Coin Relief 7 – What does it all mean?

Welcome to the seventh edition in our series of blog posts written by our Finds Advisers for ancient coins, Dr. Sam Moorhead and Dr. Andrew Brown. In this series they are exploring some of the many (many) ancient coins from the PAS database. In this special edition, Andrew reflects on what all of the numismatic information on the database can tell us about Roman coinage in Britain.

What does it all mean?

Our Daily Coin Relief blogs until now have focussed specifically on the numismatic details of the coins themselves and their position against the backdrop of Roman history. The mass of information that has been gleaned from individual coins recorded through the PAS in the last two decades is remarkable. New coin types and variants of known types have emerged on an almost yearly basis and the PAS database now represents the largest dataset of its kind anywhere in the world. The question that springs to mind, though, is what does this all mean? It’s all very well having hundreds of thousands of objects, but what is this telling us about Roman coinage in Britain?

I thought it might be useful to take stock slightly in this edition and have a brief look at what the data contained within the PAS material is beginning to tell us. This will be an introductory, high level look at the Roman data to give a flavour of what the database contains and what we can do with it. There is much work to do on the minutiae of the individual coins, but we can begin to interpret what is already present on the PAS database. This is in large part thanks to both the willingness of finders to record and the hard work of Finds Liaison Officers (FLOs), volunteers, interns, and public recorders in physically recording the material through the PAS.

Figure 1: All Roman coins on the PAS database up to May 2020.

Some high-level statistics

By the end of 2005, there were just over 34,000 Roman coins recorded through the PAS (Fig.2). Since then, extensive reporting and recording has resulted in a huge increase in the volume of material seen within England and Wales so that by 2020 there are over 320,000 Roman coins on the PAS database covering almost every part of the country (Fig. 2).

Figure 2: Roman coins recorded through the PAS: Left – up to the end of 2005 (blue dots); Right – up to the end of 2019 (red dots, with the material up to 2005 in blue superimposed)

As of today (May 12th 2020), the database holds details of 323,741 Roman coins contained within 291,991 database records – the largest group from the more than 500,000 coins on the PAS and the largest single category of object nationally. Every year between 10-20,000 new records are created and, thanks in no small measure to the work put in by Sam to encourage all Roman coins to be recorded, this means that it is not just the really spectacular that is recorded but more often than not the corroded, worn, and bashed around ‘grots’ that form the majority! This dataset also includes over 52,000 coins imported from the Welsh IARCW (Iron Age and Roman Coin finds from Wales)[1] project in 2010 – an excellent resource but complicated somewhat on the PAS database due to its mix of single finds, excavated site finds, and hoard coins. Indeed, you will have noticed that Sam and I sometimes exclude the IARCW data from searches because of this complicating factor, particularly so for more statistical surveys like this where the Welsh data tends to skew the results.

To the single finds we can also add hoard groups, recently subject to comprehensive examination by a joint AHRC funded British Museum and University of Leicester project exploring patterns of hoarding, deposition and the landscape setting of hoarded material within Britain.[2] The known hoard assemblages have been integrated with the PAS database, under the prefix IARCH, with nearly 3,000 Roman examples recorded to date.

This is a lot of data to break down in a sensible way! However, even at a very basic level, a glance at the individual coins as dots on a map (Fig. 1) reveals some distinct features. For example, there are very obvious absences of Roman coins in some parts of the country – notably the urban landscapes of the Midlands, the north west, and London, affected by the modern landscape which prevents the recovery of archaeological material. Equally, the national parks (e.g. Exmoor, Peak District, Lake District, North York Moors, etc.) are blank spots due to restrictions on metal detecting in these areas. In contrast, there are also some features that appear to be genuine reflections of the ancient landscape. Notable is the seeming paucity of coins running between Kent and Sussex – this is in fact the location of an ancient woodland, the Andredes Weald. However, perhaps even starker is the very clear dividing line between the north east and north west of England running along the Pennines (Fig. 3).

IFigureeeeFigure 3: Roman coinage in the north of England represented on the PAS.

Analysis of the coin assemblages from these regions indicates that the relatively small quantity of coinage in the north west (PAS coins from Cumbria, Cheshire, and Lancashire total 2,272 database records) contains comparatively more early Roman coins by percentage and therefore likely related to early Roman military activity. In contrast, the much large number of Roman coins recorded in the north east (PAS totals from Yorkshire number 16,817 database records), particularly those of late Roman date, points to this region experiencing extensive rural activity in the late Roman period (we’ll come back to this in passing below). Already at this very high level of analysis the extensive data contained within the PAS is demonstrating features of the Roman landscape that are only visible through our ability to fully record and map the material.

