Cite this as: Monckton, A. 2015 Charred plant remains, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.monckton
Samples of 1st to 4th-century date were analysed representing the periods and zones of the extensive settlement. The most abundant cereal was spelt with a trace of emmer and a few bread wheat type grains, and hulled barley including six-row barley, as a second cereal. Samples interpreted as fine-sieved cereal-cleaning waste were found throughout the periods of occupation and included some samples with abundant wheat glumes and arable weeds. Some evidence of larger scale cereal cleaning was found from the late 1st-early 2nd century onwards, including waste from malting spelt from rubbish deposits. Cereal waste was widely spread on the site until the middle to later Roman periods when remains were more abundant at the periphery, with malting waste in a palaeochannel. Less evidence was recovered from the Late Roman periods. Other crops included flax/linseed and indeterminate legume, while hazel nut, sloe, wild/sour cherry, blackberry, hawthorn and elder were possibly gathered for consumption. Additional fruit remains were found in the waterlogged deposits including cherry, small plums and grape, which may have been cultivated or imported. Weeds increased in variety over the Roman periods to include, for example, white bryony and henbane, possibly as weeds of the settlement. The results were compared with other sites in the region and nearby sites in the Lower Blackwater valley.
A large number of samples were taken for the recovery of charred plant remains, which include seeds, cereal grains and chaff. These can give evidence of the crops cultivated and food consumed as well as provide evidence of activities on the site. It was intended to investigate whether continuity or change was shown in the evidence from the plant remains over the period of occupation of the site. During the 1993 excavation of Area W, a total of 286 samples were assessed but were generally unproductive and were not selected for analysis. During the 1994-5 excavation 790 bulk samples were taken and of these 665 well-dated samples were selected to be processed, which form the basis of this report.
Samples were taken at the discretion of the archaeologists and the features sampled included pits, ditches, post-holes, hearths, layers, wells and other features from all periods of the site. From the samples taken, 665 dated samples were processed by wet sieving with flotation using a 0.5mm mesh and collecting the flotation fraction (flot) on a 0.5mm sieve. The residue was then dried and separated using 2mm and 5mm sieves. All the material larger than 2mm (the coarse fraction) was sorted by eye by environmental assistants and the material smaller than 2mm (the fine fraction) was saved but not sorted. This work was supervised by Peter Boyer and carried out by Essex County Council.
During the assessment, flots from 334 of the processed samples, selected to represent the feature types and periods of the site, were examined using a x10 stereo microscope. Only forty-two were found with over twenty items of charred plant remains present. However, examination of a few fine fraction residues indicated that some material had not floated, so it was necessary to include the residues in the analysis of the selected samples. In order to provide more samples for analysis, additional material was processed, where available, from samples with any evidence of charred plant remains in the assessment. Additional samples were then selected to cover under-represented periods and areas. During analysis of the selected samples the fine fraction of the residue was refloated and the material dried and sorted from the 'reflot' using a stereo microscope and this, together with the remains from the flot and coarse fraction, were identified and counted. Around a hundred samples from each period had been assessed and after completely sorting the selected samples, the twenty most productive from each period were tabulated (Table 172 (.csv), Table 173 (.xlsx) and Table 174 (.csv)).
Identification was carried out by comparison with modern reference material in the Department of Archaeology, University of Leicester. The plant names and order follow Stace (1991) and the cereals Zohary and Hopf (1993); remains are seeds in the broad sense unless described otherwise (Table 172 (.csv), Table 173 (.xlsx) and Table 174 (.csv)). The nomenclature is the same as in the charred plant remains report. The occurrence of remains by period was then summarised in Table 175.
