INTRODUCTION
Over the past quarter of a century there has been a growing realisation that items previously categorised as simple rubbish had frequently been deliberately placed, often in rites of termination and with some form of votive or placatory intent. It can be seen that this activity, now known in the archaeological record as structured deposition, was extremely common in later prehistory and the Roman period. Whilst the credit for recognising the phenomenon is often given to prehistorians working in the mid-1990s, with the seminal work of J.D. Hill often being cited,Footnote 1 it should not be forgotten that it was already recognised within a Romano-British milieu at that time even if the current terminology was not used. William Manning was laying out the options for understanding pit deposits as early as 1972, whilst Ralph Merrifield's Reference Merrifield1987 The Archaeology of Ritual and Magic drew on many Romano-British examples. A decade ago Michael Fulford summarised the development of this idea within the Roman scholarly community in this journal, and presented various case-studies drawn from sites in southern Britain to demonstrate it.Footnote 2 Regular elements in a structured deposit include complete vessels and articulated animal carcases, either complete or substantial parts of a beast such as a limb. The presence of dog skeletons is frequently a good indicator.Footnote 3 Carol van Driel-Murray has also drawn attention to the fact that shoes, especially left shoes, often form part of this regular patterning.Footnote 4 Originally the belief that similar impulses were the likely cause for the deposition of at least some hoards of valuable items was a contested area. Now when hoards are recovered as part of controlled excavations — as was the case with the Drapers' Gardens hoard of metal vesselsFootnote 5 — some can be seen to be part of a complex act of such structured deposition. At Drapers' Gardens, for example, the well fill above the vessels contained the incomplete skeleton of a juvenile deer.
Wells, deep pits and watery places seem to have been particular foci for structured deposition, but the question does arise as to whether the interpretation of such items has swung too much in favour of structured deposition at the expense of simple rubbish disposal. Complete vessels will have regularly descended into wells to draw water and it can be expected that some will have been lost through failures of the hoisting apparatus, be that a rope or a chain. Cushioned by water, such vessels may not have broken even if they are made of pottery. Some things would have been rubbish and had to be disposed of, and deep holes in the ground like wells may have been seen as useful disposal locations after they had gone out of use. The filling-in of a well after all is a health and safety issue. If they were not filled in, they would have become hazardous and animals may have fallen into them leaving behind their skeletons. It is generally accepted that the fill of wells may be the result of a range of different activities of which structured deposition is only one.
The material recovered from a well at Rothwell Haigh to the south-east of Leeds posed the question as to what activities were being represented. Although it had many of the classic markers of structured deposition, it was not the presence of any single one or even combinations of them that marked out the fill of the well as reflecting something much more than rubbish disposal. Rather it was the whole assemblage when judged against contemporary material culture use in the area. The types of item being deposited provide an interesting insight into the choices being made and it is likely that the methodology adopted to demonstrate this here will have wider application in other parts of the province, including those areas where material culture is much more prolific. As has been argued previously, special deposits always have to be viewed against the background of what was being habitually used and made by the local community.Footnote 6 Rothwell Haigh provides an excellent opportunity to explore this because there is a wealth of evidence to put the well and its contents in context.
ROTHWELL HAIGH AND ITS REGION
The site investigated was a c. 2,700 m2 square, ditched enclosure with an east-facing entrance, but with only a few internal features, including a 12.3 m-deep well. It was located to the north-east of Rothwell and to the south of the river Aire in what is now Rothwell Country Park (SE 352 295; fig. 1 ), and was identified during reconnaissance flights in the summer of 1977, during topsoil stripping at Rothwell Haigh Colliery. The local bedrock is Middle Coal Measures with glacial gravel and sandFootnote 7 and the area currently lies at about 57 m above OD. Unfortunately the exact location of the enclosure is unknown and the topography of the site pre-coal extraction is unclear. Nevertheless, it lies in an area that has seen considerable archaeological investigation in recent years, often in advance of coal extraction and the building of new roads. The work in advance of the M1–A1 road to the east of the site, for example, uncovered Roman period activity at eight sites.Footnote 8 The villa site at Dalton Parlours lies 16 km to the north, and is particularly useful for comparative purposes since another well and its contents were excavated there.Footnote 9 The fort and vicus at Castleford lie 10 km to the south-east.Footnote 10 These and other sites provide a wealth of information about both the type of occupation and the nature of the material culture in the area.
Site location, with sites mentioned in the text.
