9. Cultural Variations in Alloys Used in the Roman Period

9.1 Introduction

This section sets out some of the analytical results obtained from copper alloys in terms of the types of sites where they were found. This will allow inter-site comparisons to be made. Of particular interest is the incidence of brass on a range of Roman sites. The first large-scale use of brass in Europe was at the beginning of the Roman Empire. Brass was used for the production of certain classes of objects (such as coins and military equipment) and it has often been assumed that the state held a monopoly of production. It might be hoped therefore that the incidence of brasses on a variety of sites in northern Britain might indicate the degrees of Romanisation. This chapter examines this approach. The results are presented to show inter-site differences and similarities and these are discussed in terms of Romanisation and settlement hierarchies. In addition, the results are considered in terms of possible deposition practices (and especially variations between different sites). This section ends with a more critical examination of the sources of the samples analysed and the problems inherent in the kinds of inter-site comparisons made here. It is unfortunate that a large proportion of the samples were collected from excavations where detailed contextual data was not available (either because the excavation was very recent and the post-excavation analysis had not proceeded far enough, or because the excavation was completed some decades ago and the level of on-site recording was inappropriate for this kind of analysis). For these reasons, no attempt is made at intra-site examinations of alloy composition related to types of contexts (ditch fills, pits, etc. - cf. Hill 1994).

9.2 The Early Use of Brass in Europe

Brass is an alloy of copper and zinc which is relatively hard to produce, as zinc is a volatile metal and was not isolated in Europe until the 18th century AD (Tylecote 1992: 152). Prior to this, brass was produced by the cementation process (Craddock 1978; Bayley 1984). The best yield obtainable with the cementation process is c.28% zinc. The cementation method was first used on a large scale during the early Roman Empire. Before this, brass was a rare (and perhaps a precious) metal. The monetary reforms at the end of the Republic and the beginning of the reign of Augustus, however, saw the intensive production of brass coinage (Caley 1964). At approximately the same time large quantities of military equipment began to be produced in brass (Bishop & Coulston 1993: 191).

This sudden increase in the production of brass and the relative obscurity of the manufacturing process has suggested to some that the early Roman Empire acquired the knowledge and resources to manufacture brass and maintained a monopoly (Caley 1964, 92; Grant 1946, 88; Bishop & Coulston 1993: 191).

If the early Empire maintained a monopoly of brass production (primarily for its own use) then the incidence of brass in objects other than coinage and military equipment, and away from centres of Roman control, might indicate degrees of Romanisation.

9.3 Brass as an 'index of Romanisation'

The last decade or so has seen increasing interest in 'Romanisation' (e.g. Millett 1990; Freeman 1993). The term was originally coined by Haverfield (1912) to cover a range of social and economic transformations that occurred in Britain during the first four centuries AD. It is clear that quite important changes did occur but there is less consensus over the ways in which these changes came about. It has traditionally been seen primarily from the Roman point of view, i.e. the Romans deliberately set out to change the nature of society in Britain in order to make it easier to rule, and to this end they introduced literacy, urbanism, a cash-based economy, etc. More recently (e.g. Millett 1990), such views have been criticised and a more active role for the indigenous Britons has been proposed.

At an early stage in this research it was hoped that the incidence of different copper alloys could shed light on the Romanisation question. In particular, it was hoped that the incidence of brass may act as an 'index of Romanisation'. If brass was produced by an imperial monopoly and used largely for official purposes (coinage and the army), then the incidence of this alloy on a range of archaeological sites would reflect the degree of social and economic ties with the Roman Empire. This model would predict that the highest proportion of brass would be found on sites at the centre of the Empire and lowest on the fringes and outside. Within any one limited region, the highest proportion of brass should be found on military and urban sites with the lowest proportion on small rural farmsteads.

The view that brass was produced only by state monopoly has already been compromised, however, by Justine Bayley's analysis of Roman brooches (Bayley 1992). Many early brooches were made of brass, and include Nauheim derivative, simple Gallic, and Colchester 'A' brooches (some of which may be as early as the earliest orichalcum coins). There is no suggestion that these brooches were produced by the state.

[Alloys used on a range of Roman sites in northern Britain]
Fig.68 Alloys used on a range of Roman sites in northern Britain
(Actual numbers of samples given in brackets).

The analysis of 1212 Roman copper alloy samples for this project has allowed the comparison of the incidence of copper alloys on a range of Roman sites (Figure 68). It can be seen that the incidence of brass on a range of Romanised sites (forts, vici, towns and villas) is broadly similar (around 20% of the alloys from these sites are brasses). This suggests that there was little differentiation between these sites in terms of the availability of brass. The incidence of brass does not seem to reflect any military-civil or urban-rural hierarchy. In fact the sites with the highest incidence of brass are the small, isolated, rural farmsteads. The high incidence of brass on these sites is the opposite of what might be expected from traditional imperialist explanations of the Roman occupation of Britain. The high incidence of brass at small rural sites seen here is similar to that seen in late Iron Age and 'Celtic' metalwork (Figure 69). The widespread use of brass in the early Empire suggests that there was not an official monopoly on the production. Brass was widely available throughout northern Britain, even for indigenous manufacture.

