the later iron age belongs to the publishers oxbow books ... der veen and jones... · j. d. hill 3....

This PDF file of your paper in The Later Iron Age belongs tothe publishers Oxbow Books and is their copyright.

As author you are licenced to make up to 50 offprints fromit, but beyond that you may not publish it on the WorldWide Web or in any other form.

THE LATER IRON AGE IN BRITAIN

AND BEYOND

edited by

Colin Haselgrove and Tom Moore

Oxbow Books

Published byOxbow Books, Park End Place, Oxford OX1 1HN

© Oxbow Books and the authors, 2007

ISBN 978-1-84217-252-0 1-84217-252-0

A CIP record for this book is available from the British Library

This book is available direct from

Oxbow Books, Park End Place, Oxford OX1 1HN(Phone: 01865-241249; Fax: 01865-794449)

and

The David Brown Books CompanyPO Box 511, Oakville, CT 06779, USA

(Phone: 860-945-9329; Fax: 860-945-9468)

or from our website

www.oxbowbooks.com

Cover by Christina Unwin based on an idea by Rachel Pope; reconstruction of burial by Simon James

Printed in Great Britain atShort Run Press, Exeter

Contents

1. New narratives of the Later Iron Age 1Colin Haselgrove and Tom Moore

2. The dynamics of social change in Later Iron Age eastern and south-eastern England 16c. 300 BC–AD 43.J. D. Hill

3. Life on the edge? Exchange, community, and identity in the Later Iron Age 41of the Severn–CotswoldsTom Moore

4. Central places or special places? The origins and development of ‘oppida’ in Hertfordshire 62Stewart Bryant

5. Cultural choices in the ‘British Eastern Channel Area’ in the Late Pre-Roman Iron Age 81Sue Hamilton

6. Sea, coast, estuary, land, and culture in Iron Age Britain 107Steven Willis

7. Social landscapes and identities in the Irish Iron Age 130Ian Armit

8. Re-situating the Later Iron Age in Cornwall and Devon: new perspectives 140from the settlement recordL. J. Cripps

9. Unravelling the Iron Age landscape of the Upper Thames valley 156Gill Hey

10. Rooted to the spot: the ‘smaller enclosures’ of the later first millennium BC 173in the central Welsh MarchesAndy Wigley

11. From open to enclosed: Iron Age landscapes of the Trent valley 190David Knight

12. Realigning the world: pit alignments and their landscape context 219Jim Rylatt and Bill Bevan

13. Good fences make good neighbours? Exploring the ladder enclosures 235of Late Iron Age East YorkshireMelanie Giles

14. Putting the neighbours in their place? Displays of position and possession 250in northern Cheviot ‘hillfort’ designPaul Frodsham, Iain Hedley and Rob Young

vi Contents

15. Dominated by unenclosed settlement? The Later Iron Age in eastern Scotland 266north of the ForthMairi H. Davies

16. Artefacts, regions, and identities in the northern British Iron Age 286Fraser Hunter

17. Silent Silures? Locating people and places in the Iron Age of south Wales 297Adam Gwilt

18. Perspectives on insular La Tène art 329Philip Macdonald

19. Dancing with dragons: fantastic animals in the earlier Celtic art of Iron Age Britain 339A. P. Fitzpatrick

20. An archaeological investigation of Later Iron Age Norfolk: analysing hoarding patterns 358across the landscapeNatasha Hutcheson

21. Detecting the Later Iron Age: a view from the Portable Antiquities Scheme 371Sally Worrell

22. The end of the Sheep Age: people and animals in the Late Iron Age 389Umberto Albarella

23. To fish or not to fish? Evidence for the possible avoidance of fish consumption 403during the Iron Age around the North SeaKeith Dobney and Anton Ervynck

24. The production and consumption of cereals: a question of scale 419Marijke van der Veen and Glynis Jones

25. Making magic: later prehistoric and early Roman salt production in the Lincolnshire fenland 430Elaine L. Morris

26. Excarnation to cremation: continuity or change? 444Gillian Carr

27. Households and social change in Jutland, 500 BC–AD 200 454Leo Webley

28. Weapons, ritual, and communication in Late Iron Age northern Europe 468Peter S. Wells

29. Understanding social change in the Late Iron Age Lower Rhine region 478Nico Roymans

30. The age of enclosure: Later Iron Age settlement and society in northern France 492Colin Haselgrove

31. The polities of Gaul, Britain, and Ireland in the Late Iron Age 523John Collis

List of contributors 529

The production and consumption of cereals:a question of scale

Marijke van der Veen and Glynis Jones

Introduction

In their paper Understanding The British Iron Age: An Agendafor Action, Haselgrove et al. (2001, iv) identify regionalityand the nature of socio-economic changes as two ofthe five key areas of future research on the British IronAge. As farming formed the basis of all societies in thisperiod, and as most settlements were farmsteads andmost people were farmers (ibid., 10), any assessment ofregional differences and socio-economic change willhave to include an assessment of farming practices. Ata basic level this concerns an assessment of the scale ofagricultural production (e.g. ability to produce a surplus,intensive/extensive cultivation regimes) and of the levelof specialisation (e.g. crops versus animals, farmingversus non-farming settlements). It goes without sayingthat the success of such assessments hinges on choosingthe right methodology and the right scale of analysisand interpretation.

To date, much discussion of intra- and inter-regionalvariation in Iron Age crop production has focussed onthe level of specialisation, namely the identification ofproducer and consumer sites. A model developed byM. Jones (1985) and applied to sites in the upperThames valley was the first apparently successfulattempt to identify settlements which produced theirown crops (arable or producer sites) and those whichreceived crops that had been grown elsewhere (pastoralor consumer sites). This pioneering work broughtarchaeobotanical data into the forefront of mainstreamarchaeological debate and has stimulated much of themore recent research in this area. The model aimed tofacilitate easy comparison between sites and monitor

the movement of arable produce across the landscape,and the results allowed M. Jones (1996, 35) to suggestthe existence of ‘neighbourhood groups of agrariansites engaged in a common network of plant productionand consumption’. While the main assumptionsunderlying the model and the method of constructingthe diagrams were criticised early on (G. Jones 1987;Van der Veen 1987; 1991; 1992, chapter 8), the model,and the conclusions drawn from it, are still widely used.

