The Staffordshire Hoard Fieldwork, 2009-2010

Alex Jones (Birmingham Archaeology)

SUMMARY

Two stages of fieldwork were undertaken at the site of the Staffordshire Hoard. The first, in July-August 2009, was solely concerned with the safe and systematic recovery of the hoard. Test-pitting was followed by excavation, which confirmed that the hoard was contained within the ploughsoil. The strategy for recovery involved the excavation of a total of 155 square metres, working outwards from the initial test-pit. Within each metre square, the ploughsoil was hand-excavated in spits and repeatedly scanned with a metal detector to ensure that all objects, however small, were recovered. A total of approximately 800 objects were recovered by test-pitting and excavation. No features associated with the hoard were identified. The few features identified, comprising an irregular possible ditch, a gully and possible post-holes, were all undated. A geophysical survey was also undertaken to attempt to set the discovery within its immediate context.

Although the first stage of fieldwork was successful in its sole aim - to safely recover the hoard - it left many questions unanswered. The second stage of fieldwork, undertaken in March 2010, attempted to place the hoard within its immediate archaeological context. This stage of fieldwork involved further geophysical survey, test-pitting and trial-trenching. The further geophysical survey identified a number of linear and possible pit-like anomalies of which a selection was tested by trial-trenching. Most were found to be geological in origin. Trenching also identified two lengths of an undated palisade trench. Also investigated was a re-cut field boundary, also recorded in the First Edition Ordnance Survey map. Most significant perhaps was the total absence of any finds which could be associated with the hoard - confirming that the hoard had been fully recovered in Summer 2009. None of the features identified in March 2010 could be directly associated with the hoard.

Both stages of fieldwork were undertaken by Birmingham Archaeology. The project manager for Birmingham Archaeology was Alex Jones, and the work on site was supervised by Bob Burrows. The 2009 fieldwork was sponsored by English Heritage and Staffordshire County Council, and was project-managed for the council by Ian Wykes and Stephen Dean with advice from Bill Klemperer (English Heritage). The 2010 fieldwork was sponsored by Staffordshire County Council, and was project-managed for the council by Stephen Dean. The assistance of the landowner and Terry Herbert throughout is gratefully acknowledged.

Fig. 1: location map

INTRODUCTION

Two stages of archaeological fieldwork were undertaken at the site of the Staffordshire Hoard (Fig. 1). The first stage, leading to the systematic recovery of the hoard, was undertaken in July-August 2009 (Jones 2009). This took the form of test-pitting followed by controlled archaeological excavation of the ploughsoil within 1m squares by a team from Birmingham Archaeology (Fig. 2, Fig. 3 (below), Plate 1). This stage of work was sponsored by English Heritage and Staffordshire County Council. It was undertaken as an emergency response to the discovery of the hoard, which was under threat of theft by unauthorised metal-detecting. A magnetometer survey of the entire field was also undertaken. This stage of fieldwork also included systematic metal-detector surveys (undertaken by Terry Herbert and the Scientific Investigation Branch of the Home Office) and further test-pitting.

Fig. 2: detailed location map

Plate 1: the site under excavation, summer 2009

No archaeological work had previously been undertaken at the site, although extensive archaeological work had been undertaken nearby in connection with the recent M6 Toll scheme (Powell et al. 2008). Immediately prior to the Birmingham Archaeology fieldwork a quantity of finds had been identified and recovered by Terry Herbert, and a single 1m square test-pit was dug by Staffordshire County Council which established that the hoard was contained within the ploughsoil.

The second stage of fieldwork, comprising further geophysical survey, trial-trenching and test-pitting was undertaken by Birmingham Archaeology in March 2010, on instruction from Staffordshire County Council (Burrows and Jones 2010). This stage of work aimed to provide further details of the immediate archaeological context of the hoard.

Each stage of work was undertaken in accordance with a Design Brief (Staffordshire County Council 2009a and b), approved by English Heritage prior to implementation. Both stages of work were undertaken in accordance with the Code of Conduct of the Institute for Archaeologists. The 2009 fieldwork followed the Standard and Guidance for Archaeological Excavation (IfA 2008), although the work did not comprise a systematic excavation or evaluation of the site (sensu IfA) because of the immediate threat posed by metal detector activity, and the fact that, in the event, all the hoard finds were confined to the ploughsoil. The 2010 fieldwork followed the Standard and Guidance for Archaeological Evaluation (IfA 2008).

Further details of the geophysical surveys are contained in the Appendix.

LOCATION AND GEOLOGY

The area investigated (centred on NGR SK 406328/306396) is located to the south of the modern A5 (T) Watling Street, to the west of the M6 Toll and to the east of Hanney Road (B4155) (Fig. 2). The underlying geology consists of sandstone (Wildmoor formation) (British Geological Survey, 1:50,000 series, Sheet E154, Lichfield Solid and Drift Geology). For more on the geology of the site, see Ben Gearey's paper.

The field has been intermittently ploughed in the past, most recently in 2008. In previous years it has been laid to a variety of crops, including carrots and potatoes. At the time of the discovery the site comprised pasture.

The hoard was located towards the northwestern spur of a northwest-southeast aligned ridge (Fig. 4). The hoard was located slightly downslope of the highest point at the northwestern end of the ridge. The highest ground along the ridge was located towards the southeastern end of the ridge, away from the hoard find-spot. The centreline of the ridge was further defined by the line of a historic field boundary which turned off the ridge in a northerly direction just inside the northern field boundary (Fig. 5).

Fig. 5: First edition OS map with findspot of hoard marked

ARCHAEOLOGICAL BACKGROUND

The hoard findspot (Figs 1-2) is located immediately to the south of the modern A5 (T) road, which follows the line of Watling Street (Margary 1973, 289), possibly laid out in the AD 70s (Gould 1966; 1997, 351), as part of the Roman military communications network. Roman activity at Wall is attested by air photograph and excavation evidence. A possible Neronian vexillation fortress (Lyons and Gould 1964; Round 1983) was succeeded by Flavian enclosures (Round 1983). Further cropmarked features may represent marching camps (Welfare and Swann 1995, fig. 145). A civilian settlement is also recorded extending over 1km along Watling Street (Lyons and Gould 1964, Jones 1998). Wall itself is located midway between the Roman military complexes at Penkridge to the west, and Mancetter to the east. With the exception of the defended enclosure (Gould 1998) there is little evidence for 4th-century or later occupation at Wall (Burnham and Wacher 1990).