[Alloys used in indigenous contexts]
Fig.69 Alloys used in indigenous contexts
(Actual numbers of samples given in brackets).

The high incidence of brass in these contexts suggests that the indigenous inhabitants of northern Britain should not be seen as passively receiving those elements of Roman culture offered them. They apparently had more access to a 'Roman' metal than many more Romanised societies. The results of these copper alloy analysis should contribute to a wider discussion on the nature of indigenous societies in the Roman Empire, the nature of Roman society, and the ways in which these impinged on each other (Saddington 1991; Freeman 1993; Millett 1990).

There are a few classes of sites where much lower proportions of brass are found - especially hillforts and caves. Given the above discussion of indigenous metalwork, it would be difficult to argue that the low incidence of brass on these sites is the result of a lack of access following an absence of Romanisation. An alternative explanation is explored below in terms of varying social rules regarding the nature of deposition on different sites.

9.4 Structured deposition and copper alloys

As previously mentioned, hillforts and caves (and, to a lesser extent, hoards and burials) show little use of brass (see Figure 68). While hillforts and caves in the Roman period have been interpreted as being peripheral to Roman control and Romanisation in Britain, this explanation does not seem appropriate here. As argued above, the highest incidence of brasses actually occurs in indigenous contexts and metalwork. In addition, the evidence from hillforts comes almost entirely from Traprain Law which has traditionally been seen as the capital of the Votadini and so would have had good contacts with the Roman world (e.g. Hanson 1987). This apparent contradiction can be resolved by arguing that material on these sites was carefully selected before deposition (Hill 1994; Millett 1993).

It has become increasingly accepted that the material culture of the Iron Age which survives in the archaeological record is highly influenced by active selections made by the depositors (e.g. Bewley 1994) and so cannot be 'taken at face value' (Hill 1994). The low levels of brasses from some Roman sites in northern Britain may reflect similar structured deposition. Sites which have produced low levels of brass include Traprain Law, the Settle Caves, Coventina's Well, and the Brougham cemetery. All four of these sites can be easily interpreted as arenas for ritual activities. Millett (1993)has demonstrated that the artefacts deposited in an early Roman cemetery were carefully selected and did not reflect the full range of artefacts available. Exactly the same process would seem to be operating in the deposition of copper alloys on a range of Roman 'ritual' sites. Brass may have been seen as an appropriate metal for deposition in some circumstances. Alternatively, some artefacts (which were only incidentally made of brass) may have been regarded as inappropriate.

The use of 'Romanisation' and structured deposition both provide interesting frameworks for the explanation of the differences seen in alloy use in northern Britain. It is necessary, however, to examine the selection of samples used. The previous chapter contained some warning that chronological comparisons in the Roman period may not be entirely valid as the source of the samples analysed changes over time. Conversely, when comparing different sorts of sites the different dating means that like is not being compared with like. Some of these problems are explored further below.

9.5 The problems of inter-site comparisons

The above discussion of the differences and similarities in the alloys found on a range of Roman sites in terms of Romanisation and settlement hierarchies assumes that the samples found on each site are comparable. The structured nature of the archaeological record (even in the Roman period) illustrates that the evidence from different sites may not be comparable in this way. Further problems arise with the nature of the samples selected for analysis. Some sites (e.g. farmsteads, hillforts) have produced relatively small numbers of objects and so the validity of the results from these may be questioned.

Yet more problems arise when the date of the samples from different sites are compared (Figure 70). It can be seen that the date of the samples from each type of site varies considerably. Some types of sites discussed are not included here, most notably small rural sites which are omitted as only four samples from this type of site could be closely dated.

[Dating of samples from a range of different sites]
Fig.70 Dating of samples from a range of different sites
(Actual numbers given in brackets)

Military sites provide most evidence for the first two centuries AD, reflecting the considerable military activity that was carried out in this period. Large-scale troop movements occurred on a regular basis and new forts were frequently built and abandoned. The archaeological record is therefore rich in forts of the 1st and 2nd centuries. The later Roman period saw a decline in the number of soldiers in Britain (James 1984) and those who stayed generally remained where they were stationed. There was, therefore, less opportunity for the deposition of metalwork in the later Roman period.

