SECTION 4.5 CHARRED PLANT MACROFOSSILS
by Vanessa Straker, Martin Jones and Ann Perry

INTRODUCTION AND METHODOLOGY

THE MIDDLE IRON AGE, PHASE 1 (3rd - 1st Century BC)

THE LATE IRON AGE AND ROMAN PERIODS (Phases 2-4)
Phase 2 c AD 25-125 Late Iron Age – early Roman
Phase 3 c AD 125 - early 4th century AD (Roman settlement complex)
Phase 3D/4A late third – early/mid 4th century AD
Phase 4 4th century AD (The late Roman Villa)

SUMMARY AND DISCUSSION
The crops
Weed communities
Sowing and harvesting
Differences in agricultural and domestic activity over time
Multiple samples from linear features – macrofossil concentrations
Crop processing and harvesting: recent and new models for interpretation

ACKNOWLEDGEMENTS

BIBLIOGRAPHY

Introduction and methodology

The sites comprise a variety of late prehistoric and Romano-British settlements, fields and trackways and straddle the first gravel terrace near the confluence of the Coln and Thames c. 3km west of Lechlade and c. 18km east of Cirencester. The gravel terrace was dissected by watercourses and marshes with tongues of gravel providing better drained sites suitable for settlement. The low-lying nature of the site provided good conditions for the preservation of both waterlogged and charred plant macrofossils as well as pollen, insects and snails. Robinson (this vol.) reports on all of these apart from the charred remains, which are the subject of this report.

The Middle Iron Age site was excavated in 1979-81 and the charred plant macrofossils result from an extensive sieving programme designed by Martin Jones and Anne Perry to extract material from the non-waterlogged deposits. Perry and Jones carried out the identifications (Perry for her 3rd year placement as part of a course at Bradford University). The samples were checked and a report written in 1984 (Jones, Perry and Straker, 1984).

The late Iron Age and Roman phases (first to 4th century AD) were excavated in 1980-4 and a similar sampling programme to that for the Middle Iron Age settlement was implemented for the later phases. The charred plant macrofossils were identified and reported on by V.Straker (Straker, 1988).

Since the 1980’s, further post excavation analysis has resulted in changes to the phasing and as a result it has proved necessary to re-make all the species identification tables and this report attempts to interpret the results of that work. No doubt if time allowed, further interpretation would be possible, but the data in its current form (Tables 1-20) allows other specialists to draw their own conclusions should they wish to. The Claydon Pike data is published in full for the first time in this report.

The very thorough sample processing records made in the early 1980’s have enabled macrofossil concentrations to be calculated for all the samples. Methodology for post excavation analysis has developed since the 1980s and if the project was carried out today the floats would first be subjected to a post excavation assessment. As Tables 1-20 show, plant macrofossil concentrations are generally low. If the project took place today, many samples would not go on to receive full analysis. A further point is that by the standards of today the samples are small, many being c. 10 litres in volume. A sampling programme would collect samples in the order of 40 litres in volume, where this is possible, to attempt to recover assemblages of a useful size (English Heritage, 2002).

Linear features (ditches and house gullies) had 2-3 random samples taken along a 15-metre length, and selective samples were taken from pits, post-holes and other non-linear features. In general the samples represented about 20% of each feature with slightly less for the larger than average features. The importance of multiple samples from linear features is commented on in the discussion section at the end of this report.

The sample summary information is presented in Table 18 and Figure 4.5.1: Phases 1-4: sample summary. Seventy four samples (832 litres soil) were analysed from Phase 1 and 40 (271 litres), 50 (465 litres) 13 (121 litres) and 19 (180 litres) from phases 2, 3, 3D/4A and 4 respectively. Originally, about 300 samples had been collected from the Late Iron Age and Roman phases. Some, with the benefit of hindsight, were regarded as dubious stratigraphically and excluded from further consideration, and a few others contained no plant remains. The remaining selection was made so that a wide a range of context types within each phase was examined, and where there were large numbers of samples, as in the case of gullies and ditches, random numbers were used to select those for analysis.

Floats were recovered on a 0.5mm mesh and residues on a 2mm mesh. Flotation was effective and the residues were sorted principally to recover small bones (M Robinson pers. comm.).

Cereal identifications are only made to species when the presence of chaff allows. Some grains with a distinctive morphology have been tentatively (using cf.) assigned to a particular taxon, but it is recognised that grain morphology is variable and not reliable for close determinations. The summary information presented in the Figures classifies all wheat grains as Triticum species, apart from Figure 4.5.3: Percentage presence of crops (excluding chaff) where some grains are assigned to T. cf. aestivum type.

Nomenclature and habitat information for weed species follows Clapham et al. 1987 as this was used in the original report and is the same reference used by Robinson (this vol.). Nomenclature for crops follows Zohary and Hopf 1994, Tables 3 and 5and the rationale for wheat identification is based on Hillman et al., 1995.

