The cupae of Iberia in their monumental contexts: a study of the relationship between social status and commemoration with barrel-shaped and semi-cylindrical tombstones

a particular identity in death: the content of the epitaphs may have conformed to Roman traditions, but the style of the monuments did not. If this type of analysis is applied to the funerary monuments of the other three main sites, those of Beja, Mérida and Sintra, it will be possible to build up a picture of how the ‘barrel’ and ‘cask’ types of cupae were used by different social groups at those sites


Introduction
Attitudes towards death in the Roman period are revealed through literature, philosophy, artefacts, laws and religious practices, each of which makes a contribution to our understanding of ancient perspectives on humanity's fate, but it is the tombstone and its epitaph that provides us with the most permanent physical expression of an individual life. Preparing for the event of death and ensuring an appropriate form of commemoration were important considerations which affected people from all sectors of society, from emperors to slaves. Many people were buried anonymously, but the inscriptions of thousands upon thousands of funerary monuments throughout the Roman Empire reveal that tombstones were considered a desirable form of commemoration not only for the social elite, but also for those whose means or status might be assumed to have imposed limitations on their choice of memorial.
It is the commemoration of non-elite members of society which concerns this article, specifically that of the non-elite members of provincial Roman society in Lusitania and Tarraconensis during the first three centuries A.D. This paper focuses on a type of commemoration which has no known 'elite precedent': that is, its style is not based on the tombs of wealthy or politically important people, either in Roman Italy or in the provinces. This is unusual because tombstone forms in the provinces tended to be modelled on the traditions of Roman Italy or, alternatively, on practices that were native to the area of that province before Roman occupation. As the evidence suggests that neither of these possibilities apply to cupae, they have been considered as something of a phenomenon in their own right. The monuments concerned, which are either semi-cylindrical in form or shaped roughly like barrels, seem to have commemorated people largely at the lower end of the social scale in provincial Roman society: slaves and freedmen. The fact that the cupae do not commemorate individuals of a high status, combined with their lack of an 'elite precedent', suggests that they might represent an attempt by people at the lower end of the social hierarchy to differentiate themselves in death from those who conformed with the customs of those in power by choosing a more traditionally 'Roman' style of funerary monument. This decision was not necessarily taken by the deceased themselves, but may have been made by their commemorators.
The cupae have suffered from a lack of scholarly interest which has resulted in a very uneven state of knowledge: no publication of the cupae exists for the site at which the greatest number has been found, at Mérida (ancient Augusta Emerita: funerary monuments with portraits from this site, however, have been given a thorough treatment in Edmondson et al. 2001), and the small number of articles which have been written on the cupae have focussed almost entirely on those from Barcelona (ancient Barcino: Balil 1955-56;Durán y Sanpere 1963;Julia 1965;Bonneville 1981). The cupae present, therefore, a body of evidence which is very much in need of interpretation, offering an ideal opportunity to address the question of whether deliberate social differentiation in death was practiced by certain groups within provincial society in Hispania. This paper is intended as an introduction to this little-known type of funerary monument, and an outline of the author's approach to their interpretation.

