Environmental Report

Palaeobotanical Evidence

J.P. Huntley

Following the recommendations put forward prior to excavation (Huntley, June 1990) bulk samples were taken from occupation layers, pit and drain fills, and processed on-site. Fifteen 60-litre such samples were processed with a further two from a drain fill left 'whole' for parasite and insect work (these were sent to the Environmental Archaeology Unit at York for analysis). In addition further material was taken and kept 'whole' from the bulk sampled contexts. As requested, the flots were processed to 500µ and the residues to 1mm. Together with two whole samples these were sent to the Durham Bio. Lab. for analysis. They were from well stratified contexts and can be dated to the late-3rd - early-4th centuries.

The flots were generally small, less than 100ml, and still extremely dirty. They were consequently thoroughly rinsed over 500µ, which often reduced their bulk by nearly a half, and dried. The residues were likewise extemely dirty due to the appalling weather conditions on-site and had to be further washed and dried before sorting. This had the lead-on question as to flot 'quality' - what still remained in the residues but should have floated and more importantly, what had washed through the residue mesh and been lost?

The flots were sorted under a stereoscopic microscope at magnifications of up to 40X; seeds were identified by comparison with modern reference material held in the Durham Bio. Lab.; nomenclature follows Clapham, Tutin and Moore (1987).

The original evaluation for the site asked a wide variety of questions. Following excavation it became apparent that some of these were no longer valid and, given the processing problems and quality of material, the very basic questions of 'What cereals were present'? and 'What evidence is there for local vegetation'? are about all that should be asked.

Table 4 presents the botanical data with the samples arranged in phase order and the taxa within broad ecological category. So few remains were present that it is not possible to offer interpretations of any specific feature. The carbonised material was generally not well preserved which was particularly disappointing given the lack of intrusive material for the site as a whole. Other than pit 252 all of the samples contained only carbonised plant remains. These were mainly from cereal grains although many could not be identified. Of those that could, wheat (Triticum sp.) was most common with small amounts each of barley (Hordeum sp.) and oats (Avena sp.). Some of the wheat had the characteristic shape of the hexaploids and one was clearly from bread wheat. The barley grains were hulled and both twisted and straight embryos occurred thus indicating that at least some of the barley was 6- rowed Hordeum vulgare. One grain of rye (Secale cereale) was recovered. No chaff from any of the cereals was found.

Few other seeds were recovered. Those that were indicated ruderal communities and some grassland, with species such as ribwort plantain (Plantago lanceolata), docks (Rumex obtusifolius-type) and small vetches present. Burnt fragments of hazelnut shells (Corylus avellana) probably represent food and a few burnt sedge (Carex sp. and Eleocharis sp.) nutlets suggest wet ground communities also in the area. The heather (Calluna vulgaris) flowers probably are remnants of bedding or roofing material.

The fill of pit 252 contained waterlogged material as well as the one carbonised rye grain found. Its assemblage was dominated by seeds of orache / goosefoot species (Chenopodiaceae undiff.) with considerable numbers of blackberry pips (Rubus fruticosus) too. Bran fragments were present and one fig pip (Ficus carica) was noted. Nettle (Urtica dioica) seeds were common. The indications are that some of this material was probably faecal in origin; the blackberries, fig and bran almost certainly having been eaten. The nettles indicate high levels of nutrient close by. Whether the vast numbers of Chenopodiaceae seeds represent food, since they do contain large amounts of starch, or a few plants growing opportunistically in an alleyway for example, is more difficult to determine. If they had been part of the diet they may be expected to have been broken or at least partially fragmented. They are, therefore, probably representative of the local vegetation. This would be in accord with the henbane (Conium maculatum) which, although a drug plant, also grows as a ruderal in 'waste' ground. It is therefore suggested that this pit was probably a latrine although it also seems to have been in receipt of rubbish or it may have been in the open thus allowing seeds from adjacent plants to fall in.

Although some seeds were almost certainly lost through poor flotation due to the bad weather conditions on-site they are not considered likely to have made interpretation any clearer, particularly since so few seeds overall were recovered. There was only the occasional cereal grain in the re-washed residues.

