Family and labour in an Angolan cash-crop economy, 1910

ABSTRACT

This paper examines the composition of families in the parish of São José de Encoge, northern Angola, in the early colonial era, using a series of ‘family bulletins’ collected by the Portuguese colonial government in 1910. Encoge was one of the earliest centres of coffee cultivation in west-central Africa. While not all local families participated in this economy, the census sheds light on family-based cash-crop production after the abolition of slavery in Angola. The literature on post-slavery labour arrangements in Africa almost unanimously suggests that family-extension strategies predominantly aimed to integrate young women and girls into households, using pawnship and guardianship as important methods. The family bulletins from Encoge provide a means to evaluate this assumption on a quantitative basis in an Angolan context. They indicate that, besides natural reproduction and the adoption of second wives, the incorporation of wards, nephews, and nieces was the most common form of household extension, in which elders showed a remarkable preference for male dependants. The bulletins do not provide clear evidence of the presence of enslaved dependants, although some individuals listed as servants were possibly enslaved. Overall, the data suggest that, contrary to common assumptions about agricultural labour in Africa, family-based cultivation and trade of coffee did not discriminate against men or boys.

KEYWORDS:

• Family
• labour
• slavery
• guardianship
• coffee
• Angola
• Africa

1. Introduction

The historiography of the family in western Africa has long been concerned with how the abolition of the Atlantic slave trade, from 1807, and the subsequent expansion of commercial agriculture and colonialism affected social relations at community and household level.¹ There is a substantial literature that connects the imposition of colonial rule and the ensuing suppression of slavery and pawnship in Africa to changes in asymmetric dependency relationships (Getz, 2004; Klein, 1998; Lovejoy & Falola, 2003; Miers & Klein, 1999). Within this broad theme, several scholars have highlighted the centrality of women and children, or female and child labour, to the economic affairs of African households and kin groups (Alanamu et al., 2018; Allman, 1997; Campbell et al., 2008, 2011; Diptee & Klein, 2010; Grier, 1994; Grier, 1992; Lawrance & Roberts, 2012). A key proposition in this literature is that families are economic units whose expansion, by either natural reproduction or the integration of new members from outside, fundamentally aimed to increase their productive capacity. Furthermore, as most African societies were predominantly agrarian until the end of colonial rule, around 1960, the history of the family in Africa has largely unfolded against the backdrop of agricultural production.²

Scholarship on the ‘agricultural transition’ in West Africa following the ending of the export slave trade in the nineteenth century suggests that slave labour generally supported the early expansion of cash-crop farming (Austin, 2009; Law, 1995; Law et al., 2013). A further assumption has been that when colonial governments began to suppress slave trading and slavery within the continent, social elites turned to other forms of bondage as mechanisms to accumulate dependants, most notably pawnship, which technically meant holding someone as collateral for debt. When governments sought to outlaw this practice as well, fosterage or guardianship became an important form of incorporating outsiders and distant kin into families (Akurang-Parry, 2010; Coe, 2012; Moitt, 2011). Historians tend to agree that, because most African societies placed a premium on female dependants, the institutions of slavery, pawnship, and guardianship especially helped to integrate girls and young women into family units (Klein, 2011; Lovejoy, 2018). This idea corresponds with the widely accepted notion that women performed most agricultural and domestic labour in precolonial Africa (Robertson & Klein, 1997). The current consensus is that the exploitation of female labour generally intensified in colonial-era cash-crop economies (Robertson, 2015; Berger, 2016, pp. 18–19).

Individual studies have added nuance to such general assumptions. It is important to recognise, for instance, that enslaved and pawned people were not only employed in agriculture or domestic chores, but also in other important economic activities like trade and porterage (Akurang-Parry, 2002; Van Haer, 1982), which is not to say they dominated such activities (Rockel, 2006; Vos, 2015). Furthermore, throughout the nineteenth and twentieth centuries, family labour generally remained the basis for cash-crop production in much of Africa. Extra-familial coerced labour, first by enslaved people, then by pawns and foster children, only became relevant when the labour needs of the household exceeded the labour power supplied by wives and other family members and when hiring labour was not an option (Rodet, 2012; Austin, 2017, pp. 187–188). As Susan Martin (1995) has argued, however, bondage was not always the preferred route to expanding a household’s productive capacity. In the Ngwa-Igbo region of southeast Nigeria, a situation of labour abundance in the late nineteenth century reduced the benefits of slavery, causing entrepreneurs to adopt marriage as a strategy to exploit female labour. Jean Allman and Victoria Tashjian (2000, pp. 60–73) have shown that in colonial Asante, male cocoa farmers increasingly relied on the labour of their wives as access to bonded labour narrowed. Moreover, Elisha Renne (2005) and Cati Coe (2013) have argued that, although debt fosterage often grew out of debt pawning, fosterage tended to be less exploitative than pawning, as it usually included elements of training and care and thus instilled a greater sense of ‘belonging’ in foster children.³ These examples demonstrate that experiences of slavery, bondage, and dependency in developing cash-crop economies varied across western Africa and that, therefore, generalisations require a degree of caution.

