VALE OF GLAMORGAN

(1) Fontygary, Rhoose (NMWPA 2011.215.1) (NMGW-C14CE7) ( fig. 1).Footnote ¹⁰ A hollow copper-alloy handle bearing animal and divine faces, near-complete, with some damage to the socket and crushing on one face. It is 61.8 mm long, 31.6 mm wide, 18.8 mm thick and weighs 83.6 g. On either side of the handle is a face and at its end is a boar's head. Depending on the angle of view some elements of decoration can be read as belonging to either one of the faces or the boar's head. The youthful, clean shaven face can be identified as Mercury by the wings emerging from the hair (which also serve as the boar's ears); the god's features are schematically rendered by incision. The eyes are lentoid, the right containing a central dot to represent the pupil and the left a dot positioned at the top of the eye. Beneath the broad nose and slit mouth, the elongated face ends in a narrow chin. The hair descends in curls on either side of the face. The older face, also elongated, is that of a satyr or Pan, with moustache, long beard and deeply furrowed brow. The horns and goat ears also serve to mark the boar's lower jaw-line. The boar's head is well-shaped with careful modelling and incision is used to represent detail on the face and the underside of the jaw. On either side of the ridged mane, the flaps of the sub-triangular ears lie against the head. Both small lentoid eyes contain a central punched dot to represent the pupil. The snout has an encircling groove and the mouth is slightly open with triangular tusks emerging from the lower jaw. A circular aperture (4 mm in diameter) perforates the snout, and is worn above and below, suggesting its use for suspension. The socket of the handle, which is near oval with a raised collar, has sustained some damage. Lead survives within the socket, presumably as the remnants for a fixing. The surface is worn but is well preserved with a dark green patina. XRF analysis demonstrated that the metal is a heavily-leaded bronze consistent with Romano-British alloys. A close parallel, better preserved, from Niederbieber, with a much more fully naturalistic rendering of the figures, suggests that this was a key handle.Footnote ¹¹

FIG. 1.

Fontygary, Rhoose, handle bearing animal and divine faces (No. 1). Scale 1:1. (Photo: M. Lodwick; © National Museum of Wales)

The specific significance of the iconography is not clear. The representation of Mercury may perhaps relate to the safeguarding of the proceeds of trade. The janiform imagery, seen not only in the paired faces but also in their combination of youth and age, may be appropriate to door-keeping. An association with the boar is more difficult to interpret, though key handles in the form of zoomorphic figures are not uncommon. Parallels for objects bearing faces on opposite sides can be found in Roman iconography (e.g. on some intaglios from BritainFootnote ¹² ) and also in La Tène art.Footnote ¹³

NEWPORT

(2) Langstone (PUBLIC-A435B8) ( fig. 2).Footnote ¹⁴ A near complete cast copper-alloy cheek-piece or toggle. The toggle is 84.5 mm long, 15 mm high, 13.8 mm wide, and weighs 61 g. The mid-section is pierced by a rectangular slot 24.7 mm long and 8.0 mm high. The decorative recess on its upper face is divided into three panels. In the centre is a square panel with two raised ridges running diagonally across the middle forming a sinuous cross and defining four cells, in two of which blue enamel survives. Each end panel is subdivided into three sections, the central ones of which contain blue enamel, while the others are now empty, but perhaps originally held red enamel. The toggle, either side of the decorative panel, is waisted and then flares to circular-sectioned terminals which rise to dome-shaped knops bearing fluted decoration. At the centre of each dome are circular recesses, which may originally have held enamel insets.

FIG. 2.

Langstone, cheek piece or toggle (No. 2) (colour online). Scale 1:1. (Photo: M. Lodwick; © National Museum of Wales)

The toggle is of early Roman date — though of recognised native or late Iron Age form and inspiration — and was found near a hoard of copper-alloy bowls and strainers of similar date.Footnote ¹⁵ The use of polychrome enamel and the fluting on the terminal bosses are Roman elements suggesting a date of manufacture certainly between a.d. 50 and 150 and probably between a.d. 60 and 120. Around 30 examples of toggles are currently known. This is only the fourth known from Wales, with other examples coming from Maescar, near Sennybridge,Footnote ¹⁶ Brecon, Powys (NMGW-2362B2), and near St Brides Major, Vale of Glamorgan (NMWPA 2009.101).

CUMBRIA

(3) Carlisle (LANCUM-C840D2) ( fig. 3).Footnote ¹⁷ A first- to second-century a.d. copper-alloy mount, heavily worn and corroded, in the form of a leopard with its front right leg raised and its paw resting on what may be a human head.Footnote ¹⁸ This mount may be part of horse harness or a rein trace from a cart or wagon. The leopard, with its head high and its tail curled forward and resting on its back, stands on a near-rectangular base. Some of the mouldings representing its spots are still clearly visible on the animal's flanks. The eyes, snout and ears are finely moulded and on the head on which the paw rests some moulding to show the hair can still be seen. At one end of the object, against the leopard's hindquarters, is a shaft and loop. The inner surface of the loop is slightly worn and may have held one end of a chain or rope. The object is 39.5 mm long, 40.5 mm high, 17.5 mm wide, 8 mm thick, and weighs 42.7 g.

FIG. 3.

Carlisle, mount in the form of a leopard (No. 3). Scale 1:1. (Photo: D. Boughton; © D. Boughton)

Although lions are more common than leopards, the motif of a big feline with a human or animal head under or between its paws is also found on other objects, especially key and knife handles. An elaborate key handle depicting a panther with its paw on a ram's head was found at Silchester.Footnote ¹⁹ The existence of similar iconography on funerary monuments means that the motif on mundane objects such as this may have been susceptible to reading as a memento mori.Footnote ²⁰

NORTHUMBERLAND

(4) Glanton (NCL-A38DF3) ( fig. 4).Footnote ²¹ An elaborate copper-alloy harness stud inlaid with well-preserved millefiori enamel dating from the second to third century a.d. The object has a diameter of 41 mm, is 3.2 mm thick, and weighs 19.4 g. The upper surface of the circular stud has three narrow concentric circles of reserved metal separating bands of enamel decoration. The outer band is of blue enamel with rosettes in two rows, their white petals surrounding a white centre contained within a red circle. The inner band has a background of white enamel surrounding fourteen rosettes, their blue petals also surrounding a white centre contained within a red circle. The centre of the stud has a red background with squares in two patterns alternating in a chequer-board arrangement — one a blue square as background for a rosette in the same form as in the outer band of decoration, and the other itself a chequer board of miniature blue and white squares. Similar discs with millefiori decoration, though differing in the detail of their decoration, are known from Chepstow, Mon., and Usk, Mon.Footnote ²² PAS has recorded four studs inlaid with millefiori enamel in a similar style from Hibaldstow, N Lincs. (NLM-3F9905), Wigginton, Oxon. (BERK-555701), Thornham, Norfolk (NMS-40A082), and Freckenham, Suffolk (SF-D17023).

FIG. 4.