A number of Romano-British rural farmsteads have been excavated in advance of the M6 Toll motorway. Those sites located in close proximity to Wall appear to have been concerned with cereal production (Sites 15 and 34, Powell et al. 2008), whilst the economy of more distant rural sites was based on pastoralism (Sites 19 and 29, Booth 2008, 534), suggesting to the excavators that Wall and its immediate hinterland, most notably reflected by the excavated cemetery (McKinley 2008), remained an 'alien' establishment throughout the Roman period.

A number of excavations have been undertaken in the near vicinity of the hoard site in advance of the M6 Toll. At Site 34, west of Crane Brook, Hammerwich (Simmonds 2008, 62), the main feature identified was a Romano-British aisled building. A watching brief was maintained during the excavation of a section of Watling Street at Hammerwich (Champness 2008, 57), between Muckley Corner and Brownhills (Site 41). The excavator suggested that this section of the road may have continued in use into the Anglo-Saxon and medieval periods, before going out of use in the medieval period (ibid., 59). A watching brief was also maintained during excavation across a segment of the Wyrley and Essington canal (Powell 2008, 60).

A Welsh praise poem, the Marwnad Cynddlan (Rowland 1990, Gould 1993) describes a raid upon Caer Luitcoed, literally the 'fortified grey wood', identified as Wall because of the fortified nature of that site (Gould 1993). Rowland (1990) identifies Caer Luitcoed with Lichfield, although it is not clear that it had a discernible boundary at that time (Gould 1993). The poem describes the booty, including 15,000 head of cattle, horses and harnesses taken by an army from Powys led by Cynddlan and his brother Morfael, acting in concert with a Mercian army. It is possible that this booty was derived from a defeated Northumbrian army (see Mattias Jacobsson's paper for more on the Marwnad Cynddylan). There is little archaeological evidence for continued occupation in the post-Roman period.

There is no local Anglo-Saxon context for the hoard. A stray find of a garnet-inlaid gold pendant (the findspot is not precisely located) was found in Hammerwich parish (Evans 2004). Hooke (Dean, Hooke and Jones 2010, 150) has noted that the placename Hammerwich is OE hamor, meaning hammer, and wic, derived from the Latin word vicus, together meaning 'place associated with trade or specialised function', suggesting metalworking nearby (see David Parsons's paper for more on the place-name Hammerwich).

By the 7th century there was a clear geographical context for the Mercian kingdom (Yorke 2001, 19). During the late 7th and 8th century the Mercian kingdom was focussed in the area of the Trent: at the royal vill of Tamworth, represented by the 'palace' enclosure, bank and ditch there (Rahtz and Meeson 1992); at the nunnery of Repton; and at the historic bishopric at Lichfield (Yorke 2001, 20). Peada's conversion in AD 653 is the date recognised for the introduction of Christianity (Parsons 2001, 52).

At least one stone-built post-Roman structure has been excavated, at the Cross Keys Car Park site in Lichfield, near the Stowe Pool. Occupation of the most complete structure was dated AD 435-636 (at 95.4% confidence). This structure was overlain by two sunken-featured buildings. Destruction of the final building was dated AD 684-965 (at 95.4% confidence). Although no associated finds were recovered, the draft report suggests largely continuous occupation from AD 400 to approximately AD 950, and certainly between AD 600 and AD 800, corresponding to the period of Mercian rule (information from Stephen Dean, Staffordshire County Council).

During the medieval period the site formed part of Cannock Forest (Dean, Hooke and Jones 2010, 151), before enclosure in the early 19th century. This lack of cultivation, at least until the 19th century, will have helped the hoard to survive. The 1839 Ogley enclosure award and the tithe (1843) and enclosure (1856) awards for Hammerwich show the survival of waste until this time (Dean, Hooke and Jones 2010) (see Della Hooke's paper for more on the landscape context of the hoard).

The detailed map of the M6 Toll route in the vicinity of the site (Powell et al. 2008, fig. 5) indicates associated earthworks along the extreme northern edge of the field in which the hoard discovery was made. These presumably comprised cutting back of the southern embankment of the A5 (T). This disturbance did not extend within the hoard findspot.

AIMS AND OBJECTIVES

The aim of the 2009 (recovery) fieldwork (Staffordshire CC 2009a) was to recover the hoard and also to place it within its immediate archaeological context.

The particular objectives of the project were to:
A) Recover the entire hoard through a combination of archaeological excavation and metal detecting
B) To examine the immediate archaeological context of the hoard by means of geophysical survey and controlled archaeological excavation.

In the event, fieldwork was necessarily limited to the recovery of the hoard, supplemented by geophysical survey and a metal detector survey of the surrounding area. In consultation with English Heritage and Staffordshire County Council it was agreed to prioritise recovery of the hoard by hand-excavation. The first stage of fieldwork did not provide details of the archaeological context of the hoard.

The principal aim of the 2010 (evaluation) fieldwork (Staffordshire CC 2009b) was to provide details of the immediate archaeological context of the hoard and also to recover any further items associated with it.

METHODOLOGY

Recovery stage (2009 fieldwork, Fig. 3)

The initial discovery was made on 5th July by a metal detectorist, Terry Herbert, working with the written permission of the landowner. Over five days a total of approximately 200 objects, most made of gold, were recovered. In accordance with the procedure established by the Portable Antiquities Scheme (PAS) the finds were first reported to Duncan Slarke, then Finds Liaison Officer (FLO) for Staffordshire. In turn the discovery was reported to Staffordshire County Council, English Heritage (Bill Klemperer) and Roger Bland (British Museum/PAS) and the objects were transferred to Birmingham Museum and Art Gallery for safe keeping.

At a meeting on 21 July 2009 held at BMAG, including representatives of Staffordshire County Council and PAS, it was decided that a 1m square test-pit was to be excavated by Staffordshire County Council in order to understand the archaeological context of the hoard. This was undertaken with the permission of the landowner, and assistance from Terry Herbert.

Preliminary study of the finds by Kevin Leahy of PAS suggested that the items conformed to Salin's Style II, and were 6th-7th century in date. The range of finds suggests high-status associations - and comparison with material recovered from Mound 1 at Sutton Hoo (Carver 2005), although the range of material recovered from the hoard was largely confined to warlike objects.