This article uses a series of ‘family bulletins’ from the parish of São José de Encoge in northern Angola in 1910 to make a crucial intervention in these ongoing debates about the organisation of household labour for agricultural production in sub-Saharan Africa in the wake of slavery. Most of the evidence of changing forms of dependency relationships within African families in the early colonial period has so far been of a qualitative nature, as historians have used court cases and reports from government officials, missionaries, ethnographers, and others to construe their arguments.⁴ As Samuël Coghe (2022) has recently demonstrated, colonial officials, doctors, and missionaries were counting people at village level from around 1900; but, since these efforts were often scattered and did not produce countrywide censuses, historians have so far done little with the available data. In this context, the family bulletins, from the Encoge parish provide a unique opportunity for a quantitative assessment of family structures in an African cash-crop-growing region at a crucial point in its economic development. The bulletins do not reflect directly on the prevalence of slavery and pawnship in northern Angola in the early twentieth century, as these practices were illegal by then. Nevertheless, contrary to the female bias in the Africanist literature on slavery, pawning, and fosterage, they demonstrate that local strategies of household expansion specifically aimed to integrate junior males, notably nephews, wards, and servants, into households.

2. Historical context

São José de Encoge was a fortified Portuguese settlement on the northern fringes of the Angola expansion, established in 1759 to strengthen Portugal’s control of a main slave-trade route from the central African interior to the Atlantic coast (Thornton, 2020, pp. 257–258). While the Portuguese only partially succeeded in their endeavour to regulate the slave trade, nineteenth-century records indicate that government officials and private traders annually dispatched thousands of enslaved Africans from Encoge to the main Portuguese slave port of Luanda, where ships waited to carry them onward to Brazil and Cuba (Vos & Matos, 2021). Until the Portuguese expansion into the Kongo kingdom in the 1850s, Encoge was Portugal’s northernmost colonial outpost in Angola. However, the presence of a European-style stone fortress and Catholic parish church notwithstanding, Encoge was a thoroughly African community. Practically all its inhabitants were locally born, including those who carried Portuguese patronyms and those whom the colonial state officially recognised as residents (moradores) and electors (eleitores). Because of its isolated location about 300 kilometres inland from Luanda, the region did not attract colonial settlers, as did Cazengo and Golungo, two neighbouring districts to the southwest, when the rise of legitimate commerce created opportunities for new agricultural ventures (Figure 1).

Figure 1. Northern Angola Map based on Carta de Angola contendo indicações de produção e salubridade, Lisboa 1885, https://purl.pt/1926/2/).

When the export slave trade ended in the mid-nineteenth century, the participation of the Encoge population in the global economy came to depend mainly on coffee. The district had been the site of Angola’s first recorded coffee harvest in 1820. Local coffee production, based on indigenous robusta trees, remained small while slave trading continued to generate bigger profits. Indeed, in the 1850s, outputs from Encoge still barely surpassed 15 metric tons per year (Oliveira Beça, 1860; Salles Ferreira, 1859). However, as the slave trade to Brazil and Cuba came to an end between 1850 and 1867, and global coffee prices increased simultaneously, the number of African entrepreneurs in Encoge willing to invest time and labour in coffee cultivation grew from a few dozen to more than a hundred.⁵ At the same time, the small colonial district became a centre of regional plant transfers, as growers from across northern Angola used seedlings and cuttings collected in Encoge to establish their own coffee plantations.⁶ It was small farmers like these, in Encoge and neighbouring districts, who accounted for the rapid expansion of coffee production in Angola in the second half of the nineteenth century, making the Portuguese colony the largest coffee exporter in the African continent before 1900.

Angolan coffee exports peaked at 11,000 tons in 1895. Local outputs declined when world coffee prices began to fall that year, and by 1910, exports had slumped to 6,140 tons (Mesquita, 1918). While many mono-cropping settler estates in the Cazengo and Golungo districts struggled during this depression in the global coffee trade (Mata et al., 2017), African farmers used their multi-cropping routines as a buffer against the market. They may have paused the extension of their coffee plots, but they continued to grow the crop as a way to earn cash income. An administrative report from 1906 highlights the importance of coffee to the economy of Encoge, with local producers selling their crop to factories in Ambriz, on the Atlantic coast, or in locations midway to the coast.⁷ Besides trading coffee, inhabitants of Encoge were also involved in the wild-rubber trade, based on ‘root rubber’ collected from savannah regions further east.⁸ However, coffee was generally the main commercial product, which some local growers also used to pay the colonial hut tax.⁹

The fact that Encoge was a key node in the nineteenth-century slave trade to Luanda before it became the focal point of Angola’s burgeoning coffee economy raises interesting questions about the history of slavery in the district or, more specifically, the composition of the labour force involved in local cash-crop production. Vos and Matos (2021) have argued that after the closing of the slave trade from Angola to Brazil, in 1850, the slave population of Encoge declined rapidly because the district stopped ‘warehousing’ captives for the Luanda market. The share of captives in the overall population dropped from 6.5% in 1844 to under one percent in the 1860s. Although coffee production was expanding and slave prices were declining in Angola in this period (Ferreira, 2013), Encoge’s emerging cash-crop economy did not appear to incorporate many of the captives that were potentially glutting Angolan markets. A register of slave-ownership in Angola in 1856 indicates that Encoge only counted nineteen senhores, who together owned 253 enslaved persons (Angola, Secretaria do governo geral da província de, 1856). These slave-owners were very likely involved in coffee production, but without creating plantations based on colonial models of landownership.¹⁰ Strikingly, as the number of coffee growers went up in subsequent years, the number of enslaved people in Encoge diminished further. By 1869, according to official counts, there were only 76 slaves. A register from 1877, when servitude had been officially abolished, included 71 servants (criados de servir), which was the new term to designate people who in earlier times would have been enslaved. This trend set the smallholders of Encoge apart from the plantation-owning ‘coffee barons’ of Cazengo who, as David Birmingham (1978) has shown, relied extensively on coerced labour. It suggests that the coffee growers of Encoge primarily depended on the labour of their immediate kin, conceivably assisted by people from wider family and community networks, and sometimes on bondspeople whom wealthy farmers integrated in their households.

