If you have any comments or new information about this record, please email us.

Summary

Archaeological investigations in 2002 and 2005 on each side of Castle Street recorded two parts of a substantial ditch which had previously been noted between Holly Street and Kelvin Close in 1963. This is likely to have been the ditch surrounding the adulterine motte and bailey castle known to have been built in Luton by Robert de Waudari in 1139 but demolished in 1154.

Full Description

<1> Ordnance Survey, Ordnance Survey Archaeology Record Cards, OS: TL 02 SE 23 (Unpublished document). SBD10879.

TL 0906 2082. Castle (site of). 12th century.
Holly Lodge, Castle Street, Luton stands on a mound which may be the site of a castle built in 1139 or soon afterwards. (M.S. Gordon 02/06/1961: F. Davis, History of Luton , 1855, pp. 49, 144; W. Austin, History of Luton Vol. 1, 1928, p.74)
The suggested site is now occupied by a modern bus depot (FDC 12/11/1975)

<2> F W Kuhlicke, Bedford Museum, Annotated OS 6" map (Map). SBD12671.

"Site of Norman castle (Holly Lodge)" arrowed at TL 0905 2082

<3> Luton Museum, Luton Museum Document, 1972 (Unpublished document). SBD10952.

TL 091 208. Site of timber castle, Castle Street (Norman). (J.F. Dyer, etc.)
[Map shows area centred at TL 0912 2082].

<4> Thames Valley Archaeological Services, 2002, Excavation at Castle Street, Luton: Robert de Waudari's Castle?, Sarah Coles (Archaeological Report). SBD11889.

A substantial V-shaped defensive ditch, 6m wide and 3m deep, and of medieval date was found along with a medieval pit. Some Bronze Age pottery and prehistoric flintwork was also recovered . A gully produced a single sherd of Roman pottery.

<5> Council for British Archaeology, 2003, South Midlands Archaeology, Volume 33, Vol. 33, 2003, pp. 17-20 (Serial). SBD14146.

As for reference 4.

<6> Albion Archaeology, 2005, Land at Castle Street/Holly Street, Luton: Archaeological Field Evaluation, 2005/9, January 2005 (Archaeological Report). SBD11890.

A possible boundary ditch was revealed containing very little occupational debris which suggests it represents rural field demarcation.

<7> The Bedfordshire Archaeological Council, 2004, Bedfordshire Archaeology, Volume 25, ‘Excavation at Castle Street, Luton: The site of Robert de Waudari's Castle?’ (Coles), pp. 201-207 (Article in serial). SBD14109.

SUMMARY
A small excavation in advance of development at the former Luton Bus Depot in Castle Street, Luton, produced evidence of a ditch of defensive proportions. Previous observations and documentary sources suggest that this is likely to be a part of Robert de Waudari's castle which was built in 1139 and pulled down in 1154.

INTRODUCTION
The former Luton Bus Depot was located between Chapel Street and Castle Street to the S of Luton town centre (TL 090 207, Fig 1). As part of the planning process prior to redevelopment of the site for residential use, an archaeological evaluation (Coles 2002) followed by the excavation described below were carried out during the summer and autumn of 2002 by Thames Valley Archaeological Services Ltd.
The site lies at approximately 120m above OD, and is located on an area of glacial sands and gravels with an underlying geology of Middle Chalk (BGS 1992). The original ground surface sloped downwards moderately steeply from E to W towards the river Lea in the north. Ground to the east is relatively level.
The evaluation revealed that most of the site had been severely truncated during construction of the bus depot but trenches in a small portion of the eastern side of the site contained archaeological features. At the time it was thought that a recut ditch (1 and 2) and a gully (3) were located. Ditch (1) contained a single sherd of medieval pottery of 12-13th century date whereas ditch (2) contained two sherds of possible Bronze Age pottery. Gully (3) produced one sherd of possible Roman pottery. It was thought that some if not all of these deposits related to the site of a castle recorded in the Borough Sites and Monuments Record.
Following the advice of Bedfordshire County Council's County Archaeological Officer, Luton Borough Council requested a further phase of archaeological excavations to excavate and record the deposits revealed prior to their damage or destruction by new groundworks. The site code is CSL 02/73, and the finds and archive will be deposited with Luton Museum with accession number 2002.313.

ARCHAEOLOGICAL BACKGROUND
Luton's recorded history begins as early as AD792 when it is referred to as Lygetun. Domesday Book (1086) calls it Loitoine (Mills 1991, 228) and it was a royal demesne (Williams and Martin, 1992, 562). The place-name itself means 'farmstead on the (River) Lea' and combines the Celtic river name with the Old English (Anglo-Saxon) tun. There are documentary references to a castle on or near the site excavated and whose presence is remembered in the name of Castle Street. Robert de Waudari came to Luton as a mercenary for King Stephen and had land granted to him. In AD 1139 he built a wooden castle for the war against Matilda but this was pulled down fifteen years later in 1154 under the terms of a truce. A second castle was built by Fulk de Breauté in 1221 close to the modern centre of Luton, to the north of this site. Deeds for properties on Castle Street refer to the mound on which Holly Lodge stood as being the 'castle mound'. A mound was present in this vicinity prior to construction of the bus depot earlier in the last century and was traversed by steps (J Dyer, pers. comm.). One of the boundaries of the property was believed to include the castle ditch (Austin 1927).
In 1963 during the construction of The Luton News printing works on the eastern side of Castle Street a ditch was observed surrounding a raised area thought to be a bailey. The bailey was raised '3 or 4 feet' (c. 1 m) above the surrounding land and the ditch was 10 feet (c. 3m) wide and of an unknown depth. The inner and outer edge had been lined by post holes which supported wooden posts of a double stockade (Dyer et al 1964, 55). No pottery or other dating evidence was found. These observations were carried out on the eastern side of Castle Street, between Holly Street and Kelvin Close, and appear to have located two sides of a ditched enclosure which returned in the direction of the former bus depot (Fig 1). It was assumed that these features belonged to Robert de Waudari's Castle and that the motte was located where Holly Lodge once stood.

