1What do coin finds tell us about the past? At a very elementary level, they can be appreciated for their contribution towards dating contexts. However, since we know very little about the longevity of coins in circulation, their utility in this respect is rather more restricted than non-specialists might imagine. Our monetary histories are still largely constructed via typologies rather than contextual dating. An alternative history, making use of contextual data to inform us when major changes to the circulating medium took place, is something that is sorely needed. While interest in this approach is gaining momentum, it will require far more contextual evidence than is currently available to us in order to construct an overview of coin use in antiquity¹. If we knew more about the lifespan of coins we could be more sanguine about the precision of coins for dating contexts.

2There are, of course, more interesting things to do with coin finds than using them to provide spot dates for archaeological features. Coins are social objects, and therefore have something to tell us about ancient social relations. They might have arrived at a site for a variety of reasons: as votive offerings, or as items of personal adornment, for example. In general, however, coin finds are considered to bear some relation to the ancient economy. This is, however, a very difficult topic, since it makes a grand claim for the material. The link between the ancient economy and the coins found on a site is far from obvious. Numismatists who work with site finds are well aware of the fact that they are dealing for the most part with low value denominations². This sort of work has been characterised explicitly, if informally, as “small change studies”. It concerns itself with the sort of money used in day-to-day transactions.

3While the sort of coins encountered on a site may well have been the sort used for day-to-day exchange, is there a strong correlation between the number of coins of a given period found, and the number of exchanges taking place? Does small change give us an accurate picture of economic circumstances in a given period? These are fundamental questions that impact on the significance and relevance of much that is done in archaeological numismatics.

4It must be admitted that the humble nature of this material does not always endear it to historians of the ancient economy. They are quick to remind us that gold and silver coinage, being of much higher value, was of far greater economic significance than the small change that makes up the majority of site finds³. Furthermore, really large transactions might have been conducted through credit, at least in the Roman world⁴. If “small change studies” is an attempt to supply evidence for ancient economies, it is doing so in an environment in which most of the evidence is missing.

5Despite these shortcomings, evidence of some sort is present, and it is of a kind that cannot be obtained from other sources. The archaeological record preserves two main sorts of evidence: coin hoards, usually (but not exclusively) composed of precious metal coins; and site finds. As we are all aware, these two sorts ought to be considered separately, even when they occur on the same site. That said, distinguishing dispersed hoards from single finds is not always easy. Unless the coins are found in a mass together, excavators may not recognise a group of coins as a hoard. Hoard material can end up dispersed over several contexts. My work on the coin finds from the Packard Humanities Institute’s excavations at Zeugma revealed a number of likely hoards that had been dispersed through several different, and not even stratigraphically-related, contexts. The hoards were probably dispersed in this way not by human activity, but during the decay and collapse of buildings abandoned after the Sasanian sack of the city c. 252 AD. Some of the material appeared to have fallen from upper floors, resulting in coins from the same putative hoard being distributed over adjacent rooms of a building and even the street outside⁵.

6The only way to identify these dispersed hoards is by noting suspicious concentrations of the same issues or types of coins in a particular area of a site and checking the stratigraphic record to see how the coins were likely to have been deposited. Sometimes the record will yield a note by the excavator of a possible hoard among the coin finds. Often, however, the identification of hoard material is more problematic. Small hoards not recognised as such by the excavators can easily enter a finds processing system and become further disassociated from each other by being recorded alongside other coins from the same contexts. Some concentrations will always remain uncertain. For example, a context numbered 10,600 in the Beirut excavations yielded a variety of coins, including a slightly suspicious concentration of late 6^th and early 7^th c. Byzantine folles: 4 of Justin II (565-578 AD); 1 of Maurice (582-602 AD); and 3 of Phocas (602-610 AD)⁶. Are these a hoard? There were no other folles of Phocas from any other context excavated on the site, which is suggestive. Even so, it was not clear that all 8 coins went together. The same context 10,600 also yielded a small module follis and 20 nummi coin of Anastasius (491-518 AD), a 20 nummi coin of Justin I (518-527 AD), a 10 nummi coin of Justinian (527-565 AD) and a 20 nummi coin of Justin II⁷. Should they be considered as single finds, or part of the putative hoard? Other coins from the same context included an issue of Licinius I from the mint of Siscia, dated to 313-315 AD, a provincial issue of Berytus struck under Hadrian (117-138 AD), both of which are very unlikely to have formed part of an early 7^th c. hoard⁸. In the end, there were too many uncertainties to warrant a listing of the suspected hoard coins in the section on hoards. Nevertheless an early 7^th c. hoard from a major Near eastern city, even if composed of 8 coins, would be of some interest to specialists. In the Zeugma report, where there were many more examples of probable hoards, I attempted to discuss putative hoards, again without listing them separately⁹. Whether one decides to list them separately is to some extent a matter of taste; but such difficulties remind us how often the single finds from a site are potentially contaminated with unrecognised hoard material.

