent, gugong) only existed through the two opposite dimensions of paternalistic discourses and disciplinary prescriptions. Framing the exercise of mastery in terms of benevolence, magnanimity, and reciprocity was an absolute prerequisite because the enslaved present in the households of the late Ming and early Qing period were not criminals but were ordinary people who had abdicated their relative autonomy "willingly". However, as outsiders incorporated to "serve", their mere presence (albeit necessary) posed multiple dangers calling for constant checks and strict methods of control. Although they should know "fear" and be promptly punished when causing trouble, 89 discipline could not amount to "abuses" and be "excessive", since excessive violence resulted in "severing the feelings" (qingshu) that were supposed to cement the master-nubi relationship. 90 The parameters of nubi discipline did not differ much from those applied to other lineage members. However, what distinguishes them from other lineage constituents is that discipline outweighed all other considerations. Nubi were persons to be controlled, not simply a labour force to be managed, even less genuine lineage members to be educated, taken care of through lineage solidarity, and included in ritual activities. As noted in the Family Rules (Jiazheng) of the Zhu lineage from Kaoting (Fujian, 1620): "As to nubi, one [only] wishes to have them in good order" (bipu yu qi zhengqi). Although closer to the ground, lineage sources cannot be taken as true reflections of the everyday disciplining of "household workers". Lineage authorities could certainly be mobilized when situations escalated, or to make an example of ( just like imperial justice could be mobilized to the same effect). However, considering the threats of prosecution repeatedly made against masters who failed to "control" their nubi, 92 it is more than likely that everyday discipline was seldom delegated to lineage authorities and remained in the hands of individual masters. Bringing to life the day-to-day management of enslaved and hirelings, however, is a more difficult task than exploring normative prescriptions like the Code and lineage rules.

Punishment(s) in Practice
Unearthing actual practices related to "household workers" is indeed challenging. When sources are available, the voices that we hear are overwhelmingly those of the (male) educated elites, who usually considered private affairs too trivial to be worth writing about, and who mostly reported on practices that, in one way or another, came out of the ordinary.
The previous sections have underlined both proximity and distance between nubi and gugong. In law, nubi was the matrix from which the parameters of the legal treatment of gugong were framed. Domestic regulations reaffirmed this affinity in principle, but they did not demonstrate the same level of concern about controlling gugong as they did about nubi. Sources reporting on everyday practices like contracts, management treatises, and narratives of things "seen and heard" hardly help remove these ambiguities, especially in the late Ming period when the numbers of nubi grew significantly and when "labour" relations seem to have been widely contaminated by the axiom of the master-nubi relation.

Contracts
Contracts show both differences and commonalities between the two categories. Nubi contracts standardly open with the exposition of the "poverty" of the contracting party. This served to legitimize the self-sale of a person (alone or with their family) or the sale of a child. 93 Nubi contracts seldom describe work obligations, and they usually remain vague as to what enslaved people received in return (food and clothes at a minimum, sometimes a house to inhabit, the promise of marriage, or a lump sum of silver upon entering the relationship). 94 However, they often insist on the commitment to "obey orders" (tingcong shihuan), to "serve" (yishi, yingyi, fuyi), and sometimes to "work with diligence" (qinli yiye). And they always contain clauses absolving buyers from responsibility if something "unexpected" happened to the enslaved, including untimely and accidental death. 95 Nubi contracts also frequently end with a formal acceptance by the contracting party of their "punishment" (fa, zui) should they disobey, run away, or cause trouble. dare neither to discuss [orders] nor dawdle (weiman), desert (paoli) or run away (tao)." 101 Such a "pawning" contract is not a typical gugong contract, as it does not set a clear end to the employment and only mentions the payment of a lump sum to the parents upon signing the contract. The situation was thus more that of a child used as collateral for a loan and risking permanent enslavement in case of a payment default. However, it demonstrates that situations close to (or preceding) enslavement could be framed in the language of "hiring out" (chugu).

Management Prescriptions
Sources alluding to the management of household workers are quite rare in the literary landscape of the late Ming and early Qing period. The few examples available convey a somewhat mixed picture. Treatises of agronomy and estate management sometimes include prescriptions about the management of nubi and gugong, but they seldom insist on discipline and punishment.