On 22nd July archaeologists from Staffordshire County Council working with the FLO and Terry Herbert began to excavate the 1m square test-pit. Numerous further gold items were recovered, all derived from the ploughsoil.

In agreement with English Heritage, Birmingham Archaeology were contracted to assist with the fieldwork. The test-pit was enlarged to 2m square on 24th and 27th-29th July (Fig. 3). All finds from the test-pitting were individually numbered.

Scanning the area surrounding the test-pit with a metal-detector suggested that the hoard may be confined to an area measuring approximately 20m square (Fig. 3). Therefore, during continued excavation of the test-pit a strategy was developed by Birmingham Archaeology and Staffordshire County Council in consultation with English Heritage, the FLO, Terry Herbert and the landowner, to facilitate full recovery of the hoard.

Fig. 3: site plan

This strategy involved the layout of a 20m square grid coinciding with the concentration of metal detector signals, and centred on the 2m square test-pit. This grid was divided into 1m squares, designated by a number and letter combination (eg P3). Each grid square was hand-excavated separately, to permit later analysis of the finds distributions (Plate 2). Each individual find was allocated a small find number (recorded in a site finds register, copied to the FLO at the end of each day), and the grid square number.

Pl. 2: recording the findspot locations on the finds bags

Within each 1m square the ploughsoil was excavated by hand in spits, and repeatedly scanned with a metal detector to maximise finds retrieval. The soil was then initially 100% bagged by grid square, and wet sieved through a 1mm sieve using a York flotation tank, with collection of all finds from the flots (Plate 3). Wet sieving was intended to maximise finds recovery, including the smallest items. This methodology was abandoned due to the slow progress of wet sieving caused by the wet clay soils, poor water pressure and the continued threat of illicit metal detecting activity.

Pl. 3: wet sieving of spoil

In consultation between Staffordshire County Council and English Heritage a new strategy was defined. This involved continued excavation in spits within 1m squares, with recording of individual finds as before, but no water flotation of the soil. Spoil from hand-excavation was additionally spread over plywood sheets to enable visual sorting for finds recovery, and by using a metal-detector (operated by Terry Herbert) to maximise finds recovery.

A total of 155 metre grid-squares were hand-excavated, working outwards from the initial discovery, within an area measuring a maximum of 9m by 13m in extent (Plates 1 and 4). Finds were recovered from 85 grid squares within the excavation, and within five 1m square grids outside.

Pl. 4: the site under excavation, summer 2009

The site was assessed as extremely vulnerable to raids by illicit metal detectorists, in particular because of its visibility from a trunk route. For this reason it was decided to maintain 24-hour security on site from 24 July to 21 August 2009, when the excavation was completed.

Following completion of the finds recovery, an area measuring 50m square, centred on the original discovery, was systematically metal-detected by Terry Herbert. Possible metal-detector signals were individually marked with a cane. Each of these potential find-spots was tested by hand-excavation. As part of the same deployment, Terry Herbert scanned the topsoil during the backfilling of the excavated area, and a small number of gold or silver objects were recovered. Immediately prior to the news of the hoard's discovery being released to the national media, the hoard find-spot and its surrounds was scanned by metal-detectors operated by the Home Office Scientific Investigation Branch.

The initial test-pit dug by Staffordshire County Council, and the layout of the 20m square grid, was undertaken using a Leica GPS, working at an accuracy of +/- 0.25mm.

Throughout the field investigations the finds were taken off site at the end of each day, and either stored within a safe location within the University of Birmingham, or were delivered to Birmingham Museum and Art Gallery. A detailed, duplicate list of finds for countersignature was prepared before each handover.

Evaluation stage (2010 fieldwork)

The resistivity survey covered an area of 1ha in the northwestern angle of the field in which the hoard was located (see Appendix). The preliminary resistivity survey plot, together with the plot of the 2009 magnetometer survey, provided the basis for the positioning of trial-trenches, defined in consultation with Stephen Dean (Staffordshire CC). The trial-trenches were surveyed using GPS.

A total length of 100m of trenches measuring 1.6m in width was excavated to test a representative selection of the main geophysical anomalies recorded (Fig. 3). Each trench measured either 1.6m or 3.2m in width, in the latter case to test geophysical anomalies more intensively. In addition, a total of eleven test-pits, each measuring 1.6m square, were excavated to provide a profile of the ploughsoil and subsoil at a right-angle to the main axis of the natural ridge.

All topsoil and modern overburden was removed using a JCB excavator equipped with a toothless ditching bucket, working under continuous archaeological supervision, to expose the top of the uppermost archaeological horizon or the subsoil, whichever was first encountered.

Subsequent cleaning and excavation was by hand. Following completion of machining, the sides and base of each trench or test-pit were hand-cleaned to facilitate better definition of the features, or possible features, present. All features of archaeological (or possible archaeological) interest were sampled by hand-excavation. All spoil from machine and hand-excavation was scanned with a metal-detector.

Pl. 5: recording a section

All stratigraphic sequences were recorded (Plate 5), even where no archaeology was present. Features were planned at a scale of 1:20, and sections were drawn of all cut features and significant vertical stratigraphy at a scale of 1:10. A comprehensive written record was maintained using a continuous numbered context system on pro-forma cards. Written records and scale plans were supplemented by photographs using black-and-white monochrome, colour slide and digital photography.

The full site archive includes all artefactual remains recovered from the site. The site archive will be prepared according to guidelines set down in Appendix 3 of the Management of Archaeology Projects (English Heritage 1991), the Guidelines for the Preparation of Excavation Archives for Long-term Storage (UKIC 1990) and Standards in the Museum Care of Archaeological Collections (Museum and Art Galleries Commission 1992).

The paper archive will be deposited with Stoke on Trent Museum, subject to approval from the landowner.

RESULTS

Recovery stage (2009 fieldwork, Jones 2009)

Features identified (Figs 3 and 6)

The natural subsoil mainly comprised a reddish-brown silt-clay (1021), recorded at a depth of 0.37m below the modern surface. Identification of this natural subsoil horizon was confirmed by the excavation of a machine-cut and hand-cleaned trench measuring a maximum of 1.6m in width, 1.2m in depth, and 6m in length. This was dug after the last of the hoard finds was recovered. At the base of this sondage was revealed a yellow-orange clay (1022, not illustrated), incorporating patches of orange sand, interpreted as a localised change in the natural subsoil. In the southwest of the area investigated the natural subsoil was a red weathered sandstone (1008). Several plough scars (not illustrated) were recorded cutting the natural subsoil.