Although large flots were recovered from the wells and some charred deposits, showing that the processing equipment and methods used were efficient, recovery by flotation was often poor from the general soils because of the nature of the sediments. Only 17% of the remains in Period 3 were found in the flots, the rest being mainly in the fine residue. Refloating these residues proved efficient in recovering material from the selected samples and refloating more residues may have produced more samples with higher numbers of remains. Unfortunately, this was not possible so a selection of residues has been retained in the archive; these could be refloated if required for long-term storage. The material sorted by eye from the coarse fractions of some samples was relatively rich in grains, hence it may be more efficient in future on such sites to refloat the residue below 5mm (rather than 2mm) in order to save time and give more consistent recovery of remains.
In order to compare the samples with each other and with those from other sites, the proportions of chaff (the glumes and spikelet forks that consist of two glumes), cereal grains and seeds were calculated (from Table 172 (.csv), Table 173 (.xlsx) and Table 174 (.csv)). This was done because the composition of the remains can indicate stages of cereal processing (Hillman 1984). The proportions were shown on a barchart (Figure 686) that also shows the density of remains expressed as items per litre of soil. The barchart of Period 4-6 includes the charred plant remains from two samples from the palaeochannel of Period 4 (see Plant macrofossils from waterlogged contexts) instead of two less productive samples from Areas J and H. The ratios of glumes to wheat grains, barley rachis to barley grains and weed seeds to total cereal grains were also calculated (Table 172 (.csv), Table 173 (.xlsx) and Table 174 (.csv)) to assist with interpretation of the plant remains (van der Veen 1992). All the tabulated samples were included on the barcharts to show the range and character of the samples. However, those with less than about fifty items cannot be interpreted.
Cereals: Wheat chaff fragments were numerous in some of the samples (Table 172 (.csv), Table 173 (.xlsx) and Table 174 (.csv)) and the majority of identifiable glumes were of spelt (Triticum spelta) with prominent minor veins, one prominent wide-angled keel and wide bases. A few glumes were identified as emmer (Triticum dicoccum) because of their lack of prominent minor veins, the acute angles of the two keels and their small size. Glumes that were too short to distinguish these features or were of intermediate type were identified only as the glume wheats of either emmer or spelt (Triticum dicoccum/spelta). Small rachis segments were also identified only as glume wheat, but most were probably of spelt.
Cereal grains were not numerous on the site and it was thought that they may have been broken down by abrasion in the sand and gravel. However, when the residues were examined a few cereal grain fragments were found. The identifiable cereal grains were mainly of wheat (Triticum sp.). A few of these had the characteristic shape of emmer and a few were short, rounded grains classed as free-threshing wheat, possibly bread wheat (Triticum aestivum s.l.). Wheat grains consistent with the form of spelt wheat were more numerous although the majority could only be identified as either emmer or spelt. It is difficult to identify charred wheat grains with certainty because of wide intra- and inter-specific variation in grain characters today (Hillman et al. 1995), and because distortion can occur on charring (Jacomet 1989). Hence only the most characteristic grains were identified to type. Evidence of germination was found as detached cereal sprouts, which were numerous in one of the samples. Barley (Hordeum vulgare) of a hulled form was found, with the occurrence of twisted grains indicating the presence of six-row barley; a few germinated grains were found and a few chaff fragments (rachis segments). Oats (Avena sp.) were present, probably as a weed or contaminant of the cereals.
Other cultivated plants: Seeds of flax or linseed (Linum usitatissimum) represent a further crop and a fragment of a larger legume (Vicia/Pisum/Lathyrus) may represent food or fodder.
Fruit and nuts: Gathered wild food plants included hazel (Corylus avellana) nutshell fragments, sloe (Prunus spinosa), wild or sour cherry (Prunus avium/cerasus), and blackberry (Rubus fruiticosus), while hawthorn (Crataegus sp.) and elder (Sambucus nigra) berries may also have been consumed; all these plants may represent scrub or possibly hedgerows in the area. Thorns of blackthorn or hawthorn and a thorn, possibly of bramble, were also found. The fruit remains were mainly uncharred, originating from the waterlogged deposits.
