Introduction

The Roman Empire was one of the great empires of antiquity, spanning, at its greatest extent, the entire Mediterranean basin including North Africa, and extending from the British Isles to Western Asia. Within this vast melting pot, the Empire incorporated the customs and behaviours of each territory and, in turn, each region shows signs of incorporating (to a greater or lesser extent) novel/introduced material objects, lifeways and practices found within the wider Roman Empire (Philpott 1991; Toynbee 1996). At the time of the Roman conquest, cremation was the dominant burial practice of both the Late Iron Age population in Britain, particularly in the South-East, and also the incoming Roman army and associated followers. However, the influence of Roman burial practices began to be seen in Britain in the late first and early second century, with the presence of more formal cemeteries and an increase in personal ornamentation, together with the deposition of coins, lamps and imported luxury items such as samian pottery and glass phials. Cremations were generally carried out on funeral pyres, with partial remains often being collected into pottery vessels and deposited into, sometimes lined, grave pits (McKinley 2000). Notably, a common feature of burials from the Late Iron Age to the late Roman period are grave furnishings comprising suites of ceramic vessels associated with food and drink consumption or storage, thought to nourish the soul on the journey to the underworld. These accessory vessels include platters, dishes, beakers, flagons, jars, and amphorae although there is considerable variability in the numbers and types of vessels from burial to burial (Black 1986; Philpott 1991; Biddulph 2005; Weekes 2016), ranging from single vessels to the collections of high-status dining sets (up to c. 50 vessels per grave) placed over graves during the backfilling of a series of rich early Roman cremation burials at Alton, Hampshire (Millett 1986).

These elements of continental burial practices were slowly assimilated by the indigenous population, firstly in the newly established towns and forts where Roman influence was most prevalent, and sporadically filtering out to rural populations, although it should be noted that there are regional and chronological variations (Alcock 1980; Philpott 1991; Williams 2004).

The later second century saw a (somewhat patchy) shift toward inhumation practices in Britain. The impetus for this change may have come from the Continent and is argued to be a result of changes in fashion (Nock 1932). However, inhumation in the first century A.D. was by no means unknown and in some cemeteries, such as Pepper Hill in Kent, it was the majority rite during the early Roman period, plausibly representing a continuation of Iron Age practices (Biddulph 2006a). Nevertheless, the reintroduction of inhumation began in towns and military centres, spreading gradually into the countryside. Elements of these new practices included the use of coffins, plaster burials and tiled inhumation cists. The range of grave furniture underwent something of a change, with the inclusion of more manufactured items, although ceramic vessels remain a constant presence, with typical graves containing between one and three pots (Millett 1986; Philpott 1991). Such changes, including the increasing predominance of inhumation from the second century onwards, may relate to changing beliefs. At Pepper Hill during the later first and early second centuries A.D., brooches and other items of personal ornament were deposited unburnt in inhumation graves or burnt in cremations, in the latter case having been worn by the deceased on the pyre. In the second and third centuries, shoes and jewellery (except brooches, which were now rarely deposited) were found unburnt in inhumation graves and declined in relative frequency in cremation graves. Similarly, the burnt bones of chickens, sheep, pigs and cattle were recovered from later first century and second century cremations, while unburnt animal bone was recovered only from second century cremations. Overall, the evidence points strongly to the declining importance of pyre goods over time, suggesting a change in religious belief, in particular, concerning the afterlife. If shoes were worn for the journey to the afterlife, or they were required, along with personal items, for an existence in the afterlife, then in the second century, the journey no longer began at the pyre, but the grave. Similarly, food items placed in cremation graves from the second century onwards no longer needed to be burnt on the pyre, as they had been in the first century (Biddulph 2006b; 2012).

Accessory vessels

The presence of accessory vessels in burials in Britain has been argued to represent a table setting in which a meal is set out for the deceased, possibly in connection with the Roman institution of the funerary feast, the silicernium, which was eaten at the grave and at which the dead received a token portion rather than a complete meal (Philpott 1991; Gee 2008). It has been argued that the vessels represent the consumption or sacrifice of food and drink analogous to the consumption of the body by fire as an act of remembrance (Williams 2004). Certainly, various classical sources discuss the importance of food and drink at several stages in Roman funerary rituals, including Ovid (Metamorphoses: 6.566–570) and Cicero (De Legibus: 2.22.55–57, as cited in White 2007) although such documentary evidence is not directly applicable to Romano-British contexts (Alcock 1980; Toynbee 1996; Williams 2004; Weekes 2016). However, evidence for the consumption or deposition of food and drink as part of various processes and rituals relating to funerals in Roman Britain, including animal sacrifices as part of pyre-side rituals, the placement of food offerings during burial to feed the dead and the consumption of food during commemorative visits to the deceased, has been found in many contexts (e.g. Evans et al. 1997; Mackinder 2000; McKinley 2000; Polfer 2000).

However, as Philpott (1991) notes, by the early fourth century, beliefs that the dead needed sustenance may have been fading. Pottery assemblages in the south-east of England, which contained table settings of three or four vessels in second century cremations, shrunk to one or two vessels in the third century, ceasing altogether in the fourth century, possibly relating to the adoption of Christianity. Thus, deposition of food in graves may have declined from a selection of foods making up a symbolic meal to a token portion, to ceasing altogether, possibly due to a change in beliefs about the afterlife. Indeed, the funerary meal may have moved from the graveside to another location. Alternatively, it has been suggested that the vessels themselves become the offerings to the deceased (Philpott 1991).

The presence of accessory vessels raises questions as to how pottery was acquired for funerary use and why specific vessels were chosen (Biddulph 2005). For example, were they recycled from the domestic sphere or made specially for funerary purposes? Furthermore, what was the pottery used for? Did the vessels contain foods for the deceased, possibly for their final journey or did they simply represent a symbolic meal? Analysis of funerary assemblages –frequencies of vessel types, the selection and arrangement of vessels in individual graves, condition of the pottery, and so on – provides many clues. It is evident, for example, that while food and drink-related vessels were in general preferentially selected for burial in Roman Britain, there was considerable variation in the number of vessels per grave and the combination of functions represented; what might be regarded as a full table setting was far from the norm (Biddulph 2018).