3. Materials

A series of 204 individual family registers from the parish of São José, the nucleus of the colonial district of Encoge, dated 31 December 1910, provide an opportunity to examine the size and composition of African households in a cash-crop-growing region at the end of the first Angolan coffee boom. These registers come from two different boxes in the Angolan national archives.¹¹ They were part of a general census of the Angolan population that the government in Lisbon, in 1899, had ordered colonial administrations to conduct every ten years, in an attempt to reanimate a tradition of census taking in the Portuguese empire going back to the late eighteenth century. Instructions sent to local administrators gave some flexibility to the execution of the census: in areas near the centres of colonial power, administrators were expected to conduct a ‘nominal’ census, listing all inhabitants individually in preformatted ‘family bulletins’; whereas in areas of limited colonial control, they could produce estimated counts of villages, homes, and residents (Portugal, Repartição técnica de estatística geral da colónia de Angola, 1941, pp. 20–28). The person in charge of the population count in Encoge in 1910 was the local district commander, Guilherme Vandunem, a 49-year-old Black African born in Benguela who had been in this post since 1902.¹² The archival find is evidence that Vandunem intended to register at least several hundred inhabitants of the São José parish in family bulletins. Since only 204 bulletins have thus far been retrieved, while a countrywide census for 1910 was never published, it remains unclear how far he got with his efforts.¹³

The 204 bulletins were collected from 32 different settlements throughout the parish and listed the names of 797 individuals, 51.4% of whom were male. For comparison, a published population count of São José de Encoge in 1900 suggests the parish included 85 villages, containing 5,301 homes (fogos) and a population of 27,106 people, of whom 47.2% were male (Angola, Secretaria geral do governo geral da província de, 1903, pp. 80–81).¹⁴ Against this background, it becomes clear that in most parish settlements he visited, Vandunem only created bulletins for some families. How he selected these families is difficult to know. Different circumstances constrained his census-taking efforts: official instructions gave him just one month to collect his data, there was only a scribe to assist him, and resistance to census taking was common in Encoge, as African authorities often associated counting with people theft.¹⁵ Moreover, the neat administrative division of Angola in parishes, townships, and districts could not disguise the fact that colonial power remained feeble in the interior. Beyond the moradores, who gained social standing from their liaison with the Portuguese empire, Portugal’s claim to sovereignty in the Encoge region depended on unstable alliances with a handful of local chiefs. Most chiefs resisted colonial domination, including population counts, until 1919 (Pélissier, 1997).¹⁶

Despite the small sample size, by listing people individually the recovered bulletins provide an extraordinarily detailed snapshot of northern-Angolan family structures in the early twentieth century. For each of the 204 registered households, they provide the name of the village, the names of all household members, their sex, birthplace, age, marital status, relationship to the head of household, educational achievements (meaning whether a person could read or write), and their occupation. The final column of the form was for special observations, which Vandunem used to record people’s skin colour. Vandunem listed his own family first, which included himself, his female companion (amasia), Maria João – a 25-year-old dressmaker from the nearby district of Icolo e Bengo – and their four servants (criados): the cook João, the washerwoman Lucia, and two domestic servants, Napolião and Curiva. The only other persons living at the government’s headquarters near the old fort were the 40-year old scribe, José António de Sousa, a Black African from Luanda, and his companion, Francisca Pinheiro da Cunha, who pertained to one of the local ruling families. Apart from Vandunem, Maria João, and José Sousa, all parish inhabitants were born in the Encoge district. A few moradores were able to provide their ages, probably because a Catholic cleric had officially registered their births. Most families included a husband and wife, but since none had married in the Catholic Church, everyone was officially single. Only a few people could read or write, which were skills obtained in the local parish school. In terms of occupations, most people were considered ‘workers’ (trabalhadores), although some were recognised as ‘farmers’ (lavradores or agricultores), meaning they were officially registered with the colonial government as smallholders and therefore exempt from compulsory labour duties. The population of Encoge was considered fully Black, except for three members of the family of farmer António Bernardo dos Santos e Silva (António himself, his nephew, and his female cousin), who were mestiço (mixed-race). The most important features of the census for the purpose of this article are household sizes, relationships within households, and the size and composition of households headed by officially recognised farmers, as these were almost certainly involved in cash-crop cultivation.

The 204 families together constituted a population of 797 people, which is an average of 3.9 members per household. Families of three (67) and two people (53) were most common (Figure 2). Men were the head of household in 195 cases; 183 of these men had one or more wives (amasias) (Table 1).¹⁷ Only in nine cases were women the head of household and in all these cases there was no husband. Beyond husband and wife, other categories used to identify household members were son (filho) and daughter (filha), grandson (neto) and granddaughter (neta), stepson (enteado) and stepdaughter (enteada), brother (irmão) and sister (irmã), nephew (sobrinho) and niece (sobrinha), cousin (primo, prima), uncle (tio), ward (pupilo, pupila), and servant (criado, criada, servente). Before analysing the numbers in these categories, it is necessary to insert a few cautionary notes.