Given such few remains in total little may be said of the differences between phases of activity with any certainty. However, there are no seeds from the early timber fort phase. This may reflect short-term use of such a fort with little burnt material ever being produced. Only two unidentifiable cereal grains were recovered from phase 2, the preliminary period of construction of the second fort. The majority of seeds were from phase 4 samples and from pit fills. This was a period during which the officers' quarters were initially still occupied but with evidence for demolition towards the end of the phase. The waterlogged seeds came from the ?17th century pit which, although containing 16th and 17th century pottery close to its base also contained some Roman material. The pit was clearly in receipt of faecal material.

In summary, this extremely small assemblage of plants shows that wheat and barley were being used by the Romans at Concangis with, perhaps, oats although they may have been weeds alone. There is no evidence for cereal chaff. The locally occurring vegetation seems to have been a mix of ruderals and grassland with little wet ground. This is similar to the results of Donaldson (1979) whose data came from the inner fort ditch excavated at Middle Chare although she recorded large numbers of seeds from a variety of aquatic plants. These, she concluded, were extremely locally growing, probably within the ditch itself. It seems highly likely that the Church Chare material also represents very local vegetation - perhaps with weeds growing in less used areas behind buildings etc. The early modern pit was probably used in part as a cess pit although other rubbish was apparently deposited within it or it was in the open.

Acknowledgements

I would like to thank Mike Bishop, the site director, for the samples taken under such adverse weather conditions, and for promptly providing phasing, site matrix and structural report. I am also grateful to Shaun Doran, the laboratory technician, for wading through wet and muddy heaps of gravel and converting them into clean residues which he then sorted.

Animal Bone Evidence

S. Stallibrass

The material

A. The hand-recovered animal bones

Preservation

Most of the animal bones recovered by hand are in a poor state of preservation. Many of them have suffered breakage during excavation, due mainly to their brittle texture and to the problems of excavating material from wet clayey deposits during cold weather.

Occasionally, the bones are in a very poor condition, where the surface has been totally eroded away and much of the bone structure is infilled with mineral deposits.

On many sites, the ratio of loose teeth to total numbers of fragments recovered indicates the degree of loss due to poor preservation (since the more mineralised teeth tend to preserve better than the more organic bone). However, it is noticeable that some of the loose teeth from Chester-le-Street are in a worse condition than the bones from the same contexts. In these instances, the laminae of the teeth are being pushed apart by minerals, often at the junction of the enamel and dentine layers. This leads to cracking and exfoliation, and many of the teeth have broken up into pieces that are difficult to recover and identify. In this collection, therefore, it is possible that loose teeth are, themselves, under-represented.

Table 5 summarises the numbers of fragments, weights of bones and loose teeth by Phase.

Table 6 lists the material by context, and also names the identified species.

For the Roman levels (Phases 1, 4 and ?Phase 4), the preservation conditions of the bones tend to be uniformly poor within each context. Only one ?medieval context produced animal bones (?Phase 5: context 88) totalling five whole or partial cattle teeth. For the more recent levels (Phase 6) the preservation conditions vary considerably. Some of the contexts have material that is uniformly buff coloured, slightly brittle but otherwise robust (e.g.: context 37, which contains a dog burial) whilst others contain mixtures that appear to include some residual Roman material (e.g.: context 34/39).

Material from the different phases

The bulk of the material derives from Roman levels, accounting for approximately 5kg of the total of c.7kg bone weight.

No bones were recovered by hand from any contexts dating to Phase 2 or 3.

Material from the Primary fort (Phase 1)

Context 248 was a clay fill of the ditch (context 376) of the primary fort, and it produced five bone fragments including a substantial portion of a cattle skull with both horncores intact. The size and morphology of the skull fragment is similar to that of the indigenous 'Celtic shorthorn' cattle that were ubiquitous throughout England during the Iron Age and which continued in use into the Roman period (Luff 1982; Thomas 1989). The measurements of the horncores are given in the archive report (Stallibrass 1991, Appendix 4). No other species was identified.

Material from later Roman levels (Phase 4)

The late Roman material is dominated by bones of cattle, with a few bones from sheep/goats, pigs and dogs. The material appears to be mainly the remains of food and/or butchery waste, with no indication of any craft working activities. Some of the cattle, sheep/goat and pig bones bear butchery marks, but the eroded surfaces of many of the bones means that many more such marks may have been eradicated.

None of the dog bones show any signs of butchery or skinning.

The largest group of bones was recovered from context 129, the fill of a demolition pit (context 148), but this does not appear to differ from the material from the other contexts, either in preservation or element type.