THE EXCAVATION
An area of 300 sq m was stripped of 0.8m of overburden by machine under archaeological supervision to expose archaeological deposits located during the evaluation (Fig 2). The machined surface was cleaned by hand but no further archaeological features were found within the area. This relocated the deposits revealed in the evaluation trenches and located an additional pit (8).

THE ORIGINAL CASTLE DITCH (101)
The two ditches comprised a recut (100, excavated as slots 6 and 7) of the original ditch (101, excavated as slots 9 and 10). Neither ditch terminated within the excavated area. They were aligned NWSE, and had been truncated by modem disturbance to the W, but continued beneath Castle Street to the E. Two slots were excavated and showed that ditch 101 was 6.10m wide and 2.90m deep and cut the natural chalk (Fig 3).
The infill comprised seven layers (60-63, 68, 70-75). The primary fill was a brownish white clayey chalk (75), which slumped from the NE edge. Fill 71/63, a yellowish-brown clay, is the only deposit that slumped from the SW side and is sealed by 74 a pale brownish-white chalk layer. Layer 72/61 comprised a mottled brown clayey chalk layer whereas subsequent layers (73, 70/62, 68) were all whitish-brown silty clay deposits. Fill 73 contained two pieces of 12th-13th century sandy ware and deposit 68/60 produced one piece of 12th-13th century pottery.

PIT 8
One of the upper layers (61) of ditch 101 was cut by pit (8). This in turn was cut by the later ditch recut (100). The oval pit contained a brownish cream clayey chalk (69) and was 1.70m x.0.40m wide and 0.45m deep. It contained three sherds of pottery dating to the 12th-13th century.

DITCH RECUT 100
This ditch was slightly narrower in width than the original ditch and was 6.00m wide but only 0.80in deep. It contained four fills 56/64, 57/65, 58/66, and 59/67. The primary layer 59/67 was a mottled light brown and white silty clay chalk layer and contained one piece of possible Bronze Age pottery, which was residual. This was sealed by a light brown clay (58/66) and which also contained one piece of probable Bronze Age pottery. A white brown clayey chalk layer (57/65) lay above 58/66, and below 56/64, which was a brown clay (Fig 3). As for the original ditch, the infilling had largely originated from the north eastern side

GULLY 102
This feature was aligned N-S and terminated just short of the lip of ditch 101. It was 9.40m in length, 0.30m wide and OA 5m deep. Despite its near full excavation (90%) and sieving of soil samples, no further finds were recovered to add to the single sherd of Roman pottery found during the evaluation. It is possible that this feature is of Roman date but its association with, and apparent respect for, the large ditch (101) suggests that it might be of medieval date but containing residual material

FINDS POTTERY
Paul Blinkhorn
The pottery assemblage comprised 18 sherds with a total weight of 2 I3g. The majority of the pottery was early medieval, although a single sherd of St. Neots-type ware, of late Saxon or Saxo-Norman was also noted, as were four small sherds of a crude, flint-tempered ware which appears most likely to be prehistoric. Where appropriate, the codings and chronology of the Bedfordshire Ceramic Type Series were used, as follows:

B01 : St. Neots ware, 9th 12th C. 1 sherd, 6g.
C01 : Sandy ware, 12th 13th C. 3 sherds, 147g.
C03: Fine sandy ware, 12th 13th C. 2 sherds; 22g.
C03A Fine sand and flint, 12th 13th C. 5 sherds, 7g. A fine dark grey fabric with few visible inclusions. Highly abraded, with surfaces missing, is of possible Roman date (1 sherd, 1g). A fabric tempered with moderate to dense angular white flint up to 2mm probably of Bronze Age date (6 sherds, 30g) was also noted.

The pottery occurrence by number and weight of sherds per context by fabric type is shown in Table 1.
All the pottery was plain bodysherds, with the exception of a single sherd from the rim of a jar in feature 10, context 73. The form and fabric are typical of the early medieval traditions of the region. The presence of the St. Neot's ware sherd may be evidence of pre-Norman occupation, but the material had a use-life which extended into the post-conquest period, and thus may be contemporary with the medieval wares.
[Table 1 Pottery occurrence by number and weight (in g) of sherds per context by fabric type]

STRUCK FLINT
Steve Ford
A small collection of 15 struck flints were recovered. It is possible that one or two pieces are accidental by-products of the ditch digging in Medieval times but the majority are clearly of prehistoric date. The collection comprises 12 flakes, a scraper, a possible scraper and a spall. Two of the flakes may have been retouched and one, with a heavily crushed striking platform is from a broken hammerstone. None of the pieces are closely diagnostic and only a broad date range of Neolithic through to the end of the Bronze Age can be suggested.

ANIMAL BONE
Sian Anthony
A single piece of well preserved cow mandible (8g) with no teeth in situ was the only bone recovered. It came from ditch recut 100 (7, 66).

METALWORK
Sarah Coles
A single iron object (80g), probably part of a horseshoe, was found during the evaluation in ditch recut 100.

CARBONIZED PLANT REMAINS
Seven 40-litre samples were sieved for the recovery of small artefacts and charred plant remains. These samples contained a very little wood charcoal but no other charred plant remains, nor artifacts