Identification of the remains sorted from the coarse fractions of the samples by Peter Boyer recovered more wild fruit and nut remains as above but also some possibly introduced species typical of the Roman period. These included wild or sour cherry (Prunus avium/cerasus), small plums or bullace (Prunus sp.), a fragment of walnut (Juglans regia) from sample 2185, and a grape pip (Vitis vinifera) from sample 2501. All these may have been imported or introduced and cultivated locally.
In summary, Period 3 Well 9421 contained cherry, small plums, sloes, hazel nutshell, blackberry and elder in sample 486.1. Period 3 Well 13883 contained bullace, small plums and sloes from samples 1072, 1090 and 1092. Period 2-4 Well 6280 contained small plum, sloes, hazel and a fragment of walnut shell from samples 2184 and 2185. Period 4-5 Feature 20451 contained a few charred cereal grains and a grape pip from sample 2501. The waterlogged deposits were analysed in detail by Peter Murphy and these remains from additional bulk samples from the wells add to the evidence for consumption of fruit and nuts on the site and possible cultivation. A few flots from the bulk samples from the wells were recorded with the charred plant remains but more material is held in archive.
Wild plants: Most of the plants found were those of arable or disturbed ground, probably present as weeds of the crops. They include those typical of autumn-sown cereals such as corncockle (Agrostemma githago) and cleavers (Galium aparine). Spelt is usually considered to be an autumn-sown crop. The large grasses, including brome grass (Bromus sp.), were the most numerous in the samples and this was a common weed of cereals in the past. Weeds of disturbed land, such as is found in settlements as well as in cultivated fields, included plants such as goosefoots (Chenopodium sp.) and stitchworts (Stellaria sp.). Plants of sandy soils were represented by wild carrot (Daucus carota). A group of grassland plants is also present, including eyebright or bartsia (Euphrasia/Odontites), selfheal (Prunella vulgaris), ribwort plantain (Plantago lanceolata) and the grasses (Phleum sp. and Cynosurus cristatus). Plants of damp ground such as sedges (Carex sp.) and spike-rush (Eleocharis sp.) were also present. Some of these plants may have been brought to the site for use as fodder or flooring. However, all the above mentioned could have grown in the cultivated fields, which were much less uniform in the past. Plants used for thatching may have included bulrush or club-rush (Schoenoplectus sp.) as well as straw.
A few weeds appear from the Roman Period 3 onwards; these include the crop weeds corncockle, scentless mayweed (Tripleurospermum sp.) and particularly stinking mayweed (Anthemis cotula) identified from the charred remains from the palaeochannel. Other plants such as the mallow (Malva sylvestris), henbane (Hyoscyamus niger) and white bryony (Bryonia dioica) are possibly weeds of the settlement (Table 175). The latter is more often found in waterlogged deposits but charred specimens were recovered from this site. Roman contexts at Leicester (Monckton 1999) have also produced charred seeds of white bryony as part of the urban flora.
Types of remains in samples: In order to interpret the samples it is necessary to consider what is known about processing cereals from ethnographic studies. Spelt is a glume wheat in which the grains are held firmly in the chaff even after threshing, which only breaks the ears into segments called spikelets. This type of wheat could be stored as spikelets because the chaff protected the grains from weevil and fungal attack (Hillman 1984). Before use, the chaff could be removed by parching by heating, then pounding, followed by fine-sieving to remove the chaff (glumes) and any small weed seeds, leaving cleaned grain for use (Hillman 1981). The waste chaff can be preserved by charring if it was burnt, either as rubbish, or if it was used as fuel or kindling. Evidence for this fine sieving waste is found where the ratio of glumes to wheat grains is high because in the ear of wheat there is one glume to each grain, so an excess of glumes in a sample indicates that this waste is present (Figure 686). Similarly, a high ratio of weed seeds to cereal grains also indicates cereal-cleaning waste (van der Veen 1992). When samples are found with grain more abundant than chaff they may originate from domestic use of grain or may represent part of the product at various stages of cereal processing. Large seeds such as those of cleavers, large grasses and black bindweed were probably hand-sorted from the grain before use as they are not removed by fine sieving.