The main excavated feature was a possible ditch (1007, Plates 6-7), which appeared to be aligned northwest-southeast, recorded in the extreme southwestern corner of the excavated area. It was slightly extended to the west in order to attempt to recover the full profile of feature 1007. This possible ditch was cut into the natural subsoil. It measured a maximum of 1.5m in width, 0.4m in depth, and was cut to a very irregular profile. The primary fill of the ditch was an orange-red sand-silt (1006), containing a basal lens of orange-red silt, representing an episode of weathering of the possible ditch sides. Layer 1006 was sealed by a deposit of dark grey-brown silt-sand (1005). Because hand-excavation of this feature was limited to a single hand-dug slot measuring only 0.7m in width, its interpretation as an anthropogenic feature should be treated as provisional only.

The other excavated features comprised two possible post-holes (1012, 1016), and a gully (1013, 1015). In the north of the area investigated, the possible post-hole 1016 measured 0.14m in diameter, but only 0.01m in depth. It was filled with light-mid grey sand-clay (1017). It is possible that this 'feature' may be no more than a slight localised dip in the surface of the natural subsoil. The roughly north-south aligned gully (1013, 1015) was recorded for a maximum distance of 5m towards the south of the area investigated. The feature was cut through backfilled feature 1007, and into the natural subsoil (1008). The gully was cut to a U-shaped profile, and measured a maximum of 0.65m in width and 0.22m in depth. It was filled with orange-red sandstone rubble and sand (1014). One hand-excavated segment (1013) through the same feature was cut to a stepped profile, measuring a maximum of 0.5m in depth and 0.l45m in width. It is possible that this segment could represent a post-hole cut through the gully, rather than an enlarged segment of the same feature, although this could not to be proven because of disturbance by a later feature (1012, see below). Feature 1013 was filled with the same material as gully 1015. Feature 1013 was cut by a circular post-hole (1012), measuring 0.4m in diameter, and 0.1m in depth. This feature was filled with dark grey-black silt-sand (1009), similar in composition to the ploughsoil. It was probably modern in origin.

Fig. 6: plan of excavated features

No artefacts were recovered from these features, or possible features. None of the features investigated at this stage could be related to the hoard.

The natural subsoil and the backfilled features were sealed by a layer of ploughsoil, which comprised a dark-brown silt-sand (1000).

Three 1m square test-pits (not illustrated) were dug by hand into the ploughsoil to test a selection of geophysical anomalies. All were found to represent variations in the natural subsoil. No finds of archaeological interest were collected during this test-pit excavation.

Finds distributions (Figs 7-9)

This section of the report presents a preliminary summary of the finds distributions, based on preliminary object identifications provided by Dr. Kevin Leahy (excluding objects excavated from intact soil blocks at BMAG). The identifications at this stage are provisional only and will be subject to revision following detailed study. The distributions plot the finds recovered from the July-August 2009 excavation by 1m grid square. Items discovered during the preceding Staffordshire County Council test-pitting have been retro-fitted into the excavation grid.

As is evident from the magnetometer survey (see Appendix) the site has been heavily ploughed. The direction of ploughing is approximately east-west and north-south.

Prima facie, the recorded distributions may reflect the original locations of the finds within the hoard, their subsequent movement through plough action, or both.

Fig. 7: distribution of finds by number

Most of the excavated grids yielded at least one item (Fig. 7). The largest numbers of items were derived from an area immediately adjoining the 2m square test-pit excavated by Staffordshire CC. An area measuring 5m (east-west) by 3m (north-south) contained up to 19 objects per square metre, possibly including objects 'dragged' from their original location by plough action (Fig. 7). Outside this rectangular area the artefact numbers were reduced, although not uniformly. Areas to the southeast, northwest and northeast of the main concentrations yielded no artefacts. By total weight per square metre (Fig. 8), the largest artefact collections were again located adjoining the area of the test-pit.

Fig. 8: distribution of finds by total weight per sq m

Outside this concentration, the artefact density was generally reduced, although a number of significant 'hot-spots' may be noted (eg to the southwest, east and north of the main distributions). The plot of average weight of finds per square metre (Fig. 9) shows interestingly that a number of the grid squares containing larger items are located away from the centre of the find-spot. This could suggest that some of the larger items have been 'dragged' by the plough from their original location, and the smaller items may not.

Fig. 9: distribution of finds by average weight of items per sq m

Other distribution patterns are summarised in Table 1.

Table 1: preliminary object distributions, by category (not illustrated)

Overall, the smaller items may have remained where they were originally placed, albeit with some movement in a limited area. The distribution of the larger items suggests that some plough-induced movement has occurred.

The finds may have been dispersed as a result of several episodes of ploughing. The evidence for movement amongst some of the larger, but not the smaller, items refutes the suggestion that the hoard was very recently buried.

Out of a total of 155 one-metre square grids that were hand-excavated, a total of 85 square metres (within the area systematically excavated) contained at least one hoard find. A further five grid squares outside the excavated area contained hoard finds, but only one item per grid square.

Evaluation stage (2010 fieldwork)

Trench 1 (Fig. 3)

This trench was located to intercept two possible southwest-northeast aligned anomalies and a pit-like anomaly. Trench 1 measured 25m in length and 3.20m in width and was orientated northwest-southeast.

The natural subsoil varied from reddish-black to reddish-orange sand (4003, 4004), representing localised changes in the natural subsoil. A possible gully and an irregular depression (4006), were identified in this trench. The possible gully (4005) was backfilled with light grey silt-sand (4002). The composition of this material suggests that the feature may have been natural in origin, although a single worked flint object (see below) was recovered from the fill. The irregular depression (4006) was a very shallow hollow of geological origin. The grey-brown silt-sand subsoil (4001) measured 0.35m in depth and was overlain by 0.15m of topsoil (4000).

No archaeological, or possible archaeological, features were identified within the trench. Gully 4005 at the southeastern end of the trench corresponded with the position of the linear anomaly. Feature 4006 at the northwestern end of the trench appeared to have been caused by a localised concentration of gravel within the natural subsoil.