CONCLUSION
The fieldwork described above has confirmed the extension of deposits to the western side of Castle Street following the observations made during the building of the Luton News printworks on the eastern side of the Street in 1964. Despite the severe truncation of the ground during construction of the bus depot, the project has allowed the investigation of the remaining archaeological deposits, their characterisation and dating.
The earliest material from the site is of prehistoric date with six sherds of possible Bronze Age pottery and 15 struck flints, all of which were residual in later features. Other than pointing to prehistoric occupation somewhere in the vicinity little else can be said about this material.
The single Roman sherd of pottery would also be dismissed as of minimal interest were it not that it is the only datable find from gully 102. Although this gully respects the line of the main (medieval) ditch and the sherd could easily be residual, it is possible that it is of Roman and not medieval date as believed.
The main feature on the site, is a V-shaped ditch over 6m across and nearly 3m deep, clearly of defensive proportions. That the feature is unambiguously of medieval or later date rules out the possibility that it is a prehistoric ceremonial monument such as a Neolithic henge or a defensive feature of some other, earlier period. The convergence of various strands of evidence, including the size of the ditch, its date, the historic documentary references and the more recent references to the presence of a mound, can be taken to confirm that this feature is indeed a part of an Anglo-Norman castle, most probably the ditch surrounding the bailey. That the line of the ditch observed in this excavation does not match the projection on Figure 1 (based on Dyer et al., 1964) need not cause undue concern given the conditions under which the previous observations were made, and the uncertainties over projecting the precise line in any case. There can be little real doubt that the ditches recorded in 1963 and 2002 were parts of the same plan.
The typical form for early Norman defensive works is known as a 'atone and bailey'. This consisted of a steep, artificial mound (motte) surrounded by a ditch, and with an adjoining larger flat area (bai ley), also usually ditched. Both would also be provided with a palisade and often a rampart would flank the ditch. The motte was a defensive strong-point, usually more-or-less circular, whilst the bailey was usually more rectangular and housed stables, smithies, and stores. Many mottes were intended only as temporary fortifications (for example, several shown on the Bayeux Tapestry were clearly not permanent) but many were also originally, or were quickly replaced by, stone keeps. In these cases the bailey might become a focal point for the surrounding settlement and a town's development could be profoundly influenced by the lines of the bailey. In other instances the castle was sited purely for defence and had little influence over the development of the settlement under its protection (Laing and Laing 1996).
The infill of the original ditch (101) and the subsequent recut (100) are mostly derived from the north-eastern side and suggest slumping from a mound or rampart located there. The infill is not uniform and comprises thick alternating bands of silt followed by chalk. This could represent different origins of the mound/bank material with the silty deposits reflecting the slump of turves which formed a core, with the chalk reflecting slump from the upcast from the construction of the ditch.
It is tempting to assign the infill of the ditch to the documentary reference to the castle being pulled down in 1154. The meagre dating evidence from the fills is insufficient to provide firm support for this date, but certainly does not contradict it. The discrete nature of the banded fills suggests that this is a product of periods of natural slumping rather than deliberate slighting and backfilling in a single episode, when a more homogenous fill might be expected. That such a large ditch had become infilled so quickly such that it was subsequently cut by a pit also of medieval date perhaps points to a catastrophic and rapid collapse of a bank or mound following the removal of a retaining structure. Unfortunately no evidence was recovered for a retaining structure such as a revetment or palisade, as the relevant adjacent areas of the ditch had been disturbed by modern development.
The subsequent recut (100) also of medieval date, ceramically indistinguishable from the original, is puzzling. Despite a width almost as great as the original ditch it was much too shallow to be considered a replacement defensive structure. Its function therefore is unclear. It is possible that this feature is a property or field boundary and that by this time the remnants of the castle had become a convenient setting-out point in the landscape. Assuming the motte survived, perhaps the bailey had already become a conspicuous element in the local topography, worth redefining, even if it had lost the necessity for defence.
Despite the short life recorded for Robert de Waudari's castle, and the lack of evidence recovered here beyond the ditch, it should not be assumed that a timber castle (assuming that this is what it was) was necessarily intended to be short-lived or was in any sense inferior to a stone one, at least originally (Higham and Barker 1992). However, in this instance, it may well have been intended primarily as a campaign fort rather than a lasting visual symbol of lordship, or the lord's residence: the location seems inconvenient (as tacitly attested by the relocation when a new castle was built) and the context of building suggests it was a stop-gap measure. Recent archaeological attention has been increasingly directed at the elements of castle-building and the functions of castles beyond defence, but it is worth recalling that defence was their primary purpose (Thompson, 1995). Sites such as this, abandoned very quickly after construction, emphasize this point, and a recent call (Coad, 1994, 221) to study such abandoned sites as closely as the more obviously 'successful' castles is welcome. Although the evidence here is meagre, it is a start.

[Acknowledgements]

BIBLIOGRAPHY
Austin, W, 11927, 'The History of Luton and its Hamlets'. British Geological Survey, 1992, British Geological Survey, 1:50 000, Sheet 220, Solid and Drift Edition.
Coad, J, 1994, 'Medieval fortifications and post-medieval artillery defences; developments in post-war research and future trends', in (ed.) Vyner, B, Building on the past, 214-27.
Coles, S, 2002, 'Former Luton Bus Depot, Castle Street, Luton, Bedfordshire: an Archaeological Evaluation', Thames Valley Archaeological Services Report 02/73.
Dyer, J, Stygall, F, and Dony, J, 1964, The Story of Luton.
Higham, R, and Barker, P, 1992, Timber Castles.
Laing, L, and Laing, J, 1996, Medieval Britain, the Age of Chivalry.
Mills, A D, 1991, Oxford Dictionary of English Place-Names.
Planning Policy Guidance 16, 1990, Archaeology and Planning, Department of the Environment Planning Policy Guidance 16.
Thompson, M W, 1995, 'The military interpretation of castles', Archaeological Journal 151, 439-45.
Williams, A, and Martin, G FI, 1992, Domesday Book.

<8> Bedfordshire County Council, 1970 - 2000s, HER Photograph Archive, P1010042 and P1010049, 2002, by Lesley Ann Mather (Beds. C. C. Archaeological Officer), site visit (Photograph). SBD10506.

Two photographs of substantial ditch section as excavated by Thames Valley Archaeological Services Ltd. in the summer and autumn of 2002.

<9> The Bedfordshire Archaeological Council, 2010, Bedfordshire Archaeology, Volume 26, Abrams and Shotliff, pp. 387-404 (Article in serial). SBD14110.

‘The remains of Robert de Waudari’s adulterine castle, Castle Street, Luton’
Joe Abrams and Drew Shotliff
SUMMARY
An adulterine motte and bailey castle was built by the mercenary Robert de Waudari at Luton in 1139. It was only in existence until 1154 when it was demolished under the terms of the Treaty of Winchester, signed the previous year by King Stephen and his agreed successor, Henry of Anjou. In 1963 construction works on the Luton News printing works revealed the remains of a large ditch south-east of Castle Street. It was suggested that this represented part of the bailey ditch of Robert de Waudari’s castle. Recent redevelopment of the site has provided a second opportunity to study these remains through open area archaeological excavation.
The excavation has confirmed the existence of the castle ditch and an attempt has been made to map its known, and putative, extent. The topographical reasons behind the castle’s location are discussed and the later history of this part of Luton is considered.