We can perhaps break this down a little further and consider the Roman coinage by modern parish and look at the volume of finds for each parish in England as of end of 2018 (I’m excluding the Welsh data here as the IARCW material introduces huge biases) (Table 1).

Table 1: PAS Roman coins by parish (up to end 2018).

Although this table needs updating to take into account the material from 2019-2020, what it shows at a glance is the huge impact recording has on our knowledge of the Roman landscape. Most striking are the parishes with greater than 500 or 1,000 coins, the latter almost three times the number recorded in 2011. Many of these sites, particularly those with smaller numbers at present, would have been otherwise unknown.

If this data is in turn plotted visually (Fig. 4), we can begin to see where the concentrations of material are to be found nationally. Traditional interpretations of the Roman landscape often highlight early military and urban activity, notably in the west and north. However, what we can clearly see simply by plotting those parishes where more than 20 Roman coins have been reported is that there is extensive activity in the rural areas of England. This is especially so in the south west, East Anglia, Lincolnshire, and Yorkshire – a distribution pattern that until relatively recently would have been unheard of. The band of material running south west to east appears to reflect intensive late Roman activity, likely in relation to the rural landscape and economy (see Fig. 15 below). One important point to note, too, is that in his study of 140 Roman sites (see below), Roman numismatist Richard Reece suggested that coin use was extremely limited in the east of England (Norfolk, Suffolk, and Cambridgeshire). What the PAS data demonstrates immediately is that this was far from the case!

Figure 4: Map of England with parishes recording more than 20 PAS coins plotted. Darker blue parishes are 20 coins, moving to bright green with more than 1,000. The heat map beneath demonstrates at a schematic level the concentrations of coins generally (as of end of 2018).

Of course, one thing this map does not take into account is the internal breakdown of the coinage by type or period. Equally, the high concentrations of finds from places like Wiltshire and East Anglia also reflect the accessibility to the modern arable landscape by metal detectorists (and the resultant larger numbers of finds). We’ll come back to this a little below.

Applied numismatics

Although the growth of the PAS database provides a huge volume of material, this also means that at times finding an individual coin, particularly one that is of a relatively standard type or is poorly preserved, becomes more of a challenge. When explored in its entirety, though, this data can give us a different insight into coin use and loss within the province of Britannia and begin to resolve some of the issues noted above with regard to biases in the dataset and the refinement of distributions (e.g. by period, coin type, etc.). This is an important point to note – we are dealing with coin loss around the province, which will be affected by a number of factors both ancient and modern. By recording the full range of material that is being found (not just the nice silver and gold!) we can begin to see a bigger picture and start to interpret what this might mean locally, regionally, or nationally, on top of the vital numismatic detail provided by the individual coins.

The last few decades witnessed extensive work on Roman coinage in Britain and the various potential methods to interpret it, notably by the likes of numismatists John Casey[3], Richard Reece[4], and more recently by Sam Moorhead[5] and Philippa Walton[6]. The application of approaches that are more archaeological and interpretive in nature to large coin datasets – ‘applied numismatics’ – rather than purely numismatic has meant we can begin to look at large assemblages of coins on individual sites (or indeed larger areas) and elucidate key pieces of information about site usage, character, and longevity.

Reece’s work has been particularly significant in this regard. He divided Roman Britain into 21 distinct chronological periods (Fig. 5) – Reece Periods – based on the development of Roman coinage. By doing this he was able to drop individual coins into their correct period and therefore compare individual sites to one another based on the frequency of coins from one or other period. This approach was exemplified in his 1991 study looking at 140 sites from Roman Britain, the results of which suggest, for example, that sites with peaks of early Roman coinage are likely related to Roman military activity, sites with more radiate coinage from Reece Periods 13 and 14 (c.AD 260-296) may reflect more urban activity, and sites with extensive 4th century nummus coinage likely demonstrate rural activity within the landscape.

Figure 5: Reece periods as defined by Richard Reece.

The PAS database allows us to attach a Reece Period to an individual coin so that, even when worn beyond being able to identify an individual ruler or coin type, we can very often place it within its correct Reece Period. This in turn enables comparison with other Roman material both at a local level and nationally. By looking at the PAS dataset as a whole we can create a PAS ‘mean’, an average of the coins on the database with an associated Reece Period that demonstrates the peaks and troughs of coin loss within the Province (Fig. 6). It should be noted that this analysis uses per mill (per 1,000) rather than percent to help avoid discrepancies when comparing sites with potentially widely varied numbers of coins.

Figure 6: The PAS ‘Mean’ based on Roman coins recorded through the PAS with associated Reece Period data up to May 2020. The PAS ‘Mean’ based on Roman coins recorded through the PAS with associated Reece Period data up to May 2020. 