Roman era ‘native’ sites are by far the most commonly excavated sites in West Yorkshire, with most attesting to Roman period agriculture. Few contain evidence of dwellings, including wells, and as a result, these enclosures are often associated with livestock management.Footnote 11 Re-use of a later Iron Age rectilinear enclosure at Methley in the late third and fourth centuries, for example, is indicative of small-scale subsistence,Footnote 12 while a corner enclosure at Swillington Common representing the redefinition of a larger Iron Age enclosure was seemingly used for crop-processing.Footnote 13 At Ferrybridge, modified enclosures and field-systems of Roman date have been associated with both livestock management and crop-processing, as well as providing a rare example for the region of a rectangular building of early Roman date.Footnote 14
Further rural sites in the region also occupied in the mid- to later Roman period include two settlement sites near Whitwood, occupied from the late Iron Age and then abandoned before being reoccupied in the later Roman period,Footnote 15 and a rectilinear enclosure, trackway and secondary D-shaped enclosure of second- to early fourth-century date at Apple Tree Close, Pontefract.Footnote 16 A peak in domestic activity was noted at Parlington Hollins during the late third and early fourth centuries,Footnote 17 while at Swillington Brickworks unenclosed or insubstantial enclosures of Iron Age date were replaced by a more extensive enclosure system, probably at the beginning of the third century.Footnote 18 At Bullerthorpe Lane, north of Swillington, a series of small enclosures of Roman date (mid-second to fourth century) were investigated, as well as a rare example of a late Roman roundhouse.Footnote 19 Interestingly, a similarly-sized, sub-square enclosure of Iron Age origin at Normanton, also with an east-facing entrance, continued in use into the Roman period when it became incorporated into wider land divisions. The possibility that this enclosure represents an Iron Age shrine has been raised.Footnote 20 Also significant is a sub-square enclosure, which was clearly occupied in the second to fourth centuries, that has recently been excavated just to the south of Knottingley.Footnote 21 This was of similar size to the Rothwell enclosure, had a north-east-facing entrance and few internal features. The inhumation of an adult female, sitting upright in a pit just outside the entrance and radiocarbon dated in the range 380–130 cal b.c. attests to an earlier presence, and perhaps indicates a non-domestic role for the enclosure.
In contrast, there are very few settlements of this date in West Yorkshire that could be classed as urban, although concentrations of finds suggest minor towns or centres at Leeds and Wetherby.Footnote 22 To the north of Leeds, some 10 km from the enclosure at Rothwell, a fort was constructed at Adel close to the Roman road which ran between the fort at Ilkley and the settlement at Tadcaster,Footnote 23 but there has been little investigation and it is poorly understood. The fort at Castleford constructed in the early 70s a.d., and its associated vicus, has been more intensively studied.Footnote 24 It occupied a strategic position on the rapidly expanding Roman road network, linking the fort at Doncaster to the south, to the forts at Newton Kyme and York to the north.Footnote 25
Generally, the up-take of Roman material culture on the rural sites was low. For example, the Roman sites excavated during the M1–A1 campaign produced only 15 kg of Roman pottery in total. Only two sites revealed any metal findsFootnote 26 and the only item of glass was a bangle fragment.Footnote 27 This paucity of finds, which is the normal pattern here, has to be borne in mind when assessing the finds from the Rothwell well.
After the discovery of the site at Rothwell in 1977, the County Archaeology Unit for West Yorkshire was afforded access and carried out a rescue excavation of the enclosure and its associated features. Exposed features were rapidly excavated over a five-day period in August 1977 to reveal the plan of a 52 by 52 m enclosure with an east-facing entrance ( fig. 2 ). In addition, a limited number of discrete features were investigated, the most significant being the well. Bedrock as well as topsoil had been removed during the stripping and so it is probable that other features had been destroyed prior to excavation. Subsequent investigations, between September and October of the same year, allowed the full excavation of the 12.3 m-deep well ( fig. 3 ). While the excavations were funded by the local authority, and preliminary post-excavation, including conservation, was undertaken, it was unfortunate that the funds were not available at that time to report on the findings. The surviving archive was finally subject to detailed analysis in 2010, when funding was made available by the West Yorkshire Archaeology Advisory Service and it was at that point that the full significance of the deposits was realised. The full report is available on the Archaeology Data Service and the reader is referred there for the detailed specialist reports,Footnote 28 which are referenced in this paper as ‘X in Richardson 2011’ in the footnotes.
Plan of the enclosure and associated features. Scale 1:500.
Schematic section through Well 17. Scale 1:100.
The enclosure ditch survived to only a few centimetres in places, probably because of the unmonitored stripping, although at the entrance terminals of the ditch, depths of 1.5 m and widths of 3 m were recorded.Footnote 29 It is perhaps telling that most of the discrete features were also found in the vicinity of the entrance. The enclosure was fully exposed and the ditch terminals flanking the entrance and a further ten sections across the ditch were investigated ( fig. 2 ). There is no record of any finds being recovered from the enclosure ditch. The ditch terminals were joined by a narrow gully which, in association with three possible post-holes externally and a centrally-placed pit or post-hole just inside the entrance, may have formed part of a gateway structure. Within the enclosure, a sub-rectangular pit (25) and two further gullies were observed. Again artefacts were scarce from the discrete features and gullies, but from Pit 25 a complete beehive quern base was recovered ( fig. 5 , No. 2).
THE WELL
Well 17, within the enclosure, represents the most significant feature exposed during the investigations and fortunately is by far the best recorded. Measuring c. 12.3 m in depth and approximately 2 m in width, the well was emptied in its entirety, though in conditions that would be deemed too high risk today ( figs 3 and 4). The well was cut through bedrock, as was the case with the well at Dalton Parlours,Footnote 30 with no evidence for timber or wattle lining. The conditions were recorded as damp to c. 5.5 m, wet from 5.5 m to just over 7 m, and thereafter, waterlogged. Discrete layers were recorded in section, although the necessary records to describe all of these with confidence are not available. The level of the contemporary ground surface was unclear, although the excavators suggested that as much as 1.5 m may have been stripped prior to archaeological intervention. Unsurprisingly, no evidence for a superstructure or well-house was encountered.
Well 17 during excavation.
Querns. Scale 1:6.
