Vegetation History and Archaeobotany

, Volume 20, Issue 1, pp 41–52

New evidence for the consumption of barley at Romano-British military and civilian sites, from the analysis of cereal bran fragments in faecal material

Authors

    • Department of Human EvolutionMax Planck Institute for Evolutionary Anthropology
    • Department of ArchaeologyDurham University
  • Jacqui Huntley
    • English Heritage North East
Original Article

DOI: 10.1007/s00334-010-0245-3

Cite this article as:
Britton, K. & Huntley, J. Veget Hist Archaeobot (2011) 20: 41. doi:10.1007/s00334-010-0245-3

Abstract

Despite the abundance of barley in the archaeobotanical record at Roman military sites along Hadrian’s Wall in northern England, and the suitability of the British climate for growing the grain, contention still remains concerning the human consumption of this cereal in the Roman world. Previous experimental and archaeological work has demonstrated that cereal bran fragments in faecal material can be successfully assigned to species. Here, microscopic analysis of plant fragments is utilised to investigate the relative abundance of Triticum/Secale (wheat/rye), Hordeum (barley) and Avena (oats) from faecal deposits from two Roman military sites and a contemporary civilian settlement in Carlisle. Cereal bran was identified in all deposits, along with certain other edible plant fragments such as Coriandrum sativum (coriander) and Allium sp. (onion genus). The presence of barley in deposits from military sites appears to confirm its consumption, with the frequency and size of fragments hinting at likely occasional culinary use in soups and stews. Increased frequency at the contemporary civilian site indicates more widespread culinary use in non-military settlements. The practical and analytical limitations of this method are discussed.

Keywords

Roman military dietCereal branBarleyFaecal materialNorthern EnglandHadrian’s Wall

Introduction

Grain and grain provisioning were pivotal to successful military campaigning in the Roman world. It is estimated that over 33 tons of grain would have been consumed each day by the frontier forces when they numbered their greatest during the 3rd century a.d. (Davies 1971, p. 123), with individual rations ‘issued to him [each soldier] as grain, ground by him in portable mills, and made into bread or pulse’ according to Caesar (Edwards 1917, pp. 604–605). Granaries were important structures at military sites, and were capable of holding as much as two years’ grain supply, in case of a break in supply or siege (Bidwell 1997, p. 86). In spite of a wide consensus on the importance of grain to the Roman army in Britain, there are conflicting bodies of evidence concerning which grain species were regularly consumed and in what forms.

The analysis of botanical remains in ancient faecal material permits the reconstruction of dietary habits and food preparation methods in archaeological populations. First recognised and recorded in British archaeological sewage deposits in the late 1970s (Dickson and Dickson 1979), cereal bran commonly dominates the <1 mm fraction of environmental samples containing human waste (Greig 1981; Hall et al. 1983a; Kuijper and Turner 1992). Cereal bran consists of the outermost tissues of the grain including the testa and pericarp, which are relatively indigestible for humans. Despite the chemical, biological and morphological changes that occur during cooking, mastication, digestion and deposition, cereal bran in faecal remains can be highly diagnostic. Taxonomic determination of bran fragments has been previously demonstrated and remains can be identified to genus and, in some cases, to species, through the use of simple microscopy procedures and the creation of comparative modern materials (Dickson 1987; Körber-Grohne 1964, 1991). Furthermore, morphological assessment of the shape of fragments also allows the pretreatment and the culinary use of different grains to be inferred (Dickson and Dickson 1988, pp. 122–123).

In this study, the Roman consumption and culinary use of barley is explored directly through the analysis of cereal bran fragments in archaeological faecal remains from military and civilian sites near Hadrian’s Wall in northern England. When combined with previously published macrobotanical reports from these sites, the microscopic analysis of cereal bran (<1 mm fraction) allows an assessment of the relative importance of barley in the Roman military diet compared to other grains and compared to local contemporary civilian populations. Fragment size data provides a new insight into the likely culinary uses of barley. An evaluation of these methods and an assessment of the potential of these approaches to answer such palaeodietary questions follows.

Barley and the Roman military diet

Pollen records from throughout the Roman occupation of Britain indicate the clearing of woodland and opening up of arable land (Dark 1999; Dumayne 1994), potentially reflecting an increased need for agricultural produce in order to supply Roman troops (Dark 1999, p. 254). Botanical evidence from Romano-British civilian sites—the likely source of these staples for the frontier troops—indicates the production, processing and storage of barley, along with other grains (Huntley and Stallibrass 1995; Van der Veen 1992). Furthermore, archaeobotanical evidence from many northern English Roman military sites themselves indicates the storage of large amounts of clean hulled barley, in addition to wheat (Hall and Huntley 2007; Huntley 1997, 2002; Huntley and Stallibrass 1995; Stallibrass and Thomas 2008; Van der Veen 1992). It must be noted that these cleaned, hulled barley grains do not appear to be consistent with animal feed (Jones 1998) and also lack the sprouting characteristic of malting and brewing seen at other European sites (Renfrew 1973, p. 81; Van der Veen 1989).