7Of course, we are all aware of the fact that it can happen the other way around –that a hoard can be contaminated by alien material. There will always be grey areas, and these need to be reported, so that readers can make up their own minds.

8Some specialists who work with site finds bemoan the scholarly focus on hoards; and it is true that historians of the ancient economy, when they consider archaeological evidence at all, tend to concentrate on hoards as evidence for the monetary economy. While there is something vaguely perverse about studying coin circulation through assemblages of coins that were unambiguously not in circulation, I feel that it is important to consider hoard evidence in parallel with site finds when interpreting the latter, precisely because they preserve complementary forms of evidence. Together they have preserved evidence of ancient perspectives on the nature of coined money. This is where I think the subject of small change becomes really exciting, and where the study of coin finds has a real contribution to make: not by using finds to chart the economic fortunes of a city over time, or to describe trade links between the site and other locations; but to draw comparisons, spatially, chronologically and between different categories of evidence. The archaeological record preserves evidence of tensions between the issuing authority and users, and about local or regional solutions to problems such as undersupply of coin. Without wishing to impose an agenda on “small change studies”, it seems to me that an initial objective could be to explore and enrich our understanding of ancient monetary history rather than pursuing the broader but more elusive goal of shedding light on the ancient economy. Of course, monetary history is a part of economic history, but the subject of small change seems better suited to answering the needs of the former than the latter.

9That said, the nature of money is itself a difficult topic, but this very difficulty could be a key to understanding what we see in the patterns of hoards and site finds. It is no secret that economists and monetary theorists are unable to agree on a precise definition of money, and perspectives have shifted back and forth over time, between money as a thing with intrinsic value, and money as a signifier of credit. There are some well-known ancient views on the subject, which merely serve to highlight the fact that people in antiquity faced similar difficulties in determining the nature of money. Aristotle puzzled over how money seemed to reconcile what to him was irreconcilable: the (proper) use value of a thing on the one hand, and its exchange value on the other¹⁰. In the Severan age, Paul the jurist declared money to be a price (pretium), not a commodity (merx), deriving its power not so much by its substance as by its quantity¹¹. The fact that this needed elucidating suggests that not everyone saw it that way. His comments perplexed many medieval and Renaissance scholars and jurists, who tended to think of money as deriving its power from its intrinsic value as a commodity. Being unfamiliar with token base metal coins, some of these scholars proposed that scribal error was to blame: Paul had meant “quality”, not “quantity”¹². The Renaissance Italian jurist Girolamo Butigella disagreed, and took Paul at face value: all that mattered was the form, not the substance¹³. His opinions, and those of his contemporaries, paved the way for experiments with fiat money in base metal, among the first since ancient times.

10This debate about the nature of money continued in the centuries that followed, and moved beyond the realm of theory to impact on the real world. England, having experimented with fiat money in the 17^th c., famously reverted to the medieval commodity position in 1695, on the advice of the philosopher J. Locke: “Silver is the Instrument and Measure of Commerce in all the Civilized and Trading parts of the World. It is the Instrument of Commerce by its intrinsick value¹⁴”. Yet his near contemporary J. Law saw money not as a thing, but as transferable credit: “Gold, Silver, Copper, Bills, Shells mark’d and strung are only representative Riches, or the Signs by which real Riches are Transmitted¹⁵”. (Both views famously led to monetary catastrophe.) Here were two very different perspectives on money: either that it was material, and had value because of the commodity from which it was made; or that it was immaterial, transferable credit, which could be represented by any manner of substance.