Figure 2. Household sizes in São José de Encoge parish, 1910.

Table 1. São José de Encoge parish population by household member category.

Relationship to head of household	Total	Male	Female	% of population	Number of families
Head of household	204	195	9	25.7	204
Spouse	228	0	228	28.7	183
Child	155	74	81	19.5	94
Grandchild	12	7	5	1.5	7
Stepchild	16	10	6	2.0	7
Sibling	12	8	4	1.5	9
Nephew/niece	43	30	13	5.4	26
Cousin	5	3	2	0.6	4
Uncle	1	1	0	0.1	1
Ward	83	51	32	10.5	52
Servant (criado/a)	28	23	5	3.5	12
Servant (servente)	7	6	1	0.9	3
Total	794	408	386	99.9

Historians have long recognised that the categories Europeans used to describe African households are problematic in different ways. First, the concept of ‘family’ itself was a European construct for which there was no equivalent term in most African societies. In the Kongo-speaking region – with Encoge on its southern periphery – family translated as ‘clan’ (kanda), a matrilineal descent group that divided into ‘houses’ and lineages (Bentley, 1887, MacGaffey, 2000, pp. 62–3, 71–2). The families listed in the Encoge census were residential units, or households, where husband and wife normally belonged to different matrilineages. The Kikongo term that most closely described such household units was wantu awanso bekalanga oku lumbu lumoxi, which translates as ‘the persons of a family who live in the same compound’ (Bentley, 1887 p. 111). Furthermore, Africans used terms like brother, cousin, or uncle more flexibly than Europeans did and, indeed, one can wonder if people in Encoge understood or agreed with the strict biological meaning of such terms that the census implied.

More fundamentally, as David Sabean (1983) has argued, ‘any lists created by colonial authorities are “frames” imposed on social reality from outside. If one does not subject them to radical query, family history … fails to deal with the relational aspects of family life – mutual claims and conflicts – and remains a statistical artefact’ (p. 167). Similarly, Megan Vaughan (1983) has warned that formal notions of the ‘household’, as represented in this census, do not capture the ‘reality of family structures’ (p. 283). Lists like those created in Encoge, although detailed, notably obscure the status and origin of family members and the nature of the relationships between them. Concerning the mobilisation of labour, moreover, it is important to recognise that a narrow focus on the household risks cutting out from view the social and economic organisation of the wider community, including cooperation between families and between neighbours. In this regard, the Encoge census is a fascinating document, as the incorporation of nephews, nieces, cousins, and wards points to labour transfers within broader social networks. It is therefore a useful entry point for the study of labour organisation in Encoge, which was primarily family-based.¹⁸

Two other question marks hover around these registers. One concerns the notable absence of enslaved and pawned persons, which relates to Vaughan’s point that the language used in European models of the family covers up the status and origin of individual household members. To find an explicit reference to bondage in a colonial population count from early twentieth-century Angola would be surprising given that Portugal had legally abolished all forms of servitude in its imperial domains in 1875. However, slavery and pawnship remained core features of social life in west-central Africa (MacGaffey, 1970, pp. 215–22, 2008; Vansina, 2005; Vos, 2015), so it would also be strange if none of the 204 households in the census included slaves or pawns.

Archival documents from Encoge in the years around 1900 show that abduction, coercion, and extortion were mechanisms commonly applied by men with access to political and military power to exact labour from others.¹⁹ This might not be a fair reflection of everyday life in Encoge at the time, but such instances of plunder existed within a broader context of practices of controlling and transacting people for productive purposes and as surety for debt. In 1899, for instance, a group of Encoge residents claimed it was perfectly legitimate for an administrator to accept slaves as payment of fines and levies.²⁰ The question is where – under which category – to find the subjects of such transactions in the census. Heads of household might have listed dependants of slave origin as servants; but other family members, especially wives, were possibly obtained by purchase or otherwise enslaved as well. The same fuzziness applies to pawned persons. Leaving the problem of census taking aside, these categories were often obscure in their immediate, lived environment, as social practices could make social categories unstable. As Susan Herlin Broadhead (1997) and Gareth Austin (2003) have shown, for instance, normative distinctions between free, enslaved, and pawned wives became muddled when the development of commercial agriculture in western Africa transformed marriage rituals and labour practices.

Secondly, it is important to remember that the bulletins do not mention the economic activities households engaged in. Coffee growing was widespread in the area, but it is difficult to say which households were involved in it and which were not. The obvious point to make is that, given the labour requirements of commercial agriculture, larger families were more likely involved in it than smaller ones. However, since coffee cultivation required little capital investment and easily combined with subsistence farming, a small family could with minimal effort cultivate a small coffee bush and contribute their harvest to the collective output of the village. With these caveats in mind, historians can use the registers as a lens to study the constitution of households in an African cash-crop economy at the start of the twentieth century.