What this graph demonstrates is that the PAS assemblage has very distinct peaks in the late Roman periods, notably Reece Periods 13 and 14, then again in Reece Periods 17 and 19, whereas the early material is comparatively poorly represented. The coins from Reece Periods 17 and 19 in particular demonstrates the very ‘rural’ nature of the PAS dataset, in part reflecting what we have seen in Fig. 4 and the distribution of coin finds nationally.

Once we have this national PAS mean, we can then use it to compare individual sites or regions to see where they deviate and therefore what this might tell us about that specific site or landscape. As noted above, peaks in Reece Periods 13 and 14 tend to be associated with urban activity, whereas the later peaks in Reece Periods 17 and 19 are typically linked to the development of the Roman rural landscape. If a site, therefore has a peak in one or other (or indeed an early peak linked to Roman military activity) this helps characterise its nature.

Some preliminary examples

At a very basic level, applying Reece Period to analysis to PAS coin assemblages is a quick and relatively straightforward way to characterise our dataset. If nothing else, thinking about how an assemblage is formed helps to understand an individual site and its position within the landscape. It can also aid in highlighting oddities and/or discrepancies in the data! At a national level we can also look at the spread of, for example, early Roman coinage to demonstrate those areas most affected by early Roman activity and what this might in turn tell us about the province.

Beginning with this national perspective some interesting results can be gained simply by grouping the coins by broad periods. Taken as a whole, if we plot all PAS coins to date by modern administrative district (county, unitary authority, etc.) the result we get in Fig. 7 demonstrates what at first glance seems like a huge area of Roman activity particularly in Norfolk and Suffolk, as well as Hampshire, Lincolnshire, Wiltshire, and Yorkshire.

Figure 7: PAS coins by modern administrative district excluding the IARCW data (c.271,000 coins). Greater density is demonstrated by darker reds.

This is reflected in a similar way if we look at the volume of Roman coins recorded by individual FLO region (Fig. 8). Again, the concentrations are in south west, East Anglia, Lincolnshire, and Yorkshire. What this initially seems to suggest is that, while you are certainly more likely to find and record a Roman coin in these regions, that Roman activity was most intensive here. However, we can begin to get slightly different perspectives if we interpret the coin data in different ways.

Figure 8: Roman coins recorded through the PAS by FLO region as percentages of the total number of PAS Roman coins.

The distribution changes dramatically if we consider the number of Roman coins as a percentage of the total number of Roman objects (coins and everything else!) recorded by each FLO region (Fig. 9).

Figure 9: Roman coins by FLO region as a percentage of the total number of Roman objects recorded.

Now we see the concentrations shift, most notably moving away from East Anglia to regions like Devon, which exhibit a higher percentage of coins amongst the Roman material reported. If you are recording finds in these regions, this means you’re more likely to see Roman coins – surely a win for everyone?!

This is all well and good, but what we are really seeing here is a reflection of the processes of recovery and recording. The large volumes of material from the east of England, for example, mean that there are always going to be larger numbers of Roman coins in these areas. Thinking about some of the approaches noted above using Reece Period analysis to refine this dataset slightly, is it possible to reveal a little more detail that is more pertinent to the ancient Roman landscape?

If I take the three broad groupings noted above – ‘Early Roman’ coinage (Reece Periods 1-11), Radiate coinages (Reece Periods 12-14), and nummi (Reece Periods 15-21) – to roughly correspond to early military(?) activity, urban activity with the radiates, and rural activity with the nummi, we get some interesting results.  Plotted as simple quantities (by modern administrative district) the image is almost identical for each and this reflects where we see the most material found and reported nationally (Figs 10-12). Not essentially that helpful in interpreting the Roman landscape other than to highlight where we might expect the largest number of finds.

Figure 10: ‘Early’ Roman coinage on the PAS, Reece Periods 1-11.
Radiate coinages (left; Reece Periods 12-14) and nummi (right; Reece Periods 15-21) recorded through the PAS.

However, if we look at this data proportionally as percentages of the total within each administrative district a very different picture emerges (Figs. 13-15).  For the early period, the shift is remarkable. The areas with the highest percentages of early coin types are in the far south west – Cornwall and Devon – the west of England, Wales, and up into the north west. East Anglia is comparatively poorly represented. It is notable, too, that the regions with the highest percentages in this instance are also those that are affected in particular by early (and in some cases extended) Roman military and urban activity. It might therefore be expected to some extent for these regions to exhibit higher volumes of early material.

Figure 13: ‘Early’ coinage on the PAS (Reece Periods 1-11) by percentage.

In the radiate period of the 3rd century, the picture again shifts. The south and south east are proportionally better represented by this stage, but there are still noticeable concentrations in the south and west, albeit on a more restricted scale than in the previous period.