Location
The exact number of extant cupae is not absolutely clear, a situation which has arisen partly from lack of publication and partly from the unusual circumstances at Mérida, where up to three hundred of these monuments were re-used as building material in the walls of the ninthcentury Alcazaba fortress. Unfortunately they were built in with their long sides at a ninetydegree angle to the line of the wall, which means that their inscriptions (if they have survived) are completely inaccessible. During consolidation work over the past few decades a small number of cupae have been removed from the walls, but they have yet to be published. General inconsistencies in the published material in determining what can be called a cupa and what cannot has also led to some difficulties in assessing the actual number of monuments. For the purpose of clarity, this study encompasses those stone funerary monuments which were made either in direct imitation of a barrel, or which, like those from Barcelona and Sintra, possess a distinct curve on at least one of their long sides (those from the Sintra area are usually described as 'arciform cippi').  Fig. 1 shows the distribution of cupae across the Iberian Peninsula. In the north-eastern part of Tarraconensis, they have been found at Barcelona (25) and at Tarragona (4). Other examples in the same province appear at Ávila (5), Alcalá de Henares (3), Palencia (2), León (2), Astorga (24), Coria (15) and Alcuéscar (3). All the remaining cupae come from sites in the central and southern parts of the province of Lusitania, with a few exceptions located to the east of Beja, just over the border into Baetica. A large group are found in the Sintra area, just to the north of Lisbon (69). The southernmost examples are found on the coast, at Mexilhoeria Grande (1) and Quinto do Marim (1); 60 km or so further north there is a cluster around Mértola (11). There is a concentration of cupae in the region of Beja (31), along with a number of smaller sites nearby, each of which possesses one or two cupae: Alfundão (1); Beringel (2); Trigaches (2); São Matias (1); Baleizão (1); Pedrogão (1); Quintos (2); Santa Maria (1); São Brissos (1); and Santiago Maior (3). Immediately to the east on the opposite side of the river Guadiana are five sites which, again, have small numbers of cupae: São Salvador (2); Serpa (2); Brinches (2); Santo Agostinho (1); and Pias (2), all of which are located over the border in Baetica, whilst about 70 km further up the Guadiana lies Cerces (1). The remaining findspots are all within Lusitania, at Santa Margarida do Sado (2), Alcácer do Sal (2), Alcáçovas (1), Viana do Alentejo (1), Alvito (4) and Arraiolos (1). As this project is a work in progress, it is probable that a small number of further sites at which cupae have been found will be added to this distribution map in due course, but at present, 232 cupae from places other than Mérida are known, of which 58 are uninscribed or are too weathered for their inscriptions to be read. There are 80 accessible cupae at Mérida, along with the 300 or so mentioned above which are built into the walls. Once the Mérida examples have been taken into account, the total number of (accessible) cupae reaches 312, but as many of the cupae from Mérida do not possess legible inscriptions, the number of known inscribed cupae stands at 193.
From a stylistic point of view, the cupae fall into two main categories: those which possess a clear resemblance to a barrel in their almost cylindrical shape; and those which are arciform rather than cylindrical. These can further be divided into four principal types (the classification is my own, and includes only the four most numerous types of cupae. Other regional variations do exist, but in much smaller numbers).
Type 1, from Barcelona, form a distinctive group characterised by the sculpted border around the epigraphic field, which is placed on one of the long sides. Made of Montjuïc sandstone, these monuments generally have an elongated arciform shape, although some have a flat surface on the inscribed side and are curved only towards the back of the monument. They range from 0.66m to 1.90m in length.
Type 2 can be identified in the area around Sintra, just north of Lisbon: these arciform monuments possess a simple border at the base of each long side, and the inscription is located on one end. They are made of a pink limestone which resembles marble, and range from 0.68m to 1.30m in length.
Type 3 is found in Mérida: the majority have a double moulding around the base, and the inscription has either a plain border or is set into a tabula ansata. These are mostly made from local grey granite, and range from 0.63m to 1.33m in length.
Type 4, distributed across a large number of smaller sites around Beja and Mértola, makes the connection with barrels much more explicit, both in shape and decoration. They possess representations of the metal hoops which held together the planks of wood, and are mostly made of whitish marble from a local quarry at Trigaches. They range from 0.79m to 1.59m in length.