Glanton, stud with millefiori enamel (No. 4); for colour image see online journal: http://journals.cambridge.org/bri. Scale 1:1. (Photo: R. Collins; © R. Collins)

NORTH YORKSHIRE

(5) Askham Bryan (SWYOR-374234) ( fig. 5).Footnote ²³ A cast copper-alloy handle from a fixed-blade knife depicting an erotic scene involving two people engaged in sexual intercourse. It is 59 mm long, 31.7 mm wide, 10.7 mm thick, and weighs 64.3 g. The man lies on a couch and is straddled by a woman who faces his feet, which she holds, while the man's left hand rests on her left buttock. A small section of the handle, behind his head, has broken away and there is a circular hole for the knife's tang. The modelling is relatively crude and the features of both individuals are much worn. The handle is corroded with a rough, dark brown patina. The closest parallel noted so far is a handle of unknown provenance with somewhat better modelled features, which was sold by auction at Christie's in December 1998.Footnote ²⁴ The reverse mulier equitans scene is also paralleled on other objects, for example the discus of picture lamps.Footnote ²⁵ Other knife handles portray erotic scenes, but all those otherwise known from Britain involve variations on the theme of three individuals.Footnote ²⁶ Three other examples, all in a similarly provincial style, have been recorded by the PAS from Over Wallop, Hants. (HAMP-4D3135), Monk Soham, Suffolk (SF-A23522), and a fixed-blade knife handle from Syston, Lincs. (LIN-536F87).Footnote ²⁷

FIG. 5.

Askam Bryan, knife handle with erotic scene (No. 5). Scale 1:1. (Photo: A. Downes; © A. Downes)

(6) Fulford (SWYOR-FF96F4) ( fig. 6).Footnote ²⁸ A copper-alloy harness strap loop with a stylised bird's head forming a hook, and peltoid projections to the sides. At the centre of the fitting is a domed boss decorated with punched dots around the rim. An incised circle around the crown has radiating grooves that flare towards the base of the boss and which were possibly decorated with niello. There are traces of a white metal coating on the rest of the fitting. At the back of the dome is a rectangular loop. A flat decorative element in the form of a closed pelta crescent projects from either side of the dome. The bulbous ends of the crescent on the right join to form a closed loop, but the peltoid on the left side is broken. A curved hook projects from the top edge of the dome and terminates in the form of a stylised bird, probably a swan with a pronounced rounded forehead and a flat beak which flares outwards. The hook is twisted slightly to the right. At the bottom of the dome is a broken edge from which there would have been another projection, perhaps in the form of a loop. The break is patinated. The object is 29.6 mm long, 26.6 mm wide, 21.3 mm thick, and weighs 11.8 g. No close parallel has been found, though it bears some general similarities to military harness strap fittings.Footnote ²⁹

FIG. 6.

Fulford, harness strap loop with a stylised bird head (No. 6). Scale 1:1. (Photo: A. Downes; © A. Downes)

NORTH LINCOLNSHIRE

(7) Crowle (FAKL-9900E3) ( fig. 7).Footnote ³⁰ Three joining fragments of a small cast copper-alloy pan decorated with champlevé enamel. The handle and base are missing. The surviving fragments are 107 mm long, 41 mm high and 2.2 mm thick. Most of the enamel is now missing, but the champlevé cells show the decorative scheme consisted of a series of roundels (37 mm in diameter) set around the vessel, each enclosing a three-armed whirligig with the arms containing enamel-filled recesses. It appears that the whirligigs were inlaid with mid-blue and light green enamel and each was surrounded by a ring inlaid with red. The roundels appear to have been set touching each other in pairs. Between each pair were three horizontal bars of enamel inlay, comprising two green bars separated by a band of red. Above and below the horizontal lines were pairs of lung-like motifs inlaid with red, surrounded by green, enamel and containing what appear to be pairs of opposed comma motifs. The distortion of the fragments makes it impossible to determine the original dimensions of the vessel, but it is clear that it was originally convex with a diameter comparable to that of other bowls in the series. The similar Ilam pan has a circumference of 281 mm, suggesting that around 38 per cent of the original circumference of the Crowle bowl survives. The vessel's rim bears an external beading 3.8 mm wide by 2.9 mm thick. Below this is a plain band, 4.2 mm wide, separating the rim from the decorated zone. All of the fractures appear to be old, suggesting that the damage is not recent, and the surfaces are corroded with some loss of patina.

FIG. 7.

Crowle, fragments of pan decorated with champlevé enamel (No. 7) (colour online). Scale 1:1. (Photo: K. Leahy; © K. Leahy)

This pan is the third of its type recorded by the PAS: previous finds are the Ilam (‘Staffordshire Moorlands’) pan (WMID-3FE965) and the vessel found at Winterton, N Lincs. (NLM-F50443).Footnote ³¹ The decoration on the Crowle bowl resembles that seen on the ‘Staffordshire Moorlands’ pan in that both are decorated with ‘Celtic-style’ motifs consisting of curvilinear scrollwork laid out in roundels (eight in the case of Ilam). However, the decoration on the Ilam pan is more fluid than that on the Crowle bowl and all the roundels touch. It is known that enamel-decorated vessels were being made at Castleford (W Yorks.) and while most of the material produced there bore geometric or leaf designs, some ‘Celtic’ designs were applied; one vessel fragment from Castleford, while adorned with leaf-motifs, bears some stylistic similarity to the Crowle find.Footnote ³²

LINCOLNSHIRE

(8) Osbournby (LIN-1213A7) ( fig. 8).Footnote ³³ A copper-alloy hollow object in the form of a human head with caricatured and grotesque features, crudely modelled. The back of the head is plain. The object is 34 mm long, 16 mm wide, 20 mm thick, and weighs 17.4 g. At the top is a small loop for suspension, indicating its likely purpose as a steelyard weight (though its weight is low and traces of lead alloy within the head have not been seen). The grotesque adult male head, elongated and asymmetrical, shows remarkable facial detail. On the left eyebrow is a large projection representing a boil or excrescence. On the forehead above is one or possibly two similar, though smaller protuberances, from which a ridge runs to the nose, further emphasising the creased and lined character of the face. The figure has projecting ears of different sizes, with deeply recessed interiors; these are placed asymmetrically, the left ear being further forward than the right. The socket of the left eye is much more pronounced than the right, which contains some traces of glass. Across the bridge of the prominent and very broad nose is a thick band of flesh, accentuated on the left side of the face by furrows which extend across the cheek; the upper and longer of these may represent a scar. On the right cheek is another large boil. While the tip of the nose points to the right, the tongue lolls to the left as it emerges from the large, thick-lipped mouth. At each corner of the mouth is a hole, though the perforation does not extend across the mouth.

FIG. 8.