INTRODUCTION
As part of their redevelopment at Luton’s Holly Street and Castle Street, Bellway Homes’ commissioned Albion Archaeology to carry out a programme of archaeological investigation, consisting of watching brief, trial trenching and open area excavation. All work was carried out between 2004 and 2005 (Albion Archaeology 2005a, 2005b) in accordance with a project design prepared by Albion Archaeology (2004), in response to a brief issued by the County Archaeological Officer of Bedfordshire County Council (2004a, 2004b). The project archive will be deposited with Luton Museum, under accession number 2005/48.

SITE LOCATION AND DESCRIPTION
The development area is located c. 500m south-west of the historic core of Luton, bordered by Castle Street, Holly Street and Latimer Road (Figs 1 and 2). It is c. 0.66ha in extent and centred on grid reference TL 0920 2075. It lies at an average height of c. 115m OD.
The historic core of Luton lies on the south side of the Lea where it developed around a series of crossing points over the river (Fig. 3). In contrast to this valley bottom site, the development area lies higher up the valley side where only a relatively thin deposit of terrace gravel seals the underlying chalk (Fig. 4).

ARCHAEOLOGICALAND HISTORICAL BACKGROUND
Remains dating to the prehistoric, Roman, Saxon, medieval and post-medieval periods are known to exist within this part of Luton (Albion Archaeology 2003). However, on this particular site, interest mainly focussed on the possibility that the development area preserved part of Robert de Waudari’s early medieval, adulterine castle. Robert de Waudari himself was a mercenary who fought for King Stephen against Matilda, during the Anarchy, a period of civil war and unsettled government that characterised Stephen’s reign (1135– 1154). The king granted both the manor and church at Luton to de Waudari, who subsequently built a castle in 1139 on the strategically important south-western approach to the town (Figs 1 and 3). A large number of castles were constructed during this period; they are collectively known as adulterine, i.e. unlicensed.
Such castles were deeply resented by sections of the population, as this extract from the Anglo-Saxon Chronicle for the year 1137 (de Waudari’s castle was built just two years later) illustrates.
‘And they filled the whole land with these castles. They sorely burdened the unhappy people of the country with forced labour on these castles. And when the castles were made, they filled them with devils and wicked men’ (Austin 1928, 74; Brittania.com website).
In the Treaty of Winchester in 1153, when Stephen recognised Henry of Anjou as heir to the English throne, it was agreed that all adulterine castles would be destroyed (Higham and Barker 1992, 132). Many of them, including de Waudari’s, which had only been in existence for fifteen years, were demolished.
The reasoning behind the association of this part of Luton with the site of de Waudari’s castle is set out by Austin (1928, vol 1, 74–75 and vol 2, 81–82). Late 18th-century title deeds for a piece of land adjacent to Holly Lodge, which formerly stood on Castle Street immediately to the south of St Mary’s RC Church, state that “the castle ditch” formed part of the boundary of the plot (Figs 5 and 6), which was itself referred to as “Castle Ditches Close” (see also Kennett 1976, 18). The title deeds of Holly Lodge itself state that it stood on an ancient enclosure called “the castle close”. Austin suggested that the mound on which Holly Lodge was built might have been the remains of the motte of de Waudari’s castle. (In the early 12th century the style and materials with which castles were constructed varied by region. Here, they typically took the form of a motte, or mound, raised from material taken out of a ditch which defined a more extensive bailey, or courtyard. The motte was surmounted by a fortified, two or three storey house, surrounded by a timber palisade (Salter 2000, 5).
It has never been possible to test this hypothesis by archaeological investigation. A possible earthwork was still present on the site around 1948. It took the form of a low irregular mound, which supported a number of trees and a modern one-storey building which had replaced the 19th-century Holly Lodge (demolished in the 1930’s). Over the next decade, the site was incrementally incorporated into the Luton Bus Depot and the mound was gradually levelled (Fig. 7). Since then the site has been completely redeveloped and all traces of the mound removed. However, if castle ditches were still identifiable in the late 18th century, it is certainly feasible that the motte of de Waudari’s castle could also have survived. Unfortunately, it will probably never be known for sure if the Holly Lodge mound can be identified with the motte of the castle.
Archaeological investigation has provided some physical evidence for the location of de Waudari’s castle. In 1963, the western end of the present development area was redeveloped. The remains of a substantial ditch, cut into the underlying chalk, were observed. It was suggested that this feature might represent part of the bailey ditch of de Waudari’s castle (Dony and Dyer 1975, 40–41). More recent work to the west of the development area, during redevelopment of the former bus depot, also led to the discovery of a substantial ditch which produced a small quantity of 12th–13th century pottery (Fig. 2). Again, it was concluded that this ditch formed part of de Waudari’s castle, although a note of caution was sounded about the dangers of drawing over-ambitious conclusions from relatively slight archaeological evidence (Coles 2005, 206).
However, on the balance of probability, this combination of historical and archaeological evidence leads to the conclusion that Robert de Waudari’s castle did indeed straddle this part of present-day Castle Street. It is, therefore, argued that the large ditch running across the western end of the development area, described in this article, represents the south-eastern corner of the castle’s bailey ditch.

RESULTS OF THE INVESTIGATIONS
PHASE 1: EARLY MEDIEVAL CASTLE DITCH AND ASSOCIATED QUARRYING
The south-eastern corner of the bailey ditch of Robert de Waudari’s castle (L1) (Fig. 2) represents the most significant archaeological remains encountered in the development area. This part of the ditch became infilled in three distinct episodes, largely as a result of erosional processes but, perhaps, partly as a result of deliberate human action. It was never re-cut but did survive as a distinct depression in the landscape until well into the post-medieval period. The ditch was finally levelled before the construction of various buildings and walls in the 19th and early 20th centuries. Following extensive sampling, only a small quantity of charcoal was retrieved from its fills.
The ditch partly truncated a quarry pit (L1), which had been rapidly backfilled and is assumed to have been broadly contemporary with the construction of the castle.