The identified anatomical zones listed in Appendices 1, 2 and 3 show that cattle bones dominate the collection from Phase 4 deposits. The numbers of zones identified are: 39 Cattle, 3 Sheep/goat, 2 Pig and 14 Dog.

All of the cattle epiphyses are fused and the few tooth rows present all have full adult dentition erupted and in wear. It is possible that the poor preservation conditions have led to the loss of less robust, juvenile bones and teeth, but the fact that occasional unfused bones of sheep/goat were recovered from the same contexts suggests that most, if not all, of the cattle represented had mature bones.

The archive report (Stallibrass 1991, Appendix 1) lists the anatomical zones that were recorded for the cattle bones. Most parts of the skeleton are represented, despite the small number of fragments.

The measurements for the cattle bones are presented in the archive report (Stallibrass 1991, Appendix 4). Generally, they are similar to those for cattle bones from successive Roman forts at Annetwell Street, Carlisle dating from c.A.D.74-330 (Periods 3 and 5, circa A.D.74-140: Stallibrass 1991; Period 9, circa A.D.320-330: Stallibrass, unpublished data). These are, themselves, similar to those of the indigenous cattle of the preceding Iron Age: the so-called 'Celtic shorthorns'. However, there is a slight difference between some of the element types ie: although the measurements for the four metatarsals from Chester-le- Street are very similar to the means for the Carlisle material, the measurements for the two Roman metacarpals from Chester-le-Street are all at the larger ends of the ranges of those for contemporaneous material from Carlisle. This could simply reflect the extremely small sample size, but it might be significant, and indicate a difference in conformation of the cattle in the eastern and western sides of northern England. A third possibility is that the Chester-le-Street cattle happened to be entire males (since the sexual dimorphism between bulls and cows is more marked for the forequarters than for the hindquarters), but this is less likely to provide a full explanation, since a large hindlimb is also represented at Chester-le-Street by a large naviculo-cuboid. Unfortunately, none of the securely-dated Roman material from Chester-le-Street can be sexed. The larger measurements are towards or beyond the upper limits of the ranges given by Luff (1982) in her review of cattle bone measurements from several Romano-British or Roman sites in Britain.

O'Connor (1988) noted that people in Roman York used cattle that had two different types of horncores, one type having horncores almost twice the size of the others, which resemble those of the indigenous 'Celtic shorthorns'. Although the collection from Concangis is small, both types of horncore are present. Whereas the horned cattle skull fragment from the Phase 1 ditch is similar to the 'native' type of small cattle, there is a broken horncore from Phase 4 (context 129) which is considerably larger. Although it is broken, so that its original length cannot be estimated, its basal measurements are similar to the large type noted by O`Connor, fitting neatly into his group of ten large horncores in his Figure 11 (O'Connor 1988, 94). It seems possible, therefore, that cattle of two types were present in the Roman fort at Chester-le-Street: one similar to the indigenous 'Celtic shorthorns' (represented by the metatarsal measurements from Phase 4, and by the skull fragment from Phase 1), and a larger type (represented by the metacarpal and naviculo-cuboid measurements and the large horncore, all from Phase 4). This aspect of the material is worth pursuing in future enquiries into the site of Concangis.

Sheep/goat bones are far fewer than those of cattle. None of the Roman sheep/goat bones could be identified to species, and none of the bones was complete enough for measurements to be taken. No skull or horncore fragments were recovered, and so nothing can be said about the conformation, stature, sex or hornstructure of the sheep and/or goats represented at the site.

The archive report (Stallibrass 1991, Appendix 2) lists the anatomical zones identified for sheep/goat bones by context. It is noteworthy that at least one of the sheep/goat bones (a distal tibia) was unfused when the animal died (probably at less than 18-24 months of age: Silver 1969). Also, there is a mandible fragment with the first permanent molar (M1) only just coming into wear. Since the fourth deciduous premolar has fallen out of its socket since the animal`s death, it is not possible to state whether the animal was a kid or a lamb, but it was probably only three or four months old when it died. This contrasts with the data described above for the cattle bones, all of which had fused epiphyses.