Key to investigating these questions is the analysis of lipid residues that accumulate in the fabric of such vessels. Molecular and isotopic techniques can be used to determine the origins of products preserved within vessels, including terrestrial and aquatic animal fats, plant oils and waxes, beeswax and resins, based on their chemical composition (Evershed 2008a). Here, we present the results from two pottery assemblages, comprising complete and partially complete vessels, from two Romano-British cemeteries. The first is a mid to late first century Roman military cremation cemetery at Baginton, just outside The Lunt Fort, near Coventry (52°22'28.9"N; 1°28'45.1"W; Fig. 1), which included at least one individual, buried accompanied by 23 vessels laid out across the base of the grave, that can confidently be described as a high-ranking officer (Page 2020). The second, found at Rhodaus Town, Canterbury (51°16'28.7"N; 1°4'43.9"E; Fig. 1), represented a small, enclosed inhumation cemetery, perhaps associated with a family group, and was part of a much more extensive late Roman inhumation cemetery in use from the late third to late fourth century, or early fifth century A.D., located on the south side of the Roman town (Gollop and Helm 2015). Thus, our results provide valuable preliminary insights into the variability of use of Romano-British funerary vessels in both early and late burials.

Fig. 1
figure 1

Map showing location of Romano-British burial sites at Baginton, just outside The Lunt Fort, near Coventry and Rhodaus Town, Canterbury (© Ordnance Survey)

Full size image
Fig. 2
figure 2

A Phase 2 multi-pot burial 309 during excavation showing the layout of the cinerary urn, offering flagon and the wooden box lid, B Phase 2 multi-pot burial 311 after excavation showing the Samian vessels laid out across the southern half of the grave of a possible high-ranking military officer at Baginton, adjacent to The Lunt Fort (Image: Archaeology Warwickshire)

Full size image

Baginton site, The Lunt Fort

Recent archaeological work at Baginton, near Coventry (Fig. 1), yielded evidence for a mid to late first century A.D. Roman period cremation cemetery, located c. 400 m to the west of The Lunt fort, one of the most significant Roman sites in the region (Page 2020). Extensive archaeological excavations have shown the fort was initially constructed in c. A.D. 60–64, possibly as a response to the Boudiccan revolt. Subsequent rebuilding of the fort included a circular gyrus (horse training ring) and stables, suggesting that it was redesigned for use by a cavalry unit. The fort appears to have only been in use for a short period, to the later first century, but was briefly reoccupied during the mid- to late third century. There is also evidence for a civilian settlement outside the fort, located in the area of the current village (Hobley 1973).

The cremated bone assemblage from the cemetery contained 62 individual burials (including males, females and children) and evidence for several more almost completely removed by later ploughing. The burials varied between simple pits filled with unsorted material from the pyre to individual urned burials and urned burials with accompanying offering pots and/or grave goods (Page 2020). One of the urned burials (311), found in a greyware vessel of later first century date, includes a male, buried accompanied by 23 vessels laid out across the base of the grave (Fig. 2B), likely belonging to a high-ranking officer. The accompanying vessels included two carinated cups and a glazed ware bowl, possibly Central Gaulish, a roughcast Lyons ware beaker, two Central Gaulish glazed miniature flagons, a broken and well-used decorated oil lamp and a collection of eleven samian vessels (Evans 2019; Mills 2019; Page 2020). The samian vessels, ten of South Gaulish origin from La Graufesenque and one first century micaceous Lezoux Dr 24 cup (Mills 2019), are a highly unusual deposit at this date, and combined with the array of Gaulish finewares in this grave suggests the individual was not just of high status, but could have been an immigrant from Gaul, although these could also have been imported vessels. Several other high-status burials of a type previously unknown in the Midlands were present, including another adult in burial 309. The grave contained a copper alloy mirror, recovered from below a fragmentary wooden box lid, suggesting it had been buried in the box, or laid on the base of the grave below the box. Also present were a flagon in Verulamium region ware of late first to second century date, the remains of a copper alloy edged wooden box lid, hobnails and globules of melted glass, from a possible unguentarium (Page 2020).

Also found at the Baginton site were a number of Anglo-Saxon sunken-featured buildings (one of which yielded a miniature Saxon vessel) and graves (discussed further below).

The Baginton pottery assemblage

The Baginton pottery assemblage comprised a series of complete and partially complete vessels found in graves (see Fig. 2 for vessels found in graves 309 and 311) and a total of 2,455 Roman potsherds. Complete vessels included both burial urns containing cremated bone (seventeen of the burials were urned) and ‘offering’ pots. The sherds are mainly represented by wheel made greyware, with c. 60 per cent of pottery, all associated with the cemetery, coming from a single, likely local, source, presumably that supplying the Roman fort (Evans 2019; Page 2020).

The range of samian vessels deposited in the Baginton burial [311] may well be unique (Fig. 2B). The group includes some of the least common forms in the pre-Flavian repertoire and the selection is further distinguished by the small sizes of the vessels (Mills 2019). The samian from this burial is all plainwares, as is generally the case in burial assemblages, perhaps because the inclusion of decorated images in burials was seen as nefas (unsuitable) in some form. The vessels from grave 311 include two Ritterling type 1 plates, a Dr 18 plate/dish, six Ritterling cups and single examples of cups Dr 24 and Dr 27 (Mills 2019). This assemblage suggests three ‘table’ settings of a plate, a large and a small cup, and two ‘spare’ cups.

Of the Baginton vessels, twenty samples, from nineteen complete and partial vessels, and pot sherds, were selected for lipid analysis, specifically to investigate and compare vessel types and forms, including cinerary urns and offering pots from seven graves (301, 306, 309, 311, 314, 317 and 407). Included were nine vessels from grave eleven, including the complete Lyon ware cornice rimmed beaker and the central Gaulish glazed carinated handled cup, and (only) a further two samian vessels as these do not generally preserve lipids well due to their dense, well-fired fabric (Table 1). Also analysed, for comparison purposes, was a complete Anglo-Saxon hemispherical cup from a domestic context at the site (discussed earlier).