Osbournby, steelyard weight in the form of a grotesque head (No. 8). Scale 1:1. (Photo: A. Daubney; © A. Daubney)

Steelyard weights in figural forms are widely known from excavated and PAS examples, though they comprise a small minority of the corpus of weights.Footnote ³⁴ Anthropomorphic examples include images of gods, heroes and humans, but only one parallel of a weight in the form of a caricature has been found — an unprovenanced and much more finely modelled example from the Getty collection.Footnote ³⁵ The depiction of grotesque figures in Roman art has Hellenistic antecedents, best known in the mass-produced terracotta figures of the eastern Mediterranean representing deformity, disability and disease, though in this case no specific condition appears to be indicated.Footnote ³⁶ The Osbournby head is similar in spirit to a much more finely modelled terracotta grotesque head of second- to first-century b.c. date from Smyrna, which has a huge nose, ridged brow and lolling tongue as the most prominent of its distorted features.Footnote ³⁷ Among the small number of other instances of this genre in Roman Britain are the caricatured elderly declaimers and diners among the pipeclay figures associated with a cremation burial in Colchester.Footnote ³⁸ The Osbournby weight further serves to demonstrate the diversity of classicizing forms and genres in circulation in a provincial setting.

(9) Market Rasen (LIN-B8FA27) ( fig. 9).Footnote ³⁹ A second-century a.d. copper-alloy bust bearing a close resemblance to portraits of Antinous. The bust, 84 mm long, 62 mm wide and 35 mm thick, is hollow and the surviving remains of an iron shank suggest that it is a furniture fitting, perhaps from a chest. The bare-chested figure faces forward, his gaze slightly upturned. The very full head of hair is swept forward from the crown, from which radiate thick locks running in a neat fringe across the forehead and falling to the neck on both sides, with an exceptionally thick clump over both ears. Individual strands are delineated by incisions. The face, flat and square jawed, lacks expression, as noted on other similar representations, and there is little fine detail. The thick neck and broad shoulders convey a muscular physique. The recessed eyes and nipples lack the inlay they would originally have held. If correctly identified as Antinous, this is the third such depiction of Hadrian's favourite from Britain; a high quality bust was discovered in Littlecote, Wilts.,Footnote ⁴⁰ and another from Capel St Mary, Suffolk, was recorded by the PAS. The Market Rasen example is very similar to the latter. All three busts are comparable to an unprovenanced example set on a tripod now in Lyons Museum.Footnote ⁴¹

FIG. 9.

Market Rasen, bust in the form of Antinous (No. 9). Scale 1:1. (Photo: A. Daubney; © A. Daubney)

LEICESTERSHIRE

(10) Near Bosworth Field (LEIC-7F4CC3) ( fig. 10).Footnote ⁴² A Roman copper-alloy statue fragment, 91 mm long and 40 mm at its widest. The object is in fair condition with a green patina. The fragment represents the left hand of a third- to half-life-size statue. It is broken just above the wrist and much of the back of the hand is also missing, having broken away more or less along the line of the knuckles. The forefinger is extended while the others curl towards the palm and the tips of the middle finger and thumb touch. Had the hand originally grasped an object, such as a spear or sceptre, it is unlikely that the forefinger would be so extended, but too little is preserved to identify the specific gesture.

FIG. 10.

Near Bosworth Field, hand from bronze statue (No. 10). Scale 1:1. (Photo: W. Scott; © W. Scott)

This represents a further addition to the group of fragments of monumental Roman statues documented by the PAS.Footnote ⁴³ Other examples are more commonly from life-size images. In this case, greater contextual information than for other finds allows its original setting to be suggested. Among the associated artefacts reported from an extensive scatter of Roman period material of first- to fourth-century a.d. date is the largest assemblage of horse and rider brooches (95) yet to be documented from a single site in Britain (revising the figure of 19 given in an initial report on the site), indicating the possible presence of a temple.Footnote ⁴⁴ An (unpublished) geophysical survey has also found possible traces of a temple building.Footnote ⁴⁵ On this evidence the fragment is likely to have been associated with a sanctuary, although its size suggests that it may derive from a subsidiary rather than the principal cult image.

NOTTINGHAMSHIRE

(11) Rolleston area (DENO-BC9E66; 2011 T804) ( fig. 11).Footnote ⁴⁶ A very small gold finger-ring, intact and undamaged, which dates from the third to fourth centuries a.d. The hexagonal hoop has neatly incised decorative edges and triangular shoulders with elaborate openwork. The small hexagonal box bezel, with sloping sides and flat surface, is set with a tiny oval gem, its surface now largely degraded, but what does remain is black in colour and there is evidence of an engraved design. The ring has an external width of 16.2 mm, an internal width of 13.9 by 11.1 mm (the dimensions of the bezel are 8.1 by 7.3 by 2.9 mm), and it weighs 2.46 g. Non-destructive surface metal analysis conducted at the British Museum indicated a gold content of 90–93 per cent and silver content of 5–7 per cent. Raman spectroscopy indicated that the degraded gem is more likely stone than glass.

FIG. 11.

Rolleston area, gold finger-ring (No. 11). Scale 1:1. (Photo: C. Burrill; © Derby City Council)

(12) Maulden (BH-B6AF26) ( fig. 12).Footnote ⁴⁷ A copper-alloy figurine of the Genius-Paterfamilias type, taking the form of a standing male wearing a toga, a fold of which is pulled up over the head. The left forearm extends outward from the elbow and the poorly formed upturned hand holds a patera. The right forearm is missing, and the well patinated break suggests that the damage occurred in antiquity. The elongated face, completely framed by the toga, is somewhat stylised; the large lentoid eyes are set within deeply recessed sockets, the brow ridge extends downwards into a slightly bulbous nose, the mouth is narrow and the chin is squared. The folds of the toga's drapery are more fully modelled on the front of the figure, especially the deep fold which runs diagonally across the body to the left shoulder. The toga hangs almost to the ankles of the figure. The undersides of the heavily corroded feet display an off-white corrosion product, possibly representing a degraded solder. In profile the figure is unusually thin, emphasising, as might be expected, that it was intended to be viewed from the front. The figurine measures 65.1 mm high, up to 30.6 mm wide, 18.1 mm deep, and weighs 47.9 g. Its current appearance follows conservation which has revealed significant additional detail.

FIG. 12.

Maulden, figurine of Genius-Paterfamilias type (No. 12). Scale 1:1. (Photo: J. Watters; © J. Watters)

Parallels for copper-alloy figurines of this type have been found in Britain, including examples from a stratified context within the forum at Silchester Footnote ⁴⁸ and from the excavation of a late third- to early fifth-century Romano-Celtic temple at Lamyatt Beacon, Somerset.Footnote ⁴⁹ A plausible context for its original display is in a lararium.