The twenty selected samples were relatively productive, giving a total of 2961 items of charred plant remains ranging from 0.7-79 items per litre, average 14.9 items per litre (Table 172) (.csv). Wheat was the main cereal, mostly spelt with occasional remains of emmer. Barley was present in small numbers in only five of the samples. Most of the samples were dominated by chaff or weed seeds, showing the presence of cereal-cleaning waste (Figure 686). All the samples analysed were from Period 2A with the exception of samples D306 and K631 from Period 2B.
Northern zone samples were relatively poor, containing a scatter of cereal-cleaning waste consisting of wheat chaff, weed seeds and a few grains. This was present in the Area D samples including the storage jar oven D406, giving little indication of the use of the oven. The sample from F1540 was more productive being dominated by weed seeds, particularly large grasses, a common contaminant of cereal grain at the time. Sample D306 contained a group of charred elder seeds that may have been food waste gathered from scrub or hedgerow. This scatter may suggest domestic activity in this area.
Central zone samples include H2138 as the most productive from the site, with abundant wheat chaff and seeds representing fine sieving waste from dehusking wheat. The pit sample H2163 differs from other samples from this period in being dominated by cereal grains, which may be waste from food preparation. The inclusion in the sample of abundant hazel nutshell and a fragment of a possibly edible legume supports this interpretation. The sample from J1109, thought to be a destruction layer of the temple area, contained only a small amount of cereal-cleaning waste, probably accumulated from the general scatter of domestic waste on the site.
Southern zone pits produced a larger number of productive samples than the other areas of the site, which contained samples dominated by wheat chaff and weed seeds. Areas K, L and N all contained productive samples with Area L most productive. This material probably represents waste from dehusking batches of wheat for consumption, with the waste being burnt and the hearth cleanings dumped in the pits. The evidence would suggest that this activity took place in this area. Ditch L791 also contained this waste so was probably also used for rubbish disposal at this time.
Peripheral area sample R2433 from a storage jar oven contained only material similar to the general scatter of cereal-cleaning waste as found in the oven in Area D above. This probably represents disuse and most other hearths and ovens examined produce few remains, suggesting they were cleaned after use.
This was the most productive period with a total of 4724 items from the twenty most productive samples, ranging from 1-88 items per litre, average 16.6 items per litre (Table 173) (.xlsx).
Northern zone samples were most productive, particularly the Area D well sample D396 which in disuse was used for rubbish disposal. The sample contained abundant spelt chaff and evidence of germinated cereal from numerous cereal sprouts; these were fairly uniform, being around 4mm in length suggesting that this was waste from malting. The germinated grain was roasted to halt germination before extracting the malt for brewing. The favoured fuel for this purpose was waste chaff (Hillman 1982) and here this is mixed with detatched sprouts similar to the charred waste found in the Period 4 palaeochannel (see below). Spelt is known to have been used for malting in the Roman period, as at Catsgore (Hillman 1982), the use of barley for malt being unknown until the Anglo-Saxon period (G. Campbell, pers. comm.). Hence this deposit suggests that brewing was being carried out nearby and also provides evidence of larger scale dehusking of wheat than is seen on the rest of the site. A second sample from a ditch D410 also has a high concentration of dehusking waste. This burnt chaff may represent spent fuel from parching spelt in a kiln or oven because waste chaff was often used as fuel to process further batches of cereals, and it has often been found in corn-dryers (van der Veen 1999). However, such features are not often found in the Early Roman period and were not found here in this phase, although this waste indicates that this activity was being carried out.
The sample from the beaten earth floor of Building 54 of Area G sample G1223 contains abundant wheat chaff with a few grains and very few seeds, mainly of large grasses. This can also be interpreted as fine sieving waste from dehusking wheat. This may be part of the scatter of waste from the site mixed with the flooring material rather than reflect the activity within the building, although the waste chaff may have been burnt in a domestic hearth within the building. It does, however, show the high concentration of this waste in the Northern zone in this period.