As to gugong, they usually focus on incentives and "benevolence" as means to ensure efficient labour and to curtail laziness. The often-quoted Agricultural Treatise of Master Shen (Shenshi nongshu, Zhejiang province, c.1639) essentially prescribes "generous" food allocations and their adjustment to seasonal variations to ensure compliance to work and to prevent defection. In this text, the emphasis is placed on compliance and generosity as preconditions to "reprimand" ( jie) hirelings when necessary. 102 Defection, taken very seriously in the case of nubi, is merely mentioned as something to be prevented, not as a crime to be corrected by capture and prosecution.
Prescriptions for nubi management are found in Zhang Lüxiang's Supplement to [Master Shen's] Agricultural Treatise (Bu nongshu, published 1658). Zhang simply reproduces prescriptions formulated centuries earlier in Sima Guang's (1019-1086) Miscellaneous Etiquette for Family Life (Jujia zayi) that neither mention discipline nor refer to benevolence and generosity. The text describes a typical workday, during which nubi would get up at dawn, clean the house, and prepare food before the masters wake up, perform a "hundred services" (gong baiyi), "obey their master's orders" (wei zhuren zhi ming), and "accomplish their tasks" (ge cong qi shi) until dusk. 103 Here, the emphasis is placed on nothing else but "service" and continuous and absolute obedience.
Punishments and discipline thus do not appear central in texts alluding to labour management. The differences between the management of nubi and gugong seem rather tenuous, despite a greater emphasis placed on gaining the compliance of hirelings rather than immobilizing them by force.

The Tenuous Line Between Discipline and Violence
Discipline and violence appear in a different light in the sources produced during or in the aftermath of the late Ming social, economic, and military crisis. In the urban strikes and riots initiated by skilled hired workers at the turn of the seventeenth century, discipline is never mentioned as a trigger. For instance, the so-called "weaver workers revolt" (zhiyong zhi bian) that broke out in Suzhou in 1601 was essentially caused by an increase in the costs of raw materials and by the aggressive methods of tax collection by imperial envoys. 104 On the contrary, punishments and high levels of brutality were apparently decisive in the numerous murders of masters recorded in the sixteenth and seventeenth centuries, as well as in the nubi revolts (nubian) that sporadically broke out between the 1630s and the 1650s.
Faced with the growing presence of nubi within families and society, 105 from the mid-sixteenth century onwards, the literati increasingly voiced concerns over the laissez-faire attitude of masters, the influence of powerful nubi, and the visible excesses of slaving practices. Retrospectively, an excessively laissez-faire approach and brutality were pointed out as hallmarks of the social decay of the late Ming, which led to nubi unrest and the ultimate fall of the dynasty in 1644. Calling for the more rigorous control of nubi, promoting the ideal figures of benevolent "master-father" (zhufu) 106 and loyal enslaved ( yipu), and even sometimes questioning the social relevance of enslavement, diaries, letters, anecdotes, and jottings all offer valuable glimpses of the extensive disciplinary powers that masters exercised.
The emphasis that the literati put on punishing nubi more "humanely" 107 where necessary shows that discipline was consubstantial with enslavement and suggests, as many observers have underlined, that nubi were frequently "less well treated than dogs and swines" (zhi quanzhi zhi buruo). 108 Examples of brutality abound in the sources produced before the outbreak of nubi revolts. Caning (zhang), "cruel beatings" (kuda), 109 and the like were disciplinary methods of choice and became the epitome of the master's cruelty. As one revolt leader declared: "Why should our enslaved lives be satisfied with receiving caning and being lectured?" 110 Dong Chuance (1530-1579), who was murdered in Huating (Jiangsu) by a group of nubi on 31 May 1579, is said to have abused many with the cane: "Vice-Minister Dong Youhai [i.e. Chuance] was an inflexible and severe person who treated household slaves ( jianu) with excessive rigour. When they committed a fault, they immediately received caning. When using the cane, [Dong] counted to one hundred [blows].