In this paper we argue that the problems associatedwith M. Jones’ model are such that it cannot be used todistinguish between producer and consumer sites, andthat recent explanations of differences betweenarchaeobotanical assemblages at sites in the upperThames valley (Campbell 2000; Stevens 2003) are alsoflawed. Here we briefly summarise M. Jones’ model, andthe criticisms it has received, and review the more recentinterpretations of the observed site differences. We thenapproach the problem from a different angle, proposinglevels of analysis and interpretation more appropriate tothe data available and the questions posed. Finally, weput forward our own interpretation of the patterningobserved.

As the model is based on the interpretation of charredplant remains, our arguments inevitably involve detailedconsideration of the formation processes at work. Inthis paper we try to put our case without recourse tocomplex archaeobotanical jargon, to keep the paperaccessible to a wider readership. Some basic features ofcereals and relevant terminology do, however, need tobe explained first.

420 Marijke van der Veen and Glynis Jones

Cereal types and terminology

In the following discussions, we make a distinctionbetween emmer and spelt wheat on the one hand andbread wheat and barley on the other. The first two cerealsare ‘glume’ wheats, which, when threshed, break up intoindividual segments (spikelets) with the grain still tightlyenclosed by the surrounding chaff (glumes). The lasttwo are ‘free-threshing’ cereals, the ears of which, whenthreshed, disintegrate into free grain and chaff (Fig. 1).The glume wheats therefore require a further ‘dehusking’operation to release the grain from the glumes, which isnot necessary for the free-threshing cereals. Further-more, for the glume wheats, the chaff elements mostlikely to survive archaeologically (because of theirrobustness) are the glume bases, while similarly, for free-threshing cereals, the central ‘stalk’ (rachis) is the part ofthe ear most likely to survive.

When we use the term ‘producer site’, we follow M.Jones’ original definition (1985) of a site growing andharvesting its own crops. This is effectively the same ashis later definition of ‘biological’ production (theproduction of grain by the plant itself), rather than hisdefinition of ‘economic’ production, which includes allcrop processing activities including the later graincleaning stages (M. Jones 1996, 34), and which tends toblur his earlier distinction between consumer andproducer sites. By consumption we mean the use ofthese crops, mostly their consumption as food, asopposed to ‘everything that humans do’ (ibid.). For thepurposes of this paper, therefore, the inhabitants ofproducer sites are cultivators (as well as consumers ofcrops), who may or may not export part of theirproduce, whereas consumer sites import their cropsfrom elsewhere, although they may process them furtherto obtain clean grain (especially with the glume wheats).

Throughout the paper we will use the terms grain-rich, chaff-rich and weed-rich to refer to the relativequantities of grain, chaff and weed seeds. Where largeabsolute quantities are indicated we have described theseas ‘large’. Furthermore, the term ‘assemblage’ will beused to indicate a site assemblage of archaeobotanicalremains, while the term ‘sample’ will be used to refer tothe plant remains recovered from a single archaeo-botanical sample.

M. Jones’ archaeobotanical model ofproduction and consumption

The model

The model relies on the interpretation of ‘triangulardiagrams’ which display the broad characteristics ofcharred plant assemblages from individual sites (M.Jones 1985). M. Jones made the convincing case that theoccurrence of grain-rich samples required an

explanation, as grain is that part of the harvest leastlikely to be wasted, and argued that the ‘most likely placefor this unlikely event to occur is at its place ofproduction’ (ibid., 120). By plotting the relativeproportions of cereal grain, chaff, and weed seeds ontotriangular diagrams, on a sample-by-sample basis, andgiving a measure of seed density, the overall assemblagefrom each site could be characterised. Grain-rich sites(i.e. Ashville and Mount Farm) were interpreted asproducer sites, and sites poor in grain (but rich in weedsand/or chaff) as consumer sites (i.e. Smith’s Field andClaydon Pike). Danebury is different from both of thesecategories in terms of both composition and density,and was interpreted as being engaged in a ‘broad rangeof agricultural activities’ (ibid., 121). Finally, sites wherethe samples are concentrated in the centre of the diagram(with approximately equal proportions of grains, chaffand weeds; e.g. Iron Age sites north of the Tyne) wereseen as ‘self-contained units’ (M. Jones 1996, 35).

The critique

While the aspiration of the model was widely welcomed,three aspects were criticised from the start. Firstly, in theconstruction of the diagrams the content of each sampleis summarised without regard for context or speciescomposition. With regard to the latter, any variation inthe abundance of glume wheats (emmer or spelt wheat)versus free-threshing cereals (bread wheat or barley)affects the location of samples in the diagram (G. Jones1987; Van der Veen 1991; 1992, 98). This is due todifferences in the likelihood of chaff from these twotypes of cereal being found archaeologically. The chaffof free-threshing cereals is largely represented by rachisremains which are removed early in the processingsequence, often off-site, and is thus relatively rarelyrepresented in archaeobotanical assemblages. The chaffof glume wheats, on the other hand, is largely composedof glume bases which are removed at a later stage ofprocessing, often on a day-to-day basis in a householdcontext. Thus, variations in the proportion of chaff asindicated in a triangular diagram may reflect variationsin the relative importance of emmer and spelt wheatversus bread wheat and barley, rather than variations insubsistence strategy. Secondly, the assertion that grain iswasted more frequently on producer sites than onconsumer sites is questionable (G. Jones 1987; Van derVeen 1991; 1992, 98). It has been argued that, on thecontrary, producer sites are characterised by the wastefrom early stages of crop processing (straw and rachis)and consumption by grain-rich samples (Hillman 1981;1984a). The burning of grain usually represents anaccident, and accidents can occur on all types of site (G.Jones 2000). Finally, the diagrams make no reference tocontext or crop processing stage: variations betweensites in the contexts or stages sampled (e.g. storage areasversus ditch fills and/or products versus by-products)

Production and consumption of cereals 421

Fig. 1. Effects of crop processing on glume wheats and free-threshing cereals (after Charles 1984). Note that hulled barley is a free-threshingcereal; it behaves in the same way as bread wheat during threshing (the ‘hulls’ refer to the lemma and palea which are fused with the grain,not the glumes).


may affect the position of samples in the diagram; forexample, storage areas house the cleaned products ofcrop processing and so are likely to be richer in grainthan ditch fills receiving waste mainly from the by-products of crop cleaning (Van der Veen 1992, 98).