Top of Page

The Middle Iron Age, Phase 1 (3rd – 1st century BC) (Table 1 a-d, Figures 1-8)

The Warrens Field settlement was located on 3 gravels islands and consisted of round houses, typically sited inside a circular drainage gully with an entrance to the south east. The plant macrofossils were recovered from gullies (mainly) and some ditches and clay-lined pits excavated in trenches 6 (Island 1), 8 and 12 (Island 2) and 14 (Island 3).

The assemblages are very small and are dominated by weed seeds and chaff rather than cereal grain. Plant macrofossil concentrations (number of items per litre of soil) are shown in Tables 1 and 17, 19 and Figure 4.5.2: Plant macrofossil concentrations . They are low, with a mean of 1.5 for gullies and 1.3 for ditches. As in all phases, pits have a slightly higher density with a mean of 3.8 for the Middle Iron Age, which is still very low. These sorts of figures are difficult to interpret but relate to the nature of the activities taking place in the vicinity. They can also show the patchy nature of deposition in linear features as demonstrated by gully context 21a with over 6 items per litre compared with 0.6 for context 21c.

The assemblages tend to be dominated by crop processing waste (chaff and weed seeds). The information from grain and chaff shows that of the crops, wheat was more commonly present in contexts than barley (see Table 20 for percentage presence of different components). The wheat chaff allows identification of hulled wheats suggesting that spelt wheat predominated, though a single rachis internode of free threshing wheat, probably hexaploid Triticum aestivum sl. (bread wheat) was also identified. This is the only example of free threshing wheat rachis node from a Middle Iron Age context in the south of England (Campbell and Straker, 2003). It is likely that emmer wheat and bread wheat were minor components of the wheat crop. The barley was not well preserved, rarely allowing distinction of the (more likely) hulled from naked form, but the absence of twisted grains suggests that the 2-row form with 2 as opposed to 3 grains maturing at each rachis node predominated.

Most of the charred weed seeds are associated with arable land or grassland, though some will live in a more varied range of habitats. Numbers of weed seeds are generally low though a fairly wide range of taxa was identified. Members of the Caryophyllaceae (e.g. Stellaria media agg., chickweed) were among the most numerous and are associated with arable or disturbed conditions. Spike rush (Eleocharis spp.), was also common; this plant is associated with soils experiencing at least spring waterlogging (Walters, 1949) and pond margins. Scentless mayweed (Tripleurospermum maritimum) is an arable weed, but prefers lighter, drier cultivated soils and was found more commonly at Ashville (Jones, 1978). Sheep's sorrel (Rumex acetosella agg.) has a competitive advantage over many other species in acid soil conditions but also grows on the largely calcareous gravels of the Thames Valley today (M.Robinson, pers. comm.).

Cleavers (Galium aparine) was quite common at Claydon Pike, as it also was at Ashville, and is regarded as a weed of winter-sown cereals. Taxa associated with open grassy habitats are numerous and include Clover (Trifolium sp.), Vetches (Vicia/Lathyrus), Eyebright or Bartsia (Euphrasia/Odontites) as well as several different grass taxa.

Top of Page

The Late Iron Age and Roman periods (Phases 2-4) (Tables 2 to 17)

A range of contexts (gullies, ditches, pits, wells, layers, ovens and hearths) was examined from the three platforms, rectangular temenos and circular shrine excavated between 1980 and 1984. The dating available for the late Iron Age and Roman periods allows the main phases to be subdivided, as shown in the Tables. Brief summaries of the results for each phase are presented below. A general summary of the crops and weeds is presented in the final discussion section.

Top of Page

Phase 2 c AD 25-125 Late Iron Age – early Roman (Tables 2, 3, 4, 5 and 6)

40 samples were analysed from 10 circular gullies, 17 ditches, 3 pits and 1 layer which comprised what was thought to be a nucleated settlement, though no directly structural remains were identified from it. This may be reflected in the very small assemblages and low concentration of macrofossils (Figure 4.5.2: Plant macrofossil concentrations ), with only the assemblage from gully 1645a of more than 50 items. Cultivated plants include emmer (Triticum dicoccum), spelt (Triticum spelta) and breadwheat (Triticum aestivum sl) and the wild species are all from open and disturbed habitats. They include Field madder (Sherardia arvensis), sheep's sorrel (Rumex acetosella agg.) and Brome (Bromus, probably B. mollis or secalinus) clover (Trifolium sp.) and self heal (Prunella vulgaris). However, the range of wild plants is smaller than in both the Middle Iron Age and later in the Roman period. The impression, with the exception of gully 1645 where grain predominates, is of occasional discarding of waste from the later stages of crop processing where small chaff such as glume bases and weed seeds predominate.