11What has any of this to do with coin finds? Apart from the fact that the two perspectives still compete, in various permutations, to this day, and so shape our own thinking about the nature of coinage, it is possible that they also competed in antiquity as well, leaving traces in the archaeological record. Some have seen Paul’s declaration that money is a pretium and not a merx as a reflection of the age in which he was writing: after Septimius Severus debased the Roman silver denarius in 194 AD, the latter became a token coinage, whereas previously it had been a commodity currency¹⁶. However, it is equally possible that the legal position on coinage had always been that it represented value, rather than embodying it; in which case the precious metal was present as collateral, to convince all parties involved that the credit represented was good. However, given that the history of monetary theory is largely a history of a tension between the two positions, the coexistence in antiquity, at least unofficially, of a parallel opinion about money – that it embodied value by virtue of its substance – seems not unlikely, even if it was false from the jurists’ perspective. It would certainly explain the choice of gold and silver for hoarding by private individuals, and why token base metal coinages often did not circulate widely compared with gold and silver. It also helps to explain the operation of Gresham’s Law in antiquity: “better” coins were hoarded or exported¹⁷. Conversely, the presence of small change as single finds on sites could be explained not simply by their low value, but as a consequence of their value being regarded as representative rather than “intrinsic”, and hence not suitable for hoarding.

12In my original conference abstract, I promised to use a definition of money supplied by the Columbia economist R. Mundell, and here it is. Mundell is wary of token or fiat money, and sees it as the last stage in a succession of “bad” moneys that have driven out their “better” antecedents. He ended an essay on Gresham’s Law with the following definition of money: “In the world of exchange, debts are settled in the cheapest medium possible¹⁸”. This might seem like nothing more than an argument in favour of the efficiency of money as a medium of exchange, but it is not. Mundell argues that whatever becomes the primary medium of exchange effectively becomes the “bad” money that drives out other forms of money.

13When we use the term “bad” money, we tend to mean that such moneys have less “intrinsic” value. The perjorative term, however, belies “bad” money’s utility. It does not mean that people do not want “bad” money; on the contrary, it can even be more useful in exchanges than “good” money. Perhaps this is one way of thinking about site finds, as evidence for the medium considered most cost effective for exchanges. As long as there was a choice of denominations or coinages in different metals, one denomination or metal had a tendency to drive out other denominations from the realm of exchange, or at least from the realm of those types of exchanges that are likely to have left traces in the form of single finds.

14In this paper I would like to look at three potential examples of this presumed tendency for the cheaper exchange medium to drive out others. All date to the third or very early 4^th c. AD. In all cases the distinctions do not concern precious metal coins on the one hand and base metal on the other. The differences are more subtle than that. All the coins concerned are common issues of base metal or billon. We might readily expect them all to turn up regularly as single finds on sites, but some of the coinages do not. I suggest that the coexistence of these two ideas about the nature of money explains the patterns that we see.

15All three case studies require revisiting some of the evidence from the Beirut excavations that I published in 2003. The first example concerns coinage in circulation from roughly the second half of the 3^rd c., down to the end of the reign of Gallienus in 268 AD. The second spans the period of Aurelian’s reforms of 274 AD down to Diocletian’s reforms of c. 294 AD. The third concerns coins issued in the decade after Diocletian’s reform. The conclusions are tentative, but can be viewed as invitations to further research on the topic.