Problems with the application of the model

Whenever the model has been applied to other sites,analysts have struggled to interpret the patterns found.The classification provided by the model frequently didnot match the expectation based on other information.M. Jones encountered such a situation himself at MaidenCastle, where the assemblage was dominated by chaff(especially glumes) and to a lesser extent weeds, eventhough grain-rich samples might reasonably have beenexpected, given the type of the site: a hillfort, likeDanebury, with ample storage facilities (Palmer andJones 1991). This discrepancy was explained by: (1) asampling factor (the area excavated was small and grain-rich deposits might have lain outside this area); (2) ataphonomic factor (locally poor preservation of grain);and (3) a cultural factor (scale of storage smaller than atDanebury) (ibid., 1991, 136). In fact, in the original study,the evidence that Danebury was a site supplied withcrops from various parts of its territory was basedprimarily on the presence of weed species representinga mixture of ecological types (M. Jones 1985), ratherthan on the relative proportions of grain, chaff andweed. The suggestion that this grain would then leaveDanebury in a clean state, introduces a methodologicalproblem. The sites receiving this grain would becharacterised by clean grain only, similar to the grain-rich assemblages considered by M. Jones to be typical ofproducer sites, making the two types of site difficult todistinguish.

Similarly, the application of the model to Iron Agesettlements in north-east England led to all sites southof the Tyne being classified as consumer sites, whichwould have resulted in a complete absence of producer

sites in that region (Van der Veen 1992, chapter 8).Conversely, the Roman fort at South Shields wasclassified as a producer site, even though the assemblagewas derived from a granary destroyed by fire at a classicconsumer site (Van der Veen 1992, chapter 8).

Alternative interpretations of variation inarchaeobotanical site assemblages

Recently, two researchers have offered interestingalternative interpretations of the patterns observed byM. Jones, which provide new insights into the possiblenature and organisation of Iron Age settlement andstimulate further debate. We review both of theseinterpretations here.

Use of chaff as fodder

Campbell (2000) applied the model to several Iron Agesites excavated as part of the Danebury Environs Projectand reviewed the evidence from the upper Thames valley.She suggests that M. Jones’ producer and consumer sitesmay, in fact, all be growing their own crops, and relatessome of the observed differences to variations in theneed for fodder (Table 1). The inhabitants of the secondgravel terrace, where pasture was thought to be scarce,may have used chaff as fodder rather than as fuel, andhence created charred assemblages low in chaff.Moreover, high-status sites with more animals to feedover the winter (she mentions Danebury and SuddernFarm as examples) may also have used all the availablechaff as fodder, rather than fuel.

This interpretation is attractive in that it offers apossible explanation for the lack of chaff at some sites,in particular at M. Jones’ producer sites, which – if theywere producing (and presumably partly consuming) theirown crops – would be expected to generate considerablequantities of chaff. It does not, however, explain whysome samples are dominated by large quantities of grain,

Table 1. Alternative interpretations of grain and chaff/weed-rich assemblages.

Model Samples rich in grain Samples rich in chaff/weed

interpretation producer site consumer site M. Jones 1985

reason grain wasted at harvest time grain carefully conserved

interpretation fodder scarce fodder plentiful Campbell 2000

reason chaff used as fodder chaff used as fuel

interpretation communal storage household storage Stevens 2003

reason storage as ‘semi-clean spikelets’ storage as ‘partially threshed ears’

interpretation large scale small scale Van der Veen and Jones reason accidental charring of products by-products of day-to-day processing


a commodity that should not, in the normal course ofevents, be deliberately burnt. If grain cleaning by-products were used as fodder, any grain that wasinadvertently removed in this way would be consumedby the animals along with the chaff. This explanation istherefore partial, at best, and some other explanationmust be sought for the presence of grain-rich assem-blages.

Communal versus household storage

Stevens (2003) reinterpreted the assemblages used in M.Jones’ original model alongside some newly studied sitesfrom the upper Thames valley, and related the differencesto the stage at which crops were put into storage which,in turn, may reflect storage at a communal or householdlevel (Table 1). He argued that cleaning waste from wheatstored as ‘clean or semi-clean spikelets’ led toassemblages rich in grain compared to weed seeds (as atM. Jones’ producer sites), and were characteristic ofcommunal storage. Conversely, cleaning waste fromwheat stored as ‘unsieved spikelets’1 led to assemblagespoor in grain compared to weed (as at M. Jones’consumer sites), and were characteristic of householdstorage. While the concept of storing sieved versusunsieved spikelets is an interesting one to explore (andcould indeed have some connection with household orcommunal storage practices), it does not explain thepattern observed. Both forms of storage involve wholespikelets, and the quantity of grain relative to chaff(glumes) would therefore be the same in both cases.

Stevens maintains that the samples from his sitesconsist primarily of the processing waste associated withthe routine, day-to-day dehusking and cleaning of glumewheat spikelets, after these have been taken out ofstorage (and that samples rich in grain result from a biasagainst the preservation of chaff in this type ofprocessing waste; cf. Boardman and Jones 1990).However, contrary to what Stevens says, this wouldgenerate samples rich in weed seeds relative to chaff atsites where unsieved spikelets were stored (i.e. samplesin the bottom left corner of a triangular diagram; seeFig. 2b), and samples rich in chaff relative to weed seedswhere sieved spikelets were stored (i.e. samples in thebottom right corner of a triangular diagram; see Fig. 2c).If there were a bias against chaff, because it ispreferentially destroyed compared to grain, some of thechaff-rich samples would appear higher in the triangle.In fact, however, neither of the sites (Ashville and MountFarm) that Stevens classifies as storing cleaned spikelets(indicating communal storage, the equivalent of M.Jones’ ‘producer’ sites) have samples rich in chaffcompared to weed seeds whereas some of the samplesfrom sites (e.g. Gravelly Guy and Yarnton) classified asstoring unsieved spikelets (indicating ‘household’storage, the equivalent of M. Jones’ ‘consumer’ sites) do(Stevens 2003, fig. 3). This is the opposite of expectationsbased on his model.