The contents of the well as excavated are described here as fully as records allow from the bottom up. Given the rescue conditions under which it was excavated, and the practical difficulties of emptying such a deep and narrow feature, it is hardly surprising that challenges remained when interrogating the archive. In particular, it proved impossible to identify any near-complete or complete animal skeletons with certainty, owing to slumping, disturbance and/or a failure to identify articulated elements during excavation. Complete skeletons are assumed on the basis of the occasional identification of paired elements, and the presence of the majority of the body parts (in particular of dogs), but assigning depths to specific animal carcases has not been possible. In addition, a small quantity of finds is missing: ‘wickerwork’ from 10–11 m, part of a jet bracelet from 10.4 m, and large structural timbers from 7–7.2 m. The latter may represent the dismantling and disposal of the winding mechanism — certainly structural timbers of the kind recovered from the well at Dalton ParloursFootnote 31 are otherwise absent from Rothwell. Equally it is not possible to say precisely how most items were positioned in the well, for example upside down or on their sides.
From 12.3 to 10.9 m, a black, organic-rich clay containing sandstone rubble was observed. This layer contained little in the way of animal bones. Pottery was present but not in large quantities (less than 1 kg) and there were also fragments of tiles (tegulae and imbrices). Wood fragments, such as twigs and chippings, were few in number but wooden objects were common and included some remarkable items:
-
• Fragments of two turned beakers or cups; one of ash (Fraxinus excelsior L.), but the wood of the other was not identifiable. (11.69 m; fig. 7 , Nos 82 and 100)
-
• A spade with an iron shoe, with the blade made of ash (Fraxinus excelsior L.). The wood of the handle was not identifiable. (11.2 m; fig. 6 , No. 77)
-
• A stave-built complete bucket made of yew (Taxus baccata L.) bound with iron and retaining the lower part of the handle. (11 m; fig. 6 , No. 78)
-
• Part of a curved oak (Quercus sp.) bracket. (11.85 m; fig. 7 , No. 99)
-
• Nine pegs were recovered from between 11.4 and 12.2 m. Of these five were of oak (Quercus sp.) ( fig. 7 , Nos 84 and 87; fig. 8 , Nos 101, 110 and 114), one was of ash (Fraxinus excelsior L.) ( fig. 8 , No. 102), one was probably of willow (?Salix spp. or similar) ( fig. 7 , No. 97), and one was not identifiable ( fig. 7 , No. 90). Two of the oak pieces were turned and retained tenons and so clearly came from multi-piece objects ( fig. 8 , Nos 101 and 114). No. 110 was crudely hewn but also retained a tenon. These pegs are likely to have had a range of different functions.
Wooden objects, Cat. Nos 6, 19, 44, 72, 74, 77 and 78. Scale 1:4, except Nos 77 and 78 which are at 1:8.
Wooden objects, Cat. Nos 82, 84, 87, 90 and 97–100. Scale 1:4.
Wooden objects, Cat. Nos 101, 102, 110 and 114. Scale 1:4.
Some leather was also recovered, including the fragmentary insole of a nailed shoe, probably for the left foot ( fig. 9 , No. 4), and a fragment from a possible one-piece shoe at 11 m.Footnote 32
Leather, Cat. Nos 1, 4, 6 and 11. Scale 1:3.
At 10.9 m patches of clay were observed that clearly divided the deposits below from those above.
From 10.9 to 8.49 m, the same black, organic-rich clay matrix with rubble was present but animal bone and wood was now commonplace, and a human skull was recovered at 9.6 m ( fig. 10 ).Footnote 33 The skull had come from an adult of about 25 years of age, based on dental attrition, but who may have been somewhat older given the complete union of the cranial sutures. The remains showed features consistent with decapitation, and the absence of the mandible suggests it was placed in the well after putrefaction of the soft tissue. Structural timbers (now lost) and off-cuts were noted at this depth, as well as another fragmentary turned cup made of ash (Fraxinus excelsior L) at 8.85 m ( fig. 6 , No. 44), which is not the same vessel as No. 82 that had been found at a lower level. Other wooden artefacts include a perforated disc of oak (Quercus sp.) from between 9.1 and 9.7 m ( fig. 6 , No. 72), and another peg from between 8.9 and 9.6 m but which is now known only from a drawing.Footnote 34
Human skull from Well 17, depth 9.6 m.
It was in this deposit that most of the leather was found, including some off-cuts and non-footwear items such as a thonged calfskin patch ( fig. 9 , No. 11). The footwear recovered here includes fragments of one-piece shoes at 10.7 and 10.3 m, a complete left bottom unit and a fragment of an upper, both of nailed shoes ( fig. 9 , No. 1), and fragments from two one-piece shoes ( fig. 9 , No. 6).Footnote 35 All of these were found at 8.8 m.
A distinct layer of animal bone was observed at 8.4–8.8 m, although bone generally was noted between 8.4 and 10.9 m and, measured by Number of Identified Specimens (NISP), the bulk of it was in the 10.5–9.6 m division. The total for that depth was 1,037 NISP of which 224 were dog and 235 were ovicaprid (sheep/goat). At this level, and up to a depth of 8 m, there is good representation of all the skeletal elements, which suggests that whole carcases may have been present. Below this (10.5 m and lower), the skeletal elements relate mainly to the head and the hooves,Footnote 36 a pattern also observed in the assemblage from the Wilsford Shaft (Wilts.).Footnote 37 The other two main domesticates (cattle and pig) are present but in lower numbers. Of considerable interest is the presence of cats. A total of 44 bones was recorded representing a minimum of five individuals.