Although barley was known to ‘increase the strength and enlarge the muscles of the body’ according to Pliny the Elder (Rackham 1971, p. 18), contemporary literary sources and modern treatises deny the human consumption of this staple in the Roman period. Modern literature on the Roman military diet describes wheat as ‘the foundation of the military diet’ (Bidwell 1997, p. 84), adding that barley was used either for horses (Wilmott 2001, p. 103), or normally only given to soldiers as a punishment ration (after Davies 1971, p. 140). Such modern claims undoubtedly stem from literary sources contemporary with the period in question, where barley is described as a punishment ration in a number of works. These include Polybius’ Histories, in which he describes decimation and the punishment of deserters, which includes the provisioning with barley instead of wheat in order to make a ‘conspicuous example’ of those guilty of acts of cowardice (Shuckburgh 1889, pp. 6, 38). Plutarch’s Lives also describes similar circumstances, where defecting soldiers were forced to eat barley as a humiliating public and symbolic punishment (Langhorne and Langhorne 1898, Book 22, Chapter 60, ‘Anthony’). In his Epitome of Military Science, Vegetius also describes such practices occurring during the training of soldiers, with barley rations forced upon individuals until they demonstrated their capabilities to the satisfaction of their senior officers (Milner 1993, p. 13).

Alternative lines of evidence, along with the archaeobotanical evidence, appear to contest this. Written evidence in the Vindolanda tablets (2009), for example, indicates that huge quantities of barley were being ordered and utilized at this Stanegate fort (for example, Tablets 213, 185 and 190). In Tablet 185—an expenditure list—barley is classified separately from animal fodder, perhaps implying that it was not being fed to animals. This is reiterated in Tablet 190, with detailed accounts for the domestic administration of the praetorium and apparently provisions for a festival, including a large quantity of barley. The other items listed on Tablet 190 include large quantities of Celtic beer, fish sauce, pork and three different types of wine and are clearly domestic and not agronomic in context. Furthermore, given these large quantities of Celtic beer ordered, it may seem unlikely that this large volume of grain was intended for malting or brewing (http://vindolanda.csad.ox.ac.uk/).

The contemporary evidence for large purchases of barley at the Vindolanda Stanegate fort and archaeobotanical remains at a range of British Roman military sites indicate that this crop may have been grown locally and was being stored and eaten regularly at Roman military as well as at civilian sites (Hall and Huntley 2007; Huntley 1997, 2002; Huntley and Stallibrass 1995; Stallibrass and Thomas 2008; Van der Veen 1992). Given the range of evidence available, and contrary to what is suggested in primary literary sources, we must now question the part barley played in the diet of Romano-British soldiers in the north of England.

Materials and methods

Samples from deposits containing faecal material from a range of context types from Roman frontier sites in northern England were selected for this study. Sites included the Roman military encampments/fort sites of Birdoswald (Banna) and Carlisle (Luguvalium; Millennium Project excavations), as well as a Romano-British civilian settlement at Carlisle (Old Grapes Lane) (Fig. 1; Table 1). Sites were chosen for their well-preserved waterlogged deposits and individual contexts were selected on the basis of their composition, summarised in Table 2, see discussion 6.1 ‘Composition and likely origin of deposits’, below. These sites are from very different areas along Hadrian’s Wall and the type of military unit and the size of the two forts would have varied greatly—from the large earlier site of the Carlisle fort (1st/2nd century a.d.) to the much less significant Birdoswald (3rd century a.d.). The civilian site of Carlisle at Old Grapes Lane, contemporary with the fort, has been selected to contrast with the military sites.
https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig1_HTML.gif
Fig. 1

Map of northern England showing locations of Hadrian’s Wall, Birdoswald Roman fort and the town of Carlisle

Table 1

Details of contexts sampled from the forts at Birdoswald (Banna) and Carlisle (Luguvalium; Millennium project excavations) and the Romano-British civilian settlement at Old Grapes Lane, Carlisle

Site

Context

Context type

Date

Reference

Birdoswald (fort)

BD-1755

Upper fill of ditch

Mid-/late 4th century a.d.