Formation of the archaeobotanical recordand the scale of agricultural production/consumption

It will be clear from the above discussion that we cannotexpect a simple relationship between the observedpatterning in the charred seed assemblage and the statusof the sites in terms of production or consumption,nor do the alternative interpretations put forward byCampbell (2000) and Stevens (2003) explain theobserved differences between sites. Nevertheless,

Fig. 2. Different types of charred plant assemblages. Shaded areasindicate the position (in the plot) of the majority of samples in eachtype of assemblage.


patterning in the crop, chaff, and weed components existsin the archaeological record and the desire to interpretits meaning remains. We contend that an appreciation ofthe formation processes underlying the charredarchaeobotanical record will help in understanding thepatterning.

With the exception of deliberate offerings ordestruction due to conflict, charring events occur in oneof three circumstances: (1) when the by-products ofgrain dehusking and cleaning are deliberately burnt aseither fuel or waste; (2) when an accident occurs duringsome process involving fire, e.g. during parching, dryingor cooking; and (3) when a building containing storedproduce catches fire. The first circumstance mostly arisesfrom day-to-day processing immediately prior toconsumption and is likely to yield samples rich in chaffand/or weed seeds rather than grain. Cooking will ofteninvolve ‘processed’ grain (e.g. cracked wheat or flour) orcooking with water, neither of which is likely to generatewhole, charred grains. Drying/parching and destructionin store therefore remain the most likely events leadingto samples rich in charred grain, and more likely (at leastfor large grain-rich samples) than either the ‘wastage’ ofgrain proposed by M. Jones or the preservation biasagainst chaff in processing waste proposed by Stevens(above). Thus, although most archaeobotanical materialis probably generated during the day-to-day processingof cereals, some is generated during infrequent accidents.

The day-to-day processing of cereals takes place at allsites (producers and consumers), and the presence ofsamples consisting primarily of chaff and/or weed seedsis, therefore, to be expected at all sites. In contrast, theoccurrence of large grain-rich deposits (as M. Jones 1985says) needs explanation The answer to the question,‘where are accidents involving parching, drying andstorage most likely to occur?’ is that they will tend tooccur in places where these activities are regularly carriedout, i.e. where grain is handled in bulk. Hillman (1984a)suggested that quantities of charred grain mostcommonly occur on large sites, whether large producers(e.g. for a later period, manorial farms) or largeconsumers, since there is more opportunity at these sitesfor accidents such as the destruction of a store by fire(or for large scale parching/drying accidents).

A possible exception to this association of grain-richdeposits with accidental charring is the cleaning ofstorage pits through the deliberate burning of pit linings.This activity could generate grain-rich deposits that werenot created by accident. As these pits tend to beassociated with the large-scale handling of cerealproduce (see below), however, they also point to sitesengaged in the bulk handling of grain.

In other words, a predominance of grain-rich samples(interpreted as accidental charring of cleaned productsor the deliberate cleaning of storage pits by fire), is farmore likely to be an indicator of the scale of productionand consumption than a means of distinguishing

between the two (see also Van der Veen 1987; 1991;1992, chapter 8). The upper Thames valley ‘producer’sites may therefore represent large-scale productionand/or consumption, and the ‘consumer’ sites mayrepresent settlements engaged in small-scale activityrelating to cereals.2 Whether the sites are producers ormerely consumers of cereal grain may not be reflectedin the chaff : grain ratios because sites that import grainin bulk, and then store and dry it, might well produceplant assemblages very like large-scale producers, andsmall producers may be indistinguishable from smallconsumers (Van der Veen 1991).

Levels of archaeobotanical analysis andinterpretation

The fact that some archaeobotanical samples aregenerated during routine activities while others occurprimarily by accident also has implications for the levelat which archaeobotanical remains should be analysedand interpreted (cf. G. Jones 1991).

Level of analysis

A critical drawback of M. Jones’ model is that it interpretsa site assemblage ‘mechanically’ on the basis of broadbotanical composition. As we have indicated above, thecomplexity of the archaeobotanical record is such thatonly an analysis that takes this into account can hope tosucceed. This means that we need to understand thetaphonomic pathways of individual samples, primarilyfrom botanical composition but also taking into accountarchaeological context (Dennell 1974; 1976; Hillman1981; 1984a; G. Jones 1984; 1987). Only through thistype of analysis can a distinction be made betweenregular, routine activities, of a particular type, and rareaccidents, with their likely cause. While at the level ofthe sample this may be seen as rather mundane in itself,bringing together samples which have first beeninterpreted individually ultimately provides a morereliable interpretation at the site level and above thandoes the broad botanical site composition.

We suggest therefore that, rather than using atriangular diagram to summarise the botanical com-position of a whole site, methods are first applied todetermine the origin of individual samples. This can beachieved through a combined consideration of the ratiosof major plant components (grain, chaff, straw andweeds), the types of weed accompanying crops, and thecircumstances of deposition (see Table 2). Anotheradvantage of this approach is that it allows thecalculation of separate chaff : grain ratios for freethreshing cereals (rachis internodes : grains) and glumewheats (glume bases : grains), while still allowing thecalculation of weed : grain ratios, as it is not possible todetermine the association of particular weeds with a


particular crop type in a mixed sample (Hillman 1981;1984a; G. Jones 1984; for applications see G. Jones 1987;Van der Veen 1992; Campbell 2000; Hodgson et al.2001). As different types of weed are removed at eachstage of processing, this provides a complementary wayof assessing processing stage and is applicable to bothglume wheats and free-threshing cereals. This can beachieved through multivariate statistical methods(G. Jones 1984; for applications see G. Jones 1987; Vander Veen 1992; Charles and Bogaard 2001) or, moresimply but less conclusively, by calculating ratios, e.g. ofsmall weed seeds : large weed seeds. Circumstances ofdeposition can be assessed through: (1) the ‘density’ ofcrop remains (number of items per litre of deposit),which gives a broad indication of the rate of deposition;and (2) archaeological context, which may provide a clearindication of the nature of the deposit (e.g. a granary) orsimply indicate a secondary or tertiary context (e.g. arefuse pit). This method will also help to distinguishbetween finds of very mixed origin and those derivedprimarily from one type of activity.