Table 1 Laboratory number, grave number, SF, cinerary urn/offering pot, vessel type, fabric, vessel size (external rim diameter), vessel date, lipid concentrations (µg g−1), δ13C and Δ13C values, and attributions of pottery lipid residues from the Baginton (The Lunt Fort) Roman accessory vessels
Full size table

Rhodaus Town site, Canterbury

Recent excavations by the Canterbury Archaeological Trust (CAT) on land at Rhodaus Town, Canterbury (Durovernum Cantiacorum), identified a small, enclosed cemetery at Petros Court (Gollop and Helm 2015) containing twenty inhumation burials in earth-cut graves (Fig. 3) located immediately outside the Roman walled town and approximately 80 m south-west of Roman Watling Street, the main road between Canterbury and the Roman port at Dover (Portus Dubris). The cemetery was broadly contemporary with a more extensive, enclosed cemetery, the main part of which was located at Palamon Court, with some 449 inhumation burials so far investigated, located to its north-west, and an extra-mural shrine or temple, at Augustine House, immediately to its north. The shrine or temple, active between c. A.D. 340 and A.D. 360, comprised a polygonal timber structure of ‘Romano-Celtic type' (Helm 2014), with associated placed deposits in pits, including coins, personal jewellery and hobnailed shoes, indicative of votive offerings (Helm 2010; 2014).

Fig. 3
figure 3

Rhodaus Town burials A Group 20 grave 2, looking south-west (no scale) B Group 20 grave 8, showing in situ face vessel (SF162), looking south-west (scale 0.2 m). (Image: Canterbury Archaeological Trust)

Full size image

As well as being spatially separated from the main burial area, several other features, including the unusual depth of the graves (up to 2.20 m deep, see Fig. 3B), and the higher quantity and range of grave goods present, distinguished the burials within this cemetery group, possibly suggesting this was a grouping of high-status individuals (Gollop and Helm 2015).

Skeletal remains were present in all but one grave, with one mature adult female also being interred with a neonate (birth to 1 month). Where identifiable, both male and female adults were present, as well as juveniles. No clear health or physical difference between the individuals within the smaller, enclosed cemetery and the main burial area were determined, though a female bias in the population was noted. Results of stable isotope analysis indicated that the population had all derived from the same locality and shared a similar and consistent omnivorous diet (Jay 2016). Furthermore, ancient DNA retrieved from four individuals, comprising one adolescent (Grave 6), and three adult females (Graves 4, 9 and 12), indicated that they all shared a level of familial relatedness, with the high probability that the cemetery represented several generations of a shared family group (Pinhasi and Fernandes 2017).

Evidence for coffins either in the form of coffin outlines and/or coffin nails were identified in fifteen of the graves. Preserved mineralised wood on coffin nails in Grave six established that it had been formed of sawn oak planks. Nine graves also had evidence for internal features, such as earth-cut ledges, while soil stains and iron fittings indicated that some might also have been shored with timber planking (Gollop and Helm 2015).

A range of grave goods was retrieved from the burials. This included twenty complete pottery vessels, recovered from eleven graves, of which one, an Oxfordshire ware beaker with elaborate demi-rosette decoration, stamping and a facemask, is potentially unique (discussed further below). Also present were six finely made funnel neck beakers. Typological dating of the most complete pottery vessels provisionally suggested that the cemetery was in use from the late third to late fourth, or early fifth, century A.D. (Lyne 2016).

Notable dress accessories included copper alloy anklets in Grave one, a copper alloy bracelet and two hair pins in Grave sixteen, worked bone hair pins in Grave twelve and a glass bead necklace and shale ring in Grave seventeen. Several individuals appeared to have been interred wearing hobnailed footwear (Graves four, eight and eleven), while in Grave six, at least three items of unworn hobnailed footwear had been placed at the end of the coffin. Grave six was also noted for containing a wooden casket, potentially bound with leather, with copper alloy fittings and an iron lock mechanism, within which had been placed a copper alloy coin (minted between c. A.D. 270–290 but very worn), a copper alloy gilded plate brooch, a spherical flint pebble, and a worked boar’s tusk, all possibly originally held inside a cloth bag. A glass unguentarium had also been placed behind the head. A second potential timber casket was noted in Grave four, though no contents were recovered (Gollop and Helm 2015).

The Rhodaus Town pottery assemblage

Twenty complete vessels were recovered from the graves found in the inhumation cemetery (Fig. 4). Graves one, five, nine, ten and fourteen contained one vessel each, and the following graves contained either two or three vessels: two (n = 2), seven (n = 3), eight (n = 3), twelve (n = 3), sixteen (n = 2) and seventeen (n = 2). Graves number three, four, six, eleven, thirteen and fifteen did not contain accessory vessels. The graves cover a span of c. A.D. 270 to 420, with the pots from graves seven, ten and twelve belonging to the period c. A.D. 270–370, grave fourteen to c. A.D. 300–350, grave sixteen to c. A.D. 350–370 and graves one, two, eight, nine and thirteen to c. A.D. 370–420. Graves five and seventeen can only be generally dated to c. A.D. 270–400 + (Lyne 2016). However, it should be noted that the use of old pots in Roman burials is a recognised phenomenon (see, for example, Wallace 2006).

Fig. 4
figure 4

The assemblage of complete Roman vessels excavated from Group 20 inhumation burials at Rhodaus Town, Canterbury, including the unique face pot and the six funnel neck beakers (Image: Canterbury Archaeological Trust)

Full size image

There are some unusual features to the pottery assemblage found at Rhodaus Town (Fig. 4). The face-beaker found in Grave 8 is unique (Lyne 2016). As noted by Braithwaite (2007), richly decorated ceramics such as head- and face-pots are not usually selected for inclusion in graves. Face-pots and face-beakers are quite unlike any other type of Roman pottery, displaying ‘crude, barbaric, rather comic-looking features stuck incongruously on a well-made Roman jar’. These unusual vessels evolved on the Continent in the first century when the frontiers of the Empire were stabilizing along the Rhine and were brought into Britain by the Roman army where they took root in Eastern England and in the Northern Military Area (Braithwaite 2007). The great bulk of the late Roman vessels with applied faces from Britain are jars: face beakers are almost non-existent. They have been suggested to be used in the worship of Dionysius/Bacchus and the very large size of the beaker from Grave eight could be taken as evidence for this practice (Braithwaite 2007).