Trench 2 (Fig. 3, Plates 8-9)

This trench was located to test a curvilinear magnetometer and resistivity anomaly and a possible pit-like anomaly. The possible curvilinear anomaly was aligned northwest-southeast. The pit-like anomaly was positioned at the northeastern end of the trench. The trench measured 22.50m in length and 1.60m in width and was aligned northeast-southwest.

Plate 8: Trench 2Plate 9: Trench 2

The subsoil in this trench was a reddish orange sand and gravel (4105). It had been truncated by a feature (4104) filled with orange-brown silt-clay (4103) which was recorded in the central area of the trench. Dr. Ben Gearey (Birmingham University) has interpreted this feature as a possible ice wedge.

At the northeastern end of the trench, the natural subsoil (4105) was overlain by a thin band of natural gravel in the area of the pit-like geophysical anomaly. Towards the southwestern end of the trench the natural subsoil had been cut by a linear gully or plough furrow (4101). The gully was aligned northwest-southeast and measured 0.70m in width and 0.20m in depth. It contained a grey-brown silt-sand fill (4102) and was sealed by 0.18m of mid grey brown silt-sand subsoil (4106) which was overlain by 0.14m of topsoil (4100).

The geological feature (4104) filled with orange-brown silt-clay (4103) corresponded in orientation and location with the curvilinear geophysical anomaly. The pit-like geophysical anomaly corresponded with the location of a thin band of gravel which overlay the natural sand subsoil (4105).

Trench 3 (Figs 3 and 10; Plates 10-11)

This trench was located to target one curvilinear and two linear geophysical anomalies. The linear anomalies were located towards the middle of the trench, and the curvilinear anomaly was located towards the eastern end of the trench. Trench 3 measured 20m in length and 1.60m in width and was orientated east-west.

Fig. 10: Trench 3

The natural subsoil comprised a reddish-orange sand and gravel (4201). In the centre of the trench the natural subsoil had been cut by a mainly northeast-southwest aligned palisade trench (4203, S.1-S.2). The excavated part of the feature appeared to have been dug in two sections with a slight change in angle. It was cut to a steep-sided, U-shaped profile, and measured 0.50m in width and 0.45m in depth. The base of the palisade trench was irregular, perhaps suggesting that stake- or post-holes had been inserted along its length to retain a fence. The palisade trench was backfilled with mottled mid grey and pale brown sand-silt (4202). It had been partially truncated by a probable north-south aligned plough furrow (4205). This plough furrow had been sealed by mid grey-brown silt-sand subsoil (4206) which was 0.20m in depth. The backfilled features and the natural subsoil were overlain by 0.20m of topsoil (4200).

Plate 10: Trench 3Plate 11: Trench 3 (detail)

The palisade trench and plough furrow were first recognised as resistivity anomalies. The parallel curvilinear resistivity anomalies projected to intersect with the eastern end of the trench were not located in this trench, although they were recorded in Trench 5 (see below).

Trench 4 (Fig. 3)

This trench was sited to target a large, possibly pit-like geophysical anomaly. Trench 4 measured 10m in length and 3.20m in width and was aligned north-south.

The natural subsoil was a brownish-orange sand (4303) in the south of the trench, while in the north of the trench the subsoil comprised gravel (4302). A north-south aligned plough furrow (4306) represented a continuation of a feature (4205) also recorded in Trench 3 (see above). The subsoil and the furrow were sealed by greyish-orange silt-sand subsoil (4301) which was overlain by 0.20m of topsoil (4300).

The large geophysical anomaly corresponded with the gravel subsoil (4302) recorded in the north of the trench.

Trench 5 (Figs 3 and 11; Plate 12)

This trench was located to identify two parallel geophysical curvilinear anomalies which mainly followed a northwest-southeast orientation. The trench measured a total of 15m in length and 1.60m in width and was orientated northeast-southwest.

Fig. 11: Trench 5

The natural subsoil comprised a reddish-orange silt-sand (4402). Towards the southeastern end of the trench the natural subsoil had been cut by a northwest-southeast aligned palisade trench (4406, Plate 12). This feature measured 0.90m in width and 0.30m in depth and had quite steeply sloping sides and was cut to a 'U'-shaped profile. It was backfilled with mid grey-brown silt clay-sand (4405).

Plate 12: Trench 5

A curvilinear ditch (4404) was excavated immediately to the northeast of feature 4406. The ditch also followed a northwest-southeast alignment. It measured approximately 4m in width and a maximum of 0.40m in depth. It had an irregular profile, probably as a result of re-cutting, or periodic cleaning-out. The ditch was backfilled with silt-sand (4403), overlain by dark brown silt clay-sand (4407) which was flecked with charcoal. This layer and the natural subsoil was sealed by a grey-brown silt-sand subsoil (4401) which measured 0.20m in depth. This was sealed by dark brown silt clay sand topsoil (4400), measuring 0.10m in depth.

The two northwest-southeast aligned features corresponded with the location of the two curvilinear geophysical anomalies. The wider of the two features represents a former field boundary. The palisade trenches located in Trenches 3 and 5 are probably lengths of the same feature.

Finds from trial-trenches (identifications by Emma Collins, with Stephanie Ratkai)

No objects associated with the hoard were recovered from trenching or from the associated metal-detecting, with the exception of a single glass bead (SF1).

Excavations recovered eight sherds of pottery weighting a total of 70g from four contexts. The majority of this small assemblage was post-medieval (3000 x 1, 4000 x 3 and 4407 x 3), and one sherd was medieval (4403, Trench 5). The single medieval sherd was partly reduced (?burnt) whiteware with quite a lot of iron inclusions, reddish where oxidised, black/grey where reduced and is dated to the 13th to14th centuries.

A small flint flake was recovered from Layer 4002 (Trench 1) and it appears to have secondary working along one edge.

A small quantity (65g) of slag-like material was recovered during trenching. The material was visually examined (by Dr. Gerry McDonnell), described, identified and weighed. The material derived from two features. The material from field boundary ditch 4404 (context 4407) in Trench 5 was a single piece of burnt 'organic material' (weight 6g), probably partially burnt coal or burnt bone. The context contained 19th-century pottery and the burnt material may derive from a firebox, or manuring which contained ash deposits from a hearth. The second feature, palisade trench 4406 (context 4405) contained burnt stone (weight 9g) and un-burnt stone (weight 50g). This material may also have derived from a hearth or firebox. There was no associated dating evidence.

