Introduction

In a memorable scene from the 1975 film Monty Python and the Holy Grail, King Arthur meets two field workers and asks who their ‘lord’ is. One of them, Dennis, infuriates him by claiming to live in an ‘autonomous collective’ and questioning his kingly authority. Fed up, Arthur manhandles him:

Shut up!

Oh! Come and see the violence inherent in the system! Help, help! I’m being repressed!

Bloody peasant!

Dennis’ intellectual engagement with his ‘peasant’ status, while comical, recasts him from a subordinate object to a social agent. Scholarship has tended to treat the ancient Roman peasant as the former. Subsistence, self-sufficiency, and dependence on social superiors are conventional in discussions about the rural poor, limiting our understanding of the peasant to a spectrum of socio-economic independence and dependence.

In this article, I consolidate the key literary and archaeological evidence which suggest the importance of social relationships between peasants in the Late Republican and Early Imperial Italy. After discussing the prevailing tendencies in our theorisation of the peasant and their shortfalls, I focus on Dio Chrysostom’s Euboicus and three sites from central Italy – Site 9 from the Ager Lunensis survey, Site 154 from the Ager Capenas survey, and Case Nuove from the recent Roman Peasant Project. Despite significant methodological issues regarding both kinds of evidence, I suggest that hospitality, reciprocity, and cooperative work form a baseline of reasonably-expected horizontal social relations among peasants which have been neglected. However, better and further consolidation of evidence is needed before we can re-evaluate the significance of horizontal social relations within broader historical narratives.

Theorising the Roman Peasant

The peasant — ubiquitous in agricultural, demographic, and economic histories — is widely accepted on the basis of a dichotomy between economic independence and dependence, but little understood as a social agent (there is no direct Latin equivalent for ‘peasant’, but see Kolendo 1993: 199–200; Osborne and Rathbone 2006; for recent demographic work, see de Ligt and Northwood 2008; Launaro 2011; de Ligt 2012; Hin 2013). Narratives of the Late Republican rural landscape have long centred on opposing free peasants and slave-run villa estates (de Ligt 2006: 590–598; Launaro 2017: 85 n. 2). This draws a clean line between ‘peasant’ and ‘capitalistic’ economies (de Neeve 1985: 94–96; 1990: 364–365) with a predominant focus on production — especially factors of production and labour — which is part of a broader tendency in Roman socio-economic history from White (1967; 1970a; 1970b; 1975) to Kehoe (2007), Kron (2012; 2017) and Bowman and Wilson (2013b). On the one hand, the ‘slave mode of production’ linked to money-making, market-oriented villas, owes it pre-eminence to Marxist influences (Witcher 2006b: 41). On the other hand, the notion of the peasantry, where the household determines production and labour inputs, has been influenced by A. V. Chayanov’s work on Russian peasants (Thorner 1966; Erdkamp 2005: 58; Bernstein 2009: 56–63; Grey 2011: 42; McCarthy 2013: 7–8).

Fundamental to the peasant concept are subsistence and self-sufficiency (White 1970a: 416; Frayn 1974: 12; 1979: 15–16; Evans 1980: 144, 162–163; Garnsey 1988: 48; de Ligt 1990: 43, 49; Kolendo 1993: 200; Pleket 1993: 333; Morley 1996: 75–76; Finley 1999: 107–108; Erdkamp 2005: 50; Garnsey and Saller 2014: 101; Hopkins 2017: 218–219). In Roman studies, these are not consistently or even explicitly defined (Hollander 2019: 1–4). They sometimes refer to living conditions (‘peasants lived at subsistence levels’). At other times they refer to goals, ideals, and mentalities (to the tune of ‘peasants sought to be self-sufficient’; Finley (1999: 108) asserts that even wealthy landowners had a ‘peasant-like passion’ for self-sufficiency). In essence, subsistence farming is the minimal production for survival (Garnsey 1988: 48; 1999: 25). This made the household self-sufficient, since it needed no outside food, tools, or goods. Juxtaposed with this economic independence is market involvement or dependence, even though this juxtaposition was recognised to be unhelpful early on (Thorner 1987: 65). We find this polarisation in Finley’s and Hopkins’ influential models. The former states plainly that ‘subsistence farming is by definition not market farming’ (Finley 1999: 107). While he accepts commercial trade in the Greco-Roman world (Launaro 2016: 237–238; more recently Bang 2007: 25), the extent to which Finley thought peasants participated in it is not clear. Hopkins claims that 80 to 90% of ‘self-sufficient production always stood outside the market economy’, but that ‘this cellular autarky of individual peasant farmers… was penetrated, but not pervaded by outside demands’ (Hopkins 2017: 219).