Amongst the pottery there was a complete, very large (400 mm diameter) bowl of South Yorkshire greyware found at 9.7 m ( fig. 11 , No. 1). Tile fragments continued to be found, though mainly in the lower part of this deposit. At 10.4 m, two fragments from a tegula with five paw-prints of a small dog were found.Footnote 38 There were also two unusual stones; a large cup-marked boulder found at 10 m and part of the ‘outer shell’ (cortex) of a clay ironstone concretion at 10.8 m.Footnote 39
Pottery, Cat. Nos 1–7. Scale 1:4.
Sealing this material (from c. 8 m) was a substantial stone layer containing roofing stones and ‘building’ stone. The structural timbers were first observed in this layer of demolition material, but as they tended to be positioned near-vertically, they were not lifted until lower levels had been reached. It is tempting to interpret this ‘demolition’ deposit as the dismantling and disposal of the well-house.
Above this, a shallow stone-rich clay layer was observed, sealed by a narrow band of charcoal at c. 7.5 m. Waterlogged conditions persisted above where ‘peat-like’ vegetation was recorded, but relatively little animal bone or wood off-cuts were noted. Conditions changed from waterlogged to ‘wet’ around 7 m, and the number of organic artefacts declined sharply. Pottery and building debris continued to be deposited. Pottery from here includes sherds from a near-complete flanged bowl at 5.5–5.9 m ( fig. 11 , No. 3) and a small beaker at 5.45 m ( fig. 11 , No. 5) which was effectively complete; the rim of the beaker was missing, but the break had been deliberately rounded prior to deposition. Both of these vessels were in Crambeck reduced ware. Just above these vessels, a curious depression or ‘basin’, 0.45–0.5 m in depth, that contained an ash and charcoal layer was noted. Sitting directly on the ash were a few highly degraded fragments of animal bone and two sherds of pottery, a fragment of Nene Valley colour-coated ware and a greyware body sherd. It was noticeable that the pottery assemblage as a whole between 4 and 6 m was dominated by beakers. Ceramic building material became much rarer in this part of the well.
The well was subsequently infilled with clay-rich deposits containing frequent sandstone rubble, with a thin band of burning at c. 2.5 m. A complete but fractured disc quern rough-out (probably a Coal Measures Sandstone) was encountered at 1.65 m ( fig. 5 , No. 3), and a box-flue tile at 1 m. Two slickensided Coal Measure stones were found at 4.1 and 3.4 m.Footnote 40 These are stones of striking appearance with striated and polished surfaces caused by friction along the fault plane. They are the sort of item collected as curios.
Table 1 summarises the distribution of the main find categories. There is some overlapping of the depths depending on how the individual items were recorded. Clearly the principal area of deposition lay between c. 8.5 and 10.5 m with the main focus being in the bottom half of that division. The wood off-cuts and the like continued to be recovered up to the point that the conditions changed from waterlogged to wet, but the leather and the animal bone either stopped or decreased noticeably above the 8.4 m level, and in the case of the animal bone, the dog and ovicaprid element that had been such a major feature of the assemblage below was now lacking.Footnote 41 The change to increasingly aerobic conditions may thus not have had a major impact on what was preserved, since the organic elements ceased to be found in any numbers at some depth below this.
MATERIAL BY DEPTH
The pottery allows the infilling of the well to be dated reasonably closely. Crambeck greywares occur in the lowest fills and provide a terminus post quem of c. a.d. 280. There is an absence of late Crambeck wares and East Yorkshire calcite-gritted wares of the mid- to late fourth century which indicates a terminus ante quem of the mid-fourth century. Holme-on-Spalding-Moor type wares are well represented and, in this area, they generally show an increase in early fourth-century groups. This suggests, that within an a.d. 280 to mid-fourth-century bracket, the optimum date for the infilling of the well lies in the first third of the fourth century. The pottery in the fill showed little chronological progression which argues for a relatively rapid infill period.
It is natural to assume that the well formed part of a settlement within the enclosure, but the colliery stripping destroyed any evidence of structures, if indeed they had been present. The environmental work carried out on a selection of five soil samples representing a spread through the depth of well deposits, however, does call the presence of a settlement into question, at least for the period when this feature was being infilled. Modest-sized assemblages of plant remains included some charred material from cereals (grain and chaff), heather and bracken (perhaps from litter, most likely from stable cleanings), but only the merest trace of plants likely to represent domestic activities (a fragment of coriander seed in one sample). Overall the plant material was dominated by taxa likely to have originated in an area of scrub or even developing young woodland.Footnote 42
Interpretation of the invertebrates was limited by the rather small numbers of insect fossils recovered and by the poor preservation of a good proportion of them.Footnote 43 Nevertheless, there was a range of taxa indicating grassland, and indeed almost all of the plant-associated and ground-living taxa, as well as the dung beetles, may have originated in such a habitat. The proportion of individuals suggesting very foul matter was strikingly high, most likely originating from herbivore dung. Notably no grain pests were present, and with a rarity of ‘typical’, and lack of ‘strong’ synanthropes, it is probable that there were no occupied structures near by. Both the plant and the insect remains thus seem to indicate that the enclosure was not home to a settled community, although it may have housed livestock.