The material does not derive from metalworking and probably derives from a hearth or firebox. The material was probably deposited via manuring of the fields in the 19th/20th centuries. The material has no significance (other than evidence for agricultural activity in the 19th/20th centuries in the case of the material from context 4405) and requires no further work.

Test-Pits (Fig. 3, Plate 13)

A total of eleven test-pits were recorded. The natural subsoil mainly comprised a red-orange silt-clay. This was sealed by a layer of light brown silt-sand subsoil, measuring an average of 0.16m in depth. Above was a layer of mid grey-brown silt-sand ploughsoil.

DISCUSSION

Material distributions

As first suggested by the preliminary test-pit excavation undertaken by Staffordshire County Council, all the hoard finds were found within the ploughsoil, and not within cut archaeological features. The hoard was scattered over a total of 85 square metres within the area excavated in detail, with further finds being made within a total of five square metres outside this area (Fig. 7). This broad distribution suggests that the objects have been spread by the plough away from their original location. The distributions can be interpreted to indicate east-west and north-south ploughing, as also indicated by the magnetometer survey (see Appendix). Among some of the most readily identifiable object types (using preliminary object identifications by Dr. Kevin Leahy), sword hilt collars, sword hilt plates and sword pommels were found within a total of 17, 28 and 17 one-metre square grids, respectively. These object type distributions further evidence the movement of the objects into the ploughzone. By number of finds per square metre (Fig. 7), and total weight per square metre (Fig. 8), the largest quantities of finds were located towards the centre of the excavated area. Interestingly, the plot of average weight of finds per grid square shows a wider distribution (Fig. 9). This may indicate that the larger objects may have been moved greater distances by the plough than the smaller items. The distribution of fittings suggest plough movement within an east-west direction.

It is important to emphasise that the object identifications on which these distributions are are provisional only, and will be subject to change; they also do not include objects excavated from intact soil blocks at Birmingham Museum and Art Gallery. The subsequent trenching confirmed that the 2009 fieldwork had recovered the entire hoard, with the exception of a single glass bead.

The hoard was probably originally placed within a wooden or leather container in a pit cut into the subsoil. No trace of any container was found at excavation, nor of the presumed pit, as a result of plough truncation. Over time the pit would have been entirely scoured-out by ploughing, and the objects would have been 'moved' into the ploughsoil, and also spread widely within it.

Hoard and its landscape setting

The topographic location of the hoard is important. It was located close to the northwestern spur of a northwest-southeast aligned ridge (Fig. 4). The placement of the hoard at the northwestern (lower) spur of the ridge suggests that a location adjoining Watling Street (or facing generally to the north) was more important than occupation of the higher land towards the southeastern spur of the natural ridge.

Both the magnetometer and resistivity surveys (see Appendix) recorded a curvilinear anomaly, describing roughly the southern half of a circle, immediately adjoining the hoard location. Trial-trenching (Trench 2) established that this was probably not an anthropogenic feature. Dr. Ben Gearey has suggested that this localised change in the subsoil could have been marked above ground by a localised difference in natural vegetation. This could have made the northwestern spur of the natural ridge a particularly distinctive landscape feature.

Magnetometry (see Appendix) also identified a historic field boundary (Fig. 5). This boundary follows the crest of the natural ridge before turning to the north, close to the northern boundary of the field in which the hoard was located. This change in alignment could indicate that the field boundary was diverted from its course to respect an adjoining feature. If correct, this hypothesis has two implications. Firstly, that the field boundary was the later of the two features. Secondly, that the feature avoided by the field boundary was visible as an above-ground earthwork, such as a mound, visible (and avoided) when the field boundary (or an earlier boundary on the same alignment), was laid out.

The topographic location of the hoard, and its proximity to the change in alignment of the historic field boundary is unlikely to be a coincidence.

Two, conflicting interpretations of the hoard have been suggested. One proposes that the hoard was buried hurriedly, with a view to recovery by individuals who did not survive to reclaim their treasure. In this hypothesis the ridge was chosen as a prominent landscape feature which would facilitate its rediscovery. In the alternative hypothesis, the hoard is interpreted as a being a pagan offering, in which case the topographic location may have held a symbolic meaning. The suggested band of 'different' vegetation could have imbued the locale with a symbolic meaning, or have made the location easy to rediscover.

Archaeological features (Fig. 6)

Archaeological recovery (2009)

None of the features, or possible archaeological features, identified during the recovery of the hoard in 2009 contained any finds which could be associated with the hoard. None of these features, or possible features, contained any datable finds.

The main feature identified was a ditch, or possible ditch (1007). Only a small segment of this ditch could be hand-excavated, and its interpretation as an anthropogenic feature cannot be confirmed, in particular because of its notably irregular profile. This possible feature appeared to be aligned northwest-southeast - the same alignment as the natural ridge. This possible feature is located just to the southwest of the main concentration of the hoard, but there is no present evidence to associate this possible feature with the hoard.

Post-dating the backfilled possible ditch was a roughly north-south aligned undated gully (1015), possibly including a post-hole or deeper segment dug along its length (1013), cut by a post-hole (1012). This is likely to be a recent feature, because of the similarity of its fill to the ploughsoil. Possible post-hole 1016 may not be a feature at all.

Archaeological evaluation (2010)

Similarly, none of the features, or possible archaeological features, identified by trenching could be associated with the hoard. Few of the features identified contained dating evidence; where datable finds were recovered they could have been residual.

Possibly the most significant features identified comprised the two excavated lengths of a palisade trench (4203 in Trench 3 and 4406 in Trench 5, Figs 10-11). The similarity in size and profile of these features suggests they may have been originally part of the same feature, a change in alignment being recorded between the two excavated segments. Palisade segment 4203 contained a number of possible post-hole or stake-hole impressions in its base, which could suggest the feature formerly contained a fence. Neither hand-excavated segment contained any datable pottery. Segment 4406 contained a small quantity of burnt stone. The absence of Roman artefacts from the palisade trench fills suggests either a pre-Roman, or post-Roman, date for the feature. It is notable that these features are located away from the crest of the natural ridge, which may suggest that the location was sheltered from westerly winds. One possibility to be considered is that the palisade trench functioned to retain livestock within a farmstead, although the absence of supporting artefactual evidence or associated features makes this interpretation highly speculative.