Scholarship has broadly accepted this dichotomy. Exceptions are few (Foxhall 1990: 99–100; Witcher 2006a: 3), and one offshoot of this acceptance is the argument for peasant ‘rationality’ in pre-modern circumstances (Halstead 1987: 86; 2014: 329–354; Foxhall 1990: 113; Pleket 1993: 334; Morley 1996: 75–76). In comparison, less attention has been paid to social autarky (Evans 1980: 136; Kolendo 1993: 200; implicit in Morley 1996: 58) and to challenging it (Garnsey 1998a: 96; de Ligt 1991: 35; Pleket 1993: 333; Witcher 2006a: 5; Ghisleni et al. 2011: 135–136). While scholarly opinion has shifted towards arguing that peasants were more market-involved or dependent than previously envisaged (Evans 1980: 144; de Neeve 1985: 95; Foxhall 1990: 113; de Ligt 1990; 1991; 1993: 110; Morley 1996: 165–174; Erdkamp 2005: 101, 134–142; Hollander 2019: 3, 83–96), subsistence and self-sufficiency are themselves retained, the former as a baseline for production which was exceeded, the latter as an ideal or goal that was unattainable (Erdkamp 2005: 96; Silver 2008: 16; Hollander 2019: 3). Horden and Purcell (2000: 272), however, reject all assumptions of subsistence, self-sufficiency, and autarky (followed by Witcher 2006a; Bowman and Wilson 2013a: 22 in passing; Hollander 2019: 1–4). From their ecological point of view, the omnipresence of risk necessitates overproduction; ‘to aim at subsistence is suicidal’ (Horden and Purcell 2000: 272). Likewise, the ‘independent producer is an inherently unlikely specimen… where diversification, storage and redistribution are rendered essential’ by environmental unpredictability (Horden and Purcell 2000: 274). Their alternative conceptualisation remains, however, based on an independence-dependence polarisation, albeit on broader socio-economic terms. It involves ‘burdens of dependency’ like confiscation, tenancy, and slavery at the hands of the ‘powerful’, among whom are not just wealthy landowners, but even those we consider as ‘peasants’ (Horden and Purcell 2000: 276–277). This dependency echoes the prominent definitions of the peasantry from sociology and anthropology. These invariably stress unequal power relationships of exploitation which brought about the removal of surplus production from peasant households (Shanin 1971; 1973; 1974; Wolf 1966: 3–4, 11–17; Ellis 1993: 6, 13; for a critique, Rosenstein 2008: 193 n. 3). Romanists continue to adopt these essentially negative definitions which can coexist neatly with subsistence assumptions (Garnsey 1988: 44; Finley 1999: 105; van Dommelen 1993: 172–173; McCarthy 2013: 8; Gallego 2007 gives a Greek take), thus preserving them as unchallenged paradigms, despite recent positive re-framings of peasant identities (Bernstein et al. 2018: 701; see Bernstein and Bryes 2001 for a summary of recent peasant studies).

All this means that scholarship on the rural poor remains restricted to an interpretative spectrum between socio-economic independence and dependence. When social relations are considered, emphasis is given to dependent relations, those outlined by Horden and Purcell, and others like patronage (Garnsey 1988: 58–62; Garnsey and Woolf 1989; Witcher 2006a: 8–9; Terrenato 2007 on clan relations; Hollander 2019: 89). Despite the consistent recognition of the variety of peasant conditions, tidily put by Garnsey as ‘there was no typical ancient peasant’ (1999: 28); this emphasis constrains us to a one-dimensional image of the Roman peasant. Horizontal or peer-level social relations are understudied (Grey 2011: 16), in part perhaps because Chayanov himself ‘lacked any significant theorisation of social relations’ (Bernstein 2009: 65). When cooperation and reciprocity are analysed, they are mainly conceived of as survival strategies in desperate times: kin, neighbours, and friends are only risk-buffers when a household’s production became insufficient (Garnsey 1988: 56–57; Garnsey and Woolf 1989: 155–157; Halstead 1989: 73–74; Erdkamp 2005: 98). Exceptions include Lirb’s contribution (1993) on agricultural partnerships, while others duly note peasant participation at community events and their use of urban facilities (de Ligt 1991: 35; Erdkamp 2005: 96–97; Hollander 2019: 87–88); Grey (2011: 46–89) recently extensively analysed the community dynamics of Late Antique peasants. There remains scope to explore these in Italy and in the Late Republican to Early Imperial period. And while ‘non-elite’ encompasses a wide variety of circumstances, I concentrate on those who lived and worked on smaller rural sites rather than villas or urban areas.

Further reason to consider horizontal social relations comes from contemporary social science. The tendency to view relations as hierarchical or survivalist does not cover the full range of human motivations. Some economists recognise that a ‘positional’ motivation, that is, a desire to ‘gain a better position than other[s]’, which in turn fosters competitive interactions, is not the only kind of orientation people assume (Sacco et al. 2006: 699–703). Instead, there is also a ‘relational’ orientation, ‘a desire to get closer to someone else’ (Sacco et al. 2006: 700). From this perspective, social interactions produce relational goods, goods which are produced, consumed, and enjoyed simultaneously by participants of the interaction; examples include solidarity, friendship, social approval, and identity (Uhlaner 1989: 253–256; Gui 2000: 152–155; Sacco et al. 2006: 704–706; Donati 2019, providing an alternative definition to Uhlaner’s). This is not to say that relationality is straightforwardly positive, as someone may adopt a relational disposition for personal advantage (Gui 2000: 143; Sacco et al. 2006: 704). Nevertheless, relational dispositions alert us to other facets of socio-economic life we take for granted. To explore these in our context, I turn to our literary and archaeological evidence.