RUBBISH OR STRUCTURED DEPOSITION?
The finds from the well contain all the usual markers of structured deposition. There is the human skull, apparently from a decapitation and possibly old when it entered the deposit. Human remains are not infrequently recovered from both Iron AgeFootnote 44 and RomanFootnote 45 pits and wells. Examples may be noted from Newstead (Borders), Dalton Parlours (W Yorks.), Appleford (Oxon.),Footnote 46 Carrawburgh (Northumbd), London, Northwood (Herts.), and Headington (Oxon.).Footnote 47 As such, it seems likely that the use of human skulls for votive purposes in Roman times reflects a continuation of a pre-Roman tradition.Footnote 48 Nine cases of complete or near-complete human skulls were recovered from pits during excavations at Danebury (Hants.), for example.Footnote 49 Merrifield goes so far as to suggest a link between decapitation and watery places, including wells.Footnote 50
There are also the complete bodies of dogs and other animals. Evidence, along with the regular deposition of pots, that Fulford recognises as ‘common themes’ in the structured deposits of pits and wells.Footnote 51 Animal carcases, often dogs, but also cats, sheep, goats, cattle and horses, are known from similar contexts at Dalton Parlours,Footnote 52 Shiptonthorpe (E Yorks.),Footnote 53 Welton Villa (E Yorks.),Footnote 54 Baldock (Herts.),Footnote 55 Dorchester,Footnote 56 Winchester,Footnote 57 Oakridge (Hants.),Footnote 58 and Staines (Surrey).Footnote 59
There is also the presence of complete vessels of types which would have been unsuitable for drawing water. This depositional practice too is often observed, indicating that objects were relinquished or ‘sacrificed’, despite being serviceable or even valuable.Footnote 60 Items such as a Rhenish beaker at Neatham (Hants.),Footnote 61 a small glass bottle at Stanmore (Middx.),Footnote 62 and a greyware bowl at Skeldergate (York)Footnote 63 are good examples of the range of complete vessels for which there is no functional explanation for why they should be found complete in wells.
In addition to this, however, there is the character of the Rothwell assemblage as a whole that suggests deposition somewhat out of the ordinary was taking place. For instance, there are repeated occurrences of items that are generally unusual within Roman Britain, as well as types of material that are very uncommon amongst the material culture from the area. It is these features that indicate that we are not dealing with rubbish disposal, but with some form of structured deposition, and that much of the cultural and faunal contents of the fill were the result of a careful selection of items.
The animal bone also indicates a structured deposition. The concentration of the bone in the 8.4–10.9 m zone seems to indicate an episode of intense dumping. Rapid deposition is also suggested by a lack of gnawing (less than 1 per cent of the bones) and the presence of articulated remains. It is assumed that the well was redundant when these animal carcases were being introduced, since their putrification would have fouled the water supply. The identifiable assemblage (measured by NISP) is dominated by sheep/goat closely followed by dog, then cattle, pig, cat, and chicken respectively.Footnote 64 Complete skeletons are indicated for sheep/goat, dog, cattle and pig, but unlike the Dalton Parlours well, articulated horse remains were not represented and the species is relatively rare. From the lowest levels, primary butchery waste from the slaughter of sheep/goat is likely, with a kill-off pattern indicative of meat production. A minority of the cattle bones were butchered.
The ovicaprine assemblage is certainly unusual. Differentiating between sheep and goats from their bones is a well known problem,Footnote 65 but on the whole sheep tend to be identified in far greater numbers than goats. At Dalton Parlours no goats could be identified with certainty from the site assemblage though they were present in the well.Footnote 66 Animal bone preservation was unfortunately not good at the sites excavated as part of the M1–A1 road scheme, but the ratio of sheep to goat bone fragments securely identified was 84 to 11 for the Roman period contexts.Footnote 67 Goat meat, other than that from kids, does not appear to have been to the taste of either the late Iron Age or the Romano-British populations.Footnote 68 In the well at Rothwell, by contrast, there was a minimum number of 34 individual beasts that were either sheep or goats of which twelve were definitely sheep and ten were definitely goats. The securely identified NISP measures were sheep 71 and goats 38.Footnote 69 This is an unusual pattern and would imply selection was being carried out, and it is perhaps worth noting that goats were also identified in large numbers at Uley (Glos.), where the unusual preponderance could be attributed to the fact that they were being sacrificed to Mercury who was worshipped there.Footnote 70 At Wattle Syke (W Yorks.) it was observed that the goat bones tended to be identified as partial or complete skeletons that were part of structured deposits, rather than amongst the disarticulated bone,Footnote 71 suggesting that in this area of Yorkshire too they may have been regarded as a suitable beast for sacrifice.