The practical functions of the cupae Those of Types 3 and 4 are solid and serve as markers of the grave. Those of Type 1 are mostly solid, but some possess small semi-circular hollows towards the base of the uninscribed long sides which may have served as receptacles for offerings, and one cupa does actually have a hollow compartment which is not visible from the outside (Julia 1965: Planche XVI, 2). Some of the cupae were displayed on plinths, which were also solid. Those of Type 2, however, reveal a different function: several of them have been found together with what appear to be plinths from the outside, but which are actually hollow inside and served as containers for ashes and grave goods. Thus the inscribed sections are actually the upper parts, or lids, of twopart monuments which were intended to encase the cremated remains of the dead. The Plaça de la Vila de Madrid, in the centre of modern Barcelona, is the only site at which a number of cupae have been found in situ (Fig. 2). The site as it is today is rather misleading: not all of the cupae now positioned on either side of the road were actually found there. Only five were actually discovered there, along with nine roughly semi-cylindrical masonry tombs and a number of other monuments. The rest were found elsewhere in the city and are now displayed in the Plaça. This site allows us to see how some of the cupae of Barcelona were actually used in their cemetery context, as the site was excavated in the 1950s (Balil 1955-56;Durán y Sanpere 1963). The roughly semi-cylindrical masonry monuments found with the stone cupae acted as protective coverings for the remains of the dead: tiles were placed over the body, either flat or balanced diagonally against each other, and the masonry part of the monument was built around this, into the shape that is visible from the outside. The cupae themselves were also placed directly on top of inhumations, physically protecting the remains just as the tile and masonry monuments did, a characteristic which none of the other tombstones of the colony (such as altars and stelae) fulfilled.
This function could help to explain why these monuments from the colony of Barcino were called cupae in antiquity despite having only a very superficial resemblance to barrels. The word itself is found in an inscription from one of the monuments from Barcelona (Fabre et al. 1997: no. 219):

D(is) M(anibus) Valerio Melippo Caelia Quartula fecit patri cupa(m) bene mer(enti) et Caelia Saturnina uxor m(arito) o(ptimo)
'To the spirits of the departed. To Valerius Melippus, Caelia Quartula set up the cupa for her well-deserving father, and his wife Caelia Saturnina (set it up) for her excellent husband.' The word cupa is most frequently interpreted as a barrel or vat for storing liquids, but there is evidence to suggest that the broader meaning of 'cask' was also a correct usage, and that this could include the type of cask used for holding the remains of the dead. Lewis and Short's Latin Dictionary defines the word cupa as follows: 'a tub, cask, tun, vat etc. for holding liquids, esp. wine…; also for grain; and for the ashes of the dead, a niche.' To support this last definition, two funerary inscriptions from Rome are cited: the first (Orelli 1828(Orelli -56: no. 2697 gives the name of the commemorator, and states that this man cupam aedificavit ('built the cupa'), which recalls the inscription from Barcelona. The second inscription (Orelli 1828-56: no. 4550) states that in hac cupa mater et filius positi sunt ('in this cupa mother and son have been placed.'), which implies that the remains of the mother and son were buried in the cupa rather than underneath it. Unfortunately both inscriptions appear to have been disengaged from their associated monuments, so the exact nature of the cupae from Rome must remain unclear. However, the second inscription in particular suggests that the word cupa was used to denote a receptacle for the remains of the dead. Whilst only one of the cupae from Barcelona has a hollow compartment inside (Julia 1965: Planche XVI, 2), all of those found in situ acted as protective coverings for the remains of the dead. The strong stylistic similarities between these cupae and both the semi-cylindrical masonry tombs found with them in the Plaça de la Vila de Madrid and the arciform monuments from Sintra (which were used as containers for ashes and grave goods), suggest that the Barcelonese monuments were known as cupae because they fulfilled the function of physically protecting the remains of the dead, rather than because they possessed any visual similarity to barrels. Thus the name cupae is appropriate in each of these contexts, but for different reasons: in the Sintra area it is appropriate due to their actual function of the monuments as two-part funerary casks; at Barcelona, to their function of forming a protective covering for the remains of the dead; and at Beja at Mérida it is appropriate due to their visual resemblance to wooden barrels, especially in the case of those from Beja, which were also decorated like barrels. We have, therefore, the prospect of two separate and possibly unrelated commemorative practices: that of the 'cask' cupae, and that of the 'barrel' cupae.