Huntley 1991

BD-1499

Lower fill of ditch

Mid-/late 3rd century a.d.

BD-1976

Fill of ditch

Mid 3rd century a.d.

Carlisle (fort)

CSL-1010

Fill of feature

Phase 3b (a.d. 83/84–93/94)

Huckerby (pers. comm.)

CSL-1075

Fill of gully/drain

Phase 3a (a.d. 72/73-83/84)

CSL-6384

Fill of drain

Phase 4b (a.d. ca. 125–140)

CSL-6675

Lower fill of ditch

Phase 4b (a.d. ca. 125–140)

Carlisle, Old Grapes Lane (civilian)

OGL-732

Fill of various depressions

Period 6 (a.d. 93–94)

Huntley 1992

OGL-750

Fill of drain

Period 6 (a.d. 93–94)

OGL-1126

Fill of pit

Period 3 (Trajanic a.d. 98–117)

Table 2

Selected results from the initial macroscopic botanical analysis of contexts from the forts at Birdoswald (Banna; BD) and Carlisle (Luguvalium; Millennium project excavations, CSL), and the civilian settlement at Old Grapes Lane, Carlisle (OGL)

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Tab2_HTML.gif

Presence or absence indicated by + or − respectively. The remains were found in a waterlogged state, those marked with asterisk were also charred. Compiled from Huntley 1991 (BD), 1992 (OGL) and Elizabeth Huckerby, personal communication (CSL)

Prior to this study, the majority of materials had been subjected to flotation and the flots were analysed whilst still wet for botanical macroremains (>1 mm fraction), and then subsequently dried prior to placing in storage for up to 20 years. Sub-samples of desiccated materials were re-hydrated and re-sieved (to 1–0.5 mm). Those more recently excavated and still in whole-earth form were processed in the same way by flotation in order to separate heavier mineral deposits from the less dense organic matter and then wet-sieved to separate different sized fractions, before being sub-sampled.

Five temporary mounts on glass slides were produced for each context during analysis. Small sub-samples (~2 ml) of the well-mixed wet material were transferred to flat microscope slides using a Pasteur pipette and spread to ensure fragments were not overlapping. Excess water was carefully removed with paper towelling and samples were mounted in glycerine, using standard-sized glass cover slips (22 × 22 mm, 0.16–0.19 mm thick). Ideally, a larger quantity of each sample would have been analysed. Identification of organic fragments was made at a cellular level, a time-consuming procedure which restricted the number of slides that were studied to five. However, this method tended to yield five slides with a similar number of fragments and similar percentage composition for each context.

The material identified comprised the outer coat of the cereal caryopsis which is composed of the pericarp and the inner seed coat or testa. The pericarp generally has a single layer of epidermis outside it, covered with a cuticle. Under this lie sub-epidermal cells, crushed parenchyma, cross-cells and tube cells. The latter two types become somewhat lignified during the development of the fruit and as a result can survive deposition and, on occasions, can be identified. The testa lies inside the pericarp and typically comprises a thin cuticular layer over an inner layer of crushed nucellar cells. Further inside lies the aleuron layer which encloses the starch rich endosperm (Esau 1965). We use the term ‘bran’ to cover the pericarp, testa and aleuron layers. Identification of these layers to genus was made by comparison with modern reference material, prepared using previously published digestion-simulation methods (Dickson 1987), using both Wild M3 and Leica Diaplan transmitted light microscopes. Identifications were made at ×100–400 magnifications. Digital images were made using a Nikon 9-series digital camera attached to the Leica microscope. Nomenclature of the limited range of taxa identified follows Stace (1997).

Dickson states that the identification of Hordeum (barley) and Avena (oats) are possible to genus, although distinctions between Triticum (wheat) and Secale (rye) are only possible when the transverse cells of the pericarp are preserved (Dickson 1987, p. 100). Therefore, even where wheat and rye could be distinguished, they were counted together. This does not compromise the current investigation of the prevalence of barley bran. Furthermore, there is very little evidence that rye was deliberately cultivated during the Roman period (Huntley and Stallibrass 1995, pp. 43–47; Van der Veen 1992). Cross-references with images and descriptions from other published works were also made to ensure accurate designations (Dickson 1987; Fielding and Parkinson 1929; Hall et al. 1983a; Körber-Grohne and Piening 1980; Winton and Winton 1932). These sources also aided the identification of other materials present in the samples and the occurrence of features such as charring was also noted. However, as with plant macroremains or pollen analyses, comprehensive identification is reliant on having an extensive modern reference collection. Currently, the descriptive and photographic reference libraries for identifying plant tissues to species are small, limiting such studies (Holden 2001).