Level of interpretation

If it is accepted that differences between sites in theamount of grain charred are largely due to chanceaccidents, then the implication is that one cannot expectall sites to provide evidence for such accidents. On aprobabilistic level, site assemblages dominated by grainmay be an indication of large-scale agricultural activity.The assumption that this would inevitably be the case,however, will result in the misinterpretation of someindividual sites, such as a site with small-scale storagetotally destroyed by fire; or a site with large-scale storage,which happened never to have suffered any fire damage,or where the area excavated lay outside that where grain-

rich deposits were dumped (as is suggested for MaidenCastle). To overcome this chance element, it is necessaryto interpret charred archaeobotanical site assemblagesat a regional level where individual sites make only alimited contribution to the overall pattern. The mostuseful levels of analysis and interpretation may thereforebe those of the individual sample and the broadgeographic region, respectively.

Iron Age Britain – scale, regionality,and change

Returning to regionality and socio-economic change, anexamination, at a regional level, of the likely reasons forthe presence of grain-rich assemblages may tell us moreabout the nature of early agriculture than whetherindividual sites were importing grain or producing theirown. Grain-rich samples in the Iron Age, for example,have been contrasted with the relative lack of charredgrain from Neolithic deposits, and used to suggest thatcereals were not an important source of food in theNeolithic period (e.g. Moffett et al. 1989; Thomas 1991;Barrett 1994; Edmonds 1997). This has been questionedby several authors (e.g. Cooney 1997; G. Jones 2000;Monk 2000; Rowley-Conwy 2000), who attribute thelack of charred grain in the Neolithic to taphonomiccauses and contextual differences; indeed recentevidence suggests that the quantity of cereal grain fromNeolithic sites is not substantially less than that from theIron Age (Jones and Rowley-Conwy forthcoming). Ifwe are correct in interpreting accidentally charred grain-rich samples as representing large-scale production and/or consumption – rather than simply reflecting therelative contribution of cereals to the diet – then anyevidence indicating greater quantities of grain in the

Table 2. Variables useful for the identification of crop processing stage, and their likely meaning. For the first three variables, the terms‘high’ and ‘low’ value refer to the degree to which they differ from the ratio in the cereal plant; for the last three variables, they refer to therelative values within the site/region.

Sample variable Sample origin

Ratio High value Low value

cereal straw nodes : grains by-product from early processing stage grain product

free-threshing rachis internodes : grains by-product from early processing stage grain product

glume wheat glume bases : grains by-product from late processing stage grain product

weed seeds : cereal grains by-product from late processing stage grain product

small : large weed seeds by-product from sieving product from sieving or by-product of hand cleaning

number of crop items per litre of deposit rapid/single deposition (usually result of accident)

slow/repeated deposition (usually day-to-day activity)


Iron Age suggests that arable production in some partsof Britain had moved beyond subsistence and includeda considerable degree of surplus production.

The regular occurrences of grain-rich deposits in IronAge Britain, representing visible surplus production ofgrain, are, to date, restricted to central and southernBritain (see, for example, M. Jones 1984a; 1985; 1996;Stevens 2003). This distribution is not dissimilar to thatof grain storage facilities such as pits and, to a lesserextent, four-post structures (Gent 1983; Cunliffe 1992).While it is true that the geographical distribution ofthese storage pits is mostly conditioned by the underlyinggeology (Bradley 1978; Fenton 1983; Cunliffe 1992),and that there are alternative forms of grain storage thatleave few if any archaeological traces (Fenton 1983), it isnot just their geographical distribution, but also theirchronological spread (c. 800–100 BC in Britain, c. 900–20 BC in northern France; Cunliffe 1992; Gransar 2000)that is strongly suggestive of a link with a particularagricultural or social system. Furthermore, ethnographicevidence suggests that underground grain silos are usedfor the storage of surplus grain (Fenton 1983), ratherthan seed corn, although the latter is still the widely heldassumption amongst archaeologists (e.g. Reynolds 1974;M. Jones 1984b; Cunliffe 1992; 2000, 130). The shorttime-span between harvest and sowing for autumn-sowncrops such as spelt wheat makes the storage of seedcorn in such pits an unlikely investment, other than inperiods of major unrest. Moreover, once opened, thegrain from these pits needs to be used quickly. The factthat such storage facilities predominate in a certain typeof site, i.e. hillforts (Gent 1983; Cunliffe 1992), and thatthe amount of storage available here often exceeds theneeds of the individual site, also points to the practiceof surplus production and some form of centralisedstorage (cf. Sharples 1991; Cunliffe 1992), at least in theearlier Iron Age. By the Middle Iron Age the situation isa little different, as the bulk of the population is thenapparently nucleated into the ‘developed’ hillforts likeDanebury and Maiden Castle, rather than dispersed inthe immediate environs (see below), but even thenstorage appears to exceed need.

The unusual plant assemblage (very high density ofdeposition; grain-rich samples; weeds originating fromwide region) and the exceptionally large storage facilitiesat one of these hillforts (Danebury), led originally to thesuggestion that these sites functioned as central places,from which surplus grain was redistributed (Cunliffe1984, 556–559; M. Jones 1984b; 1985), along the lines ofthe classic redistributive chiefdoms. Problems with thismodel were outlined by Hill (1995a), who pointed out thathillforts such as Danebury did not necessarily representthe peak of the social and settlement hierarchy (given theabsence of status indicators and presence of storagefacilities at non-hillfort sites, amongst other factors). Wecan now add to this the nature of the plant assemblagesat sites in the Danebury region. The nature of the

archaeobotanical assemblage at Danebury had originallyled M. Jones to suggest that part of the crops found therederived from other settlements within its territory, whichwere then stored and processed into fully clean grain atthe hillfort. The grain would then leave the hillfort in afully cleaned state; and settlements receiving this grainshould thus be characterised by assemblages with little orno chaff or weeds (M. Jones 1984b; 1985, 122).

