8 Chronological Variations in Roman Alloys

8.1 Introduction

This section will examine the chronological changes in the proportions of Roman alloys. Most of the analysed Roman samples have been assigned to a century based on stratigraphical/contextual grounds. Occasionally context information could not be this precise. in which case the terms early Roman (first two centuries AD), mid Roman (2nd and 3rd centuries AD), late Roman (3rd and 4th centuries AD), or simply Roman were used. Where the contextual information could be improved or corrected on typological grounds, the typological dating was used.

Little previous research has looked at chronological variations in Roman copper alloys. Caley's (1964) analysis of orichalcum coins has provided the only detailed examination of progressive changes in alloy composition. The interpretation of these results is reconsidered below.

8.2 Caley's 'zinc decline'

Caley (1964) carried out the analysis of 24 Roman brass coins (sestertii and dupondii). He noticed that the zinc content was highest in the early coins and lowest in the latest coins - there was a progressive 'zinc decline' from the late 1st century BC to the early 3rd AD (when production of brass coins ceased). This led Caley (1964) to suggest that brass production started in the late 1st century BC but stopped shortly after. He argued that the brass coins of late 1st century AD date onwards were made from recycled brass. The volatility of zinc meant that the zinc content of the brass declined after each remelting (Caley 1964: 83). Caley's explanation seemed to assume that after c. AD 50 the method of manufacturing brass was 'lost' but the reasons for this were not explored.

Since Caley's initial work, the number of analysed Roman brass coins has increased dramatically (see Figure 61). Riederer (1974a) analysed another 20 or so coins, carefully selected to cover reigns which would complement Caley's research. In addition, Riederer reviewed the results of analyses from the19th century (some of which seem to be unreliable - probably due to the analysis of corrosion products and metal). A small number of analyses of Roman brass coins can also be found in Cope (1974), Carradice & Cowell (1987), Carter (1966), and Carter & Buttrey (1977). The total number of Roman brass coins analysed has been greatly increased by the analysis of over a thousand coins from the River Garonne hoard (Etienne & Rachet 1984). The vast majority of the coins which have been analysed, however, were minted in the early 2nd century AD (the Garonne Hoard [Etienne & Rachet 1984] was deposited in the mid 2nd century AD). The types of alloys used for the earliest and latest brass coins are still imperfectly understood. The lack of 3rd century analyses is particularly acute and so seven late 2nd/early 3rd century coins have been analysed for this research project (Table 8.1).

7032CommodusBMC 66517781.427.256.112.23
7033CommodusRIC 659181+78.843.634.4112.86
7034Gordian IIIas RIC 316(a)240-481.78nd8.359.8
7035Gordian IIIRIC 297(a)241-485.25nd8.596.06
7030Philip IRIC 148(a)244-788.280.858.592.06
7031VolusianRIC 249(a)251-371.76nd6.1621.98
Table 8.1. Alloy content of some late sestertii and dupondii analysed for this project

[Zinc content of Roman Brass coins]
Fig.61 Zinc content of Roman Brass coins

(Sources: Caley 1964; Carter 1966; 1971; Cope 1974; Etienne & Rachet 1984; Riederer 1974a; Carter & Buttrey 1977; Carradice & Cowell 1987).

The zinc contents of all of the analysed Roman brass coins are plotted against time in Figure 62. There are, however, some problems with this approach - many of the coins are shown as single points when in fact they could only be assigned to a date range. In these cases an average value was used. An alternative view of the same data can be gained by plotting the mean values of zinc content for each reign (Figure 62).

[Zinc content of Roman Brass coins (mean values for each reign)]
Fig.62 Zinc content of Roman Brass coins (mean values for each reign)

This is not entirely satisfactory either as there was some change in zinc content within individual reigns (Etienne & Rachet 1984). Figures 61 and 62 do, however, show that the zinc content of Roman brass coins did decline over the 1st and 2nd centuries AD. As discussed above, this 'zinc decline' was interpreted by Caley as resulting from the exclusive use of scrap brass for the production of later coins and the subsequent loss of a proportion of zinc on each remelting (Caley 1964: 83, suggested the loss of 1/10th of the remaining zinc on each remelting).