4. Results and discussion

The families listed in the census were residential units that were part of larger lineages or clans and, through clan membership, had access to one or more plots of land for subsistence or cash-crop farming. At the heart of most families stood a husband and wife, most often accompanied by a single child (Table 1). Considering that a fundamental social constraint placed on women in western Africa has long been the expectation to produce children (Cooper, 2019), this prevalence of one-child families is surprising. While a full explanation for this finding is beyond the remit of this article, potential causes, suggested by European eyewitnesses, include the smallpox and sleeping sickness epidemics that ravaged northern Angola at the time. Concerned about the impact of disease on family life and larger demographic trends in the region, these observers believed that these diseases compounded the effects of high infant mortality and low fertility rates (Laman, 1957, p. 18; Coghe, 2022, p. 46). At the same time, it is worth remembering that small nuclear families were a feature of rural life in northern Angola from at least the seventeenth century (Thornton, 1983), although families of three were usually not the norm.

Most husbands had just one wife, although there were forty men with multiple wives, adding labour and reproductive power to their household. Four of these men had three wives, while only one (a farmer named Afonso Lopes Paim, on whom more below) had four wives. Thus, while monogamous relationships were the norm, adding a second wife was an accepted way for men to expand their household, albeit not the most common one.

Unsurprisingly, the most common form to enlarge a family was through natural reproduction: 94 families included one or more biological children. In addition, 52 households adopted one or more wards; 26 households took in one or more nephews or nieces; and 14 households included one or more servants. Taking in one’s brother, sister, cousin, or uncle was not very common. Adoption strategies clearly supported the transfer of junior members between families. While strength may have resided in youth, however, grandchildren and stepchildren did not feature prominently in the everyday affairs of most rural households. The low frequency of grandchildren in the census was probably because, if people were old and fortunate enough to have them, they would normally have lived with their own parents instead of their grandparents. The sixteen stepchildren listed in the bulletins were part of only seven families, which were mostly of a larger than average size. It seems that the integration of stepchildren in households often resulted from remarriage after divorce or spousal death, as in only one of these cases did the male head of household have more than one wife.

The most striking feature of family life in northern Angola was that, leaving biological children and the adoption of second wives aside, strategies of family extension revealed a notable male bias. Households that incorporated individuals from wider kin-groups or extra-kin networks had a clear preference for male over female wards, nephews over nieces, and male servants over female ones. This destabilises the notion that in most African societies, ‘increasing production … depended more on acquiring female labour’ (Robertson & Klein, 1997, p. 10). By incorporating new dependants, this particular population incidentally produced a skewed gender ratio, overturning the female bias observed in most Angolan population counts before 1950 (Vos, 2014; Vos & Matos, 2021).²¹

Household heads in Encoge potentially valued male dependants more because young men could help to clear underbrush for coffee growing and join their elders on trade expeditions to the coast, while boys could work alongside women and girls in picking and sorting coffee. Note, here, that bush clearing was not a onetime operation to get a farm off the ground. Families often temporarily abandoned cultivated coffee bushes and then thickets and creepers quickly intertwined with coffee trees, which had to be removed when restarting cultivation. Rural families did different kinds of work, of course, including domestic chores, food and cash-crop production, and sometimes trade, to which men, women, boys, and girls all contributed in culturally conditioned ways. If the labour requirements of coffee production influenced family-extension strategies in Encoge more than other needs did, then the census data suggest that, as an agricultural activity, coffee did not discriminate against men or boys.

The categories of ward and servant merit further attention, because the Portuguese terms for them were ambiguous and they may have included pawned or enslaved people. Wards appeared in the census as pupilo (male) and pupila (female). The meaning of these terms has changed over time in the Portuguese language. In Raphael Bluteau’s Vocabulário portuguez e latino (1712), a pupilo was an orphaned minor. According to António de Moraes Silva’s reworking of Bluteau’s dictionary (Moraes Silva, 1789), by the end of the eighteenth century, the term referred to an orphan under tutelage. In early nineteenth-century Brazil, the term had a wider meaning, referring to minors under tutelage in general, not necessarily orphans (Silva Pinto, 1832). The twentieth-century Portuguese-English dictionary Novo michaelis: Dicionário ilustrado (1960) translates the word alternatively as ward, underage child or orphan in charge of a tutor, and protégé. In light of these definitions, the census clearly indicates that child fostering or guardianship was common in northern Angola in 1910, as it still is in the twenty-first century.

Servants mainly appeared in the census as criados (male) and criadas (female) and occasionally as serventes (male or female). The difference between the two terms is not entirely clear, except that criado has a personal connotation, whereas servente implies a more formal relationship between master and servant. Silva Pinto (1832) describes criado alternatively as a person who serves in return for some kind of payment or compensation, like a domestic servant, or as a person who has been raised and received education in someone’s household. Either way, the term implied a servile or subordinate position in relation to the head of the household. As António de Almeida Mendes (2022) points out, in Portugal, criado became a descriptor of Black people working in a servile relationship after the formal abolition of slavery in 1773. The term denoted a legacy of slavery in a society where slavery signified submission through attachment (afilhar) of outsiders to a household (casa), a form of enslavement that many Africans in Angola would have understood.²²

In Encoge, rather than a formal relationship of serving for payment, criado would have implied dependency of an outsider, potentially an enslaved person, on their host family. Two population registers from Encoge in 1866 and 1877 demonstrate how the Portuguese language of criado intersected the official abolition of slavery (in 1869) and servitude (in 1875).²³ The first register, from 1866, counted 91 criados de servir, none of whom were considered free (livre) or freed (liberto); instead, they were all enslaved. It is noteworthy that 54 of them were male, a situation contradicting the supposedly female nature of African slavery.²⁴ The second register, from 1877, counted 71 criados de servir. Of these, 50 were ‘free’, which probably meant they were new recruits, incorporated into households after the abolition of slavery. Furthermore, 21 served under tutelage (tutelados), suggesting they had moved from a condition of legal enslavement to a new kind of dependency, which in practice may have done little to improve their social position within the house. Note, again, that 55 of these 71 servants were male. In short, the category of criado emerged from a process of abolition that substituted enslaved people with supposedly ‘free’ servants, which started in Portugal in the late eighteenth century and continued in Africa a century later.