(13) Pirton (BH-9A1197) ( fig. 13).Footnote ⁵⁰ A lead-filled copper-alloy steelyard weight in the form of a three-dimensional male head, almost certainly representing a god, perhaps Jupiter. The object has been hollow-cast and the inside filled with lead. The face is sloping and has realistically moulded features; the eyes are defined by pointed oval grooves with recessed pupils, the brow extends downwards to the nose, the mouth is small and the lips are closed. As well as its bushy beard the face is also framed by a raised band of curls. The locks of both hair and beard are represented as large clumps. The weight is cracked down one side and this damage extends to the top where an iron prong from the otherwise missing loop is visible. The opposing end of the prong is visible on the flat underside of the weight, which is open and reveals the corroded lead infill. At 73.1 g, this weight may represent three unciae, a quarter of a Roman pound. It measures 29.8 mm high, 28.7 mm wide and 35.3 mm deep. This is a further example of a steelyard weight in figural form from Britain.Footnote ⁵¹

FIG. 13.

Pirton, steelyard weight in the form of a male head (No. 13). Scale 1:1. (Photo: J. Watters; © J. Watters)

OXFORDSHIRE

(14) Crowmarsh (SUR-A1B642) ( fig. 14).Footnote ⁵² A copper-alloy protome in the form of the front half of a prancing winged horse, i.e. Pegasus, now patinated dark grey. The object is 50.9 mm high, its greatest surviving width is 11.8 mm, and it weighs 56 g. The horse has cast detail on its mane, while the eyes have been punched. Only the right side of the neck carries decoration in the form of a group of small punch marks. The surviving part of the wings extends from the back of the neck. The ends of the forelegs, which are not separated on the figure, are missing. The object is broken at two points; at its lower extremity and above the extension on its back. Another horse protome recorded by the PAS is that from Brandon and Bretford, Warwicks. (WMID-774A77), but this, like most similar continental examples, lacks wings. A closer winged parallel of similar size, but more finely modelled, comes from Boudevilliers, Malvilliers (Neuchâtel) in Switzerland. The function of the Crowmarsh object is uncertain, but it may have been a fitting on a piece of furniture, a lamp or a vessel.Footnote ⁵³

FIG. 14.

Crowmarsh, protome in the form of a winged horse (No. 14). Scale 1:1. (Photo: D. Williams; © D. Williams)

(15) South Oxfordshire (BERK-88AAE2) ( fig. 15).Footnote ⁵⁴ A cast copper-alloy key handle in the form of a ram's head. The well moulded horns curl around small downturned ears that project from the side of the head; the tips of the horns also point slightly outwards. Although the ram's facial features are worn, the slightly recessed eyes on either side of its head can be discerned. Some modelling of the nostrils is also visible, especially on the right side of the head. The worn traces of what may be curls from the fleece can be seen between and behind the horns. At the back of the head is a slightly raised band forming a collar around an internal recess, within which is set the shank of an iron tang with traces of the lead alloy solder that held it in place. At the end of the shank, 39 mm long, the upper section of the bit is preserved.

FIG. 15.

South Oxfordshire, key handle in the form of a ram's head. Scale 1:1. (Photo: A. Byard; © A. Byard)

Zoomorphic key handles are widely known from excavated and PAS examples. The only other key handle featuring a ram was reported to the PAS from Winthorpe, Notts., but here it is held between the jaws of a lion.Footnote ⁵⁵ Ram's head terminals are, however, documented on other artefact types, such as handled pans. Zoomorphic decoration on key handles often comprises the forequarters of large felines, especially lions, and dogs, but rarely herbivores, as in this case.Footnote ⁵⁶

(16) Somerton (BERK-817A95) ( fig. 16).Footnote ⁵⁷ A heavily worn copper-alloy figurine, broken below the knees. The figurine is 41.4 mm high, 5.1 mm wide, and weighs 12.3 g. The corrosion products make the attributes of the figurine difficult to identify, but it may depict Attis, though it does not closely resemble any of the more common representations of the god.Footnote ⁵⁸ The figure appears to wear a Phrygian cap and a short belted tunic extending just below the buttocks. In the left hand the figure holds a large circular object, perhaps a tambourine, though this would ordinarily be an attribute of Cybele, while in the right hand is an object which is not easily identified, but perhaps a syrinx or pan pipes.Footnote ⁵⁹ Representations of Attis from the province are rare and it is difficult to point to a single confidently identified example in the form of a figurine.Footnote ⁶⁰

FIG. 16.

Somerton, figurine, possibly of Attis (No. 16). Scale 1:1. (Photo: A. Byard; © A. Byard)

(17) Somerton (BERK-C55677) ( fig. 17).Footnote ⁶¹ A group of Roman lead alloy (pewter) vessels was excavated and recorded in a secure stratigraphic context during excavation after its discovery with a metal detector. The group was found in an upturned pile in a burnt deposit in association with several ceramic vessels — all broken but in many cases almost complete — and several iron nails. This deposit was located within a room, which geophysical survey later showed to be part of a much larger building, probably a villa. A coin recovered from the deposit gives a terminus post quem of a.d. 332 (BERK-AB6E77).

FIG. 17.

Somerton, pewter vessel assemblage (No. 17). (Photo: A. Byard; © A. Byard)

The group comprises the following vessels: a small complete conical bowl with pellet-bead rim (Lee Type 1Footnote ⁶² ); an incomplete pedestal bowl with square flange of a previously unknown form; an incomplete large octagonal dish (Lee Type 2)Footnote ⁶³ with chicken bones found directly beneath it; an incomplete hemispherical pedestal bowl with thick rim (Lee Type 2e);Footnote ⁶⁴ an incomplete hemispherical pedestal bowl with beaded rim (Lee Type 2i);Footnote ⁶⁵ an incomplete flanged dish of under 300 mm diameter (Peal Type 4).Footnote ⁶⁶ The composition of the group is not dissimilar from that of other assemblages of pewter vessels, but such groups are typically recorded from structured deposits in wells and pits and not, so far, from a destruction deposit, as appears to be the case here, where the findspot relates more closely to the context of use.Footnote ⁶⁷

BERKSHIRE

(18) West Ilsley (SUR-BDB067) ( fig. 18).Footnote ⁶⁸ A Roman military cast copper-alloy strap fitting with niello inlaid decoration on three sides. The fitting is rectangular in plan with sloping sides and a slightly rounded top. The interior is hollow and an arching bar of circular section runs lengthwise within it. The long sides of the object are decorated with palmettes, and on the top are opposed isolated fronds. The fitting is 39 mm long, 19.5 mm wide, 29.23 mm thick, and weighs 37.45 g. The fitting, paralleled in a find from Erichem-‘Hooge Korn’, is of a type dated by Nicolay to the first century a.d.Footnote ⁶⁹

FIG. 18.