Fragments were also classified by size (small, medium and large), classes which were based on the area occupied by a fragment within the field of view at magnification ×100 (small < half total area; medium > half but < total area; large > total area of field of view). This relative sizing method was sufficient for the identification of broad trends. Subject-specific measurements using a calibrated eye-piece graticule were made during the production of photomicrographs.

Results

The results of this study are presented in Table 3 and Figs. 2 and 3. In addition to the results from the microscopic analyses of the 1–0.5 mm fraction, data from previous macroscopic and microscopic studies are also presented for comparative purposes (Table 2, Huntley 1991, 1992; Elizabeth Huckerby, personal communication).
Table 3

Results from the microscopic analysis of contexts from the forts at Birdoswald (Banna; BD) and Carlisle (Luguvalium, Millennium project excavations; CSL) and the civilian settlement at Old Grapes Lane, Carlisle (OGL) (this study)

<1 mm

BD-1499

BD-1755

BD-1796

CSL-1010

CSL-1075

CSL-6384

CSL-6675

OGL-732

OGL-750

OGL-1126

 

n

%

n

%

n

%

n

%

n

%

n

%

n

%

n

%

n

%

n

%

Cereal bran

 Triticum/Secale sp.

65

95.6

9

81.8

4

33.3

10

100.0

30

88.2

46

71.9

117

94.4

10

47.6

31

83.3

43

82.7

 Hordeum sp.

3

4.4

2

18.2

8

66.7

4

11.8

6

9.4

11

52.4

6

16.2

9

17.3

 Avena sp.

12

18.8

7

5.7

Other edible taxa

 Coriandrum sativum

2

 

 

 

 

 

9

 

1

 

1

 

 

6

 

 Allium sp.

 

 

 

1?

 

5

 

 

 

 

 

 

Other taxa

 Bromus sp.

 

 

 

 

 

+

 

 

 

+

 

 

 Sphagnum sp.

+

 

 

+

 

 

 

 

 

 

 

+

 

 Juncus sp.

 

 

 

 

+

 

 

 

+

 

 

 

 Unspecified plant remains

+

 

+

 

+

 

+

 

+

 

+

 

+

 

+

 

 

+

 

Non-plant remains

 Fly puparia

 

+

 

+

 

+

 

+

 

 

+

 

 

 

 

 Unspecified invertebrate

+

 

+

 

+

 

 

+

 

 

+

 

+

 

 

 

Charring

 

 

 

+

 

 

 

 

 

+

 

 

Where counts are not shown, presence or absence is indicated by + or − respectively

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig2_HTML.gif
Fig. 2

Proportions of different taxa identified in cereal bran from the forts at Birdoswald (BD) and Carlisle (CSL), and the civilian settlement at Old Grapes Lane, Carlisle (OGL)

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig3_HTML.gif
Fig. 3

Size distributions of wheat/rye and barley bran from the military and non-military contexts. a wheat/rye (military); b wheat/rye (non-military); c barley (military); d barley (non-military)

Slide mounts made from material from all ten of the contexts consisted of cereal bran fragments in the 1–0.5 mm fraction. Wheat/rye (Fig. 4), wheat (Fig. 5), barley (Fig. 6) and oats (Fig. 7) were all identified. Total cereal counts ranged from 10 (CSL-1010) to 124 (CSL-6675) fragments. In addition to macroscopic edible plant fragments and microscopic cereal bran, seven of the ten contexts (BD-1499, CSL-1010, CSL-1075, CSL-6384, CSL-6675, OGL-732 and OGL-1126) also contained microscopically identifiable plant food fragments other than cereal bran. Identifiable taxa include Coriandrum sativum (coriander; Fig. 8; identifications based on Dickson 1989 and Kenward and Hall 1995) and probably Allium cf. porrum (onion genus, probably leek; identifications based on Tomlinson 1991).
https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig4_HTML.jpg
Fig. 4

Characteristic transparent elongated cross-cells of the upper pericarp layer of Triticum/Secale. Scale = 50 μm

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig5_HTML.jpg
Fig. 5

The transverse cells of Triticum pericarp. These are rarely preserved and are required to differentiate between Triticum and Secale (after Dickson 1987). Scale = 50 μm

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig6_HTML.jpg
Fig. 6

The dark double cell layer of Hordeum grain testa over the endosperm. The double layer structure gives the testa a sub-rectangular, three-dimensional appearance (after Dickson 1987). Scale = 100 μm

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig7_HTML.jpg
Fig. 7

The pale, almost transparent pericarp cell layers of Avena. Cells are thin and oblong and are often arranged in a ‘modified herring-bone pattern’ (Dickson 1987, p. 99). Scale = 100 μm