8.2 The 'zinc decline' re-examined

A close look at the large number of analyses of Roman brass coins now available (Figure 61) casts doubt on Caley's explanation. The actual loss of zinc is slight in the late 1st century and early 2nd century. In the late 2nd century and the early 3rd century the gradient increases considerably. The observed relationship between zinc content and time is approximately described by the function,

[Function matched to observed zinc decline]
Fig.63 Function matched to observed zinc decline

An illustration of this function (fitted to the data in Figure 61) can be seen in Figure 63. This curve is not, however, what would be expected if the brass was suffering the loss of 1/10th of the remaining zinc on each remelting (remelting occurring once every 10 years). In this case we would expect the greatest absolute loss of zinc to be greatest in the early years. Over time the zinc content would decline - it would begin to approach zero but, it would never reach it (see Figure 64). This expected decline is described by the function,

[Theoretical zinc decline in recycled brasses]
Fig.64 Theoretical zinc decline in recycled brasses

The observed and expected zinc declines do not match each other, and so the suggestion that the zinc decline is due to the recycling of brass seems most unlikely. Further evidence for the non-recycling of Roman brass coins is provided by an examination of the tin and lead contents. The proportion of both of these elements increase as zinc decreases. One possible source of the tin and lead is leaded bronze (such as that used for paterae, etc.). It is possible that after the mid 1st century AD Roman 'brass' coins were made by mixing a fresh brass and a fresh leaded bronze. Over time the proportions of these two fresh metals changed - less and less brass was used until in the 3rd century almost no brass was added to the metal used for minting coins (Bastien 1967). The careful manipulation of the metal content of Roman 'brass' coins should come as no surprise as the silver content of denarii is progressively altered (Casey 1980: 8-11, Figure 62). While the decline in silver content has been seen as deliberate, the changes in base coinage have been seen as less important and more likely to be accidental. The 'zinc decline' cannot be due to the recycling of scrap brass coins, instead it was a deliberate alloying policy. It is possible that the reason the brass coinage was 'debased' over time was in order to keep some sense of parity with the debased silver coinage.

8.3 Chronological changes in alloying

Caley's (1964: 83) argument that the knowledge or means of production of brass was 'lost' sometime in the 1st century AD has already been challenged by Craddock (1975; 1978) who found that some late Roman objects (other than coins) were made of brass. Craddock's work (1975) was hampered by the nature of the samples available, however, as most were from museum collections which were without archaeological context dating. The existence of some brass artefacts which could be dated typologically to the late Roman period proved that brass production did not cease in the 1st century AD.

The present work has involved the analysis of 1212 Roman artefacts and allows the close examination of the change in alloy composition over time. The average and median values for each of the major alloying elements are given in Table 8.2 and the median values shown in Figure 65. The decrease in the zinc content over time is pronounced and this is accompanied by a rise in the tin and lead contents.

[Median values for zinc, tin and lead for each of the first four centuries]
Fig.65 Median values for zinc, tin and lead for each of the first four centuries AD

The changes in zinc, tin and lead content are also reflected in the changes in the types of alloys found (Figure 66). Many of the alloy types depicted are present at relatively low levels throughout the Roman period (e.g. leaded brass and leaded copper) or show relatively little change over time (e.g. gunmetal and bronze). The significant chronological changes in the alloy types are the decline in unleaded brass and the increasing use of leaded bronze and leaded gunmetal. In the 1st century brass accounts for 37% of the alloys used, while leaded bronze and leaded gunmetal together account for 27%. By the 4th century brass has declined to just 4%, while leaded bronze and leaded gunmetal have increased to 64%.