Many died from it." 111 Like many others, Dong is also said to have employed a wide variety of disciplinary and punitive methods. Upon discovering that one of his retainers had exchanged words with a concubine, he had him locked up. When the retainer finally ran away, he had him tracked down and executed with impunity. 112 Late Ming and early Qing observers widely testify to the many forms and few limitations that punishments and brutality could take. Nubi could be victims of extortion from their masters when they managed to accumulate some wealth for their own. They were frequently said to suffer from cold, hunger, and lack of decent clothes. They also experienced physical and psychological violence, sexual abuse, 113 and even mutilation. 114 Sources also provide numerous cases where masters denied nubi the right to mourn their parents or "killed [them] and burned [their bodies] without anyone daring to denounce them", 115 or with minimal consequences. 116

Revolts
Although the extent and frequency of violence against nubi cannot be evaluated with any level of precision, it is best exemplified by the violence that nubi unleashed against their masters during the late Ming revolts. Nubi revolts hardly formed a consistent "emancipation movement" (as is often claimed), 117 but they systematically targeted masters and aimed to destroy enslavement contracts. When masters and their families were not killed on the spot, they were put on trial for mistreatment and submitted to rituals of role inversion, being forced to serve others and taste their own disciplinary methods. The Baoshan district revolt (Jiangsu province, 1644) provides a typical example: Unleashing great violence, thousands and hundreds gathered to burn the houses and reclaim contracts by force. Clouds of ashes obscured the sky. While the enslaved sat down, masters stood still and served them meals. At the slightest sign of discontent, they [the masters] were thrown on the ground and beaten with zhang [the cane]. This was a catastrophe like no other in a thousand years. 118 Everywhere the revolts were triggered by excessive brutality. On the front line of the repression in 1644, Qi Biaojia (1603-1645) notes that retaliation against masters "beating nubi cruelly" (kao nu ku) was the insurgents' first motivation. 119 A year later, in Taicang (Jiangsu), several masters were judged and humiliated for systematically beating their enslaved for trifles, like coming back late from picking up tea leaves or lousy cooking. 120 Zhang Mingbi (1584-1653 extensively recounts how such trials were carried out in Jintan (Jiangsu) in 1645. Masters were made to get a taste of the instruments they had used in the past to punish nubi. While beating their masters with "sticks" (bang) or pricking them with "awls" (zhui), after each blow and after each cry of pain, nubi first asked, "does it hurt?" and then asked why they used physical violence when they knew how painful it was. 121 For challenging the social order, nubi insurgents were nonetheless punished with exemplary severity. All the revolt leaders ended up executed with their heads exposed on pikes. Their followers were beaten and sent back to their former masters, whom the local authorities left to decide how to punish them. 122 In a way, the resolution of the nubi revolts illustrates how disciplinary and punitive powers were distributed between the state and the heads of lineages and households: the state had no say and no interest in the matter as long as things remained under control and did not impact broader society.
Of course, counterexamples of loyal, rich, and influential enslaved (as well as models of benevolent masters) can be found in the sources at our disposal. 123 And in the context of an increasing and multifaceted crisis, the protection offered by voluntary enslavement was valued and preferred by many to the precariousness of autonomy and economic uncertainty. 124 The information that we can extract from the sources nonetheless leads to three conclusions.
First, the actual condition of nubi almost entirely depended on the will of one's master. The specific parameters of the relation with a master, whose extensive power was safeguarded by the whole sociopolitical order (from the state down to the local community and the lineage structures), were crucial in determining the degree of autonomy that each nubi could enjoy in practice (or could conquer by "manipulating" their masters, as many sources also underline).
Second, in a sociolegal environment designed to protect the fatherlike figure of the household head and to uphold the indisputability of his authority, discipline and violence overlapped and can barely be distinguished. Both were consubstantial with enslavement and knew very few limitations. Nubi were permanently vulnerable to violence and punishments. At any time, even the most autonomous and powerful ones remained powerless against discipline and violence unless a third party decided to stand up for them (be it a magistrate, a sympathetic neighbour, an enemy of their master, or a more influential master offering protection).