The castle ditch L1
A c. 48m length of the curving castle ditch was recorded. It was aligned WNW–ESE for c. 8m and then turned NE–SW for the remaining c. 40m (Fig. 8). To the NE its surviving extent had been greatly reduced by modern truncation (Fig. 9, Section 3). Where best preserved, it was 4.5m wide and 2.3m deep (Fig. 10, Section 7; Fig 11). The sides had a c. 45 degree slope and the base varied from V-shaped to slightly rounded. A c. 0.4m thick deposit of chalk and silt (G1.00) in the base of the ditch represents material eroded from the sides shortly after it was dug. Environmental sample (7) taken from this deposit produced only occasional flecks of charcoal.
Gradual erosional infilling of the ditch continued with the progressive accumulation of a further c. 0.6m of material (G1.01) in its base. This included both chalk and silt derived from the sides of the ditch and soil and organic matter from the surrounding land surface. Although it is assumed that the castle was occupied at this time, no attempt appears to have been made to maintain the ditch through re-cutting. However, its defensive qualities were probably not significantly impaired: it was still steep sided and at least c. 1.7m deep.
The next series of deposits within the ditch (G1.02) differed slightly from the gradually accumulated, initial silting. They were more massive in character and appear to have accumulated more rapidly. They may represent more extensive collapse of the upper part of the ditch. However, it is also possible that they represent deliberate partial infilling of the ditch, associated with the slighting of the castle following the Treaty of Winchester. Unfortunately, no dating evidence was recovered from these deposits to substantiate this possibility. Environmental sample (6) from these deposits produced only very small amounts of very fragmented charcoal.
The final non-modern series of deposits within the ditch (G1.03) were again slightly different in character (Fig. 10, Section 7). They included relatively wide bands of compacted, clean chalk, which may represent deliberate infilling and levelling. By this time, the ditch was only c. 0.7m deep and would certainly have ceased to perform a defensive function. Once more, no datable artefacts were recovered from these deposits and it is impossible to know whether or not they were associated with the 12th-century slighting. Similarly, environmental sample (1) also produced only very small amounts of very fragmented charcoal.
The overall paucity of artefacts and ecofacts from all the deposits within the castle ditch is in itself significant. The small quantities of fragmentary charcoal probably represent debris blowing around the site, from activities possibly at some distance from the ditch. The absence of artefacts in the rapidly accumulating initial fills (G1.00, G1.01) is not unexpected. However, their absence from the great bulk of the ditch fills (G1.02, G1.03) is more surprising and may be attributable to the fact that the site was some distance from any settlement or human activity of any kind.

Quarrying L1
Quarry pit G12 was only recorded in section; it was truncated by the castle ditch and a modern intrusion. With gently sloping, straight sides and a flat base, it survived to a depth of 0.90m. Significantly, it had been cut into the underlying chalk, presumably to quarry building material. Its mixed fills suggest it may have been rapidly backfilled. It is assumed to have been broadly contemporary with the excavation of the castle ditch.


PHASE 2: MEDIEVAL / EARLY POST-MEDIEVAL QUARRYING
Twelve intercutting quarry pits (L2) were dug on the site of the former castle (Figs 10, 12, 13). Several of them truncated the uppermost deposits within the infilled ditch (Fig. 10, Sections 7 and 9). They may have been dug along the edges of the ditch because it afforded easier access to the underlying chalk. They varied in size but the biggest was at least 4.5m long, nearly 2m wide and 1.20m deep.
The quarry pits were infilled with a mixture of re-deposited chalk and silts, which did not produce any artefactual material. An environmental sample (4) from the fill of one of the pits produced several charred cereal grain fragments including two tentatively identified as oat (cf. Avena sp.); the other two cereal fragments were too poorly preserved for further identification. There was also a trace amount of very fragmented charcoal. These remains are interpreted as background debris blowing around the site, possibly from cooking accidents, which may have taken place at some distance from the quarry pits.
A single shell of blind (or agate) snail Cecilioides acicula was recovered. This subterranean species is common in lowland southern England, occurring at depths down to 2m or more in unwooded calcareous places, particularly rock crevices, cracks and rootlet holes in well-drained soils. It commonly occurs in the cavities of buried bones (Kerney 1999, 168) and is commonly recovered from archaeological sites.
Despite the absence of dating evidence, a broadly medieval / early post-medieval date seems probable for these quarry pits. They were relatively small in scale, compared, for example, to the gravel pits shown in the vicinity on the first edition 25-inch OS map (Fig. 5). If they were late post-medieval or modern in date, they would almost certainly have contained brick or tile, which are generally ubiquitous on urban sites of this date. Worked-out quarry pits are also often used for the disposal of rubbish. Its absence here suggests that the pits were dug in what was still essentially a rural location, some distance from the limits of the town.

PHASE 3: LATE POST-MEDIEVAL
Isolated features
Two sub-square postholes (L3, G10) were recorded cutting a layer of buried subsoil in the eastern part of the development area (Fig. 14). They were 1.50m apart and were similar in size (0.25m deep and 0.55m wide) and shape. They contained identical deposits, which produced a post-medieval tile fragment and a 17th–18th century pottery sherd (glazed red earthenware, fabric type P01, as defined in the Bedfordshire Ceramic Type Series, held by Albion Archaeology). They are likely to have formed part of a fenceline, possibly demarcating property boundaries in the post-medieval period. A cluster of three postholes and an associated gully (L3, G21) cut the backfilled quarry pits on the south-eastern corner of the former castle ditch (Fig. 14). The postholes were 0.75m deep and 0.25m to 0.40m wide, and were similar in shape. The gully was 0.80m long, 0.22m wide and 0.14m deep. All these features were infilled with a very similar deposit. They may represent elements of a short-lived structure, built during the late post-medieval period.

Final levelling of the site of the castle ditch
A varied collection of nineteen post-medieval/ modern layers (L8, G23) represent a significant levelling episode across the development area (Fig. 15, Section 2; Fig. 9, Section 3; Fig. 10, Section 7). They had a combined maximum depth of 0.90m and incorporated significant quantities of crushed chalk. They were probably associated with the 19th century expansion of Luton. What had until then been a marginal, possibly derelict, piece of land was incorporated into the suburban residential and industrial town, as it spread south-westwards from its historic core.
The significance of these deposits is that they levelled a landscape, pockmarked not only by former quarry pits but also by the castle ditch, which even at this late date still survived (at least in its south-eastern quarter) as a c. 0.4m deep linear depression (Fig. 9, Section 7).