It is striking that people labelled servants in the census bulletins accounted for a mere 4.4% of all household populations combined. In Encoge, in 1910, servants were mainly part of larger than average families, which was partly because their inclusion helped create these larger households. In this regard, it is worth examining in more detail the households of those men the census described as farmers, landowners, and, in one case, substitute administrator, who had probably gained their privileged positions through involvement in commercial agriculture. Note that not all these ‘notables’ had extensive families and that, at the same time, there were men and women labelled in the census as workers who led households of considerable size. However, Table 2 shows that the families of the local nobility were, with an average of seven members, relatively large. Their larger size was especially due to a higher incorporation of servants (15.3% of the population) than was normal in Encoge.²⁵

Table 2. Households of Encoge farmers (agricultores), landowners (proprietários), and administrators.

List no.	Name	Age	Occupation	Family size	I	II	III	IV	V	VI	VIII	VIII	IX
3	António Bernardo dos Santos e Silva	42	Farmer	11	1		2			2	1	2	2
5	Miguel Gomes Moral		Farmer	11	1	1	1	4		3
63	Lourenço Pedro de França		Farmer	8	1	2							4
64	Jerónimo Álvaro		Farmer	8	2	3							2
106	João Moniz da Fonseca	33	Landowner	5	1							3
116	Afonso Dias d’Estrada		Landowner	5	2	2
122	Boaventura Gomes Moral		Farmer	2	1
141	Salvador Machado Vieira		Farmer	7	1	5
147	Manoel Gomes Moral		Farmer	2	1
168	João Fernandes Maquaca		Farmer	2								1
169	Afonso Lopes Paim		Farmer	12	4	5							2
174	Garcia Gomes Moral	54	Substitute administrator	14			2			4		1	6
180	André Luís Pinheiro da Cunha	48	Farmer	8	1	2				1		1	2
181	Domingos Caetano Pinheiro		Farmer	4	2							1
182	Agostinho Luis Pinheiro da Cunha	26	Farmer	4	1							2
183	Lázaro Dias Barreto	25	Farmer	4	1	1			1
197	João Rodrigues de Sá		Farmer	9	2	2						4
201	João Rodrigues da Costa	68	Farmer	8			4			1		1	1
	Total			124	22	23	9	4	1	11	1	16	19
	Average			6.9
	%				17.7	18.5	7.3	3.2	0.8	8.9	0.8	12.9	15.3

I – wives; II – children; III – grandchildren; IV – stepchildren; V – brothers/sisters; VI – nephews/nieces; VII – cousins; VIII – wards; IX - servants

A close-up look at the family lists of the eighteen local notables reveals several patterns in the dynamics of household formation, especially regarding the origins of wives, wards, and servants (Table 3). First, there were occasionally notable differences between the backgrounds of first and second wives. By studying patronyms, it is possible to identify marriages between ruling families in Encoge. Whereas some first wives were daughters of notable families, like the Gomes Moral, second wives never seemed to have such distinguished backgrounds. However, many established farmers took wives with unassuming Portuguese or African names, indicating commoner origins. In the case of Afonso Lopes Paim, who had four wives, the names of their offspring seemed to reveal these women’s modest backgrounds, as only one of his children, Afonso Lopes Paim Jr., inherited his patronym while the others carried single names: Lemba, Domingas, Branca, and Gonga.

Table 3. Names of wives, wards, and servants in farmer (agricultor) households.

List no.	First wives (15)	Second wives (7)	Wards (16)	Servants (19)
3	Suzana José		Gonçalo Lisboa da Silva; Lucia Victoriano	Pedro Jacubanga; Inveja (‘envy’)
5	Lucia Lopes Paim
63	Tumba			António; Augusto; Pandulha (‘stomach’); Gingo
64	Esperança Gomes Moral	Ana António		Panzo; João
106	Macabi		Pascoal Matoso da Cunha; Malongo; N’Dua
116	Francisca Gomes Moral	Zumba
122	Maria Manoel
141	Buanga Papa
147	Antónia Francisco
168			Quissala
169	Beia	Beia; Cutola; Milando		Bongi; Curiva
174			Ana Dias de Estrada	Joana; Pouca Roupa (‘few clothes’); Intriga (‘schemer’); Manoel; António Sebastião; José
180	Caginda		Casimiro Dias Barreto	Maria; Calo
181	Lubamba	Quizonga	Feliciano
182	Dembue		Catumbo; Teresa
183	Magdalena
197	Lucia Álvaro Fernando	Luzembo	Pedro Pereira Gomes da Cunha; Esperança Pinheiro da Cunha; Isabel de França Biança; Umba Mabiembe
201			Doqui	Gonga Culungo