Symbolic functions
The 'barrel' cupae are perhaps inevitably subject to more wide-ranging interpretations of their symbolic functions than the 'cask' cupae. If we possessed only a small number of barrelshaped cupae, they might reasonably be interpreted as references to the employment of the deceased during their lifetime, perhaps as a cooper (cuparius, a barrel-maker). However the fact that several hundred of these monuments exist (including over three hundred at Mérida alone) makes it very unlikely that all of the people commemorated with these monuments were employed in trades which involved barrels.
As it is clear that not all of the barrel-shaped cupae can have commemorated cuparii, we must seek alternative explanations for the choice of this type of monument. One interpretation is that the significance of the barrel was in its contents rather than its shape. The poor rate of survival of wooden barrels in the archaeological record places a limit on our ability to state categorically that they were commonly used to store and transport wine (for a discussion see Tchernia 1997: 121−129) but it is generally acknowledged as being the most likely case. Wine was used in funerary rituals such as the pouring of libations and in feasts held by the graveside on particular days of the year (Lindsay 1998: 74-5), and the barrel-shaped cupae may have been chosen as a way of emphasising the importance of this practice. Alternatively, the cupae could represent a thanksgiving in death to the gods who had provided the correct conditions for viticulture, particularly as it was one of the most economically important activities of the region of Alentejo, in which the barrel-shaped cupae are found. Large numbers of people were involved in the process of making wine, and if the cupae from this region are connected to this, one might have expected to find tombstones of other types relating to other parts of the winemaking or transportation process. However, there were practicalities to consider: for instance, most wine seems to have been transported and stored in amphorae, not in barrels, but the barrel shape would certainly have been much simpler for a stone carver to create than an amphora shape (amphorae themselves were actually used as containers for cremations at Ostia, but we have no evidence that wooden barrels were used in this way).
Another possibility is that the shape of these monuments was connected with a religious cult. The barrel-shaped cupae have been interpreted as a sign of following the cult of the god Sucellus, a deity whose attributes include the barrel. However, the identification of a cult of Sucellus in Lusitania relies on a small number of bronze figurines which have been identified as representations of this god (Lambrino 1965: 237) and on the presence of the barrel-shaped cupae themselves. Whilst Sucellus is certainly depicted with barrel-like objects (for instance the figurine from Vienne: Boucher 1976: no. 301), it is difficult to reconcile the theory that there was a cult of the god in Lusitania with the fact that no dedications to this god have been found there. His primary following was in Gaul, and to date no convincing evidence has been put forward to show that the barrel-shaped cupae are connected with Sucellus.
The symbolic meaning of the barrel-shaped cupae, then, is not by any means simple to explain, and there is much more work to be done. The influences on the shape of the 'cask' cupae are almost as difficult to determine, although in this case we have parallels in North Africa, Italy and Asia Minor for their form. Julia noted the similarities between the cupae from Barcelona and the semi-cylindrical North African cupulae, and suggested that the cupae were a result of the direct influence of North African settlers in the Roman colony of Barcino, citing the nomenclature of some of those commemorated with cupae to support the theory (Julia 1965). A study of semi-cylindrical masonry tombs in Dacia, Moesia Inferior and Pannonia considered each group of monuments as variants on the same theme, possessing both North African and Italian influences. The form of these monuments is compared not only with cupulae but also with the 'cassone' tombs found at Ostia (Berciu and Wolski 1970), a type of monument which is stylistically very close to the cupae of Barcelona, the primary difference being their construction in masonry rather than in stone. The theory that the Barcelonese cupae were influenced by the Ostian tombs was supported by Bonneville, who saw in the cupae a complex combination of stylistic elements from Asia Minor and Dalmatia coming through Italy, with a separate strand of influence from North Africa (Bonneville 1981). The two-part semi-cylindrical monuments from the Sintra area have not been studied in as much depth, but Julia certainly considers them as being stylistically very close to the North African cupulae (Julia 1965).