Central zone samples again show fairly high concentrations of remains in Area H although only fine sieving cereal-cleaning waste was found. The samples from Area J also compare with those from the previous period, being less productive than others on the site - perhaps suggesting a cleaner area.
Southern zone samples are less productive from Area K than in the previous period, all the samples producing few remains. Productive samples were found on Areas L, N, P and Q with N the most productive. Samples on Area N also differ in being richer in grains and having the highest proportion of barley from the site. A few germinated barley grains were found in sample N1674, but this is too little evidence to suggest malting. Some grassland plants such as eyebright, selfheal, crested dog's-tail grass and cat's-tails grass are present in this sample, which may suggest the presence of fodder. However, these could be weeds of the cereals, although barley was sometimes used as animal food. The pit may well contain a mixture of waste from different activities. The general high concentration of cereal waste in this zone shows that mainly domestic activity continued here.
Peripheral zone samples from features were unproductive in this phase; sample J919 contained only a fragment of hazel nutshell and a few charred weed seeds. However, the palaeochannel sampled for waterlogged plant remains contained a high concentration of charred wheat chaff (see Table 180) (.xlsx). The Period 4 samples R2413 and R2416 contain around 200 charred items per litre and over 95% of wheat glumes (Figure 686). Hence Murphy concludes that dehusking waste was dumped in the channel. The samples also contain numerous cereal sprouts, suggesting that malting was possibly carried out on the site. Samples assessed from the hinterland Area W included two grain-rich samples from Period 4 features identified as corn-dryers (2647) and (3042). The first contained only wheat grains, the latter was dominated by wheat grains with a moderate amount of chaff and very few weed seeds. Both appear to represent the remains of cereal product being parched, probably for dehusking. This evidence shows larger scale cereal processing was being carried out at the periphery of the settlement at this time.
For this report, these periods were considered together because some contexts were phased as Period 4-5. The productive samples were generally fewer, despite over 200 samples being processed, 128 being assessed and additional samples having residues refloated and sorted. A number of these samples produced nothing. The twenty most productive samples (including J919 mentioned above) produced 1345 items averaging 3.5 items per litre of soil (Table 174 (.csv) and Table 175). More productive samples of Period 4 are described above (and see Plant macrofossils from waterlogged deposits).
Northern zone samples were unproductive and lacked cereal remains, but charred weed seeds were found in the samples from Area E, for example E1811 Period 5.
Central zone samples were similar, the hearth H2105 Period 5 producing only a few charred weed seeds with a few fragments of hazel nutshell as evidence of food waste. Area J was even less productive despite assessment and sorting of more samples, suggesting a clean area with little accumulation of charred waste as shown in J919 Period 4 (see above).
Southern zone produced more samples with charred plant remains but much less than in the earlier periods. A pit from Area K sample K681 Period 5, was the most productive from this period containing evidence of fine sieving waste from dehusking spelt, but other samples from this area showed little potential. Of the other pits, K729 Period 4-5 contained a cherry stone giving evidence of the availability of this fruit. Area L produced most remains, from fills of a kiln and stoke-hole, Group 714 Period 5, which contained samples dominated by weed seeds but included cereal grains, probably from a scatter of domestic waste. A pit, L794 Period 5, contained samples with more grain present that probably represent domestic waste from food preparation. From Area N, kiln sample N1657, Period 4-5 Group 693, was dominated by weed seeds, as were those from the kiln of Area L in Period 5. These probably include weeds growing on the site as well as domestic waste. There was little to suggest that the kilns were used for cereal processing.
Peripheral zone samples from features were unproductive except for the waterlogged samples of Period 4 described above.
Southern zone samples L752 and L788, both from a well of Period 6 Group 722, contained samples with more grain present, including more barley and the most barley chaff from the site. This may represent domestic waste from food preparation, sorting contaminants from grain for consumption. Area M samples of Period 6 were less productive but contained a scatter of cereal waste (Table 174) (.csv).