Dogs were very well represented with a minimum of 19 individuals identified. The dogs ranged in height from 274 to 574 mm, which includes some of the smaller Roman breeds. The importance of dogs as a marker for structured deposition has already been noted. There were also 44 bones from a minimum of five cats found amongst the main deposit of animal bone in the well. Cats rarely feature in such numbers in domestic animal bone assemblages, possibly because wild cats would not normally have died within the confines of human habitation and domesticated cats, being companion animals, may have been granted a proper burial as appears to have happened at Silchester.Footnote 72 To place the Rothwell cats in their local context, it may be noted that no evidence for cat was found at Dalton Parlours; whilst at Castleford (W Yorks.), where over 40,000 fragments of mammalian bone were examined, only 18 cat bones were recovered probably from two individuals.Footnote 73 At Wattle Syke over 32,000 fragments of bone were examined and only two cat bones were recovered, probably from a single individual.Footnote 74
The bird bone assemblage also stands out as unusual for a domestic rural site. Most of the securely identified bones clearly came from chicken. A minimum of four were present and there were indications that these had been deposited whole. Additionally there are indications that pheasant and guinea fowl were also present and these are very rare in Romano-British assemblages.Footnote 75 The poor preservation of animal bone on the M1–A1 sites means that local comparanda are not available, but at Wattle Syke the role of chicken can be assessed. There it was encountered infrequently, but when present it was generally acting as a grave good.Footnote 76 This is the pattern most commonly observed on rural sites, where chicken did not form part of the normal diet and is most frequently encountered in ritual contexts.Footnote 77
As can be seen from this examination of the animal bone assemblage, the whole composition, and not just the complete carcases, requires firm consideration as to whether it was being derived from domestic rubbish fortuitously gathered to act as backfill material. The fact that there is what appears to be primary butchery waste at the base of the well sequence does not preclude it being a structured deposit. At Wattle Syke, for example, some of the special deposits consisted of what appeared to be mother and child pairings. There was one goat and kid and two instances of sheep and lamb. In two of these cases the adults had been butchered prior to disposal as if eating the meat was part of the ritual.Footnote 78
The pottery assemblage also shows some unusual aspects, both for a well context and for a general domestic assemblage. Of the complete vessels, a very large bowl found amongst the main focus of deposition had obviously been used repeatedly for a particular function that involved crushing or grinding, since it had a worn base and distinct wear patterns internally over the base and extending about a third of the way up the side. When full this vessel would have been extremely heavy, but it was clearly not a water carrier. Even to position it over 9 m down in its unbroken state must have required considerable effort. Wells in the region normally produce narrow-necked lugged jars in their lower levels and these were clearly used as water carriers. This can be seen in the assemblages from the wells at the Rudston villa (E Yorks.)Footnote 79 and Dalton Parlours where lug-handled jars and Huntcliffe jars predominated.Footnote 80 At the latter site one of the vessels was even found with a cord still passing through one of the lugs. At Rothwell such vessels (see fig. 11 , No. 4) formed less than a quarter of the assemblage stratified in the well and did not occur in the lowest level, appearing first at a depth of 7–8 m, that is to say above the main focus of deposition.Footnote 81 Instead, the pottery assemblage derives mainly from vessels associated with food preparation and consumption, though cooking jars are rare. The beakers and small dishes indicate high-status dining, though it should be noted that these are associated with a level of the fill that is above that of the main focus of deposition.
The scarcity of vessels used to draw water in the lowest levels of the Rothwell well is striking. There is one complete bucket and one stave possibly from another, but there are no disassociated iron fittings. This contrasts with the situation in the well at Dalton Parlours where there were five complete buckets and disassociated staves, base fragments and iron fittings suggesting that a minimum of 16 buckets must be present in addition to all the lugged jars.Footnote 82 Of course not all wells relied on buckets and jars to haul out water, as the discoveries in London with their complex winding gear have made clear.Footnote 83 Some wells that did appear to use buckets produce very few remains. The timber-lined well at Skeldergate, York, for example, had only a single bucketFootnote 84 and no pottery jars. Generally though one might have expected more evidence for buckets, and a greater presence of the wood that the complete bucket was made from among the more regularly occurring species. The wooden buckets in use in Roman Britain tended to be made using silver fir staves, though oak, sycamore, alder and ash also feature in the construction of some.Footnote 85 Yew, the wood used for the Rothwell bucket, is an extremely rare choice. One reason for this may have been that yew wood was regarded with suspicion as possibly being toxic. Certainly Pliny the Elder recounted that yew wood was poisonous, noting that in Gaul deaths had been reported from travellers drinking wine from flasks that had been made of it.Footnote 86 Leaves, berries and the dust from working yew are certainly toxic, but whether the wood itself is toxic is still a matter of debate. Even today wood-turners consider it better suited for furniture and artefacts that will not come into contact with food and drink.Footnote 87 It only seems to have been identified with certainty for use as a bucket in the remains found with the Welshpool (Powys) deposit of metal vessels and firedogs.Footnote 88 In that case the vessel was not a utilitarian bucket, for it had copper-alloy fittings including ox-head escutcheons, and there were hints from the surviving fragments that it may have been finished on a lathe. The nature of the Welshpool deposit is not clear, since it was found by workmen laying a drain. Boon interpreted it as a cremation burial, but clear evidence for that was not forthcoming. Certainly though it was a special structured deposit and a burial assemblage associated with a grave still seems most likely.