Case study 1

16In all three cases we need first to review the relevant monetary history as it currently stands. From about 260 to 274 AD there was effectively only one denomination produced in the Roman world: the debased radiate or antoninianus. Recent die studies suggest that production of gold coinage in this period was far from modest, but the paucity of surviving specimens suggests that gold did not survive long in circulation. Nor does it appear to have been regularly hoarded¹⁹. In contrast the ubiquity of radiates in hoards and as site finds implies that use of this denomination was widespread. In a fairly short space of time (230-270 AD) the coinage system had gone from one in which a suite of denominations was in regular production to what looks like a system involving the dominance of only one denomination. The picture may not be quite that simple, however, once we raise the question, briefly addressed at the beginning of this paper, of the longevity of coins in circulation. In the eastern empire cities had traditionally produced their own base metal denominations, but these too had all but ceased being produced after about 260 AD. How long the old denominations remained in circulation is less certain, and is one feature of monetary history that needs elucidating. At any rate, after about 260-268 (the sole reign of Gallienus) neither imperial mints nor many provincial cities in the east were investing in the production of small change. It is quite clear that the radiate went from a coin that is infrequently encountered as a single find on sites (and was therefore not treated like small change) and which instead was regularly hoarded, to one that is commonly encountered on sites as well as being commonly hoarded.

17It looks as if the antoninianus had to do the work formerly undertaken by a suite of demoninations, serving as both a store of wealth and a medium of exchange. With gold very scarce, huge quantities of antoniniani were needed to perform the functions previously undertaken by high value denominations.

18To illustrate the transition I refer to a chart originally employed for my report on the Beirut coins (fig. 1)²⁰. It shows the proportions of three categories of radiates at nine different sites. At most sites it appears that the change comes in the sole reign of Gallienus (260-268 AD). Radiates of the period 244-253 AD, which were still silvery in appearance, are generally scarce to rare on sites (though they are common in hoards). Coins of the next period, the joint reign of Valerian and Gallienus (253-260 AD), witnessed a transition from a silvery-looking coinage to one that was more obviously billon. There is much more variation in the proportions of these coins at sites, and they are well-represented at Antioch (a major mint) and Beirut. Finally in the period of Gallienus’ sole reign the antoninianus became a coin of base metal with only a minimal silver content. The proportions of this coinage dominate at all sites except Antioch.

Fig. 1. Proportions of radiates of three successive periods at nine sites in the eastern Roman empire.

19One could take the scarcity or absence of coins of a period as evidence that they were not in circulation in any quantity, but often hoards from the vicinity or region of the sites furnish us with a corrective. The hoards suggest that the coins were there, but they were not being deposited as single finds. If we follow the reasoning above, antoniniani of 244-253 AD are absent or rare as site finds because they were not the cheapest means of settling debts. There were other forms of money available to do that. Antoniniani of the following period, 253-260 AD, became a cheap way of settling debts in some places, like Antioch, but in other places like Sardis they did not.

20Naturally this raises the question as to what sort of money was the cheapest means of settling everyday debts if the radiates did not serve this function, and the pattern just described may hint at an answer. The strong showing for radiates of 253-260 AD at Antioch and Beirut may indicate that provincial base metal coinage had been demonetised, or was now harder to obtain, in Antioch and Beirut than elsewhere. Certainly there is no evidence for production of provincial coinage at Antioch later 253-254 AD, or at Berytus later than 254-255 AD, or the whole of the Near East later than 256-257 AD²¹. There are no coinages with portraits of Valerian II (Caesar 255/256-258 AD) or Saloninus (Caesar 258-260 AD) from any Near Eastern city²². In contrast, cities in western Asia Minor continued to issue coins after 256-257, since there are coinages for both Valerian II and Saloninus and, indeed, a few for Claudius II (268-270 AD) and Aurelian (270-275 AD)²³. It is possible, then, that a rise in the proportion of radiates as single finds at a site could be, at least in part, a symptom of radiates replacing provincial coinage as the cheapest medium of exchange there (either because the latter were no longer readily available, or because they were deemed of greater metallic value), but this needs further investigation to confirm or refute. If provincial coinages were “driven out” by “bad” radiates, this could explain why we have numerous hoards of provincial coins dating to the 250s-260s.²⁴.