PHASE 4: UNDATED REMAINS IN THE EASTERN PART OF THE DEVELOPMENT AREA
A significant characteristic of this part of the site is that it contained archaeological remains associated with use of the area in the pre-modern era before it was subsumed within the rapidly expanding town. Modern truncation was severe but tended to be localised. In places, extensive deposits of subsoil associated with the former ground surface survived beneath more modern levelling. In themselves, the archaeological remains were disparate and difficult to interpret. However, their very presence is important because it indicates that even previously developed parts of the town can still retain considerable archaeological potential.
Two parallel ditches L4 were recorded on a NE– SW alignment (Fig. 14). The bigger and better preserved of the two (G2) was 17m long, 1.2m wide and up to 0.7m deep (Fig. 15, Sections 1 and 2). Both ditches were sealed by subsoil. The only artefactual material for dating, however, was a small, abraded sherd of sand and grog tempered Iron Age pottery (fabric type F09) and a 2nd-century Roman flagon rim sherd (fabric type R03B), which are both likely to be residual. Four residual worked flints of probable late Neolithic/early Bronze Age date were recovered from ditch G2.
Two environmental samples (2) and (3) from ditch G2 produced a small number of charred cereal grains (see Appendix 1). These included four wheat grains, one of hulled wheat, either emmer or spelt, and one possibly belonging to free-threshing wheat. One grain tentatively identified as barley was recovered from sample (3) while the presence of an awn fragment in the same sample also pointed to the presence of oat. A charred fragment of hazelnut shell was also retrieved from sample (2). Small amounts of very fragmented charcoal were also present.
The grains may have become accidentally charred during the latter stages of crop-processing, for instance while being dried before storage or milling or de-husking, or possibly during cooking of whole grains. The small amount of remains suggests that these activities may have taken place at some distance from the ditch. It is possible that the emmer/spelt grain may be residual, like the sherd of Iron Age pottery. Hulled grains, particularly spelt, are common in prehistoric contexts and in the Roman period, with free-threshing wheat becoming the common wheat grain in the post-Roman era (Greig 1991).
Sample (2) also produced a small number of shells of blind or agate snail and a single shell of glass snail Aegopinella sp. This is a common and catholic species found under ground litter in a wide variety of sheltered places (Kerney 1999, 142).
It may be significant that both ditches are aligned perpendicular to Holly Street, and parallel to Castle Street. This suggests that they represent the remains of minor land divisions, which were in use during the post-medieval period (and possibly the medieval period or even earlier). The small quantities of artefactual or ecofactual material within them suggest that they were located some distance from any contemporary settlement.
Two relatively small pits L5, similar in size, shape and sterile fill, were located close to ditch G2 and are likely to have been associated with it. Again, the absence of any domestic refuse in the pits reinforces the conclusion that the whole of this area, including the site of the former castle, lay at a considerable distance from any settlement.

DISCUSSION
PHASE 1: ROBERT DE WAUDARI’S CASTLE
Despite extensive disturbance from the piled foundations of the former Luton News printworks, almost 50m of the bailey ditch of Robert de Waudari’s castle survived at the western end of the Bellway Homes’ development area. Where best preserved, it was 4.5m wide and 2.3m deep. Its curvature suggests that it represents the south-eastern corner of the bailey. No internal features associated with the use or occupation of the castle, nor any traces of an internal or external palisade were identified.
The sequence of layered chalk and silt within the ditch suggests that it became infilled in at least three distinct episodes. This appears to have been largely as a result of erosional processes but, perhaps, partly as a result of deliberate human action. No datable artefacts, which might have allowed some of these layers to be related to the 1154 demolition of the castle, were recovered. Similarly, extensive sampling only produced a relatively small quantity of fragmentary charcoal. This almost complete absence of evidence for human occupation is surprising and may suggest that this part of the bailey ditch lay at some distance from the main focus of habitation within the castle.
The ditch was never re-cut but did survive as a distinct depression in the landscape until well into the post-medieval period. Still at least 0.4m deep, it was finally levelled before the construction of various buildings and walls in the 19th and early 20th centuries.
The recent archaeological investigations have significantly improved our understanding of this part of de Waudari’s castle. Limited observations, in far from ideal conditions, were made on the same site in 1963 during construction of the printworks. As a result, it was postulated that the north-eastern corner of the bailey ditch had been identified (Dony and Dyer 1975, 40). However, no evidence for this right-angled corner was found during the recent work. It is, however, now clear that the south-eastern corner of the bailey ditch lay within the development area, rather than beneath Kelvin Close, as previously surmised. More significantly, the curved alignment of the ditch suggests that the bailey encompassed a far greater area than the previously estimated 3,360sqm (see below).
The recent work has also helped to elucidate the results of the 2002 archaeological investigations within the former Luton Bus Depot on the opposite side of Castle Street (Coles 2005). As Figure 2 demonstrates the c. 6m length of ditch identified on that site lines up perfectly with the ditch identified within the Bellway Homes’ development. It was slightly better preserved (6.1m wide and 2.9m deep) but is undoubtedly the same feature.
The two pieces of work also produced some contrasting results. In 2002 a shallow re-cut was identified. Its function was uncertain but it was not considered to be associated with the use of the castle as a military stronghold (Coles 2005, 207). This feature was not recorded within the Bellway Homes’ development, although it may have been destroyed by the greater degree of truncation. A small assemblage of finds (12th–13th century pottery, part of a cow mandible, and a fragment of probable horse shoe) was also recovered in 2002, although again sampling produced only a very small amount of charcoal. This minor difference may reflect the fact that this part of the ditch was closer to the motte (with, it is assumed, its associated fortified dwelling) and was, therefore, more likely to have been used for the casual disposal of domestic refuse. An alternative explanation for the paucity of finds from the ditch as a whole is that the castle was not permanently occupied, a hypothesis that could only be tested through further archaeological investigation.
The most significant result of the recent archaeological investigations is that (when combined with the 2002 work) the precise location of almost 80m of the castle ditch has been established. Given the curvature of the ditch, it has also been possible to put forward a new suggestion for the size and shape of the castle. The projected line of the ditch shown on Figures 2, 5, 6 and 7 takes in the Holly Lodge mound, which as discussed above might be identified with the castle’s motte, and allows for the fact that late 18th-century title deeds refer to the castle ditch in relation to the boundaries of a property to the north of the site of Holly Lodge.
The projected line of the ditch is highly speculative and the area defined by it does not represent an attempt to map the actual location of the castle. However, at c. 115m long, c. 90m wide and c. 1ha in extent, it is considered to be a reasonable approximation of the minimum size of the castle. It may have been considerably bigger. For example, it is possible that Chapel Street was laid out to go round the site of the castle. It is even possible, although perhaps unlikely, that the identified length of ditch actually surrounded the motte and that a far more extensive bailey awaits discovery. More precise mapping of its actual shape and extent could only be achieved through further archaeological fieldwork. The fact that the ditch has already been identified on two sites subject to repeated redevelopment suggests that this may well be achievable, even within the context of a heavily built up urban centre like Luton.
The siting of Robert de Waudari’s extra-urban castle (c. 200m south-west of the historic core of Luton) was without doubt purely short-term and military in nature. It occupied a significant, strategic position, straddling the London to Bedford road. From a ridge of high ground, it overlooked the medieval town, which had developed in the valley bottom around a series of crossing points over the River Lea. It was designed to overlook the landscape, with clear defensible vistas and to control the southern approaches to the town.
Although the castle was short-lived, it should not be considered a “failed” castle. Equally, it is misleading to compare the site unfavourably with that of Fulk de Breaute’s castle in the centre of the medieval town (Fig. 3). Built next to the church more than half a century later in 1221, the latter appears to represent a re-fortification of a pre-existing late Saxon / Domesday manorial centre (Dr Paul Courtney pers. Comm.). Unlike de Waudari’s it would have been a classic, multi functional castle: defensive residence, manorial centre and court centre. The last two, and especially the last, were perhaps the longest-lived functions, as at so many other castles. The manorial function is likely to have declined with the sub-parcelling of the manor of Luton into at least thirty sub-manors from the late 13th century onwards.
Within the context of the Anarchy, the centre of the town may not have been the best strategic location for de Waudari to site his castle. It is also possible that King Stephen did not want his mercenary to occupy the headquarters of what had traditionally been a royal manor. Whatever the reason, the castle was built beyond the limits of the medieval town and had an influence on its subsequent expansion.