Pig bones were even more infrequent than those of sheep/goats in the Roman levels. The archive report (Stallibrass 1991, Appendix 2) lists the anatomical zones identified for pig bones. Only one bone could be measured: a scapula from context 129 with the proximal tuberosity fused: LG=31.9mm and SLC=20.2mm (measurements defined by von den Driesch 1976). Context 129 also produced a pig maxilla fragment that has the third permanent molar (M3) in the process of erupting, indicating that the animal was close to attaining dental maturity when it died. The sample is small and fragmentary, but there is no indication of any particularly large bones that might indicate the presence of wild pigs.

The only other mammal species represented in the Roman levels is dog. The bones appear to be the disturbed remains of buried individuals rather than food or butchery waste. The identified zones of dog bones from the site are listed in the archive report (Stallibrass 1991, Appendix 3). An adult innominate (pelvis) and a fragment of an ulna were found in context 129. Context 64 produced the remains of at least two individuals, both of them being very small. One of them was less than 8- 9 months old when it died (the humerus is completely unfused) and had short bandy forelimbs. It may have resembled a modern Jack Russell Terrier in size and shape. One pair plus another complete mandible were found in the same context. All three jaws have complete adult dentition with only light wear on the teeth. Silver (1969) gives the age by which all of the permanent teeth are fully erupted as 6-7 months. This might indicate that one of the individuals was between 6/7-8/9 months old when it died. The other individual represented by the mandibles was probably not much older, although the other long bones (radius, ulna and femur all have fused epiphyses, suggesting an age at death of over 12-18 months. All of the long bones, together with the metapodials and the atlas are small and gracile. It is difficult to tell whether two or three individuals are represented, but all (or both) of the animals were very small. One of the two ulnas may show a slight pathological alteration where the radius and ulna meet midshaft. The muscle attachment is particularly pronounced and the shaft of the ulna is slightly bent at this point. The bone does not appear to have been broken and it might be another indication of bandy forelimbs. Unfortunately, this ulna does not have its matching radius present for comparison. None of the dog bones are complete enough to measure.

Material from Post-Roman levels (Phases 5, 6 and ?6)

Only five whole or fragmentary cattle teeth were recovered from the medieval levels (Phase 5).

Phase 6 contains a mixture of material, some of which appears to be residual ?Roman bones. In particular, context 3 (from a trial trench) appears to contain only material with 'Roman-type' preservation, and the nature of the bones themselves fits in with the general pattern for the Roman levels i.e.: all of the bones are from cattle, and their sizes and the shapes of the horncores are all typical for Iron Age/Roman cattle in the north of England (see Stallibrass 1991, Appendix 4 for the measurements).

In contrast, most of the material from context 34/49 is buff coloured and and relatively lightweight, although there are two sheep/goat humeri that may be residual. There is a wider range of species represented (including horse, which was not identified from any of the Roman contexts), and a greater proportion of pig and sheep/goat bones compared to those of cattle. A pig mandible has the permanent premolars P3 and P4, and the third molar (M3) all in their crypts or actually in the process of erupting, indicating an age at death close to dental maturity (currently circa 2-3 years: Silver, 1969). The distally fused radius could be identified specifically as sheep and is quite large (Bd=31.3mm). This is similar to modern 'improved' hill sheep.

Horse is represented only by two loose teeth, one each from contexts 34/ 39 and 10.

Most of the bones recovered from Phase 6 derive from a dog burial found in the back garden of one of the recent cottages (in context 37). All of the bones are buff coloured and lightweight. Unfortunately, breakage during excavation has caused extensive damage, and none of the bones can be measured. The skull and scapulae are missing although a fragment of one maxilla and several loose maxilliary teeth are present. Otherwise, most of the animal`s body is present (see Stallibrass 1991, Appendix 3 for a list of identified zones). All of the permanent teeth have erupted, although they are not particularly worn. Some of the epiphyses are fused, others unfused. They are all consistent with an age at death of approximately one year (circa 12-13 months). Although the long bones are broken, and were not fully grown when the animal died, it is possible to state that it was slightly larger than the two (or more) individuals recovered from the Roman context 64.

The complete nature of this post-Medieval skeleton contrasts with the Roman dog remains. Although the dog bones themselves showed little damage in the Roman deposits, the fact that so few bones were recovered from any one individual suggests either that original burials had suffered considerable subsequent disturbance, or that dogs were not afforded separate burial, but were simply discarded with other unwanted animal material.