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig8_HTML.jpg
Fig. 8

The characteristic wavy cell patterns of Coriandrum sativum. Scale = 100 μm

https://static-content.springer.com/image/art%3A10.1007%2Fs00334-010-0245-3/MediaObjects/334_2010_245_Fig9_HTML.jpg
Fig. 9

Epidermal cells of Allium sp. (probably A. porrum). Scale = 50 μm

Barley bran was identified in deposits at all three sites, in a total of eight out of the ten contexts under analysis. Counts of identified barley ranged from 0 (CSL-1010, CSL-6675) to 11 (OGL-732) fragments. Proportions were calculated based on total cereal counts per context (see Table 3) and cereal bran composition for all contexts ranged from 33.3 to 100.0% wheat/rye; 0.0 to 66.7% barley and 0.0 to 18.8% oats. The context with the highest proportion of barley was BD-1796, with almost 70% of bran fragments being identified as barley. OGL-732, rich in other edible plants, was the other context dominated by barley bran. Remaining contexts were dominated by wheat/rye bran with only small quantities of oat bran being found in some contexts.

The majority of wheat/rye fragments at military sites were classified as ‘small’ (62.8 and 68.5% at Birdoswald and Carlisle respectively; Table 4). The majority of barley fragments identified in the military contexts were larger—with more than 70% of the fragments at the two sites being classified as ‘medium’ or ‘large’. At the civilian settlement, 30.8% of barley fragments and 45.2% of wheat/rye were classified as small, with the majority of both wheat and barley fragments being classified as either medium or large.
Table 4

Size distributions of cereal bran fragments from contexts from the forts at Birdoswald and Carlisle, and the civilian settlement at Old Grapes Lane, Carlisle

Site

Birdoswald (BD)

Carlisle (CSL)

Old Grapes Lane (OGL)

Genus

Wheat/Rye

Barley

Oats

Wheat/Rye

Barley

Oats

Wheat/Rye

Barley

Oats

n

%

n

%

n

%

n

%

n

%

n

%

n

%

n

%

n

%

Small

49

62.8

1

7.7

139

68.5

5

50.0

10

52.6

38

45.2

8

30.8

Medium

27

34.6

9

69.2

55

27.1

4

40.0

9

47.4

35

41.7

12

46.2

Large

2

2.6

3

23.1

9

4.4

1

10.0

11

13.1

6

23.1

Total

78

13

203

10

19

84

26

Discussion

Composition of the deposits and their likely origin

The contexts under analysis comprised sieved material from pit and ditch fills (see Table 1) and were not macromorphologically distinguishable as discrete human coprolites. Furthermore, no chemical approaches such as sterol analyses were used to independently confirm the presence of human sewage. However, the taxa identified in these deposits, both from this study and previous macrobotanical work, can be used to understand their composition and their likely origin.

Deposits were originally selected on the basis of previously published and unpublished work, where macroscopic edible plant seeds/components, partially digested materials or large quantities of cereal bran had been recorded and where the waste deposits had been interpreted as predominately human in origin (Huntley 1991, 1992; Elizabeth Huckerby, personal communication). The results from the original macroscopic palaeoenvironmental analysis of these samples are summarised in Table 2. The small edible seeds, seeds from edible fruits and spice seeds/mericarps found, such as Ficus (fig) or coriander, are typical components of human faecal material (Knörzer 1991; Hellwig 1989, 1997). Some larger fragments, such as fish bones, hazelnut and fruitstones, may indicate the presence of kitchen refuse (Hellwig 1997, p. 113). Other edible taxa include cereals; fruits such as Prunus cerasus/avium (cherry) and Malus sp. (apple); berries such as Rubus sect. Glandulosus (blackberries) and nuts such as Juglans regia (walnut). They also include probably exotic taxa such as Vitis vinifera (grape), walnuts, Lens culinaris (lentils) and Oleaeuropaea (olive). These, along with the figs and coriander, are likely to have been grown in the Mediterranean area and transported as valuable commodities (Bakels and Jacomet 2003; Cool 2006; Kuijper and Turner 1992; Livarda and Van der Veen 2008; Stallibrass and Thomas 2008; Van der Veen et al. 2008). Mammal and fish bone remains were also identified, along with common grain crop weed seeds such as Agrostemma githago (corncockle). These weed seeds, common in Romano-British grain stores (Firbank 1988), may have been accidentally included in the diet, eaten together with cereals (Hellwig 1997), or removed during cleaning and deposited as kitchen waste.