Finding the Italian Peasant

These evidences are difficult to interpret; I first consider two methodological problems concerning the literary texts commonly used in relation to the rural poor. I discuss the written evidence here because in shaping both our theorisation of the non-elite and our interpretation of the archaeological evidence significantly, they raise issues the historian and archaeologist cannot pass over. Firstly, while the agronomists — Cato, Varro, and Columella — are peculiar to our period and place, the applicability of their descriptions and prescriptions to the non-elite is debatable (The ‘ayes’ are Kron 2012: 104–105; 2017: 119–123; Hollander 2019: 4–9. The ‘noes’ are White 1970a: 14–31, 34–37; 1970b: xii–xiv; Brunt 1972: 154; Garnsey 1998b: 108–109; Morley 1996: 109). For example, Cato advises us to be good neighbours (De Agricultura 4 and echoed by Pliny, Naturalis Historia 18.44) and to pay attention to neighbourhood conditions (Agr. 1). Varro warns us to pre-empt neighbourly quarrels (De Re Rustica 1.8.8), while Columella seems to prefer self-reliance over borrowing from a neighbour (De Re Rustica 1.8.8). While it would be crass to deny this common-sense advice general applicability in non-elite contexts, they do not provide sufficient detail to contextualise these behaviours in such circumstances securely.

Secondly, we have to be sensitive to authorial self-presentations and literary constructions in texts we once accepted as realistic portrayals of rural life. This applies not just to the agronomists (Terrenato 2001: 24–25; Reay 2005; 2012; Bertoni 2017; Doody 2017; contrasting Steiner 1955; White 1970a; Frayn 1979; overviews include Witcher 2016: 459–460; Hollander 2019: 4–9), but to other texts we commonly refer to as well, from Juvenal 14.161–171 (Evans 1980: 134) and Martial (Kron 2017: 135–136) to Virgil’s Georgics, which in particular has been consistently dismissed as unhelpful to the historian (White 1970b: xvi; Kolendo 1993: 202; Hollander 2019: 9). The Moretum in the Appendix Virgiliana and Dio Chrysostom’s Euboean discourse (Orationes 7) are at times still used to illustrate the lives of the rural poor (Evans 1980: 143; Erdkamp 2005: 55–105; Hollander 2019: 9 is more sceptical). This is despite literary assessments to the contrary. Hellmann argues that the Moretum-poet ‘integriert in diese jedoch alle zentralen Charakteristika, die in der poetischen Tradition dem Land zugesprochen wurden’, ‘incorporates in this, however, all the central characteristics which were attributed to the country in the poetic tradition’, including peace, autarky, and freedom from the adversities of urbanity (Hellmann 2004: 11; see also Gowers 1993: 46–48; Fitzgerald 1996; Dossey 2010: 69–70). Dio’s Euboicus, once cited as the best evidence for the peasant economy (Evans 1980: 143), is often deemed fictitious or a mixture of real experience and fiction, since its depiction of rural life is bound to the narrator’s rhetorical purposes and philosophical dispositions (Brunt 1973; Jones 1978: 56–61; Russell 1992: 8–13; Anderson 2000: 146–149; Brenk 2000; Ma 2000; Whitmarsh 2004: 460–463; Bekker-Nielsen 2008: 136–140; Hutton 2015: 3–13; Jackson 2017: 220–222; Bryen 2019: 133–139; Desideri 2019: 178–179).

Realist and literary readings are not mutually exclusive (Desideri 2019: 178), and we might hold that literary constructs are ultimately based on reality (Erdkamp 2005: 57). However, I prefer to make no blanket assertion about a text as a whole. For example, Erdkamp argues that the Euboicus ‘offers a wealth of realistic detail, such as… the location of the shepherd’s camp at a place offering streaming water and sufficient vegetation’ (Erdkamp 2005: 57). But the realism of such basic details does not make others equally plausible: the two shepherds are married to each other’s sisters, and later their children are married together (Or. 7.10, 69–80). This family tree, construed to an extreme degree of self-containment, accords with the underlying rhetorical message which contrasts rural virtue with urban life.

A cautious approach to texts is necessary, because a strong textual hold on historical research has side-lined pertinent social questions. For instance, the literary type of the early Roman citizen cultivator, inseparable from Republican elite moralising, has lent much credence to the theoretical conceptualisation I discussed earlier. This type is exemplified by Cincinnatus, summoned to be dictator while ploughing his four-iugera field, to which he promptly returned after his military successes (Livy 3.26.6–29.7; Dionysius Halicarnassensis, Antiquitates Romanae 10.17.3–19.4, 23.5–25.4; Cicero, De Senectute 56; Pliny, HN 18.4.19–21). His small farm (see Rathbone 2008: 307–311 on the written evidence for such farms), personal labour to feed his family, and a commitment to public service, though only insofar as it was necessary, can all be construed as confirmatory of the ideal farmer. Yet to do so uncritically accepts a glorification of agriculture by elite authors, distorting our image of the peasant (Kolendo 1993: 200–202; Garnsey 1998c: 137–138; Horden and Purcell 2000: 270; Spanier 2010; Jaeger 2015; Van Oyen 2020: 19–25).

In short, significant barriers obscure our view of rural non-elite lives. Nonetheless, literary details can act as signposts for the kinds of interactions we might reasonably expect among rural residents, rather than as direct sources of socio-economic conditions of a particular place and time. Even if the Euboicus is set in Euboea (see Whitmarsh 2004: 460), some details helpfully remind us not to take the norms of friendship (scholarship on which tends to focus on the elite, e.g. Konstan 1997: 122–148; Verboven 2002; 2011), which occurred alongside hierarchical relationships and market exchange, for granted. Working together (Or. 7.15–20), the hospitality offered to the narrator (Or. 7.5–10, 64–66), communal eating, conversation, and the religious practices of the neighbouring families (Or. 7.67–80; evidence for rural religion, particularly regarding the non-elite, is complicated: see e.g. Stek 2009: 187–202 on the Compitalia), give us the boundaries of a preliminary and non-exhaustive range of social interactions about peasants who lived close to one another. These are not trivial interactions, since they are the loci for the production and enjoyment of relational goods. The economist B. Gui notes that ‘informal interactions among neighbours… contribute to perceived quality of life and well-being’ (2000: 148), while Halstead observes that during harvest-time collaborations, ‘moments of light relief and an element of competition kept members of a team working longer and faster than they did alone’ (2014: 110).