However, the question of where the forms of the cupae originated is perhaps less important in the context of the Iberian Peninsula than the question of why these types of tombstones were considered appropriate at the particular sites at which they have been found. In some areas, these monuments continued to be used over a period of up to two centuries, which implies that whatever the original influences on the shapes of the different cupae, they took on their own meanings for those who set them up which may have had little or nothing to do with the meanings that the cupae had for the people who originally introduced them to those areas. It is not possible to interpret these meanings by thinking of these monuments as an individual, isolated phenomenon. Instead, this study advocates using an approach which takes into account the local monumental commemorative practices of each site at which the cupae have been found. By examining who set them up, who was being commemorated, and how others at the same site chose to commemorate their dead with other monumental types, we should be able to move towards a better understanding of why the barrel and cask types might have been considered appropriate in those particular areas by those individuals. The possibility raised in the introduction to this paper that cupae might be a manifestation of a desire on the part of certain people to differentiate themselves from other groups within society can only be investigated by exploring the full range of monumental commemoration at the sites in question.

A local contextual approach to the cupae
The possibility of studying tombstones within their actual archaeological context is, unfortunately, almost non-existent. It is extremely rare to find a tombstone in its original place, and it is usually impossible to reconstruct the former context of a monument which has been taken from that position. Exceptionally preserved cemeteries at sites such as Pompeii cannot be viewed as a typical examples, and even where a small number of tombstones have been found in situ, we cannot reconstruct a whole cemetery layout on the basis of one conserved section. However, the lack of tombstones in situ should not cause us to abandon entirely the theoretical principles behind a contextual approach to the material record, of which inscribed funerary monuments form a part. So we must aim for the next best thing, which is the local context.
As Hope writes in the conclusion of the published version of her thesis, 'each monument when viewed within the cemetery was affected by its design, size, surroundings, visibility, proximity to the road and the presence of other structures.' (Hope 2001: 89). An individual's view of each monument within the cemetery was also affected by their own background and perceptions which, to us, are unknowable. We can, however, approach a funerary monument from the 'other side': that is, we can study how people were represented at death through their funerary monuments, even if we cannot necessarily recover all the ways in which those monuments were interpreted by people at the time. The decision to commemorate someone with a particular type of monument was influenced by a knowledge and understanding of how others were commemorated in the area which constituted an individual's frame of reference: most probably, the site at which they lived and perhaps a small number of surrounding sites (exceptions to this would have included soldiers, slaves and merchants, whose lives could end thousands of miles from their birthplace). In other words, the monumental context of a site was as important in its influence on the decision to commemorate someone in a particular manner as it was to the way in which a monument was viewed by others.
This study considers that monuments which are visually very similar cannot be treated as though they were a coherent group, particularly when they are distributed across a large area, and that monumental commemoration can only be understood within the context of local commemorative practices. It supports the move away from the traditional separation of the epitaph and the iconography by advocating the use of both the epigraphic and the stylistic evidence to answer the question of why certain types of monuments were considered an appropriate form of commemoration by those concerned. This will take the form of an investigation into the social identities of the individuals who were commemorated with stone funerary monuments at the sites at which cupae have been found, as well as the identities of the people who were responsible for setting up those monuments. Ultimately, the social statuses of those individuals will then be related to the style and decoration of the monuments they set up, or which were set up for them, in order to establish more precisely how different social groups were represented in death. This paper, however, concentrates on the first stage of this process: an analysis of the epigraphic evidence for social status, in a case study of the funerary monuments of Barcelona.