Top of Page

Phase 3 c AD 125 - early 4th century AD (Roman settlement complex) Tables 7, 8, 9, 10, 11, 12 and 13)

In this phase the settlement was reorganised with large rectilinear ditched enclosures, 2 aisled buildings, fence lines and a cobbled trackway. Fifty samples were analysed from 3 gullies, 15 ditches, 5 pits, 4 wells, 2 layers and 4 ovens, distributed over the sub phases as shown in Table 18.

The range of farming and domestic activities taking place at the settlement throughout the early second to early 4th centuries was probably varied and is reflected by the heterogenous distribution of grain, chaff and weeds in the samples and the much greater concentration of plant macrofossils than in earlier phases (Table 19; Figures 4.5.2: Plant macrofossil concentrations and 4.5.7: Percentage presence of grassland plants). The 50 samples from phase 3 compare with 40 from phase 2 and 19 from phase 4, with more samples in phase 3 from a range of features other than ditches. Some individual features (e.g. oven 2103, see below) are very rich, but the mean macrofossil concentration is also greater. The figure for phase 3 pits, for example, at 117.5 items /litre is noticeably greater than for phases 2 and 4 (1.7 and 10.2 respectively).

The range of crops is similar to earlier phases but the very rare occurrences of beans (Vicia faba var minor) and flax (Linum usitatissumum) are found in phases 3A and 3D respectively. Neither of these crops is as likely to become accidentally charred as the hulled wheats, which benefit from heat to render the chaff brittle and facilitate dehusking.

Oven 2103 (not allocated to a subphase within 3) is very rich in chaff, notably emmer and spelt glume bases and spikelet forks which comprise over 80% of the large assemblage. These are the remnants of the oven fuel which was probably also disposed of in ditches, pits etc. Many of the other features from phase 3 are also chaff-rich and some could have originated from oven cleanings, rather than the direct burning of crop cleanings. However, as well as chaff, the oven assemblage also contains small weed seeds including over 100 seeds of stinking chamomile (Anthemis cotula) and occasional ribwort plantain (Plantago lanceolata) and sedges (Carex spp.). This is most likely to represent processing waste from a late stage in cleaning to free the grain from the spikelets and remove weed seeds. The damp ground plants could be from damp field margins or burnt animal fodder, but also from the heart of the arable fields, which often suffered from poor drainage in the Iron Age and Roman periods (Jones 1988).

Another large assemblage from pit 1398 is composed mainly of wheat grain, probably spelt, as suggested by the chaff, with some indeterminate wheat and barley grain and chaff. Among the accompanying list of taxa is a wide range of wild plants, and a relatively high number of clover seeds, perhaps indicative of animal fodder or hay as a source.

Top of Page

Phase 3D/4A late third – early/mid 4th century AD (Table 14)

Fifteen samples from 1 gully, 4 ditches and 6 pits fall within this phase. Macrofossil concentrations vary as shown in Figure 4.5.2: Plant macrofossil concentrations , ranging from 0.6 for ditch 1201k to 50.3 items per litre for pit 2385. Composition is also variable with over 80% chaff in ditches 667f and pit 1250, 55% in ditch section 1201j and over 70 % weeds in pit 1219 which suggests that a range of different crop processing activities was probably taking place and / or that crop processing waste is mixed with burnt plant debris from other sources, such as animal fodder and roofing or flooring material. Figure 4.5.4: Percentage presence of chaff shows a greater percentage presence of chaff than for other phases, but unusually, these samples are from ditches and pits rather than hearths or ovens where chaff is often found having been used as tinder.

Pit fill context 2385 was rich in plant macrofossils. Wheat and barley grain was mixed with hulled wheat (mainly spelt where identification was possible) glume bases and spikelet forks, but there were also small numbers of a range of weeds or arable land and grassland such as corn cockle, knotgrass, black bindweed, stinking chamomile and buttercups, clover and tormentil. There were also some sedges and spike-rushes characteristic of damp ground. This is a good example of a rubbish pit where waste from processing crops to release grain for consumption was dumped probably after being burnt on a domestic hearth.

Top of Page

Phase 4 4th century AD (The late Roman Villa) (Tables 15, 16 and 17)

The villa estate consisted of a masonry-footed villa and associated buildings, a small cemetery and a circular shrine. By the later 4th century (phase 4C) the villa was enclosed by the first of two ditched enclosures. Seventeen samples were analysed from 5 ditches, 2 pits, 1 layer, 1 oven and 1 hearth. The breakdown by sub phase is shown in Table 18.

Spelt and emmer wheats are still consumed, as was barley. There are internode fragments of neither barley nor free-threshing wheat. Bread type wheat may still have been a minor component of the cereal fields but there is certainly no evidence from Claydon Pike to show that it was increasing in importance.