The Euboicus points out three expressions of relationality. First is hospitality. When the shipwrecked narrator reaches the peasants’ huts, an unstinting feast is held for the guest (Or. 7.65–66). As it turns out, this generosity is the speech’s chief message (81–82), and it is also deeply embedded in a long literary tradition from the swineherd Eumaeus in Homer’s Odyssey to Philemon and Baucis in Ovid’ Metamorphosis. I am not claiming that hospitality was a behavioural propensity peculiar to the rural non-elite; altruistic gestures of hospitality are attested to in ethnographic accounts (Halstead 2014: 309–311), and recognised as the product of both genetics and socially-learned behaviours (Sacco et al. 2006: 717–722; Verboven 2014: 135–137). In the Roman context, hospitality is also discussed by elite authors as a formalised series of exchanges (Nicols 2011). But when it comes to the rural non-elite, the possibility of hospitality immediately raises the question of productivity — was there enough to share? Our answer depends in part on much bigger claims about ancient agricultural productivity (Kron 2008 and more recent approaches by Goodchild 2013; Heinrich 2017) and production goals. Erdkamp (2005: 96–97) holds that social expenditure to maintain a household’s social status within a community was factored into that household’s production targets. But the capacity to share depended on what was actually produced and stored, rather than just intended goals (on storage generally, see Cato, Agr. 3.2; Halstead 2014: 294; Hastorf and Foxhall 2017: 28). A household’s access to and consumption of different kinds of goods may have also played a part. Some scholars postulate that regulation of access and consumption of material culture were means of elite social control (Appadurai 1986; Witcher 2006a: 7–9; Dossey 2010: 68–71). In a peer-level, altruistic setting, this argument can be reconceived: altruism could have balanced differentials in access to goods between peasants, without eradicating inequality, of course. Whether or not this was the case, the tight connection between the desire to share and the possibility to do so suggests that in a non-elite context, what is ‘social’ and ‘economic’ are quite inseparable.

The thoughts just laid out also apply to reciprocal exchanges. While intensely studied (Kolm 2006; Grey 2011: 75–77), I am focusing on reciprocity as a series of interrelated gifts and less as an economic system. On the point of socially-learned behaviour, reciprocal exchanges of food are common across cultures, but differ in their social significance (Hann 2006). However, our understanding of rural reciprocity is clouded by an often-quoted passage from the Euboicus. While feasting, the guest finds out that one peasant’s daughter is married to a rich man in the village. He asks if the young couple helps the rural dwellers when they are in need, to which the wife replies:

We do not need anything… but they get game from us whenever we catch any, and fruits and vegetables, for they have no garden. Last year we borrowed some wheat just for seed, but we repaid them as soon as harvest time was come.

(Or. 7.69; Loeb translation)

Like the family tree discussed earlier, this passage presents an extreme form of the urban-rural divide fundamental to the narrator’s story. The expectation that peasants might sometimes rely on urban connections is emphatically reversed; the village folk are the ones in need, since they have no garden. This makes the text rather unhelpful in trying to work out reciprocity between rural non-elites. Reciprocity is better understood here as a long-term commitment to others (similarly Grey 2011: 77; Verboven 2014: 143–144). From this perspective, reciprocal interactions are ‘intrinsically uncertain’ (Sacco et al. 2006: 298), firstly because a certain level of trust is required in believing reciprocation will happen, and secondly because the manner of reciprocation is unspecified; different forms of repayment — a meal might be repaid in labour, for instance (Halstead 2014: 309–311) — make it difficult to value what is exchanged. This asymmetry is essential to prolonging a relationship. Being linked to another person or household in this way came in handy should dire need arise, as during poor harvests. More regular and necessary for this survival recourse were protracted relationships which meant more opportunities for time together, the building of trust, and the enjoyment of relational goods. Finally, the variety of forms a reciprocal gesture could take points again to the inseparability between the ‘social’ and ‘economic’ aspects of exchange.

Alongside hospitality and reciprocity are different forms of cooperative work, over and above wage work and other forms of employment (Erdkamp 2005: 81–88; 2016: 36–38). These include joint ownership of resources, shared infrastructural investment (including access to watercourses and drainage; see Lirb 1993: 286), temporary and informal agreements to pool resources, or simply lending a hand. These are especially relevant during harvest time, when time pressure was very high (Erdkamp 2005: 71; Halstead 2014: 110–112). These are admittedly only implicit in the Euboean narrative, for instance when the peasants’ fathers’ lives are described (Or. 7.15–20). But it is in respect of these that Lirb argues that ‘many subsistence farmers were, in one way or the other, partners in agriculture’ (1993: 285). Partnerships opened doors to economic advantages, like economies of scale and ownership of more expensive equipment and facilities unobtainable individually. In addition, the social advantages of working together already mentioned — the alleviation of boredom, a competitive spirit, and so on — also impacted a team’s productivity. In this sense, the line between what is ‘social’ and ‘economic’ is once more blurred.

The literary signposts from Dio’s Euboicus turn our attention towards a spectrum of horizontal social relations — hospitality, reciprocity, and cooperative work — that reminds us of the vibrancy of peasant lives. Alongside considerations of their socio-economic status, hierarchical dependency, and market exchange, we should not neglect these aspects of social life that are intertwined with our economic concerns.