The case of one nubi named Gu Liang can be taken as the epitome of the potential for unbridled violence and the powerlessness of nubi. Gu had voluntarily offered himself (tou) and his wife as nubi (weipu) in the service of one Zhuang Ying from Shanghai district in exchange for being housed and fed. Zhuang, however, coveted Gu's wife. One day, in 1607, he voluntarily blinded Gu and sold his wife after Gu had repeatedly objected to Zhuang having sexual intercourse with her. The prefectural judge in charge of the case, Mao Yilu (?-1629), sentenced Zhuang to penal servitude and to pay Gu ten taels of silver and a piece of land so that he could effectively "leave the household" and make a living of his own. In so doing, Mao Yilu applied the law protecting gugong from brutality (see the first section above) in the application of a sub-statute (dated 1588) stipulating that nubi owned by commoners (rather than by officials) shall be judged like gugong. 125 This can be taken as proof that the little protection offered in law to nubi and gugong was enforced by late Ming magistrates. Yet, the case was not brought to court by Gu Liang himself. It was unveiled during an investigation of theft against Gu and his relatives. One thus wonders what would have happened had Gu referred to the judge himself and thus broken the law prohibiting nubi from accusing their master.
Finally, although the sources at our disposal seem to suggest that punishments were less central to the management of gugong, due to their proximity to nubi, the importance of role performance in the assessment of statuses, and the power that was vested in household heads, gugong could easily be assimilated to nubi and subjected to similar forms of control and discipline. The late Ming period is characterized by a widespread extension of nubi status to other relations and a blurring of social and labour identities. This extension sometimes took the form of subtle contamination, as in the case of one Tang Yuan, also adjudicated by Mao Yilu. For reasons that remain unclear, Tang had moved with his aunt and her husband, two nubi in the service of the Han family in Huating district. According to Mao, Tang was of commoner status. However, for sharing the roof of his relatives' master for years and for accepting money to help him marry, he was reported as a runaway nubi after he left their household to settle on his own. Although Tang ultimately committed suicide, the magistrate concluded that he had benefited from the Han family's "benevolence". This fact alone had created a bond similar to signing a bondage contract. In the magistrate's terms, it had created a "difference between master and nubi" (zhupu mingfen), which gave ground to the Han family's claims. 126 The practice of labelling gugong as nubi and forcing them into permanent enslavement by the threat of prosecution as runaways is also widely evidenced. In a memorial written in the aftermath of the last serious revolt in Guangshan (Henan, 1658), prefect Jin Zhen (1622-1685) listed the existence of "enslaved hired workers" (yonggong zhi pu) as one major cause of the revolt. The practice he describes was a mix of hired work and uxorilocal marriage, by which men agreed to work for a fixed period in exchange for an enslaved woman to marry (one such contract set the term to 125 Mao, Yunjian yanlüe, pp. 417-418. 126 Mao, Yunjian yanlüe, p. 574. twenty-two years). 127 Once the term was over, employers threatened to report gugong as "runaway slaves" and retained them permanently. 128 Epilogue As a cornerstone of the late imperial social order, the authority of the father/household head was indisputable in late imperial China. Benefiting from high protection in law, it extended not only to children, but also to all those, kin and non-kin, who "lived together" and were part of the household. The parameters of subordination to the power of the household head nonetheless varied slightly depending on whether a person descended from the patriline, was incorporated to ensure reproduction, or was brought in (permanently or temporarily) to "serve". In the last case, of which nubi are the epitome, the analogical reference to the figure of the father was central in elaborating a fictitious filiation devoid of the prerogatives of actual membership of the family, to which only the duty of absolute obedience and subordination remained.
As "serving" outsiders, nubi and gugong were similarly placed under the direct and exclusive authority of the household head, which was protected by an asymmetric regime of legal punishments and guaranteed by the conferment of extensive disciplinary powers. The legal framework elaborated by the founder of the Ming dynasty had thus created the conditions for an almost absolute and potentially violent domination of "household workers" who, for their part, benefited from minimal legal protection and could hardly challenge the household head's power by appealing to justice. In everyday practices, the paternalistic requirement for "benevolence", reciprocity, and mutual affection prescribed by moralists and by lineage admonitions certainly played a role in mitigating violence and abuses of all sorts. However, as the nubi revolts of the mid-seventeenth century demonstrate, such requirements did not weigh much against the indisputability of the householdhead's authority.