No such settlements have however been found, eitherin the Danebury region, or for that matter anywhere elsein Iron Age Britain. The recent results from theDanebury Environs Programme make it clear that thegrain thought to have left Danebury did not go tosettlements within the region. The plant assemblagesfrom the five Iron Age sites investigated within theDanebury region all contain clear evidence for thedehusking and cleaning of grain (Campbell 2000), thatis precisely the activities that were carried out atDanebury. Thus, the grain cleaned at Daneburyapparently did not go (or return) to sites in the region,but instead was consumed in the hillfort itself or tradedoutside the region (possible scenarios already envisagedby M. Jones; 1985, 122). There is no real evidence forthe latter, in that Danebury does not have the greaterquantities of elite goods compared to surroundingsettlements that one might expect from such trade (Hill1995a). The most likely explanation for the largequantities of grain brought to the hillfort is, therefore,that this grain was consumed there. The evidence fordehusking and cleaning of the grain at Danebury alsopoints to consumption, as this processing tends to takeplace immediately before consumption. Such grain isunlikely to have been used for seed corn as the dehuskingdamages the embryo of the grains and thus reducestheir germination rate (glume wheats are usually sown asspikelets, not clean grain; Nesbitt et al. 1996). Theevidence from Danebury thus points to large-scaleconsumption, possibly during large communal feasts.

This interpretation would fit comfortably withincurrent models of social and economic change putforward for the Iron Age (e.g. Sharples 1991; Haselgrove1999; Cunliffe 2000, chapter 4). Such models haveidentified the demise of the position of the elites in theLate Bronze Age caused by the development of the newiron technology and consequent reduced role of long-distance trade of precious metals. This led to an increasedreliance on the creation of – and control over – agrariansurpluses during the Iron Age. A possible scenario is thatmany communities in southern Britain worked to achievegrain surpluses, which were stored in pits and used foroccasional feasts. Some of these feasts may have beensmall domestic feasts, others large communal ones; thelatter taking place at the regional hillforts. Both Sharples(1991) and Hill (1995a) see these hillforts as communalsites, and Hill (1995a) has suggested that Danebury mayinitially not have been occupied permanently (many ofthe storage pits at Danebury appear to have been left open


for longer periods than those at non-hillfort sites). At bothtypes of site some pits receive ‘special deposits’, offeringsto the deities associated with fertility rituals, after the grainwas removed (Grant 1984; 1991; Cunliffe 1992; 2000,130; Hill 1995b).

By c. 300 BC the leaders of certain communitiesappear to have succeeded in enhancing their status andprestige to such an extent that they could move into the(developed) hillforts and raise the required manpowerto enhance their encircling earthworks, possibly by usingcorvée labour (and involving further large-scale feasting).This would explain the evidence for increased occu-pation at these sites, the reduction in settlements aroundthem, the increased storage facilities at such sites, andthe concentration of shrines and increased evidence forritual. Finally, by the Late Iron Age the hillforts areabandoned, settlement disperses again, storage pitsdisappear from the record and towards the very end ofthe period long-distance trade (re-)emerges. One ofseveral explanations for this pattern is that instead ofsurpluses being stored for feasts, they now leave theregion in exchange for new consumer goods: Romanceramics, glass, and exotic foods such as wine and figs(Haselgrove 1999; Cunliffe 2000, 191–2).

What we appear to observe is a classic change in theway food is used either to homogenise or ‘heterogenise’the participants in the meal (cf. Appadurai 1981; Dietler1996; Van der Veen 2003). During the Early Iron Agegrain surpluses may have been accumulated forcelebratory feasts. Such feasts serve to enhance socialbonds in societies with little social inequality. Overtime certain communities or individuals manage toincrease their standing and prestige by hosting morefeasts and, by eating the food, the guests accept theobligation to give something in return, either deferenceor labour. The shift towards the developed hillfortsaround 300 BC may point to these communities orindividuals having achieved special status, andcommensal hospitality may now have been used toreiterate and legitimise growing differences in statusand power. Leaders of these communities would havebeen expected to host lavish parties, whilst participantswere expected to pay tribute and/or offer labour. Then,by the very end of the Iron Age, we see a move awayfrom the use of food to maintain and enhance socialbonds, towards the use of food to create distance. Theemphasis is no longer on the consumption of the samefoods (common staples), but on the consumption ofdifferent foods (Van der Veen forthcoming). The elitestarts to consume wine and other exotics (such as thefigs found at Hengistbury Head (M. Robinson pers.comm.); and to use imported ceramics and glass toenhance the display component of the meal. Thus, wesee a move from communal feasts to exclusive dining;in the latter there is no longer any element ofreciprocity; the ‘audience’ no longer participates(‘diacritic’ feasts; Dietler 1996).

Conclusion

In this paper we have drawn attention to two aspects ofarchaeobotanical analysis and interpretation, onemethodological, the other concerning the meaning ofcharred plant assemblages. First, the fact that grain-richsamples are generated primarily through accidents hastwo major methodological implications: (a) we need tounderstand the taphonomic pathway of individualsamples, which means that the sample is the most usefullevel of analysis; and (b) we need to allow for the ‘chance’element in archaeobotanical preservation, which meansthat the region is the most useful level of interpretation.This is in contrast to the more usual approach where thesite constitutes the unit of both analysis and inter-pretation.

Secondly, we argue that, although the use of chaff aseither fodder or fuel provides a partial explanation insome cases, the relative proportions of grain, chaff, andweeds at archaeological sites tell us more about the scaleof agricultural activity than about whether individualsites were consumers or producers, or whether storagewas at the household or community level. These latterinterpretations are based on the erroneous assumptionthat grain-rich samples are generated through charringof the waste from routine activities rather than as aresult of relatively rare accidents involving fire. We haveargued that such accidents are most likely to happen atsites where cereals are handled in bulk (be they produceror consumer sites), and that grain-rich samples thuspoint to large-scale production and/or consumption(adding an additional criterion to those discussed byBakels (1996) for the detection of surplus production).We regard this ability to assess the scale of theagricultural system, with its implication for the presenceof surplus production and its consumption, as anexciting new development – it will greatly facilitate thestudy of both regionality and socio-economic change.