West Ilsley, strap fitting with niello decoration (No. 18) (colour online). Scale 1:1. (Photo: D. Williams; © D. Williams)

NORFOLK

(19) Acle (NMS-3CECC3) ( fig. 19).Footnote ⁷⁰ A fragment of a copper-alloy figurine comprising the lower leg of a naked figure, presumed to be male, 78 mm long and weighing 101 g. On this solid cast piece, the green patina has also formed at the break below the knee, indicating that it is ancient. The fragment is of the highest quality; individual toes and toenails are separately modelled with the big toe pushed out from the others. The leg bends slightly as if the figure was leaning forwards and placing his weight upon his right leg. The treatment of musculature and the rendering of individual arteries place this piece outside the mainstream of Romano-British art. It is a striking object, both in terms of its large size (the original statuette would have stood at least 300 mm high) and its workmanship. It is impossible to identify the subject, but the slender leg conveys the impression of an athletic young male and the bareness of the foot may suggest divinity. The lower leg of a slightly smaller figurine reported in 2011 from Whittington, Northumberland (NCL-BD0923), is similarly well rendered, though it still bears flashmarks which had been filed. The sophistication of the rendering of both examples makes them unusual within the province.Footnote ⁷¹

FIG. 19.

Acle, lower leg of figurine (No. 19). Scale 1:1. (Photo: A. Marsden; © A. Marsden)

(20) Hillington (NMS-94CA46; 2011 T78) ( fig. 20).Footnote ⁷² A hollow, gold pendant in the shape of a phallus is formed from sheet metal soldered together lengthways. It is rounded at the terminal with a small aperture left open at either end. A separate transverse loop formed from triple-ribbed strip is soldered into position at the top, with separately applied solid globular testicles to either side. Separately applied wire — with irregular transverse grooves on the underside, perhaps to act as keying for the solder — defines the edge of the foreskin. The length (including loop) is 21.5 mm, the width is 3.5–4.5 mm and 10 mm at testicles, and it weighs 2.44 g. The pendant is similar to solid gold examples from Braintree, Essex (ESS-0CDDC1; Treasure case PEE 81), and Knaresborough, N Yorks. (SWYOR-E56143). In Britain the use of apotropaic amulets of this form, in a variety of materials, has sometimes been associated with the presence of the Roman army but examples, including this one, also occur in rural contexts.Footnote ⁷³

FIG. 20.

Hillington, gold pendant in the form of a phallus (No. 20). Scale 1:1. (Photo: E. Darch; © Norfolk County Council)

SUFFOLK

(21) Wetheringsett cum Brockford (SF-EE7435) ( fig. 21).Footnote ⁷⁴ A complete bone amulet in the form of a winged phallus. The wings, with incised decoration on both sides indicating feathers, project from either side of its base. The phallus is straight and cylindrical in form, and the glans is separated from the shaft by deeply incised grooves. On the underside of the phallus, the testicles are represented and hang beneath the pair of wings. There is no sign of any pendant loop or other means of suspension. The object is 28 mm long, a maximum of 35.7 mm wide across the wings, and weighs 3.05 g. It differs from most other similar examples, being made of bone and also lacking a definite means of suspension or attachment. Like the previous example (No. 20) this too derives from a rural context.Footnote ⁷⁵

FIG. 21.

Wetheringsett cum Brockford, bone amulet in the form of a winged phallus (No. 21). Scale 1:1. (Photo: F. Minter; © Suffolk County Council)

ESSEX

(22) Broomfield area (ESS-593DC4; 2011 T650) ( fig. 22).Footnote ⁷⁶ A cast silver zoomorphic brooch in the form of a leaping dolphin. The body is arched and the head terminates in a gilded triangular mouth. There are two raised and gilded circular eyes located just in front of the gilded, elongated triangular dorsal fin which extends halfway along the dolphin's back. The fin is decorated with incised vertical lines along its length. There are two short, pectoral fins with a central groove, set just back from the eyes and terminating before the end of the dorsal fin. The tail curves up and over the body of the dolphin and terminates in a triangular fin. The mouth and upper surface of the tail are damaged. The underside of the dolphin is plain and undecorated. There are double, D-shaped lugs, with central perforation for the missing hinged pin, projecting from the underside of the tail; one is damaged and incomplete. The catchplate, which is now distorted, projects from behind the mouth. The brooch is 30.4 mm long, 11.1 mm wide across the side fins, 9 mm wide across the tail, 8.9 mm thick, 19.7 mm deep, and it weighs 10.2 g. The brooch is similar to examples published by Hattatt (suggesting a date of the first to second centuries a.d.) and to a fine example from Suffolk of a dolphin inlaid with curved niello panels, with a winged Cupid sitting astride it.Footnote ⁷⁷

FIG. 22.

Broomfield area, silver brooch in the form of a dolphin (No. 22). Scale 1:1. (Photo: L. McLean; © Colchester and Ipswich Museum Service)

(23) Colchester (ESS-998F26) ( fig. 23).Footnote ⁷⁸ A hollow pipeclay standing figurine, 155 mm tall, in a cream-brown fabric, possibly representing Juno. The figure has crudely modelled and worn facial features and is heavily draped. A veil extends from the figure's headdress down the back of the neck. The figure's right arm is held against the side of her body and the right hand holds a dish. The left arm is slightly bent, and the hand is hidden beneath the folds of the robe which hangs to the base of the figure. The left leg is bent slightly and the right leg is straight. This and the impression of the body's form revealed by the folds of cloth over the abdomen and thighs give the figure a greater realism. The figure stands on a plain, hollow base and the feet are not visible.Footnote ⁷⁹

FIG. 23.

Colchester, pipeclay figurine, possibly of Juno (No. 23). Scale 2:3. (Photo: L. McLean; © Colchester and Ipswich Museum Service)

The attributes of the figure do not allow for certain identification, but stylistically it has a number of parallels with central Rhineland depictions of Juno of early second-century a.d. date.Footnote ⁸⁰ There is a somewhat similar but incomplete figure from Hofstade in Gent, Belgium, perhaps produced in the Moselle region.Footnote ⁸¹ Beyond the Capitoline triad, the cult of Juno was not widespread in Britain or elsewhere in the northern provinces of the Roman world. Pipeclay and bronze figurines depicting the goddess are uncommon and no direct evidence for her cult has previously been noted in Colchester.Footnote ⁸²

EAST SUSSEX

(24) Lewes (SUSS-0F1783; 2011 T139) ( fig. 24).Footnote ⁸³ An incomplete Roman gold hooked sheet in two joining fragments, identified either as an ear-ring or as a votive plaque or amulet. The maximum width of the object is 17.3 mm, its maximum surviving length is 33.2 mm, and it weighs 2.0 g. It comprises a simple hook at the end of a trapezoid sheet with embossed design consisting of a marginal line enclosing a central row of ribs flanked by pellets. The lower edge of the sheet is broken off and now lost, and there is also associated oblique crease-damage on one border. Immediately adjacent to the broad broken edge and pushed through from the back are three tiny perforations. They are not part of the design and are presumably related to the breaking away of the lower part of the object. The ribbed decoration may, on analogy with terracotta medical ex-votos, represent the uterus. It is possible that this is the first Roman gold votive plaque of this type found in Britain, although medical ex-votos are generally uncommon in the province, which reflects the wider decline by the imperial period in the frequency of anatomical representations as votives.Footnote ⁸⁴ If correctly identified, then the spirit of the object is similar to that of other charms and amulets from the province directed at seeking divine protection for the health of the womb and childbirth.Footnote ⁸⁵

FIG. 24.