Top of Page

Summary and discussion

The crops

Wheat and barley are identified from all phases on the basis of grains and chaff with the exception of barley rachis internodes, which were not recorded in phase 4. The figures for percentage presence (based on the percentage of samples in each phase in which the taxon occurs) (shown in Table 20 and Figures 4.5.3: Percentage presence of crops (excluding chaff) and 4.5.4: Percentage presence of chaff), demonstrate that as for the Middle Iron Age, wheat macrofossils occur more commonly than barley.

Triticum sp. Wheat
On the basis of percentage presence in samples as outlined above, spelt wheat (Triticum spelta) is the species most commonly found in all the phases. This hulled wheat is largely identified by its distinctive chaff, principally glume bases. Some wheat grains have been tentatively assigned to species, but grain morphology is notoriously variable between as well as within species. Emmer chaff (T. dicoccum) is present as a minor component in phases 2C, 3 and 3/4. Although present in the Thames Valley, the archaeological evidence shows that emmer has completely disappeared from the Hampshire chalklands by the beginning of the Iron Age, which demonstrates the local distinctiveness of farming practices. In other parts of southern England however, emmer was probably a more important component of Roman farming (Campbell and Straker, 2003).

Some rounded caryopses typical of a free threshing bread wheat type were noted occasionally in all phases but the presence of free threshing wheat is only confirmed by the presence of a small number of tough rachis internodes (probably hexaploid, bread wheat type) in phases 2C and 3.

The significant presence of spelt, with emmer and smaller amounts of breadwheat is typical of Roman assemblages. There may be an exception at Barton Court Farm (Jones 1984) but some of the material may have been wrongly assigned (Campbell and Straker 2003). However, free threshing cereals are likely to be under represented in the charred macrofossil record in relation to hulled wheats as they do not require exposure to heat to facilitate dehusking and can be removed at an early stage of crop processing (Hillman, 1984).

Hordeum sp. Barley
Most of the barley grains (as with the wheat) are poorly preserved and cannot be identified very closely. Some can be identified as hulled barley. Both twisted and straight grains were noted, but the twisted grains were uncommon and only present in phases 2 and 3. In six-row barley, the grains in the lateral florets have a twisted ventral furrow whereas the grains in the middle floret have a straight furrow. The ratio of twisted to straight grains for six-row barley is 1.5:1 or less, as it is quite usual for some lateral florets to fail to develop. In two-row barley, twisted (asymmetrical) grains do not generally occur. However, as so many of the grains are indeterminate, it is only possible to say that hulled six-row barley was consumed and hulled two-row barley may also have been. No naked barley was observed.

Avena sp. Oats
Oats are very scarce in all phases and it is not possible to tell from the grain alone whether a cultivated or wild form is present. The only floret base is from Phase 3 and identifiable to a fatua or ludoviciana wild form. It is likely that the oats are simply a minor weed component which, like the large-seeded Bromus (brome), are difficult to separate from the grain during winnowing and sieving in post-harvest crop processing. This means they often become accidentally charred with the prime product. It has been pointed out that the large seeded grasses may well have been intentionally used as a famine food, but Jones (1988, 90) notes that sometime after the Iron Age, wild grasses become much less numerous in charred macrofossil assemblages.

Other crops

Flax is known to have been an important crop in the Thames valley and it grows well rotated with other crops on wet ground. Its seeds have been found at Barton Court Farm and Farmoor (Jones 1984, and Lambrick and Robinson 1979). Only one carbonised seed was found at Claydon Pike, from an oven (Phase 3B 2113/3) but it is more plentiful in the waterlogged assemblage (Robinson, this volume). It is a useful crop for as well as providing bast fibres for textiles, its seeds are oil rich and can be used for the oil or as cattle feed.

Legumes are usually under represented in charred macrofossil assemblages as they, like free threshing cereals, do not require exposure to heat after harvest. Legumes identified as vetch or pea (Vicia/Pisum) were found in phases 2,3 and 4 but unfortunately the hila were not well preserved enough for precise identification to be made. Legumes identified as vetch or pea were also very scarce but noted in phases 2 and 3. The only example of a Celtic bean (Vicia faba var. minor) was found in ditch sample 547E from phase 3A. It is very likely that pulses were a much more important component of the diet than it would appear from the archaeological record alone and may have been an important source of plant protein. Beans, peas and other legumes are nitrogen fixing and if grown as part of a rotation, help to maintain soil fertility. Roman peas were identified from the Warwickshire gravel site at Tiddington (Moffett 1986).