The other substantially complete wooden item from the base of the well is a wooden spade with iron fittings, including the shoe and handle attachments. This is an extremely unusual and rare find. When Sian Rees was compiling her corpus of agricultural implements, she found only one wooden spade retaining evidence for its shoe and a total of 70 iron shoes, mainly from what may be termed Roman rather than native settlements.Footnote 89 A few more have emerged since then, including, for example, another complete ash example recovered from Stoney (Cambs.)Footnote 90 and the lower part of an oak spade with shoe at No. 1 Poultry, London.Footnote 91 There are also a few additional spade shoes, including one from Armthorpe (S Yorks.),Footnote 92 but it cannot be said that spades supplied with iron shoes appear to have been at all common. Rees's conclusion that wooden spades continued to be used on most native sites, as during the Iron Age, is a deduction that can probably still be considered correct. The spade in the Rothwell well is, therefore, unlikely to have been a typical example of the spades in use in the surrounding countryside.
Assessing whether wooden items are common or rare is of course problematical because of the preservation problems, but an increasing number of water-logged contexts have been excavated and published and provide a growing body of comparanda. Wooden vessels, including those from Rothwell, have been the subject of an overview by Paola Pugsley as part of her survey of domestic wood, but she found only 60 in total. Spindle-turned drinking vessels of the sort present at Rothwell were very rare and she cites only one other in Britain and three on the Continent. Interestingly two of the latter came from ritual shafts at Montmaurin in south-west Gaul.Footnote 93 So whilst it is normally assumed that there must have been a suite of wooden tablewares in use in the Roman period, these may not have been as common as generally supposed by archaeologists. Interestingly this is a conclusion that Christopher Evans also came to when considering the prehistoric evidence.Footnote 94 Thus, the wooden vessels from Rothwell can be added to the growing list of items in the well that may not have been part of everyday material culture.
The ceramic building material is also an unusual find in this area. In the M1–A1 excavations, it appeared to be used only at Parlington Hollins, a site whose material culture generally suggested a more Romanised lifestyle.Footnote 95 At Rothwell fragments of roof tiles were concentrated in the bottom 2 m of the well and appear to have been near complete when deposited. Of particular interest is the observation that two of the fragments have paw prints of what appear to be the same small dog. Paw, hoof and foot prints often feature in published tile reports because they attract attention. At sites where the entire assemblage has been studied, typically only 0.4 per cent or less will have such prints.Footnote 96 At Rothwell, however, 6 per cent of the small assemblage has prints. At the very least it is a curious coincidence that at least one of the tiles disposed of in the well should happen to have paw prints from the size of dog that dominates the animal bone assemblage. This raises the question as to whether the tile should be regarded as casual debris from demolition or items carefully selected for some purpose. Certainly the box-flue tile fragments from depths of 8.1 m (possible identification) and 1 m (certain identification) raise serious questions as to whether the tiles were being derived from the site. Box-flue tiles, after all, are associated with hypocaust systems.Footnote 97 Such tiles may be expected at a villa site, such as Dalton Parlours, but they did not occur at Parlington Hollins (W Yorks.) and it seems unlikely that an enclosure site such as Rothwell would have had such amenities. It might also be questioned as to whether it would have housed structures that needed a stone with a square mortice hole for a timber, which was found at a depth of 8.7 m, that is to say in the main area of deposition that contains so many undoubted items of structured deposition.Footnote 98 Structural stonework is not generally recovered on sites in the area.
Van Driel-Murray who has studied the leather from the well is of the opinion that the footwear fragments may be interpreted as rubbish rather than any form of special deposition. They do, however, indicate people with access to good quality shoes in the latest fashions. Her conclusion about the assemblage as a whole continues the theme, already traced in the other categories of finds, that the items are not typical of the normal range of rubbish in the locality. She notes that the shoes reflect only a few individuals, all apparently female, and that ‘the restricted size-range within this complex is unusual’.Footnote 99
As well as considering what is present in this assemblage, it is useful to consider what is not. No items of copper alloy were present, which need cause no surprise in this area. Curiously iron objects were very rare as well, other than the fittings of the spade and the bucket and the hobnails of the shoes. Iron nails are one of the few small finds that can be regularly expected on rural sites in this area. Here, with one exception, they were restricted to the upper deposits of the well. The environmental remains preserved provide only modest evidence for food remains, unlike similarly dated wells in York at SkeldergateFootnote 100 and The BedernFootnote 101 that were much richer in foodstuffs. From Rothwell, clear indicators of stable litter and herbivore dung were noted, a pattern repeated at Rudston villa (E Yorks.).Footnote 102
CONCLUSIONS
As the forgoing consideration of the contents of the well has made clear, nearly every aspect of the assemblage has demonstrated unusual features compared with what might be termed normal domestic rubbish from sites in the area. The assumption that the well and the c. 2,700 m2 square enclosure were associated with domestic occupation in the late second to late third century is a possible scenario, despite the scant evidence for the settlement itself. Given the nature of the stripping, it is distinctly possible that evidence for habitations could have been removed, though the lack of the insects normally associated with human occupation might argue against that. Certainly the possibility that this was an enclosure whose principle role lay in the religious realm, such as appears to have been the case at Ashill in Norfolk,Footnote 103 is something that could be considered. The items being brought onto the site, such as the slickensided stones and the cup-marked stone, suggest this was a site where curios were accumulated. Whether this was a settlement or a focus of religious activity, it is clear that arable production was important to the people who used it given the recovery of beehive and disc quernstones in both a finished and unfinished state. Indeed, the recovery of the complete beehive quern base in Pit 25 could be regarded as an act of structured deposition in its own right, given that something that was still serviceable was being disposed of. The unfinished roughout is an unusual find away from a production site and might hint at special attention with regard to the consideration of querns by users of the site.