At the end of this overview of the many dimensions of punishment in the relationship between "household workers" and their masters/employers in late Ming and early Qing China, we are nonetheless left with ambiguities, uncertainties, and pending questions, in particular regarding the actual differences in vulnerability to punishments, discipline, and violence between the two legal categories of "household workers" examined in this article. Although treated like cognates in law, we have seen that in the sources alluding to lineage organization and day-to-day practices, controlling, disciplining, and immobilizing gugong by force seemed a far lesser concern than in the case of nubi. This apparent difference can be explained by the fact that hirelings entered their employer's household on different terms. They were neither owned by their employers nor "debased" beyond the boundaries of their employer's household and the time frame of their employment. As one commentator of the Great Qing Code noted in 1715: 127 Zhang, Zhongguo lidai qiyue huibian kaoshi, p. 1063. [Gugong] only perform services for others (weiren zhiyi) in exchange for wages (guzhi). Their activity (shi) is demeaned, not their person (shen). Once their employment and wages are over, the head of the household is like an ordinary person [to them] (tong fanren). They are thus different from those who, as persons, are nubi (shen wei nubi). 129 The temporary and lesser incorporation of gugong into the household might thus be the sign of a significantly different labour regime (control of "activities" versus domination of "persons"), of an evolution of labour relations in the late Ming period, and maybe of an "emancipation" of hirelings from the paternalistic power of the household head, as various historians have already argued. 130 This leads us to make several concluding remarks. Although cross-examining normative sources with documents like contracts and court cases is methodologically more relevant than focusing on one single dimension, taking "China" in a centurieslong perspective as a research object presents serious limitations that can only be overcome by narrowing the focus on multiple local micro-social histories of work. Such studies already exist at the scale of specific prefectures and provinces, but they usually remain limited, in great part due to an overwhelming focus on labels and "labour" categories. 131 Legal and "customary" categories are, indeed, far too limiting to understand the dynamics of labour relations in practice. Labels reflect practices and norms, but those two dimensions never perfectly align, while "labelling" can be as much a legal as a social practice. Exploring the contextual plurality of punitive practices might thus be a more relevant method to overcome the bias of label-centred approaches in the study of social practices of labour and coercion (and to overcome the classical question of the "free"/ "unfree" nature of hired work in late imperial China). 132 Finally, we are also left with the lingering question of the relationship between the domination of people and work. Historians of slavery would probably agree that the exercise of "mastery" cannot be reduced to the single dimension of the exploitation of the workforce of other human beings, even with a broad and encompassing definition of work in mind. In this respect, the sources used in this article remain ambiguous because the labour dimension is never the most salient, to the point that the relation between punishments and the management of work remains tenuous. In the case of nubi, labour is undoubtedly never too far away in the prohibitions against "laziness", 133 in the statements demanding constant "diligence" (qinjin), 134 in those promoting reciprocity as a means to benefit from their "strength" to "develop one's household", 135 and in those explaining that nubi were brought in "to work in one's place" (dailao). 136 However, in the control of outsiders made into permanent or temporary insiders to the household, what mattered to lineages and families was to keep "household workers" in their place, to maintain hierarchies, and to avoid "disturbance" (luan). Instead of work and labour, the ubiquitous keyword in late imperial Chinese sources is "service", which could encompass any productive, non-productive, and reproductive task. 137 In a court case of the nineteenth century, we read the following remark: "Gugong simply make a living by being hired as workers (shougu yonggong). Since they are serving and ordered (yishi), a difference between superior and inferior must [nonetheless] be demonstrated (shi yi shangxia zhi fen)." 138 The meaning of "service" thus went beyond simply performing labour (even in the case of gugong whose main functions were more closely related to work). 139 It was far more saturated in the language of obedience and anxieties than in the language of work and management. What "pluralization" and contextualization could bring to the study of labour, coercion, and punishments should thus be a broader reflection on the categories that we use as entry points to the history of social practices. 136