Peripheral zone Hearth R2434 Period 6 contained very few remains, similar to the general scatter of domestic waste.
The main cereal is spelt wheat throughout the periods examined, with a trace of emmer and occasional bread wheat type grains as found at many other Iron Age and Roman sites in England (Greig 1991). Barley is also present as a second cereal but is poorly represented in the samples. Many of the weeds are present throughout all periods, such as brome grass, goosefoot, docks, spike-rush, sedges, clover type plants, blinks, black bindweed and cleavers. This may suggest continuity in the basic methods of cultivation; this group of weeds appears to have more similarities with the weeds of extensive cultivation in fields rather than intensive garden-type cultivation as described by van der Veen (1992). However, this could only be investigated by detailed statistical analysis which is beyond the scope of this project. Autumn sowing of the wheat is suggested by the presence of cleavers in all periods. In Period 2 wild radish and wild carrot occur, suggesting cultivation of light sandy soils for at least some of the crops, although damp ground plants are also more numerous in this period. Roots and tubers are only present with the cereals in this early period, perhaps indicating more uprooting of cereals during harvesting, with the practice being discontinued in the later periods. In Period 3 more chickweed-type weeds were found than previously and the additional weeds included corncockle and mayweeds, with common mallow and club-rush, and the continued presence of cleavers. The presence of short, as well as tall, weeds indicated that the cereal was probably reaped low on the straw.
In the middle to Late Roman periods the arable weed corncockle and stinking mayweed also occur; both are known to become more abundant in England from Roman times onwards. The latter is usually associated with clay soils (Greig 1991) but it is also known to occur on soils of medium texture where drainage is poor (Kay 1971). It occurs in small numbers in the waterlogged deposits of Period 4. Plants of damp ground, such as sedges and blinks, occur throughout the periods of the site, probably growing in those areas of the cultivated fields that were much more variable in the past. In the later periods lower numbers of arable weeds may be explained by less cereal-cleaning waste being found in general, except in Period 5 pit sample K681 Group 1147; however, the most common weeds are still present. In Period 6, samples representing only a scatter of probable domestic waste from food preparation were found. Barley was the most numerous cereal and possible evidence for cleaning the cereal was found, suggesting that it may have been used as food for people, although fodder for animals is possible.
Weed seeds dominate many of the samples from features such as kilns, where abundant charcoal provided evidence of wood being used as the fuel. The weeds in the kilns may originate from dried weedy vegetation used as kindling or incidentally included with the fuel, so probably include the weeds of the settlement mixed with a scatter of domestic waste. Additional weeds include henbane, a poisonous weed of organically polluted ground such as that near rubbish pits and latrines, perhaps present here because of the accumulation of organic material in the soil during the lengthy occupation of the site. Henbane together with mallow and bryony are thought to have belonged to the urban flora of Roman towns, which has been described as being similar to that of a traditional farmyard environment (Hall 1988). Parts of the site here may have been used as farmyards, although the conditions may not have been very different to those found in parts of the towns at this time, particularly in areas around rubbish pits.
The spelt-dominated assemblage compares with that described from other Roman sites in Eastern England, where many areas seem to have been agriculturally productive (Murphy 1997a). The main cereals of the Iron Age were the glume wheats, emmer and spelt, with a shift towards spelt as the Iron Age progressed (Murphy 1997a). Iron Age sites in the region are typified by generally low densities of remains with concentrations in some domestic contexts (e.g. Murphy 1992a). In contrast, Roman rural sites in the region tend to have more concentrated deposits of dehusking waste in features associated with crop processing, or with the waste as spent fuel dumped or accumulated in features, often at the site margin. Rubbish pits are also a common feature on many Roman sites.