Understanding the significance or meaning of the varied deposits within the well, however, remains far from straightforward. Fulford's study of wells, pits and their contents concludes that structured deposits were a multi-period phenomenon with its origins in the Iron Age (e.g. Danebury) and continuing through the first century a.d. (e.g. Newstead) to the fourth and fifth centuries a.d. (e.g. Barton Court Farm).Footnote 104 He also recognises that they were found on both rural and urban sites. Fulford proposes rites of propitiation to chthonic deities or to ensure fertility as possible explanations, but acknowledges that such deposits may be ‘accidental/functional’, ‘ritual’, or indeed both.Footnote 105 James Morris, in his recent studies of ‘Associated Bone Groups’ (ABGs), concurs,Footnote 106 but also warns that while human depositional actions may form structural deposits, this does not necessarily mean they were formed owing to ritualised activity.Footnote 107
Despite this uncertainty, rites associated with the opening and closing of wells are widely attested. Merrifield has suggested that complete pots, holding libations, were placed into freshly dug wells as part of a ritual of ‘commencement’.Footnote 108 Once the useful life of the well was finished, a further offering of ‘termination’ would be made prior to backfilling. This would explain the completeness of many of the vessels from wells, while closure deposits could have included animal carcases, winding mechanisms, buckets and other water-carrying vessels, all of which would have effectively terminated the usefulness of the well. Such an explanation was used by Stuart Wrathmell to account for the discarding of intact water-drawing materials down the well at Dalton Parlours,Footnote 109 and ritual closure was also surmised for the backfilling of a well in the mid-second century at Welton Villa.Footnote 110 Such an explanation might be appropriate at Rothwell, for it could indeed have been a working well, but given the relative speed with which it was filled as judged by the pottery, and the lack of any convincing evidence for it having had a long working life before the infill, a possible ritual aspect to both the digging and the backfill has to remain a possibility. As an alternative, we are faced with the scenario where a previously utilitarian working well was thoroughly cleaned out prior to the deposition of the fill found. The logistical problems associated with such an act would have been great, not to mention the safety aspects. As noted earlier in this paper, the 1977 excavations were carried out in conditions that would not be acceptable today. In 2010, for example, excavation of a Roman well in Staffordshire ceased at the 7 m mark owing to safety concerns,Footnote 111 whereas excavation of the Rothwell well continued for more than 5 m beyond that depth. Cleaning it out thoroughly in the Roman period would have had to have been done in the dark, at great depth and with water accumulating rapidly.
This study of the well contents has certainly thrown an interesting sidelight on the material culture in use during the early to mid-fourth century in the region. As previously noted the take-up of Romanised material culture on rural sites in this area tends to be low. Normally a small number of hobnails will be one of the few small finds present. Now we know from the Rothwell evidence that they could well have been attached to shoes of the latest fashion. The well also suggests that apparently mundane items might have been imbued with hitherto unsuspected meaning. Why else bring fragments of flue tile to the site? If nothing else, it suggests that whatever deities or spirits were being propitiated, they were seen as having different requirements than the people making the deposits.
Within the wider field of Roman Britain, it is hoped that we have demonstrated that a careful examination of the more mundane elements of assemblages, where structured deposition may be suggested by the more obvious markers — such as complete vessels and dog skeletons — might well be rewarding. Indeed Morris insists that to identify structured deposits, all categories of material culture must be considered together and not in isolation, as is so often the case.Footnote 112 If the patterns within the material are typical of what is recovered in domestic settings in an area, then an interpretation as rubbish may well be appropriate. If they are repeatedly at odds with the domestic pattern, as at Rothwell, then they probably were not. A more nuanced view of whole assemblages will undoubtedly produce regional differences. It is not to be expected that flue tile will have the same meaning within an early fourth-century southern urban context as it apparently had on this Yorkshire hilltop at that time. By exploring all these differences, both in different regions and between the domestic and the structured deposits in particular areas, it is likely that a better understanding of the motivations that led particular communities to make special deposits will result.
ACKNOWLEDGEMENTS
West Yorkshire Archaeology Advisory Service funded ASWYAS to publish the results of the Rothwell Haigh excavation. English Heritage and the Department of Archaeology, University of York kindly supported the environmental and faunal analyses. Jon Prudhoe drew the site plan and section, pottery, and the wooden artefacts; Ian Atkins produced the location plan. Illustrations of the querns drawn in the 1970s have been reused here. The authors are grateful to Stuart Wrathmell, Ian Sanderson and Ian Roberts for their valuable input and also to the anonymous readers for their comments.
The specialist contributions for the grey-literature report form an integral part of this synthesis. Copies of the full specialist reports, including methodologies, catalogues and illustrations are held with the site archive, which has been deposited with Leeds Museums and Galleries (accession number: LEEDM.D2011.2). The grey-literature report, which includes the full specialist reports, has been deposited with the Archaeology Data Service as part of their OASIS programme. This can be accessed at http://archaeologydataservice.ac.uk/archives/view/greylit/details.cfm?id=9075. Copies of the grey-literature report and the full specialist reports have also been deposited with the West Yorkshire Historic Environment Record and are available for public consultation.
The co-operation of the National Coal Board and the colliery staff in allowing archaeologists to access and excavate features is acknowledged.