The Contribution of Archaeology

Like our written evidence, our archaeological material is not straightforward to use. Firstly, texts have shaped archaeological research in Italy significantly. The free peasantry versus slave-run estates dichotomy prevalent in historical studies and ultimately derived from Plutarch (Vitae Parallelae, Tiberius Gracchus, 8) and Appian (Bella Civilia, 1.7–13) continues to inform research questions based on a restrictive set of concerns, like conflict, agricultural production, and demography, which omits social relations (Witcher 2006a: 4; 2006b: 39–41; 2012: 14–15; Ghisleni et al. 2011: 95–96; Launaro 2012: 123). At the same time, landscape archaeology, flourishing since the 1950s, continues to classify sites based on a tripartite categorisation of village, villa, and small/medium farm (although this is changing, see e.g. de Haas 2011: 26–31). This elides the variety of sites in the rural landscape, making it harder for us to discern how relationships between sites played out. The initial derivation of this classification from Varro (Rust. 2.10.6) and Columella (Rust. 12.15.1) by the South Etruria survey (Potter 1979: 122) is based on a contentious interpretation. The cited passages show no tripartite division between villa, casa, and tugurium. We have at best a distinction between permanent and temporary residence: Varro differentiates shepherds protected from the rain by villae and makeshift (repetinae) casae. The latter appear hardly distinguishable from Columella’s description of a curved roof ‘in the tugurium-manner’ which covers figs at night. The unsatisfactory reading stresses the point that there is a significant textual hold on our archaeological interpretations, perpetuating the equally-unsatisfactory polarisation between subsistence production in small farms and market farming by villas (de Ligt 2006: 598; Rathbone 2008: 306; Launaro 2012: 22).

As it stands, surveys have identified hundreds of supposed ‘small farms’. Only a few have been excavated, either to supplement survey interpretations or, just recently, to focus on the lives of the rural non-elite (Rathbone 2008: 323; Ghisleni et al. 2011: 95–96 as part of the broader Roman Peasant Project). While various examples are available (Rossiter 1978; Kron 2017), often only ground plans are provided, which by themselves do not allow for cross-site comparisons. As a result, previous syntheses of these small sites have concentrated on architectural form and structural change (Rossiter 1978: 4–17; Rathbone 2008), although Kron (2017) also considers the relationship between small and wealthy farmers in game farming. The data we have do not allow us to discern, with much precision, the interactions I am discussing. To illustrate this, I consider three projects in central Italy (Figure 1).

Figure 1

Map of the Italian Peninsula showing sites discussed (Adapted from LaFleur and Elliott 2000–2001).

First is the Ager Lunensis survey (1979–1981), during which a ‘farmstead’, Site 9, was excavated. The survey conditions were rough: significant soil erosion on the foothills and mountains meant few findspots (Delano Smith et al. 1986: 88–90, 103, 106). Six Roman-period scatters were identified on stratum 2(i) — the ‘lower slopes and ridges facing the coastal plain or dominating the river valleys’ (Delano Smith et al. 1986: 94; Figure 2). Four scatters were approximately 1,000 to 1,500 m² large, but erosion prevented similar data from being obtained from sites 46 to 48 (Table 1).

Figure 2

Ager Lunensis Roman findspots with stratum 2(i) sites labelled in red bold type (Adapted from Delano Smith et al. 1986: 112).

In one cluster (Sites 9, 39, 46/47, and 48), sites were 500 to 800 m apart from each other (Delano Smith et al. 1986: 111). But given the myriad factors affecting the recovery rate of sites (a vast topic, for which see Witcher 2011: 37–42 for starters), we cannot evaluate how these were related to one another. The number of potsherds per site is given, but the quantities are small, and no pottery sections are available. And while a wine or olive press base was found on Site 39 (Delano Smith et al. 1986: 103–104), the precise nature of this site is unknown, as is whether the other sites had similar infrastructure.

Similarly, the excavation of Site 9 yielded information with limited applicability to our questions. The site was chosen because it seemed typical of the hillside settlement (Delano Smith et al. 1986: 111). It was identified as ‘the kind of structure a simple independent peasant household could occupy’ (Delano Smith et al. 1986: 116), with which Rathbone’s assessment, ‘a middling family farm of say 20 iugera’ (2008: 314) agrees. However, the plan is incomplete (Figure 3). A whole wing may lie under a modern road to the north. No floors were identified, and the ceramic finds were not published in full (Delano Smith et al. 1986: 111–117). The six discernible spaces total about 60 m², but not all of them might have been covered, and there may have been a second storey. This estimate differs from Rathbone’s of 150 m² (2008: 314): sizes for rooms B (8.30 × 4.00 m), D (2.50 × 1.00 m), and E (2.00 × 1.50 m) are given, while I estimate the internal dimensions for A (1.80 × 2.40 m), C (3.30 × 3.00 m), and F (2.20 × 3.20 m), giving a total of 59.66 m². Guessing that the missing wing stretched the length of the building and was 3.00 m deep, the whole structure would have been about 90 m².

Figure 3

Plan of Site 9, Ager Lunensis survey (Adapted from Delano Smith et al. 1986: 112).

Site 9 illustrates our need for better evidence to determine the nature of inter-site dynamics. Despite the present limitations in our inherited data, tentative possibilities regarding the interactions under discussion can be furnished for my next examples.