Another unusual characteristic of the pots from the Rhodaus Town cemetery is that many are either miniature or very small (see Table 2 for rim diameters and Fig. 4). For example, eleven vessels have rim diameters of up to and including 100 mm (not including the five funnel-neck beakers which were not sampled), three were up to 200 mm in size and one was 240 mm (Lyne 2016). A recent study of miniature Roman pots from Britain (Lyne 2017) divides such pots into those from temple sites and burials, pottery production sites and the stone linings of wells, although it has been suggested that the majority were made for religious purposes (Graham and Graham 2009). They likely have more than one function, for example, the very small and underfired ones, such as those from Frensham Common and Uley temple in Gloucestershire (Woodward et al. 1993) may have been used for the inhalation of fumes from burning narcotic substances to communicate with the Gods (Graham and Graham 2009). The somewhat larger miniature pots from the Rhodaus Town burials are paralleled by those from the stone sarcophagus found at Wheatley near Alton in Hampshire (Millett 1975) and the cemetery at West Tenter Street, London (Whytehead 1986).

Table 2 Laboratory number, grave number, context, SF, vessel type, fabric, vessel size (external rim diameter), lipid concentrations (µg g−1), δ13C and Δ13C values, and attributions of pottery lipid residues from the Canterbury (Rhodaus town) cemetery Roman accessory vessels
Full size table

Twenty complete, or nearly complete, vessels from eleven graves at Rhodaus Town, Canterbury underwent lipid analysis (Table 2), although five (of the six) funnel-neck beakers were not sampled due to the restricted nature of their openings. However, one beaker (from grave seven), which had a small piece broken off the lid, was sampled, making a total of fifteen vessels sampled. Furthermore, two sherds were sampled from the shallow bowl found in grave eight (5365, 161), named CRV02 and CRV03, thus total sherds sampled, n = 16.

Lipid residue results

Baginton lipid results

As discussed above, twenty potsherds from the Baginton assemblage underwent organic residue analysis, as described in earlier publications (Correa-Ascencio and Evershed 2014) although a modified sampling procedure was adopted for the complete vessels, as described in supplementary information, to minimise destructive sampling. Also analysed, for comparison purposes, was a complete Anglo-Saxon hemispherical cup from a domestic context at the site (discussed earlier).

Notably, vessels from the Romano-British Baginton pottery assemblage yielded (mostly) low and non-interpretable lipid profiles (Table 1). In summary, all samples, save the Anglo-Saxon vessel, contained either concentrations below the limit at which such extracts can be reliably interpreted (i.e. > 5µg g−1, Evershed 2008b) or those with a low concentration which do not display an interpretable lipid profile or contained either plasticiser or compounds indicative of modern contamination (or both). Two vessels (BAG01, supplementary Fig. 1 and BAG06) display lipid profiles possibly more indicative of input from terrestrial vegetation, from plant epicuticular leaf waxes denoting leafy plant matter and/or modern contamination (see Whelton et al. 2021). Similar classes of compounds have been widely detected in soils and reflect contributions from terrestrial vegetation (van Bergen et al. 1997; Bull et al. 2000), and likely suggest these lipids have originated from the burial environment.

Thus, the eight vessels thought to be offering pots (BAG04, BAG05, BAG08, BAG09, BAG012, BAG017, BAG018 and BAG019) either did not contain lipid-rich offerings or, alternatively, were present in low concentrations which did not survive over archaeological timescales. Similarly, the cinerary urns (BAG02, BAG03, BAG06, BAG07, BAG13, BAG14 and BAG20), aside from BAG01, also only yielded either low or non-interpretable lipid profiles. The general absence of lipids in these vessels confirms their likely use as cinerary urns. Lastly, it is not clear whether BAG015 is an offering pot or cinerary urn.

Notably, analysis of the Saxon vessel (BAG016, Table 1, lipid concentration 1.4 mg g−1) from a primary fill, found to have processed dairy products, confirms that the absence of lipid profiles from the Romano-British assemblage is unlikely to be an issue of preservation.

Rhodaus Town, Canterbury, lipid results

Of the twenty Rhodaus Town vessels supplied for analysis, thirteen vessels (fourteen sherds) yielded sufficient concentrations (> 5 µg g−1) of lipids that can be reliably interpreted (Evershed 2008b). All extracts were dominated by palmitic (C16) and stearic (C18) fatty acids (FAs), typical of a degraded animal fat (Fig. 5A–F) e.g. Evershed et al. 1997; Berstan et al. 2008). Only two vessels, a small-necked jar, from grave five (CRV06; 5294,125) and the funnel-necked beaker from grave seven (CRV16; 5320,132) did not yield lipids. Five vessels contained high concentrations of lipids (see Table 2), maximising at 5.5 mg g−1 (CRV05, a deep straight-sided dish from grave sixteen), demonstrating excellent preservation and suggesting these vessels underwent sustained use in the processing of high lipid-yielding commodities, i.e. were likely used in cooking. An example of similar lipid concentrations can be seen in a recent study of 571 Romano-British cooking vessels from a major urban site and a range of small towns, villas and farmsteads within its hinterland (Greenwood et al. 2023).

Fig. 5
figure 5

Partial gas chromatograms of the acid-extracted FAMEs from the Canterbury (Rhodaus town) cemetery Roman accessory vessels pottery extracts A CRV04, B CRV07, C CRV08, D CRV09 E CRV11 and F CRV12. Circles, n-alkanoic acids (fatty acids, FA); triangles, n-alkanes; IS, internal standard, C34 n-tetratriacontane. Numbers denote carbon chain length

Full size image

Four vessels (CRV02, CRV07 Fig. 5B, CRV12 Fig. 5F and CRV14) contained a series of long-chain fatty acids (LCFAs), containing C20 to C26 carbon atoms, with lipid profiles from vessels CRV02, CRV07 (Fig. 5B) and CRV12 (Fig. 5F) also containing the C27 n-alkane. Together, these particular distributions of LCFAs and n-alkanes are strongly indicative either of an origin in leaf or stem epicuticular waxes (Kolattukudy et al. 1976; Tulloch 1976; Bianchi 1995; Kunst and Samuels 2003) although these are not diagnostic to families of plants and so cannot be used as anything other than a general indicator for plant processing. However, this specific distribution of lipids (LCFAs and n-alkanes) is entirely consistent with the processing of plant material, likely leafy plants, together with animal products, within these vessels (see, for example, Dunne et al. 2016, 2022; Dunne 2022).