It is furthermore interesting that two male elders, Garcia Gomes Moral (aged 54) and João Rodrigues da Costa (aged 68), declared to have no wife – although they had grandchildren and were thus likely widowers – but listed a female servant second on their bulletins. Usually servants appeared lower down the lists, which suggests that these two men had taken women of servile status as female companions, but did not regard them as wives. Garcia Gomes Moral appears in the colonial archive more often, showing a career with plenty of opportunities for plunder. He was a corporal in the colonial army for six years before becoming a substitute administrator in 1899, a position he temporarily relinquished under accusations of corruption. In the early 1900s, he conspired with another notable character in the Encoge community, military officer Pedro Rodrigues Barroso, someone repeatedly tried for crimes like theft, extortion, and kidnapping. Although Moral had local enemies and enjoyed a poor reputation among the Dembo chiefs near São José de Encoge for his association with Barroso, he retained his powerful position within the parish community. His household was the largest in the census and included six servants, which, in light of his military, political, and criminal activities, he had likely obtained by enslavement.²⁶

Secondly, wards often came from well-known families. The nature of such transfers of dependants between elite families remains obscure, but they probably involved consent rather than coercion. By contrast, the names of servants generally point to humble origins. Many went by a single name and some were primarily known by a nickname – ‘Envy’, ‘Few Clothes’, ‘Schemer’, ‘Stomach’ – thus lacking an identifiable family connection. This suggests there was a clear distinction between child-fosterage and servility, with wards holding a more respectable position in Encoge households than servants did. In fact, a criminal process in 1903 against morador Gonçalo Rodrigues Barroso, who had killed his ward, Manoel Barroso Comboio, indicates that wards – or their families – enjoyed legal protection against harm.²⁷

Finally, it is worth pointing out that several farmers made sure that key members of their household – mainly close relatives – also gained official recognition as farmers and became exempt from colonial duties. Such was the case with the mestiço family of António Bernardo dos Santos e Silva, whose wife, nephew, and cousin all enjoyed special status in the colonial system. The same happened in the families of Lourenço Pedro da França, Jerónimo Álvaro, Boaventura Gomes Moral, Manoel Gomes Moral, Garcia Gomes Moral, and João Rodrigues de Sá. However, not all farmers had the means or the motivation to provide close relatives with protected status. In some families, the male head of household was the only person not listed as ‘worker’; in others, privileges only extended to a few dependants. For instance, the wife and daughter of Miguel Gomes Moral had the status of farmer, but his stepdaughters did not. The wife of Lázaro Dias Barreto was officially a farmer, but his 10-year old brother was listed as a worker. In both cases, the wives descended from established families and might have gained protected status before marriage. However, the fact that many households in the Encoge census pertained to elite families without enjoying the benefits of farmer status is a reminder that a famous patronym, which in past times symbolized an affiliation with the Portuguese empire, was not a guarantee for protection against colonial exactions.

5. Conclusion

Colonial archives paint a picture of São José de Encoge as a tightly knit rural community in northern Angola, where wealthy families intermarried, but also competed with one another, sometimes violently, for political and economic resources. Despite low world-market prices, coffee was the main source of wealth in 1910, which gave farmers access to foreign goods and a means to pay the hut tax. Coffee was a cash crop many of them cultivated alongside basic food crops like manioc, beans, and maize. Some households were also involved in trading wild rubber, selling this along with coffee down the caravan routes to the coast. The families that made this happen ranged in size from two people to more than a dozen. The most common composition was a husband with a wife and one child. Further extensions mainly happened by natural reproduction (another child), the integration of wards, nephews, and nieces, or the acquisition of a second wife. Enslaving others was not very common. Indeed, when it occurred, it was the privilege of a select group of successful farmers and traders.

The series of family bulletins drawn up in the São José parish in 1910, forming the empirical basis of this analysis, have shed a unique light on African household formation in northern Angola in the early colonial period. First, they have helped demonstrate which practices of family extension were most common. Secondly, this article has used them to assess whether there was any gender bias in such practices. It has argued that, as families exchanged wards, parents placed their children under the care of relatives, and the wealthiest households incorporated vulnerable outsiders as servants, host families showed a remarkable preference for receiving young males rather than young females. This conclusion runs counter to a long historiography of slavery, pawning, and child fostering in nineteenth- and twentieth-century sub-Saharan Africa that has assumed that these practices especially aimed to integrate women and girls into households. The labour tasks involved in coffee and rubber production – clearing underbrush, picking berries, trading, and porterage – are a possible explanation for this male bias. It is too early to conclude that coffee cultivation created its own distinctive labour regime, but the census data from Encoge suggest there was more regional variation in the way that African families in this period organised, structured, and gendered rural labour than previously assumed.

The article agrees with scholars who have tried to disentangle fosterage or guardianship from servility. The Encoge census materials indicate that people taken in as wards often came from the wider kin and community networks of the host family, implying a duty of care and training. By contrast, household members labelled as servants often lacked such connections with their hosts and in some cases even appeared disconnected from kin altogether, revealing a basic vulnerability in a society where one’s lineage provided a measure of safety and welfare. A fuller, network-based statistical analysis of the census data might enable future scholars to map out in more detail the intricate connections between Encoge families manifested in the multiple transfers of nephews, nieces, and wards.