The first comes from the Ager Capenas survey conducted between 1959 and 1961 (Jones 1962; 1963). On the Monte Forco ridge (Figure 4), six surface scatters represent ‘the sum total of ancient settlement’, with limited modern ploughing preceding the survey (Jones 1962: 172; Jones 1963: 155 for the tomb found 35 m from Site 154). As Table 2 indicates, only the presence or absence of the different types of finds were published. While coarse wares and brick-and-tile remains were found on five of the six sites, we lack many details to make meaningful comparisons between them. The number and forms of pottery are not specified, and size comparisons between scatters like ‘very heavy’ and ‘small’ are imprecise; singular finds, such as a dolium on Site 151, a Domitianic quadrans on Site 154, and a loom weight on Site 155, do not allow for secure inferences. Based on the absence of potsherds from Site 156, the surveyors suggest it was a work-station or outbuilding (Jones 1963: 147). Allowance must be given for other factors affecting the recovery rate of sites, including depositional processes, and the destruction of material by erosion and even limited ploughing. In toto, the survey data from Monte Forco reinforces the difficulties we face for such small sites.

Figure 4

Map of Monte Forco sites from the Ager Capenas survey in red, with surrounding landscape (Adapted from Jones 1962: 130).

One findspot, Site 154, was excavated because — like Site 9 previously — it was deemed representative of a ‘typical small farm’ (Jones 1963: 147; a smaller ‘farmhouse’ is the unexcavated Site 204 on Monte Cuculo (6.70 x 6.35 m, for which see Jones 1963: 147). The building (10.95 m x 5.1 m; Figure 5) and its immediate surroundings furnish some additional evidence: 15 coarse ware fragments, with a cooking pot and mortarium indicative of food preparation; three dolia within and without the building; a small rubbish pit; and clinker blocks hinting at an external oven. Taken together, these do not necessarily confirm that the site was a permanent dwelling (Samuels 2019: 82 citing Cambi on the lack of hearth). We may only be seeing evidence of domestic activity on a different kind of site (Ghisleni et al. 2011: 133–134; McHugh 2017: 1–9 on the Greek side).

Figure 5

Plan of Site 154, Monte Forco, Ager Capenas survey (Adapted from Jones 1963: 149).

Nevertheless, assuming that the site was a dwelling, interpretations have thus far focused on the original occupants’ statuses. Given the approximately equal size of the findspots, Augustan dating of the reticulate walls, and the two fragments of terra sigillata, Jones linked the structure to Caesar’s veteran allotments in 46 BC (Jones 1963: 147, 156–157). However, there is a supposed disparity in quality between the well-built masonry and poorer-quality ceramics; Jones recorded that ‘the local workmen helping with the excavation thought the standard of construction higher than that of their own homes’ (Jones 1963: 150). This led Foxhall (1990: 109–111) to argue that only a wealthier landlord could have constructed the building; dependent farmers lived on-site instead. Rathbone (2008: 323–324) returns to Jones’ hypothesis, but suggests that such buildings were ‘built by the state for settlers in viritane schemes’, while noting that Keppie (1983: 168) contests a Caesarian settlement of veterans. Most recently, Kron (2017: 124) accepts Rathbone’s suggestion as plausible.

One immediate limitation of this discussion is that independent veterans and dependent tenants are treated as either-or scenarios. The site’s broad dating, from the late-first century BC to the second century AD, makes both kinds of occupation possible if, for example, veterans became dependent tenants, or if Caesar’s veterans left soon after settlement after being recalled by the triumvirs (Broadhead 2007: 161).

But the most interesting question raised by these arguments is who built the structure rather than what was built. Assuming further that the sites were contemporaneously occupied, I pursue just one line of speculative argument. I suggest that soldiers experienced in building camps and defensive structures (Phang 2008: 67–70) could have built, or assisted in building, these structures to a high standard. In this context, combined efforts to construct each other’s homes might then be one avenue towards neighbourly, reciprocal relations between veteran settlers even before there were ‘homes’. The possibility of viritane schemes raises two further points I can only briefly touch on. First, peer-level interactions were more likely to be formed based on shared military camaraderie stemming from common experiences of soldiers from the same regions, legions, and battles. Caesar’s veterans were settled in small groups (Rawson 1994: 450) and were a cohesive group (de Blois 2007: 173–176). In social science terms, these meant homogeneity and short social distances, that is, one’s perceived closeness to others, which in turn made social participation more likely (Akerlof 1997: 1010–1011; Sacco et al. 2006: 711). Homogeneity in this sense is matched with the heterogeneity of the settlers’ skillsets: differing expertise among neighbours furnished further reasons for cooperation and assistance. The second point relates to gender dynamics. While the involvement of families in viritane schemes is unclear, we should bear in mind that norms about gender roles in the division of labour on farms (Rosenstein 2004: 95–100; Erdkamp 2005: 87–90) may have shaped who in particular engaged in what sorts of neighbourly interactions.