Lewes, gold hooked sheet ear-ring or plaque (No. 24). Scale 1:1. (Photo: S. Smith; © Sussex Archaeological Society)

HAMPSHIRE

(25) Twyford (HAMP-FE1374) ( fig. 25).Footnote ⁸⁶ An incomplete and slightly damaged copper-alloy object on a low stand. The object is 39.5 mm high, has walls up to 2.3 mm thick, and weighs 36.55 g. On the convex upper body (38 mm in diameter) the sides rise to four peaks formed by globular terminals. The outer surface is decorated in relief with a vine scroll and elongated leaves beneath each peak. Below this scroll is a narrow band of cable decoration formed by fine, diagonal nicks. The lower body is waisted (13.4 mm in diameter), before flaring to a circular base (c. 29 mm in diameter), which is supported on three small globular feet, a fourth having been lost. The base is decorated with a row of extended triangular cells in red (and another) enamel. On the underside is a small patch of orange-brown corrosion. The object otherwise has a largely blue-green patina. There are some resemblances to the decoration on some rare examples of open copper-alloy lamps, but the purpose of the object is unknown.Footnote ⁸⁷

FIG. 25.

Twyford, copper-alloy object on a low stand (No. 25). Scale 1:1. (Photo: R. Webley; © R. Webley and Winchester Museums Service)

(26) Kingsworthy (HAMP-069741) ( fig. 26).Footnote ⁸⁸ A cast copper-alloy lid from a Roman seal-box with phallic decoration, 24 mm long and 18 mm wide. The lid is in the shape of a teardrop or leaf with a protruding suspension loop which has now broken, as has the tip at the other end. The upper surface is decorated with a faint incised line and within this, placed slightly off-centre, is a phallus, separately cast and riveted in place. The rivet can be seen on the underside of the lid. At the tip of the phallus a circular setting, originally filled with enamel, delineates an outsize glans. Areas of tinning survive around the phallus; the metal is otherwise a dark grey colour. The internal surface bears transverse file marks. Four similar examples have been recorded by the PAS, including instances from Piercebridge, Co. Durham (NCL-40721N6), Ditchingham, Norfolk (NMS-6FF112), Barton Bendish, Norfolk (NMS-1BA274), and Essex (ESS-861674). This form of seal-box adorned with a phallus is widely documented across the North-West provinces.Footnote ⁸⁹

FIG. 26.

Kingsworthy, seal-box lid with phallic decoration (No. 26). Scale 1:1. (Photo: R. Webley; © R. Webley and Winchester Museums Service)

ISLE OF WIGHT

(27) Calbourne (IOW-5FEAA4) ( fig. 27).Footnote ⁹⁰ An incomplete Roman cast copper-alloy double terret of an unusual elaborate form, 93.5 mm high, 66 mm at its greatest width and weighing 221.1 g. It comprises the following elements: lozenge-shaped ornaments forming a domed or croissant-like ‘skirt’ with the remains of an iron shaft on the underside; two conjoined circular loops above; a top ornament formed by a flat oval plate surmounted by a baluster-style knop with mouldings. Finds of terrets with croissant-like ‘skirts’ have previously been confined mostly to Raetia and the two Germanies.Footnote ⁹¹ However, the PAS has now recorded significant numbers of such terrets with a distribution extending from Surrey to Lincolnshire. A double terret with conjoined loops is known from Chinnor, Oxon. (BERK-41B625), but no direct parallel has been found for this piece.Footnote ⁹²

FIG. 27.

Calbourne, double terret of unusual form (No. 27). Scale 1:2. (Photo: F. Basford; © F. Basford)

(28) Shalfleet (IOW-1402E8) ( fig. 28).Footnote ⁹³ An incomplete cast copper-alloy enamelled leopard brooch dating from the second century a.d. It is 20.9 mm high, 31.8 mm long, 16.2 mm thick, and weighs 13 g. The leopard's large head is turned to the right and its features are crudely modelled. Its small ears point upwards, the mouth is a small horizontal groove, and both eyes are circular recesses; the right eye contains what may be traces of decayed enamel. Each of the three short legs represented in a sidelong view of the leopard terminates in a sharp point. The tail is curled around the body and extends horizontally towards the head. Decoration on the body consists of ten unevenly distributed circular cells which are filled or partially filled with decayed enamel. On the back of the brooch are two D-shaped lugs, the corroded remains of a copper-alloy pin, and an incomplete, short catch-plate. The brooch lacks a patina and the surface colour is reddish brown. Two other leopard brooches, both of which represent females, have been recorded by the PAS, one from Little Waltham, Essex (CAM-925392), and the other from Bradfield Combust with Stanningfield, Suffolk (SF9384). This rare type may be a continental import, but less than a dozen known examples have been found from the Netherlands to the Black Sea, while a small number are known from Britain.Footnote ⁹⁴

FIG. 28.

Shalfleet, enamelled leopard brooch (No. 28). Scale 1:1. (Photo: F. Basford; © F. Basford)

DORSET

(29) Tarrant Rushdon (DOR-6E73F1) ( fig. 29).Footnote ⁹⁵ A small copper-alloy mount in the form of a male bust which appears to rise out of a calyx.Footnote ⁹⁶ The hair is swept forward into a short fringe, the ears and eyes are well defined, the latter with clearly delineated lower lids and brow line. The nose is triangular and somewhat flattened, perhaps as a result of damage to the mould. The mouth is narrow, but both lips are visible, the chin is rounded and the head is supported by a broad neck. On the chest is a small, flat oval disc with a crescent beneath it, perhaps a bulla and another amulet, but the means of suspension for neither is indicated. The back is flat with a slightly uneven surface and a rectangular-sectioned integral rivet projects from it at neck level. The features, executed in a fairly naturalistic manner, are reminiscent of portraiture of the first century a.d. The object is 27.3 mm high, 21 mm wide at the shoulders, 14.4 mm deep (excluding the rivet), and weighs 19.3 g. A similar mount has been recorded from Hadham, Herts. (BH-84F731).

FIG. 29.