Throughout the Roman periods at Elms Farm, pits were the most productive type of feature and the least productive were post-holes. The Late Iron Age/Early Roman remains in Period 2 are more abundant than those from many Late Iron Age sites in the Midlands although very productive extensive Iron Age sites are known from the Thames Valley, often with abundant cereal grain (M. Jones 1985). Some sites in the Eastern region are also productive, with a sample from Caistor St Edmund having a density of around 73 items per litre (Murphy 1992a). A Period 2 sample has a comparable density of 79 items per litre and the average of 14.9 items per litre for the selected samples is quite high, with the samples mostly dominated by chaff and seeds. The distribution, type and the amount of cereal-cleaning waste found in Period 2 is more similar to that from Roman sites, and the spread of material compares with that found at the Romano-British village at Tiddington, Warwickshire (Moffett 1986). In Period 3 the remains are generally similar but with the additional evidence of larger scale processing in one area, probably including waste from malting spelt wheat, as found at a number of Roman sites (Hillman 1982; Murphy 1999). The main change found in the samples occurs in the middle to Late Roman periods, which have little material on the settlement site but a high density of charred remains at the periphery with c. 200 items per litre from the palaeochannel. This deposit also produced evidence of malting spelt with remains similar in character to those from corn-dryers associated with malting at Stebbing Green, Essex (Murphy 1999). The requirements for malting and brewing include an ample water supply, a tank or trough for soaking grain, drains for surplus water, a malting floor within a building and ovens to parch the malt. Remains possibly of these features were found at Stebbing Green (Murphy 1999, 21), but were not found here.
It is difficult to distinguish between consumer and producer sites from the type of cereal remains alone (Smith 2001). It has been pointed out that there are a range of possibilities between types of economy and most types of occupation site consume cereals (van der Veen 1996). Concentrations of grain may therfore be found on any site where cereals are consumed. First threshing waste is rarely found charred because straw is a useful material itself. Furthermore spelt is a glume wheat which can be transported and stored in the chaff and processed in batches before consumption, hence the waste can be found on consumer and producer sites (Hillman 1984). Therefore spelt could be traded in the chaff from sites where it is produced which is possible here. Chaff is also a useful by-product, as well as being used for fuel it can be used for other purposes such as for fodder (van der Veen 1999) when it is unlikely to be preserved. Chaff may even be traded for use (Smith 2001), although perhaps this was less likely where grassland and woodland resources were available. However, deposits of abundant cleaned cereal grain found in Colchester (e.g. Murphy 1984) have led to the conclusion that cereals were processed at some rural sites to supply towns (Murphy 1990). At Duck End Farm, Stansted, samples interpreted as cereal-cleaning waste included a sample containing 580 glumes per litre, and that site was identified as a producer and large-scale processor of grain for supply to other settlements (Murphy 1990). The waste chaff was very much more abundant there than found at Elms Farm.
The samples here have more abundant remains, particularly chaff, than found at Stonea, Cambridgeshire, which was thought to be a small-scale consumer or subsistence producer (van der Veen 1996). The Roman site at Pakenham, Suffolk, produced chaff and seed-dominated samples of a similar density to those from Elms Farm, and Pakenham was thought to be a producer and processor of cereals, possibly for distribution (Murphy and Wiltshire 1989). There, pollen and waterlogged plant remains gave evidence for pasture and meadow, as well as cereal production suggesting a mixed economy.
Sites in the Blackwater Valley studied by Wiltshire and Murphy (1998) have produced waterlogged evidence for an increasingly open landscape from the Bronze Age onwards with large areas of grassland, leading to the conclusion that pastoral farming was always important in the area. Charred cereal remains present from the Neolithic period onwards increased markedly in density in the Late Iron Age and Early Roman period (7.08 and 287.3 maximum number of glumes per litre respectively, from less than one per litre in the earlier periods). This corresponds with an increase in cereal pollen as evidence of arable expansion, together with the date of the ditched field system, drainage being necessary to increase cereal production on the local soils (Murphy 1998). A mixed economy was suggested for the area; although the importance of arable may have varied between the sites, there was evidence for cereal and flax production or processing at both sites studied (Wiltshire and Murphy 1998). The maximum densities of charred cereals from Elms Farm fall into the same range (33.4 glumes per litre in Period 2 and 203 in Period 4) following the pattern of arable expansion. Grassland and possibly hay meadow are also indicated in the vicinity of Elms Farm from the plant macrofossils in the waterlogged wells. This suggest that a mixed economy of arable and pastoral farming was probably also carried out at Elms Farm. Animals would have been necessary for traction and manure for cultivation.