Top of Page

Weed communities

The ecological analysis of weeds can either proceed through a single species (indicator) or plant community (phytosociological) approach. The latter relies largely upon observations of modern plant communities. Some reservation in the use of this approach is advisable for two reasons. First, it may be that the ecological preferences and tolerances of weeds have changed somewhat over the last few thousand years. Second, and more importantly, the actual soil environment of arable agriculture has radically changed with each change in cultivation technology, and such innovations as the mouldboard plough transformed the selection pressure on arable weeds. Indeed that was the central purpose of such innovations (Jones 1988).

In relation to the first of these points, the oceanic climate of Britain reduces the clarity of distinctions observable on the continent, between, for instance, weed communities of spring and winter-sown cereals. This said, many of the arable weeds from Claydon Pike (shown by phase in Figure 4.5.5: Percentage presence of plants of disturbed ground and arable), do correspond with the taxa included in the Class Stellarietea, described by Silverside (1977) to which most present-day arable communities are referable. Some species, such as Lapsana communis, Anthemis cotula, Atriplex patula and Odontites verna are more particularly characteristic of cereals (the order Centauretalia cyani). In recent years, a series of volumes describing British Vegetation Communities has been published. Volume 5 (Rodwell, 2000) describes the vegetation communities of open habitats including arable land and waste ground. It is clear from this that aspects of the Claydon Pike weed flora are reminiscent of several of the communities described (e.g. OV3, 0V7, 0V9, OV10, OV13), but close comparisons are not possible. This may be largely due to the increasing use of herbicides on open ground vegetation in the 20th century, especially in arable fields, which is a major contrast with former farming methods.

In relation to the second of these points, certain taxa appearing first in phase 3 have been associated by one of us with the transition from shallow ard cultivation to deep ploughing (Jones 1988). Stinking chamomile (Anthemis cotula), a cornfield weed typical of heavy soils, makes its first appearance in Phase 3 and another introduction, corn cockle (Agrostemma githago) is first seen in phase 3D/4A. More widely in Britain, these two species together with Centaurea cyanus/nigra, become visible in the late Roman period at a time when metal items associated with heavier ploughs appear in the archaeological record (Jones 1988). It is interesting to note that three taxa conversely associated with shallow ard cultivation are at their most frequent, and their most abundant, in the earliest phase of the site, phase 1.

The range of arable weeds is most diverse in phase 3, as shown in Figure 4.5.5: Percentage presence of plants of disturbed ground and arable, but there are some notable differences in phase 4 with the presence of corn gromwell (Lithospermum arvense) and particular abundance of field madder (Sherardia arvensis) and stinking chamomile. The last two plants prefer slightly different growing conditions, with field madder typical of dry soils such as sands and gravels whereas stinking chamomile is most common on heavy soils. This suggests that the crops processed at the late Roman villa were grown on a wider range of soil types than those exploited by the earlier communities.

A single seed of ground elder (Aegopodium podograria) was found in pit 1398 in phase 3. Although now regarded as a pernicious garden weed, this species is long thought to have been a Roman introduction and can be used as a pot herb (Clapham et al. 1989, 283; Godwin, 1975, 225). The Claydon Pike example appears to be the sole archaeological record for the period.

Clovers (Trifolium spp.), black medick (Medicago cf. lupulina), buttercups (Ranunculus acris/repens or bulbosus) and yellow rattle Rhinanthus sp.) increase from phase 3 onwards (Figure 4.5.7: Percentage presence of grassland plants). All of these taxa (and others) are recorded by Mark Robinson in the waterlogged assemblages. From his waterlogged data, he infers that grassland was managed and included the production of sedge hay. The plants of damp soils such as the sedges may have formed part of a burnt grassland assemblage, however the wide range of taxa that survives in the waterlogged assemblages is not reflected by the charred plant remains although the percentage presence analysis does show a wider range of weed species in Phase 3 than in earlier and later phases. The integration of the different strands of evidence in the waterlogged assemblages led Robinson (this vol) to the conclusion that large quantities of arable products must have been brought to the site, as they did not appear to have been grown locally. Certainly the evidence for the use of arable crops is much better in phase 3 than in the preceding periods or the 4th century villa complex. As large areas were excavated, the increase in evidence for phase 3 should be a reasonable reflection of the activities taking place rather than being biased through sampling of features closer to the main areas of activity in one phase compared with another.

Plants of damp ground such as sedges and spike-rushes in particular are rather more commonly found in phases 3D/4A and 4 than elsewhere (Figure 4.5.6: Percentage presence of damp ground plants). As suggested above, these may relate to a variety of sources, and as Robinson (this volume) has shown, damp grassland was a major feature of the local environment. However, the association of seeds of plants today associated with damp and wet ground, with crop and arable weed communities has been noted for a number of Iron Age and Roman sites in the Thames Valley (e.g. Ashville, Jones 1978, and Barton Court Farm, Jones 1984) as well as being noted among cereals on the continent (Groenman van Waateringe and Pals, 1983). It has been suggested (e.g. Jones 1988) that the frequent finding of the seeds of these types of plants charred with arable weeds seeds provides evidence that fields were being cultivated in areas that became increasingly subjected to flooding. This may be the case for the fields of the villa estate, alongside the possibility of other taphonomic routes from neighbouring grassland.