21There is another aspect to consider here: the relationship of the finds to the supply of coinage. In other periods the absence of gold and silver among the site finds would be a severe handicap to any such analysis, but in the 260s we are dealing with a currency predominantly composed of a single denomination. It is often argued that a vastly increased supply of debased radiates caused high inflation. Such arguments are often advanced following the observation that huge hoards of this material exist, but, as we have seen, huge hoards of radiates may be a symptom of a general absence of gold coinage for storing wealth. Furthermore, all that huge hoards tell us is that coinage was being removed from the monetary economy and left to languish in hoards, where potentially it was not put to work. It is hard to see how hoarded money can have contributed significantly to high inflation. On the other hand, there might be a relationship between the frequency of radiates as site finds and how hard that money was made to work. So an increase in the frequency of radiates as site finds could potentially be an indication that the quantity of money being used also increased, which might have resulted in rising prices. This is, however, conjecture, and the possibility that a high incidence of a coinage as site finds represents demonetisation could easily disrupt such an analysis (see below).

Case study 2

22The second case study concerns the pattern that appears after Aurelian’s reform in 274 AD. The precise character of Aurelian’s reforms have long been debated. He introduced a new radiate coin of better quality than the ones issued in the decade immediately preceding the reform. The relationship of this new “aurelianus” or “aurelianianus” to the old radiate is far from clear. The late 4^th c. historian Zosimus mentions that Aurelian gave out new silver in exchange for debased coin, but provides no further details²⁵. Hoards seem to suggest that old radiates continued circulating in many places, although they may well have been successfully demonetised in Italy and the Balkans²⁶. Fortunately there are other ways of exploring the relationship between the “aurelianus” and other coinages that do not require full knowledge of their relative face values.

23Fig. 2 shows the proportions of three categories of coin minted between 253 and 294 p.C. (the date of Diocletian’s reforms) ²⁷. The three categories comprise: 253-274 p.C., the period of the debased radiate; “aureliani” of 274-294 p.C.; and “barbarous radiates” and coins of the Gallic empire. As is immediately apparent, while proportions of radiates of 253-274 p.C. are roughly similar at all nine sites, those that are rich in “barbarous and Gallic” are poor in “aureliani”.

Fig. 2. Radiates and “aureliani” of the period 253-294 AC at nine sites in the eastern Roman empire.

24To make full sense of this we need to consider the hoard evidence as well. While “aureliani” are rare as single finds at Troy, there was a major mint for these coins at nearby Cyzicus, and a hoard of these coins was found at the site (which was not included in the chart data)²⁸. As was the case with the earlier radiates, the aureliani were probably present everywhere in the east, but are not commonly found as single finds at the three sites in western Asia Minor. The case of the barbarous and Gallic radiates is different. While hoards of these are known in the western Asia Minor, where they also commonly occur as single finds, there are no known hoards of these coins outside that region of the Roman east. They would appear to have been imported from the north western provinces of the empire to western Asia Minor at some point in the 270s or later, and somehow they were sustained in circulation there, and nowhere else did they have a significant role to play²⁹.

25The presence of these barbarous and Gallic radiates seems to have given coin users in that region a choice that coin users at the other sites did not have: to use barbarous and Gallic coins (and perhaps old radiates) as the cheapest medium of exchange. Aureliani were driven into hoards, presumably because they were deemed to be more valuable as a store of wealth than a means of exchange. The pattern of single finds in western Asia Minor would appear to replicate that in the north western provinces of the empire, where barbarous and Gallic radiates predominate and aureliani are rare³⁰.

Case study 3

26The final case also requires a little review of the background. The reforms of Diocletian are sometimes regarded as a further stage of the changes initiated by Aurelian³¹. The old aurelianus continued to circulate but was no longer produced; instead a radiate coin with no silver content was issued (mainly at eastern mints)³². The latter is hardly ever encountered in the west, either among the single finds or in hoards, and presumably it did not form a major component of the coinage circulating there. A larger, heavier billon coin, the nummus, was also introduced. The nummus is plentiful in hoards, but rare among single finds (fig. 3)³³. In contrast, the post-reform radiate is abundant in the east. The finds imply that the post-reform radiate drove the nummus out of circulation and into hoards. As has been proposed, the nummus would appear to have been undervalued at some point, at least with respect to the post-reform radiate³⁴.