Acknowledgments

I would like to thank my colleague Alex Shepard and the participants in the Rebuilding Life after Slavery workshop at Radboud University (28 March 2022), especially Wouter Raaijmakers, for their feedback on a first draft.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

This work was supported by the Leverhulme Trust under grant RF-2019-194.

Notes

1. The historiography of the family in Africa has generally concentrated on the nineteenth and twentieth centuries, as Europe’s growing presence in the continent in this period produced the sort of documentary evidence historians needed – beyond other methods – for researching African domestic affairs. For a longer-term perspective, see, Osborn (2011).

2. Slave-trade studies have produced some alternative histories of the family in precolonial proto-urban settings. See, for instance, Havik (2015) and Curto (2021).

3. Compare ‘belonging’ with the social isolation in child slavery as stressed by Greene (2017).

4. For quantitative evidence of female pawnage in 1940s Asante, see Allman and Tashjian (2000, p. 72).

5. Arquivo Nacional de Angola (henceforth ANA), caixa 4999, Mapas estatísticos do presídio de Encoge, segundo semestre de 1855; ANA, caixa 3925, Mapa estatística da população, Encoge 31 July 1866.

6. Arquivo Histórico Ulramarino (Lisbon), SEMU-DEGU, no. 645–1, Relatório do concelho de Encoge, 1 December 1875, enclosed in Governador geral interino to Ministério do Ultramar, no. 328, Luanda 24 December 1875; ANA, caixa 4997, Chefe do concelho to Secretário geral, no. 127, Encoge 22 July 1901.

7. ANA, caixa 3387, Chefe do concelho, Relatório do concelho referido ao mês de janeiro de 1906, Encoge 31 January 1906.

8. ANA, caixa 3387, Nota dos preços dos géneros coloniais neste concelho (Encoge 31 March 1900), enclosed in Administração do concelho de Encoge to Secretário geral do governo geral, no. 46, Encoge 31 March 1900.

9. ANA, caixa 5000, avulsos 1911–1916, imposto de cubata.

10. ‘There are no plantations in this district’, said the administrator of Encoge in 1887. ANA, caixa 4998, Pedro Rodrigues Barroso to Secretário geral, no. 247, Encoge 12 September 1887.

11. ANA, caixa 375, Listas de família, São José de Encoge, 31 December 1910 (nos. 65–204); idem, caixa 4999, Listas de família, São José de Encoge, 31 December 1910 (nos. 1–46, 48–64).

12. ANA, caixa 4997, Guilherme Vandunem, tomada de posse, Encoge Sep. 1902.

13. A later colonial source claimed the 1910 census never produced any returns (Portugal, Repartição técnica de estatística geral da colónia de Angola, 1941, p. 29), a claim contradicted by the archival evidence, but it suggests the census overall was a failure.

14. It is not clear to what extent these figures were estimates. According to a count in 1906, the parish consisted of 163 settlements, containing 7,505 homes and 34,264 inhabitants, of whom 43.7% were male. ANA, caixa 3387, Mapa de população, freguesia de S José de Encoge, 1906.

15. ANA, caixa 4999, pasta 4, Questões entre gentios, 1901.

16. In 1916, for another census, the commander of the Encoge district was only able to count 490 out of an estimated 10,000 homes. ANA, caixa 3902, António Albino Aleixo, Relatório, Ano económico de 1915–1916, Encoge 30 September 1916.

17. Table 1 shows a total population of 794. For one family (no. 12), headed by Sebastião Dias d’Estrada, the relationship to his three dependants (Maria Rodrigues de Sá, Luis Dias d’Estrada, and Lobo) is unknown. These three dependants have therefore been excluded from the table.

18. Low coffee prices in this period provided growers little margin to hire seasonal labour, but they may have relied on neighbourhood labour during the harvest.

19. For example, ANA, caixa 814, Francisco António da Silva to Governador geral, Requerimento, Luanda 16 June 1902, enclosed in Chefe de Encoge to Secretário geral, no. extra, Luanda 27 June 1902; ANA, caixa 4998, Juiz ordinário to Procurador da Coroa e Fazenda, no. 19, Encoge 11 January 1887.

20. ANA, caixa 4999, Moradores to Governador geral, Encoge 15 June 1899.

21. A census taken in 1913 showed an Encoge population of 1,273, of whom 646 (50.7%) were male (Ferreira Diniz, 1914, p. 14).

22. For an interpretation of slavery as a process of decreasing marginality, see, Kopytoff and Miers (1977).

23. ANA, caixa 3925, Mapa estatística da população, Encoge 31 July 1866; idem, Encoge 22 December 1877.

24. A possible explanation for male bias in the servile categories is that households could integrate female slaves as wives, whereas male slaves were at best ‘perpetual cadets’ (Meillassoux, 1991, p. 14). In the absence of historical evidence, however, this hypothesis is no more than speculative.

25. If the extensive household of Garcia Gomes Moral, the substitute administrator, were excluded from these tabulations, the proportion of servants would be lower (11.8%).

26. ANA, caixa 3387, Administrador substituto to Secretário geral do governo, Encoge 10 July 1902; ANA, caixa 3902, Chefe Luiz António Nobre Leite to Secretário geral, no. 78, Encoge 1 May 1901; ANA, caixa 4998, Requerimento, Encoge 16 February 1887.

27. ANA, caixa 4997, Processo-crime contra Gonçalo Rodrigues Barroso, Encoge 21 February 1903.