Figure 14: Radiates (Reece Periods 12-14) on the PAS by percentage.

In the Late Roman period of the 4th century we can clearly see a band of material running from Wiltshire east and north to East Anglia, Lincolnshire and Yorkshire. This reflects similar conclusions reached by Philippa Walton (2012) in her study of the numismatic data for Roman Britain and correlates with the sudden growth of the rural landscape during the 4th century AD. What we are seeing by separating the material in this way, even with three very rough, broad periods, is not only the shift in coin distribution but also in patterns of Roman coin loss and by extension potentially the changing use (or at least focus) of the landscape during the Roman period.

Figure 15: Nummi (Reece Periods 15-21) recorded through the PAS by percentage.

Although this is a very cursory glance at three quite broad periods, the same principles can be applied to individual sites, landscapes, and the national picture, to produce some very interesting and emerging data. Recent work completed by Sam and I on the Roman coinage of Devon[7] as part of a collaborative project with the University of Exeter, for example, has allowed us to apply these same methods to a wider region. Here, we have been able to compare, for example, the coins from Devon to those of Cornwall by Reece Period (Fig. 16), highlighting that the two are quite different in their coin use and loss during the Roman period – Cornwall is much better represented in the early Roman period and Devon in the late. Similarly, we have been able to compare the data from individual sites and areas within the county to demonstrate the development of the Roman landscape here as demonstrated by its coinage. More to follow on this in due course!

Figure 16: The Roman coinages of Cornwall and Devon on the PAS by Reece Period (per mill).

You will have no doubt also seen the maps that Sam and I have used at various times to demonstrate the distribution of different coin types or periods represented within the PAS data. This is the next stage of analysis that I am working on slowly at the moment, looking at how the coinage relates to the landscape as much as modern administrative districts. Hopefully, by looking at the breakdown of material by Reece Period or coin type we can begin to refine some of the broad patterns noted above.

Of course, all of this is entirely dependent on the fantastic work being carried out in reporting and recording the coins on the PAS database! It really is important that in doing so Reece Periods are attached to the coin records where possible. As I mentioned above, with more than 320,000 coins it can be very difficult to work the data into a meaningful output (at least, in a short space of time!), but by attaching this all important data we can relatively quickly interpret the material – sometimes even with maps like the above!

This has been only a very small foray into a vast topic. In a later edition I hope to come back to it again and perhaps look at one or two examples in greater detail of how we can use the PAS data to explore individual sites within their landscapes. In the meantime, keep doing what you are doing in terms of recording the material – and don’t forget the Reece Period!

References and further reading:

[1] P. Guest and N. Wells, Iron Age and Roman Coins from Wales, (Moneta 66), 2007

[2] R. Bland, A. Chadwick, E. Ghey, C. Haselgrove, D. Mattingly, Iron Age and Roman Coin Hoards in Britain Oxbow, 2020.

[3] See for example Casey, Understanding Ancient Coins: An introduction for Archaeologists and Historians, London, 1986; J. Casey ‘The interpretation of Romano-British site finds’ in J. Casey and R. Reece Coins and the Archaeologist, 1974.

[4] Multiple contributions, notably R. Reece, Coinage in Roman Britain, 1987; R. Reece Roman Coins from 140 Sites in Britain (Cotswold Studies vol. IV), 1991; R. Reece, The Coinage of Roman Britain, 2002; R. Reece, Roman Coins and Archaeology: Collected Papers (Moneta 32), 2003.

[5] See for example S. Moorhead ‘Roman coin finds from Wiltshire’ in P. Ellis ed. Roman Wiltshire and After, 2011; S. Moorhead, A History of Roman Coinage in Britain, 2013; with P. Walton: S. Moorhead and P. Walton ‘Roman Coins recorded with the Portable Antiquities Scheme: A Summary’ Britannia 42, 2013; S. Moorhead and P. Walton, ‘Coinage at the end of Roman Britain’ in F.K.Haarer et. al AD 410: The History and Archaeology of Late and Post-Roman Britain, 2016; S. Moorhead and P. Walton ‘Coinage and the Economy’ in TheOxford Handbook of Roman Britain, 2016; S. Moorhead and P. Walton ‘Coinage and Collapse? The contribution of numismatic data to understanding the end of Roman Britain’ Internet Archaeology 41:

[6] See notably P. Walton Rethinking Roman Britain: Coinage and Archaeology, (Moneta 137), 2012; See also above contributions with Sam Moorhead.

[7] A. Brown and S. Moorhead ‘The Roman Coins from Exeter and its Hinterland’ in S. Rippon ed. Exeter: A Place in Time (Forthcoming, 2020)