Tarrant Rushdon, mount in the form of a male bust (No. 29). Scale 1:1. (Photo: C. Hayward Trevarthen; © Somerset County Council)

MILITARY OBJECTS RECORDED WITH THE PORTABLE ANTIQUITIES SCHEME FROM 1997 TO 2011: A SUMMARY

The opportunity is taken in this section to summarise and review the metal objects likely to be associated with the Roman army as reported to the Portable Antiquities Scheme. Individual finds with a military connection have been among the most noteworthy reported to the PAS, in particular the Crosby Garrett helmet and the Ilam (‘Staffordshire Moorlands’) pan, as well as fragments of a diploma found near Scarborough.Footnote ⁹⁷ Among the more commonly occurring artefacts to be reported, some items stand out because of their good preservation — for instance a third-century openwork belt-plate decorated with enamel and millefiori glass from Kingsclere (Hants.)Footnote ⁹⁸ — and others for their typological or iconographic interest — for example a phalera from Yscir, Powys,Footnote ⁹⁹ an unusual late fourth-century openwork buckle from Chepstow showing a stylised helmeted rider,Footnote ¹⁰⁰ a looped strap slide from Cambridge bearing a three-dimensional representation of a horse's head,Footnote ¹⁰¹ and an early Roman stud from Nottinghamshire with repoussé decoration in the form of a male head in profile, reminiscent of examples from Caerleon.Footnote ¹⁰² Finds of military objects — namely weapons, vehicle fittings, belt and harness fittings and mounts and other horse gear — quantified as categories I and J in previous annual summaries, comprise in toto a small percentage of the artefacts reported in any one year (typically c. 1 per cent of metal objects). Nonetheless by accumulation over the last 15 years a substantial body of more than 2,000 objects which may be associated with the Roman army has been documented. Military metalwork has received renewed scrutiny in recent years for its potential not only as a possible proxy form of evidence for the distribution of Roman army units or for insights into soldiers' equipment, but also in relation to the social identities of soldiers and their relationships with the communities amongst which they were stationed.Footnote ¹⁰³ The following provides a summary of this category of artefacts from the first to fourth centuries a.d. recorded by the PAS — the types of object found, their quantities, date and distribution — and briefly considers aspects of their significance and interest.Footnote ¹⁰⁴ Particular emphasis is placed on the frequent occurrence of ‘military’ objects of mid- and late imperial date in a rural setting in the province, especially in midland and eastern Britain.

Table 2 summarises the distribution of metal objects recorded by county, organised into PAS regions. It is based on a review of the descriptions and images of all instances of object types relevant to this category recorded on the database up to the end of 2011. In order to define relevant objects, this survey has used the major studies of Roman military equipment from Britain and adjacent areas. For presenting the data in summary form, the studies by Chapman and Nicolay, respectively of military finds from Roman Wales and non-military sites in the Rhine delta, have been especially useful as a guide, in particular the latter because of its extensive use of objects found during metal-detecting.Footnote ¹⁰⁵ Catalogues of major relevant finds assemblages from excavations have also been consulted.Footnote ¹⁰⁶ In general, artefacts identified as the equipment of Roman soldiers are interpreted as such either by function, representation in images of soldiers and their mounts, epigraphic evidence (especially ownership inscriptions), occasional textual references, and in particular their occurrence on military sites. As scholars working in this area have noted, it is often not easy to attribute individual object types — especially some belt or harness fittings — to a specifically military context, especially in the mid- and later imperial period, because they have also been shown to occur on non-military sites. This is most emphatically demonstrated in Nicolay's recent survey.Footnote ¹⁰⁷ For the purposes of tabulating and mapping finds, the definition has therefore been drawn broadly and not all scholars would accept some of the artefact types presented here as having an association with the Roman army. Nicolay, for example, includes the looped strap mount, a type of harness fitting otherwise not usually considered in this connection.Footnote ¹⁰⁸

TABLE 2.

NUMBERS OF MILITARY ARTEFACTS RECORDED BY THE PAS FROM 1997 TO 2011 BY COUNTY AND TYPE

KEY

A. Arms and armour

B. Buckles and belt plates (1st–3rd century a.d.)

C. Strap ends (1st–3rd century a.d.)

D. Cart fittings

E. Phalerae

F. Misc. harness

G. Pendants (1st–4th century a.d.)

H. Phallic pendants

I. Mounts

J. Phallic mounts

K. Other looped strapped mounts

L. Button-and-loop fasteners

M. Late buckles

N. Late strap ends

O. Crossbow brooches

The categories chosen for Table 2 present the data in approximate chronological order. The offensive and defensive weapons include swords, spears, daggers, chapes, armour and helmets. The vehicle fittings specifically comprise artefacts of first-century a.d. date in the form of an eagle's head above a hexagonal socket with a projecting hook depicting a water bird's head, to which reins may have been secured.Footnote ¹⁰⁹ The other categories are: buckles, plates and strap ends from belts and baldrics of the first to third centuries a.d.; phalerae, i.e. as elements of horse harness; miscellaneous harness elements, principally junction loops, strap unions, strap fasteners; mounts,Footnote ¹¹⁰ primarily decorative fittings for leather strapping in a variety of forms — including square, rectangular, circular with or without a central boss, conical, vulvate, openwork, peltiform, Trompetenmuster, swastika forms — with a separate listing for both looped strap mounts (i.e. with one or two rectangular fixing barsFootnote ¹¹¹ ) and phallic mounts; button-and-loop fasteners; pendants, comprising elements from the military apron and horse harness, with phallic examples listed separately; buckles and strap ends from belts and baldrics of the ‘long’ fourth century a.d.; crossbow brooches.Footnote ¹¹²

The 2,202 objects presented in Table 2 represent a substantial addition to the corpus of military objects. Allowing for variations in definition, the quantity is equivalent, for example, to those used in major surveys of Roman military equipment, such as the c. 1,300 objects examined in Chapman's corpus of evidence from military sites in Roman Wales, the 1,665 instances documented by Oldenstein in his fundamental study of second- and third-century a.d. metalwork associated with the Roman army in southern Germany, or the c. 2,700 objects from over 300 non-military sites in the lower Rhine region assessed by Nicolay, of which 44 per cent were collected by metal-detectorists outside a formal fieldwork setting.Footnote ¹¹³ When individual object categories are scrutinized the significance of the new PAS data is also apparent. In 2008 an additional 177 examples of button-and-loop fasteners were available to compare with those discussed in previous surveys.Footnote ¹¹⁴ Since that time the number reported to the PAS has increased by a further 160 examples. Equally the 21 bird mount vehicle fittings represent a significant number of new finds of objects of this type,Footnote ¹¹⁵ as do the 164 crossbow broochesFootnote ¹¹⁶ or the many more strap ends of fourth-century belts recorded by the PAS which are much more widely distributed than previously known examples.Footnote ¹¹⁷

As in the case of excavated military metalwork, it is also essential to take taphonomic factors into account in assessing the character and distribution of metal-detected material.Footnote ¹¹⁸ The nature of the material and the circumstances of discovery determine some significant differences from excavated assemblages of similar artefacts.Footnote ¹¹⁹ The number of weapons, offensive or defensive, is very small and such items, with occasional exceptions such as the Crosby Garrett helmet, occur in only very fragmentary form. This must in part at least reflect the absence of iron (as well as bone) objects (for obvious reasons) from metal-detected finds, and also the lack of examination of context types that might produce complete or near-complete examples of such objects, for instance graves and votive deposits in pits or rivers. The only example of riverine deposition of arms recorded by the PAS is at Piercebridge; the finds of scabbard elements (slides, runners and a chape) and of fragments of armour (especially scale armour) account for many of the mounts and strap ends as well as the unusually high number of weapons in County Durham.Footnote ¹²⁰ Some individual finds may also derive from votive deposits or graves, perhaps for example the Ilam pan or the Crosby Garrett helmet, but without excavation this remains supposition. The difficulty of identifying, and perhaps consequent non-reporting, by detectorists of fragmentary remnants of shields, helmets and armour (including, for example, tie loops) is also likely to have influenced the low representation of this category.Footnote ¹²¹ The predominance of horse harness and, to a lesser extent, belt elements is not unexpected; fittings certainly or probably related to horse harness — especially mounts and pendants — were also by far the commonest objects among the finds from the Dutch eastern river area.Footnote ¹²² The large number of button-and-loop fasteners has already been noted. The high representation of belt buckles and strap ends of the late Roman period among the objects reported to the Scheme is a phenomenon already known from other studiesFootnote ¹²³ and may well reflect the greater recognisability to detectorists and perhaps also to Finds Liaison Officers of these highly decorated objects, often bearing zoomorphic iconography.