GC-C-IRMS analyses were carried out on fourteen extracts (Table 2 and Fig. 6) to determine the δ13C values of the major fatty acids, C16:0 and C18:0, reflecting their biosynthetic and dietary origin and allowing non-ruminant and ruminant adipose and ruminant dairy products to be distinguished (Dudd and Evershed 1998; Copley et al. 2003). Nine vessels (ten sherds) plot in the ruminant dairy region (Table 2 and Fig. 6B) confirming these were used to process dairy products. However, three of these plot at the extent of the range for dairy products (CRV04, CRV07 and CRV08) suggesting some minor mixing with ruminant carcass products, or possibly, in the case of Vessel CRV08, plants. A further three vessels plot within the ruminant adipose region, confirming they were primarily used to process carcass products from domesticated cattle, sheep or goat and the remaining vessel, CRV010, plots firmly in the non-ruminant adipose region (pig or chicken), with a Δ13C value of 0.6 ‰ (Fig. 6B).

Fig. 6
figure 6

Graphs showing: A δ13C values for the C16:0 and C18:0 fatty acids for archaeological fats extracted from the Canterbury (Rhodaus Town) Roman vessels. The three fields correspond to the P = 0.684 confidence ellipses for animals raised on a strict C3 diet in Britain (Copley et al. 2003). Each data point represents an individual vessel (except for CRV02 and CRV03 which are from the same vessel). B shows the Δ13C (δ13C18:0 – δ13C16:0) values from the same extracts. The ranges shown here represent the mean ± 1 s.d. of the Δ13C values for a global database comprising modern reference animal fats from Africa (Dunne et al. 2012), UK (animals raised on a pure C3 diet) (Dudd and Evershed 1998), Kazakhstan (Outram et al. 2009), Switzerland(Spangenberg et al. 2006) and the Near East (Gregg et al. 2009), published elsewhere

Full size image

As noted, the rim (CRV02) and base (CRV03) sherds from the shallow bowl of class C45 found in grave eight were sampled (as the vessel was broken into several pieces). Comparison of the lipid content of individual sherds from different parts of the same vessel can provide insights into the possible function of vessels (Charters et al. 1993; Evershed 2008b). Sherd numbers CRV02/03 from this vessel displayed a ruminant dairy profile (Table 2) with lipids predominating in the base area, at 138.0 µg g−1, in comparison to the rim where the lipid concentration was 39.6 µg g−1. This suggests the presence of a dairy offering, likely in solid form, e.g. butter or cheese, in contact mainly with the base of the bowl, where lipids were preferentially absorbed. Interestingly, the rim sherd plots very differently to the base sherd, with CRV02 plotting between the ruminant dairy and non-ruminant ellipses and CRV03 plotting within the ruminant dairy ellipse (Fig. 6A). This occurs because of the c. 4 per mil difference in the δ13C values of the two sherds and may relate to the presence of likely plant compounds (long-chain n-alkanoic acids and n-alkanes; as discussed previously) in sherd CRV02, not seen in CRV03, although this does suggest a plant with more enriched δ13C values, presumably of C4 origin. Lipid concentrations are low in both so a possible explanation for this is that this dish held a solid dairy product offering, possibly butter or cheese, with leafy plants being placed on the surface of the offering, subsequently being absorbed into the ceramic fabric, although this is purely speculative and experimental analysis would be needed to confirm this.

Vessel CRV10, a small-necked jar from grave twelve (a young female), was used to process pig (or, possibly, chicken) products, whether that be for cooking of pork/bacon or the rendering of lard. Lipid concentrations from this vessel (497.1 µg g−1) suggest that it had previously been used to process non-ruminant products and was therefore not specifically made for funerary purposes.

As noted, nine vessels were used to process primarily dairy products. Of these, five of the vessels (CRV02/03, CRV07, CRV11, CRV12 and CRV14, shallow bowl, miniature Young’s class bowl, miniature everted-rim jar, small everted-rim jar and another small everted-rim jar, respectively), yielded very low lipid concentrations (Table 2). This suggests that these vessels may have been made especially for funerary purposes, holding offerings of dairy products. The unusually small or miniature nature of four of these vessels adds weight to this theory. Interestingly, four vessels, CRV02, CRV07, CRV12 and CRV14, containing dairy products, also included specific distributions of lipids (LCFAs and n-alkanes) discussed above, suggesting a possible plant offering, likely leafy plants or, possibly, plant oils, within these vessels. The remaining four vessels (CRV04, CRV05, CRV08 and CRV13, small everted-rim jar, deep straight-sided dish, convex-sided dish/bowl and small jar-Lyne type, respectively) were also used to process dairy products but displayed very high lipid concentrations (1.9, 5.5, 1.9 and 5.0 mg g−1, respectively), suggesting these were vessels which underwent sustained use over a long period (Evershed et al. 2008b). However, it is noteworthy that these four vessels were all grog-tempered, in contrast to the lipid-yielding colour coated vessels, where lipid concentrations were uniformly low. It is thus possible that the slip may have hindered lipid absorption or that the vessels were used in different ways. Although the four high lipid-yielding vessels were clearly not made for funerary purposes and must have been used on a regular basis, possibly by the deceased, three of these (CRV04, CRV05 and CRV13) are very small (rim diameters 70, 130 and 105 mm, respectively), and it seems unlikely that they would be used for cooking, although CRV08, with a rim diameter of 200 mm, may have been a cooking pot.