Evidence from military sites forms the largest proportion of all the analyses carried out for this research. The range of alloys used on different sites can be seen in Figure 71. Many 1st or 2nd century sites, such as Castleford, Carlisle, Vindolanda and High Brunton, show a high proportion of brass in use. Some of those sites which have a low proportion of brass are those where occupation was later (such as Ravenglass and Piercebridge). This strengthens the suggestion of a chronological zinc decline, but some 1st century sites also show low levels of brass, e.g. Elginhaugh. The reasons for differences in the alloys used on similar sorts of sites at approximately the same date are not clear. The lack of brass at Elginhaugh may reflect a real situation resulting from supply problems in the fort. On the other hand, differences in preservation conditions and excavation strategy may be imparting a bias to the results. At Carlisle and Vindolanda large quantities of thin copper and brass sheet and wire were recovered. This is mostly scrap metal, probably awaiting remelting or reworking, and reflects the activities of smiths on the sites. Such quantities of scrap sheet and wire were not found at Elginhaugh. This may be because smithing was not common at Elginhaugh (occupation was short-lived), smithing areas were not excavated, or the soil conditions were too harsh to allow the preservation of thin fragments of metal. All three explanations are possible.

[Proportions of alloys used in Roman forts]
Fig.71 Proportions of alloys used in Roman forts
(Actual numbers given in brackets)

Vicus settlements provide most evidence in the 2nd century, but no evidence in the 4th century (see Figure 70). In large part this reflects the history of these sorts of sites in northern Britain. The lower proportion of 1st century evidence (compared to forts) may reflect an occasional time-lag in the setting up of vici in newly conquered territory. The lack of 4th century evidence from vici results partly from problems of definition and classification. Daniels (1980) has argued that changes in late Roman barrack architecture arise due to soldiers' dependants moving into forts and so vici may have been largely abandoned by the 4th century. At some sites the status of the vicus may have changed over time. Extra-mural settlement at Carlisle has been interpreted as a vicus in the first two centuries AD but as a town in the later Roman period (McCarthy 1991). There is a remarkable similarity in the range of alloys used in different vici (Figure 72). This may reflect the fact that the date range for these settlements is not as wide as it is for military sites.

[Proportions of alloys used in Roman vici]
Fig.72 Proportions of alloys used in Roman vici
(Actual numbers given in brackets)

The relationship between the definition of vici and towns in northern Britain can be seen in Figure 70. Little evidence comes from towns in the 1st century AD (because they hardly existed at this time), but there is some 4th century evidence (in contrast to the vici). The dating of the evidence from towns shows parallels with villas (this may not be surprising as these two classes of settlement are socially and economically related, Salway 1965; Casey 1982; Millett 1990).

The evidence from Roman towns dates mostly from the 2nd century onwards (see Figure 70). This reflects the fact that any 1st century settlements displaying any urban characteristics are found outside forts and so would be classified as vici. The range of alloys used in different towns is shown in Figure 73. There are considerable differences between the three sites but this may reflect the relatively low number of samples from each site.

[Alloys used in Roman towns]
Fig.73 Alloys used in Roman towns
(Actual numbers given in brackets)

The evidence from burials comes entirely from the 3rd century. This is because the only large cemetery available for study was that at Brougham where all the burials belonged to the 3rd century. The range of alloys used in the cemetery at Brougham are similar to those found on other 3rd century sites (see Figure 74). The differences seen between burials and other sites (Figure 70) may be caused by the restricted date of the burial evidence.

[Comparison of the alloys used at Brougham with those from other 3rd century sites]
Fig.74 Comparison of the alloys used at Brougham with those from other 3rd century sites

Larger rural settlements (Figure 70) provide considerable 1st and 4th century evidence but little from the 2nd or 3rd centuries. This reflects the occupation on those sites examined (Dragonby and Old Wintringham have considerable activity in the 1st century, while Shiptonthorpe and Bainesse Farm see most activity in the 4th century). It is unclear to what extent the restricted date of the evidence is a reflection of the limited excavations in northern Britain and limited sampling for analysis, and to what extent this class of site generally sees more activity in the early and later Empire but not in the middle years. The alloys used are shown in Figure 75. It is striking that Bainesse Farm (which is occupied from the 2nd century onwards) has just as much brass present as earlier sites, such as Dragonby. The only site which has noticeably lower brass levels is Shiptonthorpe. This may reflect the fact that most activity on the site belongs to the 3rd or 4th century. Alternatively, the low brass levels at Shiptonthorpe may be another indication of the ritual nature of some of the activities on this site (Millett personal communication).