Probably the most recent feature to be identified by geophysics and trenching is the curvilinear field boundary (4404, Trench 5, Fig. 11). If not intrusive, the small sherd of pottery recovered from this feature could suggest that this field boundary was medieval in origin. The sherd was very small and abraded and could have been introduced into the shallow boundary as a result of manuring. It has been suggested (above) that the field boundary deviated from its course along the natural ridge in order to respect an above-ground earthwork in the approximate location of the hoard which is no longer extant.

The remaining features identified by trenching were probably formed by changes in the composition of the natural subsoil. A possible exception was feature 4005 in Trench 1 which contained a single worked flint object, which may, of course, be intrusive. Many of the linear and pit-like anomalies were identified and tested by hand-excavation. The two linear anomalies, and the pit-like anomaly, tested by Trench 1 were successfully identified, as were the anomalies intercepted by Trench 2. The palisade trench and plough furrow excavated in Trench 3 were apparent as resistivity anomalies, although the two curvilinear anomalies projected to intercept with the eastern end of the trench were not recognised here, although they were found in Trench 5. The pit-like anomaly intercepted by Trench 4 corresponded with an area of gravel subsoil. Trench 5 was positioned to intercept two parallel curvilinear anomalies, which were both tested by hand-excavation.

No finds or datable features of Roman date were identified, which is perhaps surprising given the proximity of Roman Watling Street (modern A5). The earlier fieldwork (Jones 2009) also failed to uncover any evidence of Roman activity at this location.

Despite extensive trial-trenching, combined with careful scanning of all the machine- and hand-excavated spoil, no further items of Anglo-Saxon date were uncovered (with the exception of a single glass bead). This negative evidence suggests that all items of Anglo-Saxon date had been successfully recovered during the August 2009 excavation and in the preceding metal-detector investigations.

Subsoils and weathering

The natural subsoil exhibited considerable variation. It comprised sand (4003-4004) in Trench 1, and sand and gravel in Trenches 2 and 3 (4105, 4201). Trench 4 was located at the interface between areas of sand (4203) and gravel (4302) subsoil. The subsoil in Trench 5 was a silt-sand (4402). The subsoil exposed in the base of the test-pits was similar in composition.

A subsoil was recorded in the excavated test-pits. Towards the crest of the ridge, and along the slope down the southwest-facing slope of the ridge, this subsoil measured an average of 0.2m in depth. At the northeastern slope of the ridge this subsoil measured 0.4-0.42m in depth (test-pits 9-10), presumably as a result of accumulation caused by prolonged erosion. The topsoil was generally shallow, measuring an average of 0.16m in depth, although the topsoil in Trench 4 (adjoining test-pits 9-10) was greater in depth, possibly as a result of erosion. For more on the implications of topsoil erosion, see Ben Gearey's paper.


CONCLUSION

The hoard was probably buried in a wooden or leather container, within a pit. Over the last two centuries ploughing has progressively truncated the feature until it was entirely scoured-out. This process has 'moved' the hoard itself into the ploughzone, and scattered the finds over an area in excess of 85 square metres. The 2010 evaluation fieldwork confirmed that the hoard has been safely recovered.

Geophysics and trenching has failed to identify any surviving direct evidence of the immediate archaeological context of the hoard. If the hoard was buried hurriedly with a view to its later recovery no such associated features would, of course, be anticipated. The natural ridge would have provided a 'marker' facilitating the recovery of the hoard. An alternative interpretation is that burial of the hoard was symbolic. Following this interpretation the topographic location could have imbued the site with particular significance. The apparent deviation of the historic field boundary (Fig. 5) off the crest of the natural ridge could perhaps suggest that the boundary respected an adjoining above-ground earthwork feature. The proximity of the natural ridge, the hoard's location, and the change in alignment of the historic field boundary may not be coincidental.

Little can be said about the other features identified by trenching, in particular the possible palisade trench. In the absence of a chronological context, further speculation is probably not worthwhile.


Acknowledgements

The 2009 fieldwork was commissioned by Staffordshire County Council and English Heritage, and monitored by Stephen Dean and Bill Klemperer, respectively. The project manager throughout was Alex Jones, and the fieldwork was supervised by Bob Burrows. The illustrations are the work of Nigel Dodds. Terry Herbert is thanked for his considerable help with metal detecting. Specialists to whom thanks are due are Eamonn Baldwin (geophysics) Emma Collins (finds) and Dr. Gerry McDonnell (slag-like material).