Top of Page

Sowing and harvesting

As far as time of sowing can be ascertained, the presence of species such as Galium aparine (cleavers), an autumn germinating plant, suggests that at least some autumn sowing of crops was taking place from the Middle Iron Age onwards. Black bindweed (Fallopia convolvulus), chickweed and stitchwort (Cerastium sp. and Stellaria media agg.) are typical indicators of spring sowing. Many of the weeds are low-growing species suggesting that reaping was low on the straw. Examples include eyebright or red bartsia (Euphrasia sp. / Odontites verna), and members of the Caryophyllaceae family such as chickweed and stitchwort. The Caryophyllaceae are particularly frequent in the Middle Iron Age samples, as shown in Figure 4.5.5: Percentage presence of plants of disturbed ground and arable. In contrast, black bindweed seeds would have been harvested with the ears or ears and straw.

Top of Page

Differences in agricultural and domestic activity over time

Pie diagrams divided to show the proportions of grain, chaff and weeds were constructed for all assemblages with more than 50 items and were matched by eye to their locations on the respective site plans to see any parts of the site may have specialised in particular activities at any time. The diagrams are deposited with the site archive. No particular spatial patterns could be seen in either concentration or composition with the exception of oven 2103 in phase 3, which has been discussed. Many samples probably result from the mixing of waste from different activities or storage practices leading to charring and as Stevens (2003) suggests, this could have taken place over months or years.

Top of Page

Multiple samples from linear features – macrofossil concentrations

The figures for macrofossil concentration given in Tables 1 to 17 for the multiple samples from different l5 metre stretches within the linear features, demonstrate well the patchy nature of deposition in ditches and gullies. This was more noticeable for the Roman than earlier periods. These differences may have been more noticeable had larger samples been taken; the results obtained here cannot be used to infer differences in farming or domestic practices in different areas. However, the data serves as a reminder that sampling strategies for sites with large numbers of linear features should include multiple samples, each kept and processed separately, in order to ensure that assemblages are as representative of what was originally present as possible. This is summarised well by Orton (2000, chapter 6) who noted the lack of homogeneity within features and examined the approaches to sampling by several specialists. He comments that in order for calculations made should be valid, several samples from the same feature are preferable to one large one from a single location.

Top of Page

Crop processing and harvesting: recent and new models for interpretation

Since the excavations at Claydon Pike in the late 1970’s and early 1980’s, there has been much discussion on the use of charred plant macrofossil assemblages to try to identify the sorts of post-harvest crop processing activities that have taken place in the past and establish the existence of specialist farming settlements of an arable or pastoral nature.

The first model to be put forward was by Hillman (1981, 1984), and a further model also based on ethnographic evidence but using the characteristics of weed seeds, was published by G.Jones (1984). In 1985, M. Jones published a model for the interaction of Iron Age communities in this part of the Thames valley based on the composition of archaeobotanical assemblages. He suggested that sites dominated by fine chaff and weed seeds (presumed to be derived from final cleaning of partially cleaned spikelets), rather than large quantities of grain, could represent ‘consumer’ economies of predominantly pastoral groups. He suggested that the Middle Iron Age assemblage from Claydon Pike was an example of this, as was Smith’s Field. Both sites were located at the junction of the river floodplain and first gravel terraces. Subsequently (Jones 1993), the settlement at Mingies Ditch was also interpreted as a ‘consumer’ site. Settlements such as Ashville and Mount Farm, located on the second (higher) terrace more suitable for arable cultivation, were grain rich and could be viewed as ‘producer’ settlements.

Van der Veen (1991, 1992) pointed out that the approaches of Hillman and M. Jones were in conflict and went some way to try to resolve the differences. She concluded that the ‘producer and consumer’ model was rather too simplistic and suggested at least four types of settlement. These were those engaging in subsistence production, production for a surplus, small consumer sites and large urban complexes. Van der Veen suggested that there would be a continuum between these types of site. Van der Veen’s surplus production, typified by large, grain rich assemblages, would look similar to assemblages thought to represent a producer settlement using M.Jones’s interpretation (Van der Veen 1991, Fig 27.2 and 355, 357). Assemblages from small consumer sites however, could look very like production on a small scale.