Three vessels, including the unique ‘face beaker’, held ruminant carcass products (cattle, sheep or goat). Significantly, the lipid concentration from the face beaker was very low (Table 2), suggesting either that it had not previously been used for cooking/processing of food products (and may have served a decorative/some other purpose) prior to deposition in the grave, where carcass products (meat or, more likely, liquid tallow) were left in the vessel as an offering. Alternatively, it may have been specially made for funerary purposes. The other two vessels displaying ruminant carcass product signals (CRV08 and CRV09, convex-sided dish/bowl and miniature Lyne type jar, respectively), both displayed high concentrations of lipids (1.9 and 2.8 mg g−1, respectively), suggesting that these had previously seen sustained used in processing high lipid-yielding commodities. Again, these vessels were clearly not made for funerary purposes but although the convex-sided dish/bowl, which had a rim diameter of 200 mm, may have been used for cooking/processing animal products, it seems unlikely that the miniature Lyne type jar, with a rim diameter at 80 mm, was a stewpot or similar cooking vessel, although it could have seen use as a storage vessel for tallow.

Discussion—Meals for the Dead

Rhodaus Town cemetery, Canterbury

Eleven of the late Roman graves at Rhodaus Town cemetery each contained between one and three very well-preserved ceramic vessels, providing a total of twenty complete vessels, of which fifteen were analysed. Of these, a small-necked jar, from grave five and the funnel-necked beaker from grave seven, did not yield interpretable lipid profiles. The remaining thirteen vessels were used to process animal products, mainly of dairy origin, often with the addition of plants. The high concentrations of lipids found in five vessels suggest these may have previously been used in the domestic realm, possibly belonging to the deceased themselves, prior to their placement in the funerary realm. The remainder seem to have been new vessels specifically purchased for use in the funerary realm (or may have seen previous use as containers for low lipid-yielding products). As noted, three vessels contained ruminant carcass products, which, at least in the case of the face beaker (which has a small opening), must have been in the form of a liquid/semi-solid (as opposed to a joint of meat), possibly either tallow or a gruel or stew. Alternatively, these may have comprised entrails (including intestines, bowel, and other viscera), the portion of a sacrificed animal traditionally given to the gods.

As noted, the low lipid concentration from this vessel suggests the face beaker had not previously been used for cooking/processing of food products, perhaps serving a decorative purpose, prior to deposition in the grave where meat products were left in the vessel, perhaps as an offering or a meal for the dead. As discussed, these could represent a token portion rather than a complete meal (Philpott 1991; Gee 2008). In contrast, the other two vessels displaying ruminant carcass signals (CRV08 and CRV09, convex-sided dish/bowl and miniature Lyne type jar), both displayed high concentrations of lipids suggesting that these had previously seen sustained use in processing high lipid-yielding commodities. Again, these vessels were clearly not made for funerary purposes but likely saw everyday use in food preparation, cooking or serving, although the miniature jar is very small, with a rim diameter of 80 mm and it is hard to imagine that it would make a practical cooking vessel. Another small jar (CRV10), rim diameter 90 mm, contained pig products, perhaps in the form of lard or a pork stew, or, possibly, chicken (Colonese et al. 2017). These striking results bear comparison to our existing knowledge of food grave offerings found across the Roman Empire.

Food offerings across the Roman Empire

As Davies (2002) notes, eating and drinking are common themes associated with death and the afterlife in numerous cultures around the world. However, many grave offerings, comprising foodstuffs such as meat served off the bone, bread, cakes, fruits and vegetables and also liquids, including wine and oil, which could once have been present, will rarely leave traces (Philpott 1991). Despite this caveat, a diverse range of food stuffs are evidenced in burial rituals across various provinces of the Roman Empire (White 2007). These include plant offerings of cereals and pulses from central France, and more exotic species, such as olive, fig and almond from Mediterranean France (Bouby and Marinval 2004). Bones from cattle, pig and domestic fowl were found on plates in fourth century graves in Nijmegen, the Netherlands. In one instance, a fully articulated skeleton of a domestic fowl (missing the head and feet) was found on a plate, together with the rib bone from a cow, and the humerus of a suckling pig was present on another, suggesting these were meals for the dead. Cattle, pig, fowl and sheep/goat remains were also found in 20 Roman cemeteries or individual graves across Belgium, Germany, Austria and Switzerland (Lauwerier 1983). In the northern provinces of Gaul, burials are often rich in cuts of meat, for example, a first century A.D. burial at the ‘Actiparc’ site near Arras, France, which included at least twelve pottery vessels, some of which were miniature, held pig ribs, a half head of pig and numerous chicken bones (Lepetz 2017). A further study of Roman cremation burials of the first century B.C. to third century A.D. in 27 northern Italian cemeteries showed that cereals, pulses, fruit and vegetables were found as offerings. Also present were prepared foods such as bread and cakes (Rottoli and Castiglioni 2011). Various animal remains found in a burial at the Porta Nocera cemetery in Pompeii, included fish bones (NISP = 41) comprising a small perch-like fish, associated with a pig metatarsal (Lepetz 2017). In Algeria, North Africa, in a burial in a cemetery in Sétif, one plate contained dried grapes, wheat grains and olive pits, and a whole fish lying on a plate together with olive pits and pomegranate was found in a first century inhumation grave at Tipasa. Nuts, figs and pine nuts were found in burials at Thaenae and burnt olive pits at Hadrumetum, Tunisia (Stirling 2004).