Finally, we have interpreted the evidence for theproduction and consumption of grain surpluses insouthern Britain as surpluses that may have beenproduced for – and consumed during – feasts. Thesemay initially have functioned to maintain the social bondswithin and between communities, but may over timehave been increasingly used by the leaders of certaincommunities to enhance their own prestige and status,resulting in particular hillforts becoming centres ofpower. By the end of the period we see a major change:the grain surpluses are apparently no longer stored andconsumed within the region, but possibly exported outof the region in return for items of elite display. Thus,during the Iron Age grain surpluses in southern Britainwere used to mobilise prestige and status through locallarge-scale feasting; by the Late Iron Age we start to seethe mobilisation of grain surpluses across the landscape,something that became increasingly common during theRoman period. To conclude, during the Iron Age grain


surplus (the economic capital) was used to acquire socialpower (prestige, status); by the end of the period it startedto be used to acquire cultural power (exclusivity, elitism).

Acknowledgements

We would like to thank Debbie Miles-Williams and JoeSkinner for preparing the figures.

Notes

1. In fact Stevens uses the term ‘partially threshed ears’ here but, asglume wheat ears inevitably break up into individual spikeletswhen threshed (Hillman 1981, 1984a, 1984b), it is not clear how‘partially threshed ears’ would ever be generated. As Stevenshimself implies that both types of storage product are still in astate where the glumes (chaff) tightly invest the grain (i.e. theyhave not undergone the dehusking process), the term ‘unsievedspikelets’ is used here in place of ‘partially threshed ears’.

2. We agree with Stevens (2003) that it is unlikely that M. Jones’‘consumer’ sites in the upper Thames valley are purely ‘pastoralist’sites, as this would imply a level of agricultural specialisation (incrops or animals) not known in Britain until after World War II.Non-farming pastoralists are rare, and are typically found only inextreme environments (e.g. deserts). Instead, we interpret thesesites as having little emphasis on arable production, or as occupiedfor a short period of time only. Indeed, the concept of consumersites of cereals is one we would see as having little relevance forrural settlements in Iron Age Britain, with the possible exceptionof ‘special’ sites such as the port-of-trade at Hengistbury Head.

Bibliography

Appadurai, A. 1981. Gastropolitics in Hindu South Asia, American

Ethnologist 8, 494–511.Bakels, C. 1996. Growing grain for others or how to detect surplus

production?, Journal of European Archaeology 4, 329–336.Barrett, J. 1994. Fragments from Antiquity. Oxford: Blackwell.Boardman, S. and Jones. G. 1990. Experiments on the effects of

charring on cereal plant components, Journal of Archaeological Science

17, 1–11.Bradley, R. 1978. The Prehistoric Settlement of Britain. London: Routledge

and Kegan Paul.Campbell, G. 2000. Plant utilization: the evidence from charred plant

remains, in Cunliffe 2000, 45–59.Charles, M. and Bogaard, A. 2001. Third millennium B.C. charred

plant remains from Tell Brak, in D.Oates, J. Oates and H.McDonald (eds), Excavations at Tell Brak, Vol. 2, 301–326.Cambridge: McDonald Institute.

Cooney, G. 1997. Images of settlement and the landscape in theNeolithic, in P. Topping (ed.), Neolithic Landscapes, 23–31. Oxford:Oxbow Books.

Cunliffe, B. 1984. Danebury: an Iron Age Hillfort in Hampshire, Vol. 2.

London: Council for British Archaeology Research Report 52.Cunliffe, B. 1992. Pits, preconceptions and propitiation in the British

Iron Age, Oxford Journal of Archaeology 11, 69–83.Cunliffe, B. 2000. The Danebury Environs Programme. The Prehistory of a

Wessex Landscape, Vol. I: Introduction. London: Oxford UniversityCommittee for Archaeology Monograph 48

Dennell, R.W. 1974. Prehistoric crop processing activities, Journal of

Archaeological Science 1, 275–284.Dennell, R.W. 1976. The economic importance of plant resources

represented on archaeological sites, Journal of Archaeological Science

3, 229–247.Dietler, M. 1996. Feasts and commensal politics in the political

economy: food, power and status in prehistoric Europe, in P.Wiessner and W. Schiefenhövel (eds), Food and the Status Quest. An

Interdisciplinary Perspective, 87–125. Providence: Berghahn.Edmonds, M. 1997. Taskscape, technology and tradition, Analecta

Praehistorica Leidensia 29, 99–110.Fenton, A. 1983. Grain storage in pits: experiment and fact, in A.

O’Connor and D.V. Clarke (eds), From the Stone Age to the ‘Forty-

Five’, 567–588. Edinburgh: John Donald Publications.Gent, H. 1983. Centralized storage in later prehistoric Britain,

Proceedings of the Prehistoric Society 49, 243–267.Gransar, F. 2000. Le stockage alimentaire sur les établissements ruraux

de l’âge du Fer en France septentrionale: complémentarité desstructures et tendances évolutives, in S. Marion and G. Blancqaert(eds), Les Installations Agricoles de l’Âge du Fer en France Septentrionale,277–297. Paris: École Normale Supérieure, Études d’Histoire etd’Archéologie 6.

Grant, A. 1984. Survival or sacrifice? A critical appraisal of animalburials in the Iron Age, in J. Clutton-Brock and C. Grigson (eds),Animals and Archaeology 4. Husbandry in Europe, 221–227. Oxford:British Archaeological Reports International Series 227.

Grant, A. 1991. Economic or symbolic? Animals in ritual behaviour,in P. Garwood, D. Jennings, R. Skeates and J. Toms (eds), Sacred and

Profane, 109–114. Oxford: Oxford University Committee forArchaeology Monograph 32.

Haselgrove, C. 1999. The Iron Age, in J. Hunter and I. Ralston (eds),The Archaeology of Britain, 113–134. London: Routledge.

Haselgrove, C., Armit, I., Champion, T., Creighton, J., Gwilt, A., Hill,J.D., Hunter, F. and Woodward, A. 2001. Understanding the British

Iron Age: an Agenda for Action. Salisbury: Trust for WessexArchaeology.

Hill, J.D. 1995a. How should we understand Iron Age societies andhillforts? A contextual study from southern Britain, in J.D. Hilland C.G. Cumberpatch (eds), Different Iron Ages. Studies on the Iron

Age in Temperate Europe, 45–66. Oxford: British ArchaeologicalReports International Series 602.