A recent paper by Stevens (2003) summarises the discussion in some detail and the reader is referred to it for further information. Stevens (2003) re-worked some of M. Jones original data from Ashville, Mount Farm, Mingies Ditch and the Middle Iron Age samples from Claydon Pike. He also used data from Gravelly Guy (Moffett 1989) and Yarnton. He used a different approach to interpreting charred assemblages and also suggested that social organisation including the availability of post harvest and pre storage labour, could have played a part in different storage practices, which in turn affect the composition of the charred assemblages. Stevens used two different methods to analyse the data from the sites. One of these (the percentage of large weed seeds from all classified seeds plotted against the percentage of weed seeds to grain, Stevens, 2003, Fig 6) produced clear variations between the sites included in his study. The interpretation he proposed was that waste from a number of crop processing stages, including processing of spikelets, was present at Claydon Pike, with a similar situation for Abingdon, Yarnton, Gravelly Guy and Mingies Ditch. Stevens comments that different storage practices could also produce the observed pattern. In contrast, the assemblages from Ashville, Mount Farm and Danebury, were dominated by grain as opposed to weeds and of the weeds, large seeds in preference to small and intermediate-sized ones. This fitted the expected pattern from his model for waste produced from the processing of semi-cleaned spikelets.

The method of analysis described above has been applied to the data from each of the different phases at Claydon Pike, Stevens (2003) having only applied it to the Middle Iron Age data. The number of large seeds is expressed as a percentage of all the seeds classified as large to small, plotted against the number of all weed seeds expressed as a percentage of weed seeds plus grain. The results are presented in Figure 4.5.8: Weeds as % of weeds and grain plotted against large weed seeds as % all classified seeds . Stevens’s methodology was followed in terms of using samples with the minimum required diagnostic components (15) per assemblage. Time has not permitted development of the technique to include other large or small weed taxa which would be diagnostic for Claydon Pike. The information presented in Figure 4.5.8 is therefore limited in this respect and for most phases there are very few ‘qualifying’ assemblages. Further analysis of the data would add to the number of assemblages that could be included and probably also identify further ‘large’ weed seeds which would change the emphasis somewhat. Using Stevens interpretation, in very broad terms, the analysis suggests that over all the phases, more of the waste is likely to result from crops stored ‘uncleaned’ i.e. as sheaves or partially threshed ears, rather than from clean grain or semi cleaned spikelets and that waste from a number of processing stages may be represented. There is a little more indication from phase 3 of clean grain or semi-cleaned spikelets.

A potential problem with this (and other models) is that, inevitably, plants which may or may not have been growing as arable weeds (such as some damp ground and grassland taxa) will be included in the analysis and all or some of them may have originated from other activities and circumstances, such as roofing or flooring debris. A further problem with all the models is that they cannot take into account that chaff and / or stable waste such as straw could have been present on the site from the outset, but used as fuel or simply burnt deliberately to dispose of it. With Stevens’s model, if this had taken place, interpretation of crops stored as sheaves or partially threshed ears could actually be the result of deliberate burning of chaff as fuel. These qualifications must be borne in mind when exploring Stevens’ model.

One approach to the use of models is to use them to develop a hypothesis about possible status and activities, which can then be considered in relation to all the other forms of bioarchaeological data from a site. This emphasises the importance of a multidisciplinary approach to bioarchaeological studies. At Claydon Pike, we are fortunate to have waterlogged preservation of plant macrofossils and insects, as well as animal bones. The importance of management of wet grassland at Claydon Pike is well demonstrated by these studies. It still remains the case as Jones (1985) originally postulated for the Middle Iron Age, that pastoralism, rather than arable is likely to have been the main emphasis of local agriculture with semi-cleaned crops brought in, even if only from a short distance away, necessitating some processing in the form of dehusking and cleaning to remove chaff and smaller weed seeds. The crops might well have been in sheaf form at times, or straw may have been supplied separately as required. This does not exclude the possibility that small-scale arable plots were also cultivated on occasion at the settlements, with additional supplies brought in from outside suppliers when needed. Some of the damp ground plants so often noted above as possibly originating from animal fodder or even roofing waste, could have grown on the edges of small plots at Claydon Pike.

Top of Page

Acknowledgements

We are very grateful to Ann Perry for her work on the Middle Iron Age macrofossils, and Mark Robinson for assistance with the identification of some seeds and Gill Campbell and Mark Robinson for helpful comments on an earlier version of this report. Sheila kindly reformatted the tables.

Top of Page

Bibliography

Campbell, G and Straker, V 2003 Prehistoric crop husbandry and plant use in Southern England: development and regionality. In (ed K.Robson Brown) Archaeological Sciences 99. BAR 1111 Int. Series, 14-30. Oxford.

Clapham, A.J. 1995 Plant remains. In I. Barnes, W.A. Boisnier, R.M.J. Cleal, A.P. Fitzpatrick and M.R. Roberts, 1995 Early Settlement in Berkshire: Mesolithic-Roman Occupation Sites in the Thames and Kennet Valleys , Wessex Arch. Report 6, 35-45.