The monumental environment of the colony of Barcino
Before the inscriptions of Barcelona (ancient Barcino) are analysed, the monumental environment in which they were set up must be defined. For the purposes of this article, the stone funerary monuments published in Inscriptions Romaines de Catalogne (IRC: Fabre et al. 1996) have been used. This publication does not include every single tombstone from the colony, as some are uninscribed, but the number of uninscribed monuments is relatively small and the exclusion of these for the purposes of this case study should not affect the overall picture. Four main types of monument were used for inscribed funerary commemoration in the colony of Barcino: sandstone blocks (moulded or plain); altars; cupae; and plaques (which were not monuments in themselves, but were attached to tombs such as the semi-cylindrical masonry monuments found at the Plaça de la Vila de Madrid). Some individuals were commemorated with pedestals, but it is often difficult to determine whether an inscription was intended to honour someone while they were still alive, or whether it was set up after the person's death as a memorial. This study includes only those inscriptions which are definitely identifiable as funerary. The sandstone blocks could also be used for honorific inscriptions, and altars were used for dedications to gods as well as to people. The use of some of these monuments was not, therefore, confined entirely to the funerary sphere. The plaques present a particular problem which does not apply to the other forms of inscribed commemoration: we cannot say for certain what the plaques were originally attached to. We know that some of the semi-cylindrical masonry monuments had plaques, but as only three of the plaques were found in situ, we can only speculate about the original location of the others.
Not all of these monument types were in use at the same time. If we examine the dating of the funerary monuments, we find that the monumental environment changes over the course of the first three centuries A.D. Of the Augustan and first-century monuments, the overwhelming majority (forty-three, or 90%) are sandstone blocks. The earliest funerary altar is dated to the late first century; one plaque can be dated to the Julio-Claudian period, another to the late first century and two more to the first century in general (the dating used is that of Fabre et al. 1997). Thus there was a rather uniform approach to commemoration in the first century of the Roman colony at Barcino, one which diversified quite considerably during the second century. By the mid-second century, altars and funerary pedestals became more common, and cupae began to become established. By the end of the second century, the sandstone blocks which were so prevalent in the first century now made up less than a third of the funerary monuments being set up. The late second and early third centuries were the most diverse, and the majority of the cupae (seventeen, or 70%) were set up within this period. The last pagan funerary monuments identifiable as such were set up in the early third century: thus we have a period of approximately two hundred and fifty years of (pre-Christian) commemoration with inscribed funerary monuments in the colony, and a clear diversification of the funerary monumental environment over the course of this period.

Epigraphic analysis of the funerary monuments of Barcino
In order to establish whether there is a correlation between the type of monument chosen and the social status of the deceased, the inscriptions have been analysed by monument type, and include both the status of the deceased and that of the people who were responsible for setting up the monument. It is rare to find actual declarations of status in the epitaphs after the first century A.D., so we have to rely on other indications, usually names and family relationships, to provide clues. For this analysis, only those inscriptions which contained names which were sufficiently complete have been used, and where social status has been assigned, this has been done only where there are definite statements of status in the epitaphs or where the possibility of an individual belonging to any other social category is extremely small (hence the figures given are the minimum number of people of each status involved with commemoration with each monumental type).
The number of monuments in each category on which the names of the deceased are totally or almost entirely preserved are as follows: sandstone blocks, 45; altars, 15; plaques, 9; cupae, 24. The number of individuals commemorated on each type of monument are as follows: sandstone blocks, 79; altars, 15; plaques, 16; cupae, 25. The number of individuals named as commemorators are as follows: sandstone blocks, 36; altars, 17; plaques, 10; cupae, 23. Figures 3, 4, 5 and 6 show the proportion of people commemorated with each type of monument belonging to each social status. Freeborn citizens commemorated: Ingenui make up at least 27% of those commemorated with sandstone blocks; at least 20% of those commemorated with altars and at least 19% of those commemorated with plaques. No definite ingenui are commemorated with cupae.
Freedmen commemorated: Liberti make up at least 33% of those commemorated with blocks; at least 20% of those commemorated with altars; at least 19% of those commemorated with plaques; and at least 16% of those commemorated with cupae.
Slaves commemorated: Servi make up at least 1% of those commemorated with sandstone blocks; at least 20% of those commemorated with altars; at least 6% of those commemorated with plaques, and at least 32% of those commemorated with cupae.
Ascertaining the status of the commemorators is equally as important as ascertaining the status of the deceased, as ultimately it was the commemorators who had the final say on which type of monument was used.