In addition to the macroscopically discernable food fragments, some edible taxa were also identified at a microscopic level during this study, through the identification of cellular patterns in epidermal/pericarp fragments. These taxa, including Coriandrum sativum and Allium sp., had been previously described in the literature, making identification possible. Members of the genus Allium include onion, garlic and leek, all of which were cultivated in the Roman world. All Allium fragments found were of leaf epidermis as indicated by the presence of stomata (Fig. 9). Although morphologically not easily distinguishable from the leaves of other Allium species, it is likely that these fragments are Allium porrum (leek), as the green leafy portions of this plant, as opposed to onion or garlic, are most commonly consumed. Umbelliferous mericarps such as coriander were used to flavour foods or may have been used medicinally to prevent flatulence (Greig 1981, p. 273). These plant foods, along with the presence of edible taxa identified from previous macro- and microbotanical studies, confirm that the deposits contained human faecal material or food waste, together with varying quantities of additional detritus.

Microscopic analyses of deposits from the three sites included in this study also revealed non-useful plants such as seeds of Bromus sp. (brome grass) and non-plant materials such as puparia and other invertebrate fragments. Some non-edible plant remains are likely to originate from initial humanly deposited materials. For example, bryophytes (mosses) have been used throughout history for their packing, bedding, insulation and absorbent qualities. Their abundance at Bearsden, Vindolanda and other northern Romano-British sites (Dickson 1973; Seaward and Williams 1976), including those in this study, and also at York (Hall et al. 1983a, p. 92), may be indicative of their intensive economic and domestic use as, for example, toilet paper. The faecal material and other waste matter found in the ditches and pits sampled in this study is unlikely to be in situ but is likely to have been secondarily deposited. Other taxa identified, such as the peat bog, heathland, shallow water, woodland, grassland plants and weeds, are likely to have blown, fallen or washed into the open, waterlogged and silting-up ditches, as determined at other sites such as Bearsden (Knights et al. 1983, p. 13). The presence of non-edible and non-identifiable plant remains, which probably correspond to the meadow plants identified macroscopically, may also have been used as domestic flooring in an area where food was being prepared or eaten. This may also account for the inedible food remains such as bones and fruit stones that were identified, as observed in medieval deposits from Worcester (Greig 1981).

It is unlikely that these materials represent the addition of significant quantities of animal bedding or dung, as the assemblage does not correspond with the classic stable manure package (Kenward and Hall 1997). Even in the unlikely event that livestock were being fed milled grains, such material would also be expected to be rich in dung-feeding beetles, ‘house fauna’, brushwood, horse shoe nails, complete grain pericarps (caryopses) and an abundance of grazing flora and compressed straw (Greig et al. 1982; Kenward and Hall 1997; Wilson 1978). Furthermore, the garrison stationed at Birdoswald during the 3rd/4th century a.d. (the cohors I Aelia Dacorum) were peditata (infantry) with no evidence of a cavalry component (Jarrett 1994, pp. 45–46). This makes the presence of animal droppings less likely. The fragments, either in the macro- or microscopic fraction, in the Carlisle military deposits (for example context CSL-6384) do not resemble those consumed by horses, given the lack of whole grain caryopses or other features of the stable waste package discussed above, and there is no evidence to suggest that the ala Augusta Gallorum Petriana bis torquata milliaria civium Romanorum who were present at Carlisle from the reign of Trajan onwards, was equitata (cavalry) (Jarrett 1994, p. 38).

The military sites: Birdoswald and Carlisle

Barley was found in the 1–0.5 mm fraction of deposits from the military sites of Carlisle and Birdoswald, although in different quantities, with a larger proportion of barley bran identified in the contexts from Birdoswald than in those from Carlisle. There are temporal and spatial differences between the forts, as well as differences in their size, garrisoned troops and influence, which may account for these differences.

Barley bran fragments were identified in two of the four contexts analysed from the military site of Carlisle. The deposits from Carlisle date to the 1st and 2nd century a.d. and come from the far west of the wall, with most of these contexts pre-dating the Hadrianic fortifications. As an early military encampment at the unsecured extreme northern frontier of the Empire, this site may have relied heavily on supply lines from elsewhere in the Roman world as opposed to local supplies. This seems to be confirmed by the presence of a large amount of foreign and exotic ‘luxury’ food stuffs such as coriander in the macro- and microscopic portions of the samples. Alternatively, this could indicate a fidelity to more traditional Roman eating habits amongst the military in the early Roman period in Britain. This corresponds well with observations from Britain and other areas of the empire (Bakels and Jacomet 2003; Cool 2006; Livarda and Van der Veen 2008; Van der Veen et al. 2008) and may also reflect the fact that these early troops at Carlisle were civium Romanorum (Roman citizens). The almost total lack of barley bran in these deposits (4.3% of all cereal bran fragments identified) appears to concur with the primary literary sources discussed above—that barley was not a grain that was consumed by the Roman military, except perhaps in extreme circumstances.