Clapham, T G, Tutin, T G and Moore, D M 1987 Flora of the British Isles. 3rd edn. Cambridge: University Press.

English Heritage 2002 Guidelines for Environmental Archaeology. Centre for Archaeology Guidelines 2002/1.

Godwin, H. 1975 History of the British Flora. Cambridge: Cambridge University Press.

Greig, J.R.A.1991 The botanical remains. In S. Needham and D. Longley Excavation and salvage at Runnymede Bridge, Berkshire, 1978, 234-6 London: British Museum.

Groenman-van Waateringe, W. and Pals, J.-P. 1983 The Assendelver Polders Project: integrated ecological research, In (ed M.K. Jones) Integrating the subsistence economy, Oxford: BAR 181, 135-182.

Hillman, G C 1981 Reconstructing crop husbandry practices from the charred remains of crops, In (ed R.Mercer) Farming Practice in British Prehistory, Edinburgh: University Press, 123-162.

Hillman, GC 1984 Interpretation of archaeological plant remains: The application of ethnographic models from Turkey, In (eds W.van Zeist and W.A.Casparie) Plants and Ancient Man, Rotterdam: Balkema, 1-42.

Hillman, G.C., Mason, S., de Moulins, D. and Nesbitt, M. 1995 Identification of the archaeological remains of wheat: the 1992 London workshop. Circaea, 12.2, 195-210.

Jones, G E M 1984 Interpretation of plant remains: ethnographic models from Greece . In van Zeist W and Casparie W A (eds) , Plants and Ancient Man, Studies in palaeoethnobotany. Rotterdam: AA Balkema., 43-61.

Jones, M K 1978 The plant remains. In The excavation of an iron age settlement, bronze age ring ditches and Roman features at Ashville Trading Estate, Abingdon (Oxon.) 1974-1976, (ed. M. Parrington) CBA Research Report 28, London, 930-110.

Jones, M.K. 1984 Towards the model of a villa estate and The carbonised plant remains, In (ed D.Miles) Archaeology at Barton Court Farm. Abingdon. Oxon, London: CBA, Research Report 50, 38-42 and Microfiche Chapter V11.

Jones, M.K. 1988 The arable field: a botanical battleground, In: (ed M.Jones) Archaeology and the Flora of the British Isles, Oxford: University Committee for Archaeology 14, BSBI Conference Report 19.

Jones, M.K. 1993 The carbonised plant remains. In T.G. Allen and M.A. Robinson The Prehistoric and Iron Age Enclosed Settlement at Mingies Ditch; Hardwick-with-Yelford, Oxon (Thames Valley Landscapes: the Windrush Valley 2), 120-3. Oxford: Oxford University Committee for Archaeology for the Oxford Archaeology Unit.

Jones, M.K., Perry, A. and Straker, V. 1984 Charred cereals from Claydon Pike, Gloucestershire, 979-81: The Iron Age phases. Unpublished report.

Lambrick, G H and Robinson, M A 1979 Iron Age and Roman riverside settlements at Farmoor (Oxfordshire). London: CBA Research Report 32.

Moffett, L. 1986 Crops and crop processing in a Romano-British village at Tiddington: The evidence from the charred plant remains. Ancient Monuments Laboratory Report 15/86.

Moffett, L. 1989 The evidence for crop processing products from the Iron Age and Romano-British periods at Gravelly Guy and some earlier prehistoric plant remains. Unpublished Ancient Monuments Laboratory Report 46/89.

Orton C. 2000 Sampling in Archaeology. Cambridge: Cambridge University Press.

Rodwell, J.S. 2000 (ed) British Plant Communities. Maritime communities and vegetation of open habitats. Volume 5. Cambridge: Cambridge University Press.

Silverside, A.J. 1977 A Phytosociological survey of British Arable-weed and related communities. University of Durham, Ph.D. Thesis.

Stevens, C J 2003 Agricultural consumption and production models for prehistoric and Roman Britain. Environmental Archaeology 8, 1, 61-76.

Straker, V. 1988 Claydon Pike, Gloucestershire: carbonised cereals from the late Iron Age to Roman periods. Unpublished report.

van der Veen, M. 1991 Consumption or production: agriculture in the Cambridgeshire Fens. In (ed J. Renfrew) New Light on Early Farming, Edinburgh: Edinburgh University Press, 49-61.

van der Veen, M. 1992 Crop husbandry regimes: an archaeobotanical study of farming in Northern England 1000BC – AD 500. Sheffield Archaeological Monographs 3. Sheffield: Department of Archaeology and Prehistory, J.Collis Publications.

Walters, S M 1949 Biological Flora of the British Isles. Eleocharis R. Br. Journal of Ecology 37, 192-206.

Zohary, D. and Hopf, M. 1994 Domestication of Plants in the Old World. 2nd edn. Oxford: Clarendon Press.