Closer to home, the cremation burial of a female at the Roman cemetery at 165 Great Dover Street, Southwark, London, containing various goods such as pottery lamps and tazze, also included an impressive array of plant remains, including typical Mediterranean species such as date, almond, fig and pine nut (Giorgi 2000). Excavations at the eastern cemetery of Roman London yielded nineteen clear instances of animal bones, as partial or complete skeletons, deposited in inhumation burials as grave goods. Graves with single species include fourteen with chicken (one found within a pot) and two each with pig and goose, while a single grave produced multiple species, comprising chicken, goose and pig. Also present in many of the graves were charred plant remains including lentil, peas and Celtic beans (Barber and Bowsher 2000). At Pepper Hill in Kent, beef joints, whole pigs and a shoulder of mutton were among the food items burnt on the pyre, the burnt remains being subsequently deposited in graves. At the same cemetery, charred plant remains were recovered from two early Roman graves. Unurned cremation grave 11,801 yielded a diverse assemblage comprising cereals, weeds such as dock, and foods, including grapes, lentils, horsebean, pea or similar and figs. The cereals and weeds may have been incorporated accidentally or as pyre fuel, but the food items are more significant, having been placed on the pyre in a raw state, so that foods were ‘cooked’ on the pyre (Biddulph 2006a). Bones from domestic fowl were found in two inhumation burials at West Tenter Street, London, in one case, deposited within a jar (Whytehead 1986). Similarly, casketed cremation remains found at Baldock, north Hertfordshire, yielded several burials with assemblages of vessels that contained possible food offerings. For example, in burial seven, calcined pig bones, domestic fowl, the remains of a calf scapula and some sheep vertebrae were found in some pots within a ten-vessel assemblage and similarly, the eight samian vessels found in the grave held domestic fowl bones, the right hind-leg of a pig and the foreleg of a sheep (Stead and Rigby 1986). Occasionally, paraphernalia within graves are laid out as if for a meal, as at grave II at Winchester, where two joints of pork and a bird bone were laid on a shale trencher, alongside a spoon, two knives and a samian cup and platter (Biddle 1967).

Romano-British ‘meals for the dead’

Lipid analysis has rarely been applied to pottery from funerary contexts (although see Oras et al. 2018; Van de Velde et al. 2019; Weber et al. 2020). Thus, it is significant that here, for the first time, we have been able to confirm that accessory vessels found in Romano-British burials were, in some instances, used in the laying out of food as part of various processes and rituals relating to funerals in Roman Britain, possibly providing a token portion to nourish the soul on the journey to the underworld or perhaps consumed by mourners at a funerary feast.

Notably, together with the identification of meat products in accessory vessels at Rhodaus Town, Canterbury, the lipid results have been able to detect the presence of previously unidentifiable grave offerings, namely dairy products, often mixed with plants.

The dairy lipid signal identified in the majority of vessels could comprise a cheese which may have been either the ‘soft’ curd, green or cottage cheese which had to be eaten fresh or, less likely, the hard longer lasting cheese which involved more complicated preparation. Cheese was popular in Roman cookery, often being used with milk to create both sweet and savoury dishes and was also part of standard rations for the army (Alcock 2000; Deŕy 2000). Pliny describes both a national and international trade in cheeses and other Roman writers, including Cato, Varro and Columella, wrote on cheese manufacture and quality and on the culinary uses of cheese. Palladius wrote a treatise on Roman-British farming in the fourth century A.D., including a description of, and advice on, cheesemaking (Fox 1993; Thurmond 2006).

Alternatively, butter or soured milk products may have been used as offerings. In his Natural History, 28.35.133, Pliny notes that butter was a choice food in barbarian nations and both Columella and Apicius discuss recipes for sour milk products called oxygala (literally, ‘sour milk’), which seems to have been similar to yoghurt (Deŕy 2000).

The dairy and plant lipids in the accessory vessels may originate from cheeses flavoured with herbs and other plants. The Romans are known to have eaten a wide variety of leafy vegetables including beets (grown for their leaves, not the root), chard, cabbage, broccoli, kale, mustard, chervil, mallow, lettuce, endive, chicory (arugula), purslane and watercress. Various herbs, including capers, parsley, rue, alexanders, fennel flowers and stalks, wild parsnip flowers and stalks, flower of bryony, house leek, pennyroyal, calamint, hoary mustard and samphire were commonly used (Thurmond 2006). Columella discusses the addition of fig sap, nettles and wild thistle to coagulate milk to make cheese and the addition of pine nuts, crushed thyme and various other herbs to the fresh milk to impart a pleasant flavour (Alcock 2000; Thurmond 2006).

Written sources suggest plant material could also have been deposited (with food) in burials for symbolic reasons, including, for example, poppy (the bringer of sleep and death), dates (symbols of reincarnation), and leaves from evergreens for eternal life (Kreuz 2000). Milk may also have been used for ritual purposes, being sometimes offered as a libation in Roman funeral rites and also playing a part in the cult of Mercury. Pliny and Galen both note that butter was used as a skin moisturiser after bathing so may have been deposited in burials for cosmetic purposes (Deŕy 2000).

Finally, the dominance of dairy products in the Canterbury vessels is worthy of consideration. It has been suggested that cattle dairying was of little importance in Roman Britain (Hesse 2011, Allen 2017), in contrast to the British Iron Age (Copley et al. 2005), especially by the third/fourth centuries A.D. However, a recent large-scale comparison of organic residues preserved in cooking pots from a major urban site (Cirencester) to those found in a range of small towns, villas and farmsteads in the hinterland, suggests that this belief should now be challenged (Greenwood et al. 2023). The study showed that dairy products were absent in vessels from the urban centre of Cirencester after c. AD 100, and relatively uncommon in vessels from smaller towns close by. In contrast, dairy products were prevalent at a range of rural settlements, both farmsteads and the high-status fourth century A.D. villa at Chedworth. Greenwood et al. (2023) suggested this divide was driven primarily by economics, and that milk products were consumed where they were produced, the vast majority in rural settings. A small residue analysis study (n = 29) at the first century BC Silchester oppidum also found no evidence of dairy products in any of the sampled vessels (Colonese et al. 2018). The prevalence of milk products in fourth century pottery used for funeral purposes in the centre of the major urban site of Canterbury is therefore noteworthy. Were milk products specifically associated with ritual, sacred or funeral contexts in urban centres, even into the fourth century? Or did the inhabitants of Canterbury rely on dairy products as a greater part of their diets than those in other urban centres across Britain? Further analysis of vessels from domestic contexts at other sites in Canterbury and from comparable funereal contexts across Britain would be needed to clarify the role of dairy products in ‘meals for the dead’. On current evidence, the authors are unable to determine whether the dead at Rhodaus Town were offered a special, or even sacred food, or whether the offerings reflected the contemporary diet of the living who buried them, although the unusual nature of many of the vessels is perhaps suggestive.