Fig. 3. Proportions of tetrarchic nummi and post-reform radiates, 294-305 AC, at six sites in the eastern Roman empire. The numbers in the columns denote the percentages for the two respective categories.

27This pattern makes it hard to accept K. Harl’s judgement that public confidence in the nummus rapidly waned and that its value declined because it was overvalued³⁵. The hoarding of these coins seems to imply the opposite: that these coins were too highly valued to form the cheapest medium of exchange. Diocletian’s currency edict of 301 AD mentions the doubling in value of a denomination, or some denominations, or perhaps all denominations. Opinions vary³⁶. One position has it that only the value of the post-reform radiate was doubled. If this is so, it can only have exacerbated the situation; and perhaps the pattern that we see is the consequence of the revaluation. Whatever the case, if doubling of values in the currency edict was an attempt to deal with a reluctance to use nummi in everyday exchanges the rarity of nummi as single finds would seem to suggest that the problem was not resolved by the reform.

The bugbear of irrelevance: demonetisation and the archaeological record

28So far we have been considering public treatment of “good” and “bad” coin, but there are also occasions on which certain coins can cease to function even as the cheapest medium of exchange³⁷. Perhaps some of the coinages that predominate on sites, even some of those we have considered above, are common partly because there came a point when they ceased to have official backing from the state and became more difficult to use. There seems to be no easy way of distinguishing between presence due to frequent use and presence due to demonetisation of an issue. A horizon in which a certain coinage predominates could be an indicator, but the contextual evidence is rarely so closely dated so as to permit one to draw such a conclusion.

29The possibility that coins could simply be scrapped and end up in the archaeological record through redundancy is yet a further obstacle to linking coin finds with the ancient economy, which perhaps explains why there is a degree of scholarly resistance to it. However, even if demonetised coins are largely irrelevant to the ancient economy, demonetisation is still a part of monetary history, more so, perhaps, than other forms of non-economic numismatic evidence such as votive offerings of coin or coins used as items of personal adornment. It would be helpful to acknowledge likely cases. The chief objection to the idea of redundancy seems to be that people would not discard something that had value and could be melted down for its metal content³⁸. A quick examination of the metal small finds from any site should be enough to convince that the objection is invalid. Scraps of bronze and broken bronze objects are hardly great rarities among finds on Mediterranean sites, and many of these fragments are more substantial than individual specimens of the commonest classes of coins. The idea that coins could be present because they had become irrelevant to economic exchanges does not mean that demonetised coins were always wilfully thrown away, but that is also possible, given that larger fragments of broken bronze objects were sometimes discarded. Such objects, like demonetised coins, no longer had use value, let alone exchange value.

Conclusions

30Jurists like Paulus may have had a clear understanding of the symbolic nature of money, but the finds seem to me to speak of two conceptions: one, characterised by site finds, of money as largely symbolic of value; and a popular conception of money as a thing with intrinsic value, characterised by hoard coins that were removed from the realm of exchange, probably much to the chagrin of the issuing authorities. Symbolic money is most successfully sustained by strong institutions, and therefore the spatial and chronological patterning of site finds may have something to tell us about underlying institutional structures, providing a more finely-grained image than the evidence derived from hoards of “good” coins. Assuming that demonetised coins did not travel in large numbers after they had been rendered redundant, the “bugbear” of demonetisation would be no obstacle to the identification of these institutional structures.

31Nothing in this paper should be read as an attempt to disparage the notion of a correlation between coin use and coin finds. If anything, it strengthens the links. The “bad” money predominates in the finds record because, perversely, it was “good” for exchange. One might even say it was desirable, and to that extent, it could have been worth importing, as in the case of barbarous and Gallic radiates in western Asia Minor. The finds record preserves public ideas about which coins were suitable for a particular purpose, suggesting to us the media in which exchanges were conducted; what it probably cannot do is help us to quantify the number of those exchanges over time. It perhaps hints more at the form of the monetary economy rather than its performance. Yet an ability to delineate with confidence the monetary economy of antiquity in any given period would be a significant advance.