[Chronological changes in the proportions of Roman alloys]
Fig.66 Chronological changes in the proportions of Roman alloys
(Actual numbers of dated samples in brackets)

It would seem, therefore, that there is good evidence for a general decline in the use of brass in northern Roman Britain. It may be argued that the total amount of zinc in all copper alloys remained largely constant because of the increased use of leaded gunmetal. The increased use of leaded gunmetal cannot offset the decline in the use of brass as the average zinc content (i.e. a measure of the total amount of zinc in all Roman copper alloys) also declines (Table 8.2).

1st Century AD2nd Century AD3rd Century AD4th Century AD
Number of samples19124514578
Table 8.2 Mean and median values for the alloy elements in Roman alloys over four centuries

Any interpretation of the changing proportions of bronze and brass seen in Roman alloys has to be qualified by a consideration of the provenance of the samples. Strictly speaking the samples from each century are not comparable with each other (see Figure 67). First century evidence is dominated by that from forts and larger rural settlements while the 2nd century evidence is provided mostly from forts and vici. In contrast, the 3rd century is the only period to have provided evidence from burials and the 4th century evidence comes primarily from forts, towns and villas. Some of these differences are the result of modern archaeological activities (e.g. the only large excavated cemetery available for study was Brougham which is exclusively 3rd century). Other differences reflect changes which occurred during the Roman period (e.g. towns and villas were genuinely rare in the earlier Roman period). These differences make chronological comparisons difficult. If there are variations in the sorts of alloys used and deposited on particular sites, then the observed chronological changes may simply reflect the socio-economic differences between the sites, rather than the general chronological changes in alloy production..

[Chronological changes in the provenance of samples]
Fig.67 Chronological changes in the provenance of samples
(Actual numbers of samples given in brackets)

In addition, the types of artefacts produced and the methods used in their manufacture also changed, e.g. many early Roman military fittings were made from riveted sheet metal while most late fittings were cast in moulds. Generally, casting is a more straightforward method of manufacture than working with sheet and wire. Wrought alloy working requires considerably more skill. The increasing proportion of cast alloys could be interpreted as resulting from a decline in the number of skilled smiths working with copper alloys. Such a change could occur due to an increased volume of production. Most explanations of the production of Roman military equipment in the late Empire suggest that production processes were simplified (James 1988). The 'rationalisation' of production in order to meet an increased demand is, however, difficult to reconcile with the archaeological record. Most Roman finds can be dated to the early rather than the later Empire. It is not clear to what extent changes in the alloys available forced changes in fabrication techniques or changes in typology encouraged changes in alloy composition.

Despite the decline in the use of brass this alloy did not disappear. A limited number of 4th century artefacts (defined on typological ground) were produced from cementation brass (i.e. at least 23% zinc, Lindberg 1973; Bishop & Coulston 1993: 183).

8.4 Summary

The role of brasses in the Roman period has long been seen as important. Brasses first become widely used in the early Roman Empire. Caley, using the evidence from coin analyses (Caley 1964), suggested that brasses were only produced in the early Empire. The only source of brass in the later centuries was recycled brass. The decline in zinc content (Figure 61 and Figure 62) apparently reflected the decline of the Roman Empire. Craddock (1975) suggested that at least some late Roman artefacts were made of brass and suggested that brass production may not have gone into such a decline. The re-examination of the analyses of orichalcum coinage (Figure 63 and Figure 64) suggests that such coinage was not regularly recycled.

Analyses presented in this chapter show that brasses were used through each of the centuries of Roman control in northern Britain. The use of brasses did, however, decline over time. Many of the late Roman brasses contain only moderate levels of zinc and could have been produced by remelting old brass scrap rather than being freshly produced by the cementation process. Remelting of scrap metal may have been important throughout the Roman period.

Chronological changes in alloys used can be identified but it is difficult to be certain of their meaning. Changes may reflect general chronological changes in alloy production and use, or they may arise due to socio-economic differences between different sorts of sites and the changing nature of the available data set.


© Internet Archaeology
Last updated: Thu Apr 3 1997