In contrast, all three contexts under analysis from the smaller, later fort site of Birdoswald contained barley. Here, barley accounts for 16.7% of the total number of identifiable cereal bran fragments. A large number of larger fragments of barley was also identified in the original archaeobotanical report (Huntley 1991). There is also a lack of exotic/foreign plants in the Birdoswald contexts, indicated by the macrobotanical analysis and the other archaeobotanical evidence from Birdoswald (Huntley 1991). This may be due to the lack of availability of anything but local foods in this area and perhaps also to the unimportance of this small fort site, compared to other fortifications. The contexts from this site also date to a later period than contexts from the fort at Carlisle under analysis here and could perhaps indicate the adoption of more local eating habits.

In addition to the small quantities of barley bran, small quantities of oats were also found in the Carlisle military contexts. This could indicate that oats may also have played a part in the Roman military diet and may not just have been reserved for horse feed. This occasional human (as opposed to animal) consumption of oats appears to be confirmed by the fact that oat abundance in charred grain assemblages does not correlate with forts that housed cavalry units (Huntley and Stallibrass 1995, p. 58).

A non-military site: Carlisle, Old Grapes Lane

The non-military (civilian) contexts sampled from Old Grapes Lane display some compositional similarities to the material from the nearby contemporary military site of Carlisle. As at the military site, a large number of exotic and edible taxa were found in the deposits, for example in contexts OGL-750 and OGL-1126 (Huntley 1992). The microscopic analysis in this study also identified coriander and evidence of charring. Macroscopic remains included Prunus spinosa (sloe), P. insititia (damson/bullace), cherry, figs, Malus/Pyrus sp. (apple/pear), walnuts, olives, grape, coriander mericarps and burnt lentil (Huntley 1992). The presence of these exotic taxa, similar to those seen at the fort, may indicate trading associations and the availability of luxury goods in this early hub of the Roman north.

However, cereal consumption at this site shows a different pattern from the nearby military site. It is clear from all contexts that barley was being consumed at the civilian settlement at Carlisle. Barley bran was far more frequent in the civilian contexts, with a larger proportion of bran in all contexts at Old Grapes Lane identified as barley, 23.6% compared with 4.3% at the Carlisle fort site. This incorporation of a larger proportion of barley into the diet of civilians may reflect local or less discerning dietary habits or it may be reflective of their status. Bromus mollis (lop-grass), corncockle and other common cornfield weeds were also recorded in the Old Grapes Lane deposits. They are common in grain that has not been thoroughly cleaned during processing (Huntley 1992) and this may be indicative of different supply sources or quality of grains for civilians or the military.

Size of bran fragments: uses of grains

Fragments of identified cereal bran in each of the contexts were also classified according to their size (Fig. 3). Assignments were broad (small, medium and large) but the systematic placing of each piece of bran within one of these categories has allowed a tentative indication of differences in the culinary use of the different grains.

The majority of wheat/rye fragments were classified as small at the military sites. Grinding grains into flour tends to yield small, uniform fragments. These have been shown to pass through the digestive tract in a relatively unchanged state (Dickson and Dickson 1988, p. 122). This indicates that wheat at Roman military sites was mostly ground into flour, and subsequently eaten as bread or other farinaceous foods.

In contrast, most of the barley fragments were larger (classified as medium or large). It has previously been suggested that larger fragments of bran present in human faecal material may be due to a different type of grain processing and use (Dickson and Dickson 1988), such as leaving grains whole after hulling (removing the glumes, lemmas and paleas) or subsequently cracking (crushing) or pearling them. In the process of pearling, some of the bran layer is also removed. Larger, less uniform bran fragments would therefore be expected in human waste after consuming hulled, cracked or pearled grains than would be expected after consuming grains milled into flour (Dickson and Dickson 1988; Knights et al. 1983). Fragmentary remains of chaff (palea, lemma) may also be found in faeces after the consumption of grains processed in these ways, although no chaff was identified microscopically in these deposits. The size of the barley fragments at Birdoswald may therefore indicate the use of whole, cracked or pearled barley in soups or broths. Groats (cracked grains; American: grits) can also be used to make thick grain porridge. This may also be true of the identified oats which, although appearing very infrequently, also show similar fragment size trends.

