Roman Owls at the Villa


Back in the summer of 2002, a large deposit of small mammal bones was uncovered on the chalk floor of the large square room that abuts the hypocaust to the west and the tessellated corridor to the south at the villa. A first thought, perhaps, was that here was a grain storage area that was heavily infested with mice, but a closer look at the remains revealed the presence of voles and shrews in addition to mice. What the bones represented, in fact, were the typical components of barn owl pellets. This was not a domestic assemblage.

Owl pellet location

Fig. 1. The large square room at the Roman villa; the owl pellets were found in the central part of the western (left) side of this room, on the triangular area of exposed chalk floor.

The bones were excavated in the centre of a triangular area that was exposed on the east side of Trench 7 that comprised the entire western edge and northwest corner of this 6 x 6 m room (Fig. 1). Most of the bones were found immediately above the surface of the chalk floor, at the lowest level of the overlying deposit (context 7147) which consisted of a compact mixture of mortar and chalk and heavier building rubble including chalk blocks, flints and tile. As the bones began to be uncovered, they were at first carefully placed in small plastic bags, but there were so many of them that eventually they were simply tipped into a bucket.

Sample analysis

The sample received for analysis consisted of a mixture of tiny bones in a loose dry matrix that completely filled a 2-litre capacity ice-cream container. Fragments of mortar and painted wall plaster in the matrix confirmed the Roman association of the context.

The bones were removed from the matrix by sieving in two stages. A 4-mm mesh removed all the mandibles, crania and other identifiable skull fragments, the larger long bones, scapulae and pelves. Any concretions were soaked in water for wet sieving later. The residue was sieved again using a 1-mm mesh, which removed individual teeth, vertebrae, the smaller long bones, toe bones, calcanei (heel bones), unfused epiphyses and small fragments. The residue from this sieving contained ribs and tiny fragments that could not be identified and these were discarded.

small mammal mandibles drawing

Fig. 2. Small mammal mandibles: a) field vole, with large ridged molars; b) wood mouse, with multicusped molars; c) common shrew, with red-tipped teeth. All left side; scale=10mm

After separation from the matrix, the bones were sorted into categories. The mandibles and skulls of small mammals allow the most precise identification (Fig. 2) and these were picked out and counted to provide the Minimum Number of Individuals (MNI) of each species represented. Although most of the other bones were not used for identification, they were carefully examined for the presence of non-mammalian bones, which were removed and identified.


The total weight of the extracted bones was 90g. Most of the assemblage consisted of long and other bones and fragments. Small mammal skull fragments and mandibles accounted for just 16g of the total, and bird bones accounted for 6g.

SpeciesLeft mandibleRight mandibleOtherMNI%
Water shrew (Neomys fodiens)6363.3
Common shrew (Sorex araneus)22192212.1
Pygmy shrew (Sorex minutus)5352.7
Wood mouse (Apodemus spp.)56656535.7
Harvest mouse (Micromys minutus)0442.2
Field vole (Microtus agrestis)60646435.2
Water vole (Arvicola terrestris)21 + juvenile31.6
Rat (Rattus sp.)0221.1
Birds (various species)53105.5
Frog (Rana temporaria)110.5

Table 1. Identified bones from the owl pellet assemblage in the square room adjacent to the hypocaust at the villa. (MNI = Minimum Number of Individuals)

The results are shown in Table 1. Eight species of small mammals were identified from their mandibles, including three species of shrews, two species of mice, two species of voles and a rat. There was a large number of unfused epiphyses from the ends of small mammal long bones, suggesting that a significant proportion of the animals represented were immature. There were also a few fragments of rabbit bones (one vertebra and three rib fragments) and one sheep- or dog-sized rib fragment, all of which are considered to be intrusive.

In addition to the mammal bones, there was a single left femur of a large frog and an assortment of bird bones. These comprised a variety of skeletal elements, of which the best preserved and most numerous was the humerus: 17 of these were recovered (seven left and ten right) ranging in size from that of a small finch to that of a starling. The single mandible was starling-sized and the single premaxilla (upper beak) was that of a seed-eating bird about the size of a sparrow.