[Alloys used on larger rural sites]
Fig.75 Alloys used on larger rural sites
(Actual numbers given in brackets)

In northern Britain, villas do not appear before the 2nd century (Figure 70). This is in line with the accepted view that such sites usually date to the 2nd century or later. Southern England sees the emergence of some villas in the 1st century but the development of villas in northern Britain is slower. The range of alloys used at different villas (Figure 76) varies considerably from site to site (Welton Wold sees high incidence of brass while Kirk Sink has no brass). Some of this variation may be the result of problems associated with defining these sites. The site at Welton Wold includes a rectilinear building which generally conforms to the villa type but most of the site consists of the sorts of enclosures and buildings which are found on non-villa sites. All of the finds from Rudston have been treated as if they were from the villa, but the villa buildings themselves belong to the 4th century. The status of the site before this period is uncertain. None of the finds from Kirk Sink could be context dated and some of the finds came from unstratified layers.

[Alloys used on villa sites]
Fig.76 Alloys used on villa sites
(Actual numbers given in brackets)

The above discussion of the range of alloys on different sites has highlighted some of the problems involved in inter-site comparisons. Differences between two sorts of sites may reflect 'real' differences between the alloys used on the two sites, or they may reflect differences in the chronologies of the two sites. It is possible that the Roman period in northern Britain sees chronological and cultural variations in alloy use. Such variation in two fields simultaneously is difficult to disentangle. Further complication arises if the nature of the artefacts compared is considered. Not all sites produce the same range of artefact types (Figure 77) and the sorts of finds recovered change over time. Attempts to subdivide the total number of analyses by an ever increasing number of archaeological criteria (site type, chronology, artefact type, deposit type) leads to a fracturing of the data base until the number of results in any one field is so small that reliable comparisons cannot be made.

[The categories of objects analysed from different sites in northern Britain]
Fig.77 The categories of objects analysed from different sites in northern Britain

9.6 Summary

This section has examined the variations in alloys used on different sorts of sites in northern Britain. Particular attention has be focused on the incidence of brass. This metal is first used in Europe from the late 1st century BC onwards and first appears in Britain at the time of the Roman conquest (or shortly before). The production of brass for Roman military equipment and coinage has led many to suggest that it was produced by and for the state under an official monopoly. This model would therefore suggest that access to brass was restricted and controlled by the Empire. The incidence of brass on provincial Roman sites might therefore act as an 'index of Romanisation' for that site. The more Romanised the site was, the more a part of the Roman (as opposed to indigenous) economy it was, the more the site would have access to and use brass.

Analysis has shown that a wide range of artefacts were made of brass and that these are found on an equally wide range of sites. There is no correlation between a traditional hierarchy of settlement types and metal use, i.e. those at the top (towns) having the highest levels of brass and those at the lowest (farmsteads) having the lowest. Most Roman settlement types have similar levels of brass. The one settlement type to have particularly high levels of brass is the un-Romanised small rural farmstead - those settlements at the 'bottom' of the assumed settlement hierarchy. The high incidence of brass in this indigenous context is also mirrored by the high incidence of brass in late Iron Age and 'Celtic' metalwork. This suggests that the indigenous inhabitants of northern Britain were not as passive as they are sometimes made out to be. The indigenes were able to select artefacts and materials from the Roman world with considerable freedom. This may be of wider European significance in the light of a recent paper by Stos-Gale (1993), which indicates that the range of alloys from a Roman-period site in southern Poland included a higher proportion of brass than is usually found inside the Empire.

There is no evidence for a state-held brass monopoly in the early Empire but there is the possibility that state metal production in the late Empire did involve the production of brass. Nevertheless, brass continued to be produced and used in non-official circles.

Central to the sorts of inter-site comparisons carried out in this chapter is the assumption that the artefacts from each site are in some way representative of the activities which took place there. One school of thought (e.g. Binford 1983) assumes that the archaeological record is passively formed through the unconscious loss and discard of material culture. It is assumed that objects enter the archaeological record near their place of use and in the proportions in which they were used. This approach allows considerable reliability to be attached to the archaeological remains - social life can be 'read off' from the archaeological record. While this has been qualified to a certain extent (e.g. Middle Range Theory), it is still assumed that there are straightforward links between past life and the present archaeological record. While it has long been acknowledged that some parts of the archaeological record were deliberately constructed (e.g. deposits at temple sites), recent research (e.g. Hill 1994) shows that large proportions of the archaeological record may be deliberately constructed, even on settlement sites. Artefacts were carefully selected for inclusion in deposits and so may more accurately reflect the ideology of the society and beliefs of the individuals than the economy, raw materials, or technologies used. Such ideologies and beliefs may vary from site to site and change over time.

A final problem when attempting to consider similarities and differences between sites (and between centuries) is the incommensurability of different sites. Differences may be identified in the alloys recovered from various sites but do these indicate differences in the alloys used? Inter-site and chronological comparisons, such as those attempted here, can lose some of their power if the samples from different sites and periods are not directly comparable.


© Internet Archaeology
Last updated: Thu Apr 3 1997