Table 2 also shows that the highest numbers of objects are found in the counties of eastern England from Yorkshire to Essex, with frequent finds also in the central and north-east Midlands, i.e. a zone between the Humber, Severn and Thames estuaries. The quantity of objects recorded from Norfolk is striking, since the overall number of objects from this county entered on the PAS database is otherwise much lower than for other counties in East Anglia.Footnote ¹²⁴ With the exception of an arc from the Solent to the Severn few are reported in the coastal counties of southern England or from upland northern and western England and Wales. This is very similar to the distribution of all Roman period objects reported to the PAS, as comparison with the summaries of object distribution annually reported in Britannia and with the numbers and density of coins by county demonstrates, which must reflect similar biasing factors of regional variability in ancient societies, modern agriculture and detecting practice.Footnote ¹²⁵ An additional influence on the distribution of military objects as recorded by the PAS is of course the scheduled ancient monument status of many garrison sites in northern England and Wales.

The distribution of individual object types is, with some exceptions, similar to that of the general category. For example the distributions of buckles of different dates recorded by the PAS ( figs 30 and 31) have only minor differences in emphasis, with a greater concentration of findspots of first- to third-century a.d. buckles in Lincolnshire and slightly fewer examples in Norfolk and Suffolk than might otherwise be expected.Footnote ¹²⁶ There are very few first-century a.d. buckles or strap ends in this group, especially examples of pre-Flavian date: first- and second-century hinged buckles and second- to third-century buckles with trapeziform extensions comprise the majority of instances. The distribution of late Roman zoomorphic buckles ( fig. 31) shows a slightly greater emphasis in the West Country and East Yorkshire, though the clustering around the Severn Estuary apparent in other recent studies — drawing on metal-detected material not reported to the PAS as well as excavations — is not manifested here.Footnote ¹²⁷ The distribution of pendants of first- to fourth-century a.d. date is also generally similar to that of all the object categories ( fig. 32). A regional emphasis is not visible in the distribution of single pendant types, except perhaps for openwork, leaf-shaped and sub-triangular examples which cluster to some degree in Norfolk. The object with the clearest regional profile in its distribution is the button-and-loop fastener, the majority of findspots being in the east Midlands and north-east England, especially North and East Yorkshire and Lincolnshire. This has previously been noted, but can now be documented on the basis of a very much larger sample.Footnote ¹²⁸

FIG. 30.

Distribution of first- to third-century a.d. buckles recorded by the PAS.

FIG. 31.

Distribution of fourth-century a.d. buckles recorded by the PAS.

FIG. 32.

Distribution of pendants of Roman date recorded by the PAS.

Although the objects discussed here represent a substantial new dataset, it will of course be important in future study to take account of the caveats expressed above. There are significant biases in the PAS dataset, both in the types and in the distribution of finds documented: in particular the upland areas where permanent garrisons were sited have only very limited representation among PAS records. While the overall numbers of objects are quite large, there are very few substantial assemblages from the same site, the main exceptions being the River Tees at Piercebridge, as already mentioned, or the numerous fittings found outside the fort at Brecon Gaer, Yscir, Powys, which also represents an otherwise unusual collection of objects from a recognised military site, on which work is currently in progress.Footnote ¹²⁹ From an initial evaluation, however, it is otherwise only occasionally possible to identify groups of objects from the same or related sites, for example at Brancaster in Norfolk: otherwise even in regions where they are abundant, objects of the types considered here occur in only very small numbers or in isolation amongst other categories of finds, though there is scope for closer examination of local distribution patterns.Footnote ¹³⁰ Nonetheless some points of wider interest can be made. This survey points to the abundance of military objects in rural areas of the province, which after the conquest phase of the mid-first century a.d. lacked garrisons stationed in close proximity to them. Some forts may remain to be discovered here, but while the sites from which these objects originate are difficult to define on the basis of unstratified evidence alone, most are likely to be farms or, more occasionally, villas or small towns, and perhaps occasionally shrines or disturbed burials. Some finds will also be part of the widespread ‘background noise’ of Roman finds produced by ancient and post-antique agricultural practice and geomorphological processes.Footnote ¹³¹ Some of the arguments advanced to explain the presence of military items in towns may apply here, including chance loss by passing troops, the stationing of soldiers either in units or as beneficiarii, the manufacture of objects of this type, and veteran settlement.Footnote ¹³² In some cases their presence may be a product of a scenario outlined by Nicolay, i.e. the return of discharged auxiliaries to the places from which they had been recruited or with which they had become associated during military service, having retained items of equipment to be disposed of in rite of passage rituals on leaving the army or curated as heirlooms.Footnote ¹³³ But as Nicolay also observes, it seems unlikely that such a wide distribution of objects of this type in a broad array of forms can only be the product of military activity. His argument that ‘the terms military and civilian cannot be satisfactorily applied’ in the case of several object types — especially in the belt and harness fittings used for men and animals, particularly horses — after the first century a.d., seems to be strongly reinforced by these data.Footnote ¹³⁴ Unlike Nicolay's case study area, immediately adjacent to the frontier, this province-wide survey shows a distribution of artefacts of this category — especially dress and above all harness items — extending hundreds of miles from the frontier. Objects of this type may prove in future to be more abundant on garrison sites than on any others when excavated data are taken into account; nonetheless the evidence considered here suggests the existence of a general koine of decorative and apotropaic metalwork for dressing and adorning people and horses that extends across military and civilian communities. On the evidence of their distribution this shared repertoire includes items which are influenced by Iron Age tradition, such as the button-and-loop fastener, and those which come from a Classical tradition, such as the phallus motif. It should also be noted that the distribution in the PAS dataset of the many belt buckles and strap ends of the later fourth century a.d. — so extensively discussed for their possible insights into the distribution of late Roman soldiers — extends across a zone in which the practice of adorning the bodies of men and horses with metal ornament had already been well established in former centuries. In this respect their presence arguably represents the changing expression of an established practice. On this basis, it could be argued that the general distribution of items of harness and suspension gear should be established as a reference point against which that of specific artefact types may be compared in order to assess their importance.