The counts of identified anatomical zones given in Appendices 1, 2 and 3 show that the post-Medieval collection is dominated by the remains of this one dog skeleton. For the three major domesticates, the collection is extremely small. The rank order is the same as that in the Roman collection, but Sheep/goat bones are relatively more common. The counts of identified zones are: 9 Cattle, 6 Sheep/goat, 1 Pig and 25 Dog.

The animal bones recovered from wet-sieving

Seventeen bulk samples each of 60 litres of soil were taken for flotation and wet sieving (see above for a report on the botanical remains). The flots were collected over 500µ mesh and the residues over 1mm. These were sorted for animal bones and the quantities are listed by context in Table 7. The bulk sample from context 253 (sample no.26, Phase 6) did not produce any animal bones.

The total hand-recovered and wet-sieved collections are too small for quantifications of species ratios, although qualitative comments are made, below, when the sieved sample adds to or contrasts with the observations made on the hand-recovered collections.

In particular, it is notable that the quantities of material from Phases 1 and 2 are miniscule. No bones were recovered by hand, and only 42.3 grammes of bone were recovered from 420 litres of wet-sieved soil. Also, as expected (Payne 1975), bones from smaller species are better represented in the sieved collection than in the material recovered by hand in bad weather conditions.

Table 8 lists the identifications of bones from the bulk samples.

The burnt bone tends to consist of tiny fragments (circa 4-10mm long) of calcined bone. The fragments appear to derive from sheep/goat-sized animals more commonly than from cattle-sized animals.

Context 129 produced mainly cattle or cattle-sized bones from the wet- sieved bulk sample (as in the hand-recovered collection), and some of these bones bear tooth marks from canids, probably dogs (see Stallibrass 1986). In addition, there are a first and second phalange of a neonatal lamb or kid that articulate and which both appear to have passed through the gut of a carnivore, possibly a dog (see Payne and Munson 1985). It is interesting to speculate whether sheep/goats were being raised in the vicinity of the fort during the late Roman period (in which case, dogs may have been able to kill livestock, or to scavenge natural neonatal deaths), or whether dogs had access to butchery waste from suckling lamb/kid killed for human consumption within the fort.

Also dating to Phase 4, context 76 produced bones identified to sheep/ goat and pig from the bulk samples, whereas only cattle bones were identified from the hand-recovered collection. With regard to the ageing of the species in Phase 4, the sieved material confirms that from the hand-recovered collection i.e.: the cattle bones and teeth are all mature, whilst those of sheep/goat and pig tend to be immature. The cattle ulna is proximally fused whereas the loose third molar (M3) of pig has not yet erupted, and the second molar (M2) in the sheep/goat mandible fragment is only slightly worn, indicating that the third molar would have been in the process of erupting when the animal died.

Most of the bones from context 64 consist of more loose teeth, plus small carpals and phalanges of the dog identified in the hand-recovered collection.

Context 189 (Phase 2) contained mainly bones of sheep/goat, including a neonatal phalange.

None of the burnt bones could be identified to species except for two unerupted pig teeth from context 110 (sample 3).

The bird bone and small mammal bones are few and very fragmentary. None of them have been identified to species.

The few fish bones have been identified by Mrs Alison Locker and the identifications are listed in the archive report (Stallibrass 1991, Appendix 5). They include small salmonids and some flat fish (including possible plaice or flounder). The site is located a few hundred metres from the River Wear, twenty-one kilometres upstream from the river mouth at Sunderland.

The shell fragments are minute (totalling less than 1g in weight) and have not been identified.

Discussion and Conclusions

Previous excavations in Chester-le-Street have produced small quantities of animal bone from within the Roman fort of Concangis (Gidney in press). This collection, too, is very small and probably raises more questions than it answers, but it does show that the site has potential for further studies.

In particular, further studies might be addressed towards (1) the general question of how the fort inter-related with its hinterland, and (2) the specific question of whether or not 'improved' types of livestock (particularly cattle, but also sheep) were introduced to the fort during the Roman occupation. The processing of bulk soil samples has shown that future excavations might also seek to investigate (3) the types of fish exploited in the Roman fort. Early excavations of Roman forts did not consider the role of fish in the Romano-British military diet and, due to the absence of sieved material, had no evidence for their use. The few fish bones recovered from Concangis in 1990-91 are, in fact, well preserved, and suggest that future sieving programmes might produce useful samples of material.