Hill, J.D. 1995b. Ritual and Rubbish in the Iron Age of Wessex. Oxford:British Archaeological Reports British Series 242.

Hillman, G. 1981. Reconstructing crop husbandry practices fromcharred remains of crops, in R. Mercer (ed.), Farming Practice in

British Prehistory, 123–162. Edinburgh: Edinburgh University Press.Hillman, G. 1984a. Interpretation of archaeological plant remains:

the application of ethnographic models from Turkey, in Van Zeistand Casparie 1984, 1–41.

Hillman, G. 1984b. Traditional husbandry and processing of archaiccereals in modern times: part I, the glume wheats, Bulletin on

Sumerian Agriculture 1, 114–152.Hodgson, N., Stobbs, G.C. and Van der Veen, M. 2001. An Iron Age

settlement and remains of earlier prehistoric date beneath SouthShields Roman Fort, Tyne and Wear, Archaeological Journal 158, 62–160.

Jones, G. 1984. Interpretation of archaeological plant remains:


ethnographic models from Greece, in Van Zeist and Casparie1984, 43–61.

Jones, G. 1987. A statistical approach to the archaeologicalidentification of crop processing, Journal of Archaeological Science

14, 311–323.Jones, G. 1991. Numerical analysis in archaeobotany, in W. van Zeist,

K. Wasylikowa and K.E. Behre (eds), Progress in Old World

Palaeoethnobotany, 63–80. Rotterdam: Balkema.Jones, G. 2000. Evaluating the importance of cultivation and

collecting in Neolithic Britain, in A.S. Fairbairn (ed.), Plants in

Neolithic Britain and Beyond, 79–84. Oxford: Neolithic StudiesGroup Seminar Papers 5.

Jones, G. and Rowley-Conwy, P. forthcoming. On the importance ofcereal cultivation in the British Neolithic, in S. Colledge and J.Conolly (eds), The Origin and Spread of Domestic Plants in South-West

Asia and Europe. London: University College London Press.Jones, M. 1984a. Regional patterns in crop production, in B. Cunliffe

and D. Miles (eds), Aspects of the Iron Age in Central Southern Britain,120–125. Oxford: Oxford University Committee for ArchaeologyMonograph 2.

Jones, M. 1984b. The plant remains, in Cunliffe 1984, 483–495.Jones, M. 1985. Archaeobotany beyond subsistence reconstruction,

in G. Barker and C. Gamble (eds), Beyond Domestication in Prehistoric

Europe, 107–128. London: Academic Press.Jones, M.1996. Plant Exploitation, in T.C. Champion and J.R. Collis

(eds), The Iron Age in Britain and Ireland. Recent Trends, 29–40.Sheffield: J.R. Collis Publications.

Moffett, L., Robinson, M. A. and Straker, V. 1989. Cereals, fruits andnuts: charred plant remains from Neolithic sites in England andWales and the Neolithic economy, in A. Milles, D. Williams and N.Gardner (eds), The Beginnings of Agriculture, 243–261. Oxford:British Archaeological Reports International Series 496.

Monk, M. 2000. Seeds and soils of discontent: an environmentalarchaeological contribution to the nature of the early Neolithic,in A. Desmond, G. Johnson, M. McCarthy, J. Sheehan and E. SheeTwohig (eds), New Agendas in Irish Prehistory: Papers in Com-

memoration of Liz Anderson, 67–87. Wicklow: Wordwell.Nesbitt, M., Hillman, G., Pena-Chocarro, L., Samuel, S. and Szabo,

A.T. 1996. Checklist for recording the cultivation and uses ofhulled wheats, in S. Padulosi, K. Hammer and J. Heller (eds), Hulled

Wheats. Proceedings of the First International Workshop on Hulled Wheats,

21–22 July 1995, Castelvecchio Pascoli, Tuscany, Italy, 234–245. Rome:International Plant Genetic Resources Institute, Promoting theConservation and Use of Underutilized and Neglected Crops 4.

Palmer, C. and Jones, M. 1991. Plant resources, in Sharples 1991, 129–139.

Reynolds, P.J. 1974. Experimental Iron Age storage pits: an interimreport, Proceedings of the Prehistoric Society 40, 118–131.

Rowley-Conwy, P. 2000. Through a taphonomic glass, darkly: theimportance of cereal cultivation in prehistoric Britain, in J. Huntleyand S. Stallibrass (eds), Taphonomy and Interpretation, 43–53. Oxford:Oxbow Books.

Sharples, N. 1991. Maiden Castle: Excavations and Field Survey 1985–6.London: English Heritage Archaeological Report 19.

Stevens, C.J. 2003. An investigation of agricultural consumption andproduction: models for prehistoric and Roman Britain,Environmental Archaeology 8 (1), 61–76.

Thomas, J. 1991. Rethinking the Neolithic. Cambridge: CambridgeUniversity Press.

Van der Veen, M. 1987. The plant remains, in D.H. Heslop, The

Excavation of an Iron Age Settlement at Thorpe Thewles, Cleveland, 1980–

1982, 93–99. London: Council for British Archaeology ResearchReport 65.

Van der Veen, M. 1991. Consumption or production? Agriculture inthe Cambridgeshire Fens, in J.M. Renfrew (ed.), New Light on

Early Farming. Recent Developments in Palaeoethnobotany, 349–361.Edinburgh: Edinburgh University Press.

Van der Veen, M. 1992. Crop Husbandry Regimes. An Archaeobotanical

Study of Farming in Northern England, 1000 BC–AD 500. Sheffield:Sheffield Archaeological Monograph 3.

Van der Veen, M. 2003. When is food a luxury? World Archaeology 34(3), 405–427.

Van der Veen, M. forthcoming. Food as an instrument of socialchange: feasting in Iron Age and early Roman southern Britain, inK. Twiss (ed.), We Were What We Ate: The Archaeology of Food and

Identity. Carbondale: Southern Illinois University Center forArchaeological Investigations Occasional Paper 34.

Van Zeist, W. and Casparie, W.A. (eds) 1984. Plants and Ancient Man.Rotterdam: Balkema.

the later iron age belongs to the publishers oxbow books ... der veen and jones... · j. d. hill 3....

Documents