The species listed in Table 1 and their relative proportions are typical prey of the barn owl (Tyto alba), with voles strongly represented and shrews frequent, a small but definite presence of birds and the occasional amphibian. Other owls rarely include shrews in their diet to any extent.

A closer look at the species represented in these owl pellets and a comparison with what might be expected in barn owl pellets from the same area today offers some tantalising hints about the environment at the villa when the pellets were deposited. Voles predominate in modern barn owl pellets and usually account for about 60% of all prey items; mice are generally less frequently taken and represent about 25% of the assemblage, often less (Lawrence and Brown, 1967). Two species of small voles are usually present, the field vole and the bank vole (Clethrionomys glareolus), but the latter was not represented at the villa. As their names suggest, the field vole prefers open fields, mainly rough ungrazed grassland, whereas the bank vole inhabits hedges, woodland and scrubby areas. However, these are the habitats favoured by wood mice, which were strongly represented at the villa, so perhaps the bank vole was simply uncommon in the area at the time. Field voles and wood mice are present in similar proportions in the villa pellets (35.2 and 35.7%, respectively). This relative paucity of field voles in the barn owl diet could possibly represent overgrazing of the fields around the villa (Yalden, 1977).

Owls are rather sedentary and have limited feeding ranges (Corbet, 1975), so the species represented at the villa will all have come from the vicinity of the villa. Although the water shrew accounts for only 3.3% of the assemblage, this is considerably more than might be expected in modern pellets, and suggests the presence of weedy and unpolluted backwaters connected with the Thames in Roman times. Today the harvest mouse is a rare component of owl pellets and its presence at the villa suggests that it was more common there in the past. This tiny rodent inhabits tall dense vegetation including reed beds and may be another indicator of the state of the Thames at the time (Corbet and Southern, 1977).

The house mouse (Mus musculus) was absent at the villa: this non-native species was probably introduced into Britain during the Iron Age, but the earliest evidence so far for its presence in Oxfordshire dates to the Saxon period. The presence of two rat jaws in the owl pellet assemblage is problematical. For many years it was thought that the black rat (Rattus rattus), another non-native species originating in Asia, was introduced into this country by returning Crusaders in the Middle Ages; it is now known to have been here in Roman times but so far it has only been recorded from Roman York, Wroxeter and London (Yalden, 1999). The earliest record of the black rat in Oxfordshire is from the Early Medieval period. It is possible, although perhaps unlikely, that the villa owl pellet assemblage had been contaminated by later deposits. There are other rat-sized bones in the assemblage, but these could have come from water voles. The context appears to be secure, so either the owl pellets as a whole were deposited after the Roman period or the rat had spread up the Thames from London, possibly, to reach Oxfordshire much earlier than is generally thought to have been the case. The owl pellet deposit may continue beneath the unexcavated area in the centre of the room in which it was found (Fig. 1). Further investigation of this area may show whether rats were present in the assemblage in situ.

From the point of view of elucidating developments at the Roman villa, the owl pellets are important because they show that this room at least remained standing and roofed after the building had been abandoned by its occupants. The concentration of the pellet deposit in the middle of the western side of this room strongly suggests that a major roof beam crossed the centre of the room in an east-west direction. The presence of fragments of mortar and wall plaster in the matrix surrounding the bones indicates that the building was in an active state of decay while the owls were in residence, and the fact that the owl pellets were sealed beneath a layer of rubble suggests that the building then collapsed or was demolished.

Hazel Williams (2004) has described evidence for change of use in this part of the building over time: the adjacent hypocaust was deliberately in-filled and the presence of hearths in the corridor suggests that this part of the villa became a working area or was perhaps used by squatters after other parts of the building were abandoned. Owls will not take up residence in a building where there is a constant human presence, so how long did the building remain standing after it was abandoned for the last time?


Janet Sharpe

This article is due to be published(summer 2007) in SOAG Bulletin No.61 (2006). The SOAG Bulletin is published yearly and is free to members.