References
  • Booth, P., 2008. 'Romano-British period discussion', in Powell et al. 2008, 516-535.
  • Brown, M. P., and Farr, C. A., (eds.) 2001. Mercia, An Anglo-Saxon Kingdom in Europe (Continuum Studies in Medieval History).
  • Burnham, B. C., and Wacher, J., 1990. The Small Towns of Roman Britain (Batsford, London)
  • Burrows, B., and Jones, A. E., 2010 The Staffordshire Hoard, Archaeological Evaluation 2010 (BA report no. 1971.1).
  • Carver, M. O. H., 2005. Sutton Hoo, a Seventh Century Princely Burial Ground and Its Context (Reports of the Research Committee of the Society of Antiquaries of London, No. 69, British Museum Press, London).
  • Champness, C., 2008. 'Watling Street, Hammerwich', in Powell et al. 2008, 57-59.
  • Dean, S., Hooke, D., and Jones, A., 2010. 'The 'Staffordshire hoard': the fieldwork', Antiquaries Journal 90, 139-52.
  • Donnelly, L., 2009. Staffordshire (Anglo-Saxon Hoard): Geoforensic Search (Halcrow, Manchester).
  • Evans, A. C., 2004. 'Hammerwich, Staffordshire: Anglo-Saxon gold necklace pendant' Treasure Annual Report 2004, 82, cat. no. 115 (Department for Culture, Media and Sport, London).
  • Gould, J. 1966. 'Excavations in advance of road construction', Transactions of the Lichfield and South Staffordshire Archaeological and Historical Society 6, 1-16.
  • Gould, J., 1993. 'Caer Luitcoed: its significance in early medieval documents' Transactions of the South Staffordshire Archaeological and Historical Society 33, 7-10.
  • Gould, J., 1997. 'Letocetum: an early vexillation fortress' Britannia 28, 350-2.
  • Gould, J., 1998. Letocetum, The Rise and Decline of a Roman Posting Station (privately printed).
  • Institute for Archaeologists, 2008. Standard and guidance for archaeological field evaluation; Standard and guidance for an archaeological excavation.
  • Jones, A. E., 1998. 'Excavations at Wall (Staffordshire) by E. Greenfield in 1962 and 1964 (Wall Excavation report no. 15)' Transactions of the Staffordshire Archaeological and Historical Society 37, 1-57.
  • Jones, A. E., 2009. Staffordshire Hoard, Archaeological Recovery 2009 (BA report no. 1971).
  • Lyons, F., and Gould, J., 1964. 'A section through the defences of the Roman forts at Wall' Transactions of the Birmingham and Warwickshire Archaeological Society 79, 11-23.
  • McKinley, J., 2008. 'Ryknield Street, Wall (Site 12)', in Powell et al. 2008, 87-190.
  • Margary, I. D., 1973. Roman Roads In Britain (Baker, London).
  • Museums and Galleries Commission, 1992. Standards in the museum care of archaeological collections (Museums and Galleries Commission, London)
  • Parsons, D., 2001. 'The Mercian Church: archaeology and topography' in Brown and Farr (eds.) 2001, 50-68.
  • Powell, A. B., 2008. 'Wyrley and Essington Canal (Site 8)' in Powell et al. 2008, 60-1.
  • Powell, A., Booth, P., Fitzpatrick, A. P., and Crockett, A. D., 2008. The Archaeology of the M6 Toll 2000-2003 (Oxford Wessex Archaeology Monograph No. 2).
  • Rahtz, P., and Meeson, R. A., 1992. An Anglo-Saxon Watermill at Tamworth: Excavations in the Bolebridge Street area of Tamworth, Staffordshire, in 1971 and 1978 (Council for British Archaeology Research Report 83).
  • Round, A., 1983. 'Excavations at Wall 1968-1972' Transactions of the South Staffordshire Archaeological Society 23, 1-23.
  • Rowland, J., 1990. Early Welsh Saga Poetry (Brewer, Cambridge)
  • Simmonds, A., 2008. 'West of Crane Brook Cottage, Hammerwich (Site 34)' in Powell et al. 2008, 62-79.
  • Staffordshire County Council, 2009a. Project Design. The Phase 1 Archaeological Investigation of the Site of the Staffordshire Hoard (unpublished).
  • Staffordshire County Council, 2009b. Project Design. The Phase 2 Archaeological Investigation of the Site of the Staffordshire Hoard (unpublished).
  • Welfare, H. and Swan, V., 1995. Roman Camps in Britain: the Field Archaeology (RCHME, London).
  • Yorke, B., 2001. 'The origins of Mercia' in Brown and Farr (eds.) 2001, 13-22.

Appendix: Geophysical Investigations (by Eamonn Baldwin, Birmingham Archaeology)

Background

In July 2009 a findspot of substantial Anglo-Saxon gold was declared by a local metal-detectorist to the Staffordshire Finds Liaison Officer. The significance of the find was at once recognised and an immediate scheme of archaeological recovery implemented. Accordingly, geophysical investigations centering on the find spot formed part of this scheme. These were undertaken by Birmingham Archaeology on undisclosed farmland near Wall in Staffordshire in the summer of 2009, and were designed to place the findspot within an archaeological context as no other finds or remains had been previously noted in the immediate vicinity. The project was commissioned by English Heritage after consultation with Ian Wykes of Staffordshire County Council.

Geophysical Techniques

Magnetometer survey was chosen for its speed over large areas and for its ability to detect the magnetic response of a wide range of archaeological features, while resistance survey was chosen to follow over a smaller target area on account of its ability to reveal the presence of structural features such as walls, foundations and surfaces. Variations in response (either magnetic or electrical resistance) from subsurface features are normally measured and graded against a local 'background' value of surrounding soils and geology, resulting in contrasting positive or negative values.

Magnetometers detect variations with highly sensitive fluxgate sensors arranged within each probe. Typically, they detect burnt or fired features such as brick walls, hearths, kilns and disturbed building material; as well as features with high or low concentrations of iron oxides such as pits, ditches or foundation trenches. Less magnetically enhanced anomalies such as stone walls will also be discernable.

Resistance meter survey maps soil variation by sending an electrical current between probes inserted into the earth. Subsurface materials (and hence features) vary in their capacity to conduct currents. While the moisture content of the subsurface may effect the strength of the response, resistance survey will detect both features of typically low resistance such as pits, ditches and foundation trenches, as well as features of typically high resistance such as walls, mounds, voids, rubble concentrations and metalled (paved or cobbled) areas.

Data values are recorded simultaneously during survey in a field logger and subsequently represented graphically with a variety of display plots, the most common of which is the shaded greyscale to enhance visually the variations across the survey.

Magnetometer survey

Magnetic survey was undertaken with a Ferex 4.032 magnetometer, manufactured in Germany by Foerster. It was assembled with three 650mm long fluxgate probes mounted at 0.5m separation to a wheeled frame, with samples collected at 0.10m intervals. An integrated Leica 1200 RTK GPS receiver controlled a navigation bar on the datalogger and enabled the survey areas to be surveyed and located in reference to the National Grid to within 200mm accuracy. Survey data were captured in 1.5m wide zig-zag traverses.

Resistance Survey

Resistance survey was carried out with a Geoscan Research RM15 Resistance Meter with MPX15 multiplexer configured with 3 electrodes separated at 0.5m. At each sample station two 0.5m separation readings and one 1.0m separation reading were collected. Sample proceeded every 1m along the traverse line 1.0m apart. This resulted in two final data sets of slightly different depths.

Processing

All raw data were downloaded, processed and transformed into image maps. These were subsequently exported into a Geographical Information Systems (GIS) environment for presentation and interpretation using greyscale plots.

Results

Magnetometer survey was undertaken over a total area of 5.3 hectares. The results revealed a curvilinear ditch-like feature in close proximity to the Anglo-Saxon findspot. The situation of the feature on the crest of a prominent ridge and its proximity to the findspot is noteworthy.

A very straight linear feature was also identified cutting across the survey area, which correlates precisely with the line of a former field boundary evident in historic mapping. Several pit-like responses have been tentatively noted scattered elsewhere in the results.

Resistance survey was undertaken over a total area of 1 hectare. The results correlated precisely with the magnetometer survey by revealing the same curvilinear ditch-like feature and the same former line of the field boundary evident in historic mapping.

The results of the geophysical survey were used to target features for the subsequent phase of trial trenching.