Study and methodological evaluation

The present study is clearly limited by its scale and also by the nature of the deposits under analysis. In order to characterise the consumption of barley in Roman northern England more adequately, more sites and a greater number of contexts must be incorporated into future analyses. The time-consuming nature of the analysis is also one clear limiting factor of this type of research. In addition, although the compositional and morphological evidence for the origin of these deposits is compelling, no external method was available to independently confirm the presence of human waste in each of the contexts analysed here. Furthermore, even where sterol analysis or other methods, such as presence of human parasite eggs, could be utilised to confirm the presence of human faecal material, the timescale, number of individuals or number of ‘meals’ the deposits incorporated is unlikely to be quantifiable. Given the rare nature of finds such as the ‘Lloyd’s Bank Turd’ (Hall et al. 1983b), quantification of data from faecal remains, even from closed latrine or sewer contexts, also poses difficulties. Furthermore, it has not yet been established how well bran of the different cereal genera are preserved during digestion or within archaeological deposits. Until experimental work confirms this, we cannot assume how the relationship between the percentage composition of bran within faeces compares with that of the diet. Lastly, it was possible to meet the aims of this study given the specific nature of the research question, the limited number of genera under investigation, and the existence of previously published reference materials for these taxa. Although some other plant foods could be identified microscopically in addition to the cereals, such as Coriandrum sativum and Allium cf. porrum, until larger published reference collections have been created, these methods cannot allow a more comprehensive reconstruction of the diet as a whole.

Conclusions

This study has incorporated primary literary evidence with macro- and microbotanical analyses in order to assess whether or not barley featured in the diet of Roman soldiers and civilians at the most northern frontier of the Roman Empire. Archaeobotanical evidence from a range of sites across northern England indicates the growing and storage of barley, although both primary written evidence and modern historians have claimed that this grain was rarely consumed by humans.

Through the microscopic analysis of cereal bran fragments in faecal deposits, this study has explored the consumption of this grain at the Roman military sites of Birdoswald and Carlisle, as well as at the contemporary civilian settlement of Carlisle. Barley was identified in eight out of the ten contexts under analysis, many of which also contained other edible and/or exotic taxa.

However, there was very little evidence for the consumption of barley at the 1st/2nd century a.d. military site of Carlisle. This is in contrast to the later contexts of the 3rd/4th century a.d. from the fort of Birdoswald, where a larger proportion of the cereal bran identified was barley. This directly contradicts the primary literary evidence and confirms that barley was eaten by Roman soldiers in the north of England. The differences between the sites of Carlisle and Birdoswald could be due to temporal or spatial factors, or differences in the size, occupants and functions of these two fortifications. The majority of wheat/rye fragments were small at the military sites, with the majority of barley fragments being larger. This possibly indicates the more common use of wheat flour for bread-making and the more occasional use of cracked or pearled barley in stews and soups (Dickson and Dickson 1988, p. 122).

The largest proportion of barley in the contexts sampled were found at the 1st/2nd century a.d. civilian site of Carlisle, indicating that barley was being consumed more frequently, or in larger quantities, amongst this population. This could be due to cultural or socio-economic differences between native and immigrant military populations. Given the large amounts of exotic and foreign edible taxa in these deposits, it does not seem that the dietary choices of this population were restricted, but instead suggests that people chose to consume a greater quantity of this grain.

Despite limitations of scale, this study has successfully utilized an under-used archaeobotanical tool in order to gain insight into the consumption of barley in the Romano-British world. The identification of cereal bran fragments in human waste has allowed the consensus to be addressed that barley was not consumed by people. These techniques have given an indication of the proportion of barley and the nature of its use at three Romano-British sites, and has offered a glimpse into one of the most fundamental aspects of the Roman diet.

Acknowledgements

Research for this paper was completed in partial fulfilment of the requirements for the degree of Archaeology B.Sc. at the University of Durham by KB. Samples were provided by Durham Archaeological Services, Durham University and Oxford Archaeology. Thanks to Elizabeth Huckerby (Oxford Archaeology), Louisa Gidney and Charlotte O’Brien (Durham), Petra Dark (Reading) and also to Mike Church (Durham), Alex Brown (Reading) and two anonymous reviewers for comments and Stefanie Jacomet (Basel) for editorial critique. Thanks also to the Max Planck Institute for Evolutionary Anthropology, Mike Richards and Jean-Jacques Hublin (MPI-EVA) and to the Natural Environment Research Council and Deutscher Akademischer Austausch Dienst for professional and financial support during the preparation of this manuscript.

Copyright information

© Springer-Verlag 2010