Regarding cooperative work, Site 154 raises the possibility of animal resources held in common. Jones (1963: 152–153) hypothesises that the internal postholes may indicate animal stalls in the last phase of occupation (c. the mid-second century AD); the uppermost layer of floor, in which these were found, contained animal manure (of unspecified origin), but the incomplete number of postholes and their asymmetrical positions make reconstruction difficult. Estimating the space enclosed by them — 2.75 × 1.25 m — allows for just a single cattle stall (Applebaum 1975: 120 similarly estimates 1.25 m for one animal); of course, other animals may have resided here (Lirb 1993: 291). The agronomists’ prescriptions are for double stalls (2.07 to 2.37 m × 2.66 to 4.44 m; Vitruvius, De Architectura 6.6.2; Columella, Rust. 1.6.6; Palladius 1.21. For the different terms for ‘stall’ in the literary sources, see Ikeguchi 2017: 32), and it is unclear how common single stalls were; Rossiter (1978: 60–61) lists only four Italian examples. Without further information about changes on the neighbouring sites, we cannot say whether one site consolidated its holdings and converted Site 154 into a barn, or several households group together to use the building collectively. Either possibility remains.

With Monte Forco we can raise merely speculative hypotheses about the specifics of horizontal social interactions. The limited survey and excavation data echo the call for more refined evidence by leveraging archaeological advances in the non-elite sphere while broadening our range of explanations for the resulting patterns. For instance, the presence of coarse wares across most sites compared to fine wares in our two cases so far signals the potential of this type of evidence (Launaro and Leone 2018). In particular, Site 154 is only 250 m and 130 m from Sites 153 and 155 respectively (Jones 1963: 147, 155), but fine wares are apparently absent from the latter. Establishing typologies and fabrics of coarse wares will help refine site occupational chronologies to clarify the issue of contemporaneous occupation, and distribution patterns to test hypotheses about the proportion of imported goods and the correlation between proximity of sites and their access. These patterns should not be understood just in economic terms, but as embedded in social relations. This approach offers one response to Witcher’s call for a ‘new focus on the cultural constructions of demand and consumption’ (2006b: 50). Seeing new equipment or ceramic wares on a passing visit or as a guest may induce the spread of fashions or even ownership of new items via gift-giving. As already mentioned, we should not be thinking about power dynamics as the sole way of deciphering the social significance of material culture (Witcher 2006b: 51–52).

One project which seeks to give us better evidence is the Roman Peasant Project, the final report of which is due to be published in 2021 (Bowes Forthcoming). Following an earlier-conducted survey in Cinigiano (South Tuscany), the project excavated eight sites with scatters of 0.25 ha or less, including one ‘off-site’ (Bowes et al. 2017: 173–174). These include a potential Late Republican farmstead at Pievina (Ghisleni et al. 2011), two temporary work-huts or animal shelters at San Martino and Poggio dell’Amore (Rattighieri et al. 2013), a small farmstead at Podere Terrato, and a field drain at Colle Massari (other studies include Vaccaro et al. 2017 and Collins-Elliott 2018). One other site, Case Nuove, is of especial interest because it appears to have been a locus for communal working and neighbourly interactions.

Case Nuove is a hilltop site and 500 m from the region’s only villa at Santa Marta (Figure 6). Its context is not entirely clear. Initially surveyed as three scatters, the area was not gridded and re-surveyed (Vaccaro et al. 2013: 131–132; Bowes et al. 2017: 173–174 n. 11). Furthermore, a magnetometry plot is provided but not discussed (Vaccaro et al. 2013: 133), even though work in the same region demonstrates the value of combining such complementary methods (Campana 2017). Excavated were an opus signinum-lined basin with a 1.650 L capacity, cuts to the north and south of it suggestive of a press, and an opus signinum work surface only 0.1 m high and 8 m from the basin (Figure 7). A well and cistern may have been coeval, but the construction dates are unknown (Vaccaro et al. 2013: 134–142). The project team interprets that these made up an open-air, small-scale processing installation for olive oil and wine, an interpretation supported by analysis of residue found in the basin and Late Republican dolia of the dump (Vaccaro et al. 2013: 142–145). The Late Republican site was abandoned in the Late-Augustan to Tiberian period, and it was later reoccupied in the Late Antique period (Vaccaro et al. 2013: 139, 142). However, in addition to uncertainties about the survey methodology, the site’s limited stratigraphic record, due to erosion on the exposed hilltop and ploughing, and the absence of dating evidence from between the excavated features (Vaccaro et al. 2013: 133–134), mean that we should not rule out nearby dwellings that might have been built of perishable material. As Rathbone (2008: 323–324) suggests, many farmsteads may be archaeologically invisible because of flimsy material, as opposed to more ‘permanent’ infrastructure — pits, basins, and presses, and so on — on sites like Case Nuove; factors like poverty and mobility may have also contributed to fewer traces in the material record.

Figure 6

Case Nuove (Cinigiano) in context (Adapted from Vaccaro et al. 2013: 132).

Figure 7

Plan of Case Nuove (Adapted from Vaccaro et al. 2013: 134).

There are two suggestions regarding the Late Republican site (Bowes et al. 2017: 173–178, 186–197). First, the site’s small-scale capacity and accessibility via nearby routes and tracks hint at it being a ‘collective pressing site’ for small farmers within a catchment area of up to 2 km; this distance is based on hypothesised land use according to Late Antique pollen data (Vaccaro et al. 2013: 162–171; Bowes et al. 2015; 2017). Shared investment was attractive because the press was only required seasonally at the harvest, and its hilltop position meant the facility could be monitored by all (Vaccaro et al. 2013: 171–172). Second, the nearby villa at Santa Marta means Case Nuove could have been a ‘brand of rural euergetism or… a subtle means of controlling territory by offering and centralizing, panoptical processing point’, with either scenario broadening the villa proprietor’s influence (Vaccaro et al. 2013: 173), a striking suggestion since euergetism is ordinarily considered in urban and monumental terms. The uncertainty here rests with the precise status of Santa Marta in the first century — Bowes et al. (2017: 178) claim that it was probably still a large farm at the time– leaving both options open.