Baginton cemetery, The Lunt military fort

Damaged or imperfect vessels

The quality of manufacture of the vessels from Baginton military site suggests that some may have been ‘seconds’, and, bearing in mind their imperfect nature, that their status as wasters or damaged goods meant it was expedient to use them in burials. Possible seconds are the Central Gaulish glazed flagons and two of the samian plates, vessel number 517 Ritterling 1 plate and vessel 511, the Drag. 18 plate, which were visibly distorted. Interestingly, most of the vessels do not display traces of prolonged usage, save for vessel DR24, which is worn from use internally. A number of vessels display signs of damage, for example, vessel 487 (Drag. 24 cup) has two ancient large chips missing from one section of the rim, vessel 513 has areas where the slip had not been applied and vessel 514 has a large-ish chip spalled from the outer wall.

This correlates well with suggestions that pottery may have been specifically purchased at the time of the funeral, possibly bought at market or at stalls set up near cemeteries (Tuffreau-Libre 2000) or from burial societies which may have kept a supply of potters’ seconds or rejects, or that they may even have been disinterred from earlier funerals and re-used (Biddulph 2005).

There are certainly many instances of the use of old, repaired, imperfect or damaged vessels in burials. Vessels thought to be wasters or seconds were identified at Ospringe, Kent (Whiting et al. 1931), Pepper Hill, Kent (Bird 2006), Silchester (Boon 1974), Welling, Kent (Tester 1953), Skeleton Green, Puckeridge (Partridge 1981), West Tenter Street, London (Whytehead 1986) and St. Pancras, Chichester (Down and Rule 1971). For example, at Pepperhill, near Springhead, Kent, 10 of the 32 complete or near complete samian accessory vessels were considered sub-standard products (Bird 2006). A cup and a dish from separate cremations both with distorted rims and no signs of wear were also recorded at Heybridge (Biddulph 2015). Of the 200, mostly complete, vessels recovered at the large East London Roman cemetery (first to fifth Century), a total of 46 (23 per cent) were damaged in some way. Whilst this does not necessarily mean these vessels were not previously used for cooking, storage or other functions, Barber and Bowsher (2000) suggest that it does imply that when choosing vessels for burial, damaged pieces were frequently selected. Philpott (1991) suggests that the frequent use of pottery wasters and repaired vessels may represent a deliberate choice of items already partway to the otherworld, rather than the more prosaic clearing out of damaged household items. Broken pottery may also be linked with grave-side rituals, such as the feasting events of the silicernium and the cena novendialis, or the annual festivals of the dead, the Parentalia and the Lemuria (Toynbee 1996, 63–4). Pottery fragments that do not belong to formally deposited grave-goods are not uncommon finds in grave backfills (Biddulph 2015). In the Agrimensores, Ex libris Dolabellae (222.26–28), the presence of cooking-pots or broken earthenware vessels were regarded as an indicator of tomb boundaries, and Propertius (Elegiae 4.7.33) refers to the practice of leaving a broken flask or bottle, originally used as an offering on a tomb (Peña 2007). Caution in identifying such pottery must be applied, however, as distinguishing between pottery that was deposited in a broken state and pottery that has been fragmented and denuded through post-depositional processes, such as the movement of the plough, is not necessarily straightforward.

Also interesting is the lack of evidence (animal bone or plant remains) for either storage or processing of foodstuffs in any of the Baginton vessels, possibly suggesting that funerary rituals at The Lunt Fort did not include the leaving of food offerings or meals for the dead, and that the vessels themselves may have represented a ‘symbolic meal’. However, four of the cremation pits (301, 311, 331 and 334) all contained fragments of burnt bird bones, suggesting at least some feasting at the pyre. This may have been for socio-economic reasons in that ordinary soldiers may not have been able to afford lavish or costly funerals (although they may have been members of burial clubs; Biddulph 2005) or, possibly, that they would not have their own source of ‘domestic’ pottery, being fed from military kitchens. It could also be related to the fading of beliefs that the dead needed to be fed, or, indeed, that the funeral meal may have taken place elsewhere. Further, whilst the burial of containers for food and drink shows that the dead were thought to continue to need nourishment, as Rook (1973) notes, often the way the pots were placed in the grave shows that they were empty. Another possibility is that any lipids present may have been burnt out by the high temperatures reached during firing of the pyre, although none of the vessels show any signs of burning and are thus unlikely to have been part of the pyre ritual (Cool and Leary 2012).

Conclusion

Our investigation of accessory vessels from a mid to late first century Roman period cremation cemetery at Baginton, located close to The Lunt military fort, Coventry, and a late third to late fourth century A.D. family cemetery at Rhodaus town, Canterbury, has provided valuable preliminary insights on vessel use and burial practices across the span of the Roman occupation of Britain.

The absence of interpretable lipid profiles at Baginton, The Lunt military fort burial ground and the consequent lack of evidence for food offerings, together with the presence of seconds or damaged vessels in the graves correlates well with the suggestion that pottery was sometimes specifically purchased for use at a funeral, possibly representing a symbolic meal, rather than being re-purposed from the domestic sphere.

This contrasts with the direct chemical evidence for ‘meals for the dead’, contained in accessory vessels found in late Romano-British burials in Canterbury and comprising dairy products, possibly cheese, sometimes prepared with the addition of leafy plants, together with meat from both ruminants (cattle, sheep and goat) and pigs, likely in the form of a stew or gruel. The high lipid concentrations in some of the Canterbury vessels suggests they may have seen use in the domestic sphere prior to deposition in sacred or profane contexts. These results are in good agreement with offerings of meat, fish, fruits, nuts, pulses and vegetables found in graves across the Empire and provide the first chemical evidence for previously ‘invisible’ grave gifts such as milk, butter or cheese, extending our understanding of the range and extent of funerary ‘meals’.

Whilst these two sites encompass a military and urban setting, more work is clearly needed to investigate possible diversity in burial practices across the Roman Empire, both temporally and spatially. A larger chronological study incorporating rural, urban and military sites across both Britain and continental Europe would surely provide further insights into Roman funerary and ritual customs.