PHASE 2: MEDIEVAL/EARLY POST-MEDIEVAL QUARRYING
On the basis of the excavated evidence, there is no indication that the site of the castle was re occupied in the medieval period. It lay some distance from the centre of the medieval town (Fig. 3) in what may, for centuries, have remained an essentially rural setting. A series of intercutting quarry pits were dug along the line of the castle ditch, presumably because it afforded easy access to the underlying chalk. Even when quarrying ceased, the pits were not backfilled with domestic refuse, reinforcing the conclusion that they lay some distance from a centre of habitation.

PHASE 3: LATE POST-MEDIEVAL EXPANSION OF LUTON
The first useful, complete map of the town, the 1842 Tithe map (MAT/30/1 and 2), gives a good indication of the state of this part of Luton on the brink of its enormous expansion in the early industrial period. Holly Lodge is clearly visible in its wooded grounds, which were to remain a feature of this part of the town for the next century. The adjacent stretch of Castle Street displays an intriguing curve, which is still a feature of the road today. It is tempting to see in this the influence of its passage through the site of de Waudari’s former castle.
New Town Street (later Newton Street and present-day Kelvin Close) was laid out between 1841 and 1844 for two rows of terraced houses. Speculative developments of this kind, fuelled by a fluid land market and the rapid industrialisation of the hatting trade, were characteristic of the growth of early 19th-century Luton (Bunker 1999, 12, 65). It is noteworthy that the plot of land occu pied by the castle ditch on the Bellway Homes’ development was not built on at this time. There are many possible reasons for this but it may be that the partially infilled castle ditch and quarry pits made it a less attractive site to prospective, speculative developers.
By 1880 the southern part of the plot was occupied by a row of houses fronting onto Castle Street (Fig. 5). The construction of these buildings may have been the occasion for partial levelling of the site, as seen in the excavated evidence on the Bellway Homes’ development. However, the northern half of the plot (i.e. what would have been the south-eastern quadrant of the castle’s bailey) remained a tree-lined piece of open ground, as did the corresponding plot of land on the opposite side of the newly established Holly Street. The latter had been built on by 1924 (Fig. 6) but as late as 1959 the site of the south-eastern corner of the bailey ditch remained undeveloped (Fig. 7). Only four years later any surviving above-ground evidence for the ditch was swept away by the construction of the Luton News printworks.

CONCLUSIONS
There are still many unanswered questions about Robert de Waudari’s castle. However, since its physical remains were first recorded in 1963, archaeological fieldwork has significantly augmented our knowledge of its size, location and state of preservation. It is clear that it was positioned to provide strategic command of the southern approaches to medieval Luton; it was built for short-term, military reasons. Once the castle had served its purpose, the site slipped back into relative obscurity and, apart from an episode of small-scale chalk quarrying, was probably re-incorporated into the extensive tracts of pasture to the south and west of the town. By combining the results of the investigations on the Bellway Homes’ development with cartographic, documentary and photographic sources, it has been possible to demonstrate that, although short-lived (fifteen years), the castle’s influence on local land boundaries and the later development of Luton lasted for over 800 years.
If Robert de Waudari’s castle remains something of an enigma, the same is true of Fulke de Breaute’s castle to the south of St Mary’s church, where a substantial mound of earth still survived at the beginning of the 20th century (Dony and Dyer 1975, 47). No archaeological work has yet been carried out on this site. However, the Bellway Homes’ development has demonstrated that both substantial features (like the bailey ditch) and smaller features (like the ditches sealed by former subsoil at the eastern end of the site) can survive within modern Luton. It is clear that archaeological investigation can help to redress the effects of what Bunker (1999, 14) has described as a historical lack of appreciation of Luton’s heritage, manifested through the wholesale demolition of later buildings and the meagre extent of documentary records.
[Acknowledgements and references]

<10> Archaeological Services & Consultancy Ltd, 2008, Watching brief: 63-65 Castle Street, Luton (Archaeological Report). SBD11895.