Both interpretations imply that Case Nuove was a locus for horizontal social interactions. From this angle, a third interpretation may be more tantalising: it could have been a shared investment by the villa proprietor and smaller neighbouring farmers. A sense of ownership and responsibility for the facility could have strengthened ties between the part-owners, encouraging interactions outside harvest time. Beyond this formal cooperation, working together during the harvest meant a condensed period of close interactions. I earlier referred to collaboration based on different skillsets among veteran settlers on Monte Forco. Similarly, performing different functions in the processing of wine and oil here may have required cooperation based on individual expertise. This can also be appreciated in terms of people’s interactions with the natural landscape and its spread-out resources through concepts like ‘locale’ (Vaccaro et al. 2013: 130), or the Ingold’s ‘dwelling perspective’ (Witcher 2006b: 59; for a prehistoric archaeological perspective, see Foley 1981: 163–166). Land use hypotheses were made using Late Antique pollen data to map variations in cultivation around the site (Vaccaro et al. 2013: 162–166). But these suggest in parallel opportunities for interactions and collaborations in, for example, gaining access to use pastures or gather needed resources like wood.

Furthermore, on-site communal meals are suggested by dining wares and faunal remains found in the northern dump (Vaccaro et al. 2013: 172). Compared with the Late Republican period, the ceramic assemblage from the mid- to late-Augustan to Tiberian period, that is, the pressing installation’s abandonment phase, had a higher percentage of fine wares at 30% versus 17.7%. In absolute terms, this increase was almost double, from eight minimum number of individuals (MNI) to 15. This, plus a wider range of functional forms, amphorae from further away, and the presence of younger animals including domestic fowl, all point to a wealthier diet and more sophisticated dining practices (Vaccaro et al. 2013: 145–150; Vaccaro and MacKinnon 2014: 232–241). While this is ‘almost certainly… due to the expansion of and perhaps first articulation with the villa at Santa Mara qua villa’ (Vaccaro et al. 2013: 148), a significant unknown is whether these changes reflect use of the pressing facility only by Santa Marta personnel, or continued shared use by increasingly prosperous Santa Marta workers and other small farmers. The latter is possible if we pursue the shared ownership hypothesis. If so, we may be seeing a sharing and transfer, perhaps altruistically or reciprocally, of cultural ‘pretension’ among neighbours of differing socio-economic circumstances. The combination of a common work environment and shared dining experiences here presents some preliminary evidence for the production and consumption of relational goods among the rural non-elite in Italy.

Conclusion

In discussing the Italian peasantry, I combined insights from social science theory with literary snippets from Dio’s Euboicus and three sites from central Italy to study horizontal social relations. This approach was prompted by dissatisfaction at the restrictiveness of current scholarship to a socio-economic independence-versus-dependence polarisation. Significant methodological problems had to be addressed. Our written sources do not straightforwardly give us the ‘real’ socio-economic conditions of the rural non-elite; I treated details from the Euboicus as signposts for a non-exhaustive range of peer-level interactions. These included hospitality, reciprocity, and cooperative work. The archaeological material is also difficult to handle. Site 9 from the Ager Lunensis underscores the need for better evidence and different ways of making sense of the material — like common ware analysis — so that we can uncover the role of social relations in shaping distribution networks and rural consumption. Rather tentative hypotheses were then made regarding Site 154 of Monte Forco, while Case Nuove offers the tantalising possibility of a shared work environment fertile for formal and informal collaboration and shared experiences.

There are inherent limitations to this intensive deep dive into a couple of cases as opposed to extensive and large-scale study. I am not making claims about the representativeness of my examples, either across site and regions in Italy or in our period. In any case, any attempt at generalisation would be limited by the diversity inherent to our evidence. Monte Forco and Case Nuove are distinctly different in their physical and human landscapes, and they are best appreciated on their own terms. For example, it may be tempting to extrapolate from Site 154 on Monte Forco that (veteran) colonisation in general increases the likelihood of horizontal social interactions and hence cohesion within a local community. But in my discussion, I left the scenario of Caesar’s veterans — if they did settle there — being recalled soon after open, which would have limited the impact of such relations.

At the same time, heterogeneity in skillsets among members of a community may not have only fostered cooperation in work, but tensions and hierarchies. This point is important. I isolate horizontal social interactions and relations as an understudied aspect of the rural non-elite, I do not claim that these are necessarily more important than vertical relationships, but that they should be considered alongside the latter. This more holistic approach will allow us to draw out nuanced distinctions in our larger narratives. For example, at Case Nuove we might be seeing the potential ‘rise’ of a local villa changing the nature of access to markets and material goods, a change manifested in more complex and ‘wealthier’ dining habits. But by foregrounding horizontal interactions, I argued that there could have been a situation where increasingly affluent neighbours mingled with those less well-off, and that these interactions could have been vehicles of cultural transfer. A more conventional take would have simply opposed a slave-run villa at Santa Marta with surrounding silos of independent peasant farmers in terms of power relations.

These considerations, however, do not yet lead to substantive changes to our broader historical narratives, in terms of both chronological and geographical scale. More work needs to be done. Nonetheless, I hope to have shown that our theory of the peasant needs a more holistic approach, that archaeology can provide insights into peer-level social relations, and that our understanding of the ancient countryside can be nuanced and deepened by first finding the social life of rural non-elites in Roman Italy.

Competing Interests

The author has no competing interests to declare.