Freeborn citizens as commemorators: Ingenui make up at least 28% of those setting up sandstone blocks; at least 12% of those setting up altars; and at least 20% of those setting up plaques. No definite ingenui set up cupae.
Freedmen as commemorators: Liberti make up at least 42% of those setting up sandstone blocks; at least 17% of those setting up altars; at least 10% of those setting up plaques; and at least 13% of those setting up cupae.

Slaves as commemorators:
No definite servi set up sandstone blocks or plaques. They make up at least 12% of those setting up altars and at least 22% of those setting up cupae.
Declaring the precise relationship between the deceased and the commemorator seems to have been very important for those who set up the cupae in Barcelona: a statement of how the commemorator is related to the deceased appears in an unusually high proportion of the epitaphs (80%). Most of the relationships stated are family connections: almost a quarter are by parents for their children; a fifth are commemorations of a husband or wife by their spouse; and a sixth are by children for their parents. In only 10% of cases do brothers and sisters commemorate each other. In the case of those cupae which were set up for slaves, twice as many of them were set up by fellow slaves as were set up by slave-owners.

Discussion of the results
We have, therefore, an entire category of monuments from which we must exclude slaves: that of the sandstone blocks, which apart from a single exception (which makes up only 1% of the individuals commemorated) were used only by freedmen and the freeborn. Slaves are commemorated with a very small proportion of the plaques and only one-fifth of altars, but make up at least one third of the people commemorated with cupae and almost a third of the people who actually set up the cupae. Of the total number of individuals who can be assigned definitively to each social status in Barcino on the evidence of the epitaphs (deceased: 27 ingenui, 48 liberti and 13 servi; commemorators: 14 ingenui, 22 liberti and 7 servi), the majority of freeborn people (77%) are commemorated with sandstone blocks, as are the majority of freedmen (79%). On the other hand, the majority of slaves (61%) are commemorated with cupae. Likewise, more than two-thirds of freeborn people responsible for commemoration set up sandstone blocks (71%), as did almost the same proportion of freedmen (68%). But the majority of slaves who were responsible for commemoration in Barcino chose to set up cupae (71%).

Conclusion
There is, then, a strong correlation between slave status and commemoration with cupae at Barcino: just under two-thirds of the slaves who received a monument in Barcino were commemorated with cupae, and more than two-thirds of those known to have set up monuments in the colony chose to set up cupae. What is surprising is not the proportion of slaves commemorated with cupae, but the proportion responsible for commemoration with cupae. Slaves were not just commemorated with these monuments by their masters, they also appear to be paying for them to be set up, which raises important questions about the resources that were available to slaves before manumission. The proportion of slaves involved in commemoration with cupae, compared with the apparent lack of freeborn involvement, also suggests that cupae were considered as an appropriate form of commemoration for slaves but not as an appropriate form of commemoration for freeborn citizens. This being so, an analysis of the epitaphs of the funerary monuments of Barcino shows that there are grounds to support the theory that in this colony, the cupae may represent a intentional form of social differentiation by the commemorators of people in the lowest classes of society. The impetus to set up this type of monument, which had no elite precedent, seems to have been a conscious decision not on the part of freeborn commemorators of slaves, who might have had an interest in commemorating them with a type of monument very different from those which the freeborn used, but on the part of the slaves themselves. This is not merely a case of choosing the least expensive type of monument: many cheaper forms of commemoration were available. The cupae in Barcelona were chosen by slaves and, to a lesser extent, freedmen, because they fulfilled the needs of those groups in expressing a particular identity in death: the content of the epitaphs may have conformed to Roman traditions, but the style of the monuments did not. If this type of analysis is applied to the funerary monuments of the other three main sites, those of Beja, Mérida and Sintra, it will be possible to build up a picture of how the 'barrel' and 'cask' types of cupae were used by different social groups at those sites, and the question of whether they represent more widespread social differentiation can be addressed.