No significant archaeological finds or features were observed during a watching brief at 63-5 Castle Street between March 2007 and February 2008.

<11> Council for British Archaeology, 2009, South Midlands Archaeology, Volume 39, Vol. 39, 2009, p11 (Serial). SBD14152.

Details as per ref (10)

<12> NMR/AMIE, HE NRHE Monument Inventory, 359711 (Index). SBD12367.

Holly Lodge, Castle Street, Luton (TL 09062082) stands on a mound which may be the site of a castle built in 1139 or soon afterwards.

<13> Thames Valley Archaeological Services, 2002, Former Luton Bus Depot, Castle Street, Luton; An Archaeological Evaluation (Archaeological Report). SBD13336.

Two ditches were located traversing Trench 12 east to west. Ditch 1 appears to have been a re-cut of an earlier ditch. One piece of medieval pottery was recovered from ditch 1 along with two struck flints. Two small sherds of prehistoric (possibly Bronze Age) pottery were recovered from ditch 2.
A gully was also located, aligned north-south A single sherd of possibly Roman pottery was found in its fill. There was no direct relationship between the gully and the two ditches, but the gully terminated just short of the ditch line, suggesting it was cut while the ditch was visible.

Trenches 15 and 16 were dug to establish the surviving extent of the features located in Trench 12. Both trenches exposed ditches 1 and 2 and revealed their western extent, with extensive modern truncation further into the site.

The features revealed do not automatically invite interpretation as related to Robert de Waudari's castle, in that they are not obviously of defensive proportions (unless considerably truncated) nor are they sufficiently distinctive to be considered as representing castle architecture. It is possible that they may relate to ancillary structures surrounding or within a castle, though it is equally possible that they represent activity completely unrelated to such a structure.

<14> Luton Museum, Accession Register, 2002/313; 2005/48; 2007/127; 2008/25 (Unpublished document). SBD10775.

2002/313 - Document and excavated finds from Bus Depot site, Castle Street, Luton. Excavated by Thames Valley Archaeological Service. Medieval.
2005/48 - Medieval Castle Site, Castle Street.
2007/127 - Watching Brief at 63-65 Castle Street, Luton. Site code 838/CAS. TL 091 208. Victorian and 20th century material. [EBD 98].
2008/25 - Material from excavation at the former Luton Bus depot, Castle Street, Luton. Site Code: CSL02/73. Grid Reference: TL090 207. Material expected is medieval.

Protected Status:

• Archaeological Notification Area (AI) HER186: SITE OF CASTLE, Castle Street/Holly Street

Monument Type(s):

• DITCH (Unknown date)
• PIT (Unknown date)
• GULLY (Roman/Romano-British - 43 AD? to 409 AD?)
• BOUNDARY DITCH? (Medieval - 1066 AD? to 1539 AD?)
• DITCH (Medieval - 1066 AD to 1539 AD)
• PIT (Medieval - 1066 AD to 1539 AD)
• QUARRY (Medieval - 1066 AD to 1539 AD)
• QUARRY (Medieval to Post Medieval - 1066 AD to 1900 AD)
• CASTLE (12th Century - 1139 AD to 1154 AD)
• POST HOLE (Post Medieval - 1540 AD to 1900 AD)
• POST HOLE (Post Medieval - 1540 AD? to 1900 AD?)
• STRUCTURE? (Post Medieval - 1540 AD to 1900 AD)

Associated Finds

• FBD6226 - ANIMAL REMAINS (Unknown date)
• FBD12389 - NON MARINE MOLLUSCA REMAINS (Unknown date)
• FBD12388 - PLANT MACRO REMAINS (Unknown date)
• FBD12046 - SCRAPER (TOOL) (Early Prehistoric - 500000 BC to 4001 BC)
• FBD6225 - SCRAPER (TOOL) (Prehistoric - 500000 BC to 42 AD)
• FBD6222 - SHERD (Prehistoric - 500000 BC to 42 AD)
• FBD17979 - FLAKE (Early Mesolithic to Late Bronze Age - 10000 BC to 701 BC)
• FBD6224 - FLAKE (Early Neolithic to Late Bronze Age - 4000 BC to 701 BC)
• FBD12394 - WORKED OBJECT (Late Neolithic to Early Bronze Age - 3000 BC to 1501 BC)
• FBD17976 - SHERD (Bronze Age - 2350 BC? to 701 BC?)
• FBD12392 - SHERD (Late Bronze Age to Late Iron Age - 800 BC to 42 AD)
• FBD12047 - SHERD (Late Iron Age - 100 BC to 42 AD)
• FBD17977 - SHERD (Roman/Romano-British - 43 AD to 409 AD)
• FBD6223 - SHERD (Roman/Romano-British - 43 AD to 409 AD)
• FBD12393 - RIM SHERD (2nd Century - 100 AD to 199 AD)
• FBD6227 - HORSESHOE (Medieval - 1066 AD to 1539 AD)
• FBD6221 - SHERD (Medieval - 1066 AD to 1539 AD)
• FBD17978 - SHERD (12th Century to 13th Century - 1100 AD? to 1299 AD?)
• FBD12390 - TILE (Post Medieval - 1540 AD to 1900 AD)
• FBD12391 - SHERD (17th Century to 18th Century - 1600 AD to 1799 AD)

Associated Events

• EBD186 - Excavation at Castle Street, Luton: Robert de Waudari's Castle? (Ref: 02/73b)
• EBD98 - Watching brief: 63-65 Castle Street, Luton (Ref: 887)
• EBD984 - Land at Castle Street/Holly Street, Luton: Archaeological Field Evaluation (Ref: 2005/9)
• EBD1065 - The remains of Robert de Waudari's adulterine castle, Castle Street, Luton
• EBD2284 - Former Luton Bus Depot, Castle Street, Luton; An Archaeological Evaluation (Ref: 02/73)

Sources and Further Reading

Search results generated by the HBSMR Gateway from exeGesIS SDM Ltd.