The charred plant remains found here compare with those from Pakenham mentioned above, which suggests that this site could also have been a producer and processor of cereals. Wheat was probably grown on lighter well-drained soils such as those on the gravels on the higher terraces to the north of the site (Atkinson and Preston 1998). However, when the size of the settlement at Elms Farm is taken into account, considerable resources would have been required for self-sufficiency. It is possible that the cereals were used to support the settlement in order to manufacture other products to trade, or to support pastoral farming, although export of grain is also possible, particularly in the later periods.
In Period 2 cereal-cleaning waste, representing processing batches of spelt on a domestic scale, was found mainly in pits. It was present in the Northern zone but more concentrated in the Southern zone, showing more intense domestic and agricultural activity there in this period. Evidence of waste, possibly from food preparation, was found in Area H although few remains were found in Area J samples. In Period 3 differences in activity are shown in the Northern zone, where evidence of larger scale dehusking of wheat and malting waste were found. In the Central zone disposal of cereal waste continued in Area H but Area J continued, as in Period 2, to have fewer remains, perhaps as a reflection of the ritual use of this area. Activity on a domestic scale continued in the Southern zone from the evidence of cereal-cleaning waste disposed of in pits, with less evidence from Area K than previously. Area N differed in having evidence for the use of barley and possibly waste fodder being burnt. The spread of wheat chaff over these areas of the site shows continuity with Period 2.
In the middle to Late Roman periods much less cereal-cleaning waste was found within the settlement, which may be explained by the processing being carried out on the periphery of the site. Abundant waste chaff was found by Murphy in the palaeochannel on Area R, along with evidence of malting, and grain-rich samples were noted from corn-dryers in Area W. This processing was probably on a large scale for the whole settlement, rather than in batches on a domestic scale, which seems to have been the case in the earlier periods. Hence the main difference in cereal-related activities on the site appears to occur between Periods 3 and 4, with the replacement of generally domestic scale processing of cereals with larger scale processing on the periphery of the site.
In all periods the main cereal was spelt wheat with a trace of emmer, and barley as a second cereal; the majority of the weeds were probably present as weeds of the crops. There is insufficient evidence to suggest that cereals were grown or processed for export because spelt can be stored and transported in the chaff, but the wheat is likely to have been grown on the better drained soils in the vicinity to supply the settlement. Other crops were represented by a legume fragment in Period 2 and flax or linseed in Periods 2 and 3. Hazelnuts and sloes were present throughout the periods of the site, with bramble, hawthorn and elder present sporadically on various areas. These could all have been consumed and may have been gathered from hedge or scrub nearby. From Period 3 onwards additional fruits were recovered from the bulk samples of the waterlogged deposits including wild or sour cherry, small plums, grape and walnut. These may have been introduced and cultivated or were possibly imported. Additional weeds appeared in the Early Roman period onwards, including crop weeds and some plants probably present as weeds of the settlement, which are also found in towns of this date. Similar conditions to the towns seem to have occurred here as a result of similar activities, such as the disposal of rubbish in pits being carried out over a long period of occupation, probably made necessary because of the large size of the settlement.
Internet Archaeology is an open access journal. Except where otherwise noted, content from this work may be used under the terms of the Creative Commons Attribution 3.0 (CC BY) Unported licence, which permits unrestricted use, distribution, and reproduction in any medium, provided that attribution to the author(s), the title of the work, the Internet Archaeology journal and the relevant URL/DOI are given.