PREVIOUS   NEXT   CONTENTS   ISSUE   HOME 

Objects of personal adornment and dress (Finds Function 1)

Introduction

Items of personal adornment formed one of the largest groups of finds from the settlement. Most are Roman, but the brooches in particular include a good range of pre-conquest forms.

A very large number of brooches were found on the site. The majority are pre-conquest and Early Roman types, as would be expected in the south-east, but there is also a good representative sample of 2nd-century and later forms present. It is evident that before AD 43 the Elms Farm population had access to as wide a range of continental forms as the tribal centre at Camulodunum. Brooch use drops from about AD 70, but there is a steady trickle of brooch loss through to the end of the Roman period. A number of the brooch-types represented (Knee, P-shaped and Crossbow) are associated with the military, and perhaps the civilian, administration, and they are indicative of an 'official' presence of some form at Elms Farm during the later life of the settlement. There is also an unexpectedly large number of Late Roman military metalwork items from the site, although some of this may have been brought to the site for recycling. There is a small amount of evidence for brooch manufacture at the settlement, consisting of a number of unfinished items.

Hairpins were fairly abundant within the settlement, with examples made from copper alloy, bone, glass and jet. Hairstyles requiring the use of hairpins were a Roman fashion, and evidently one that quickly became popular in southern Britain after the Roman conquest. Overall, there were slightly more hairpins made from copper alloy than bone, although it is not clear whether this represents a regional variation in depositional practice at the temple site or is whether it is simply a reflection of the relative wealth of the settlement (i.e the inhabitants could afford them).

The majority of the beads found would have formed parts of necklaces, but they were also used in earrings and bracelets, often threaded onto wire. As is usual, glass beads were most common, with a lesser number in jet and shale (Table 57). Apart from one from a 2nd-century layer, all the jet and shale beads were from mid-3rd to 4th-century contexts, reflecting the general popularity of jet jewellery at that time. The glass beads were of standard forms, though one of the annular beads is purple, an unusual colour for glass beads of this form. There is a concentration of glass beads in the temple area, although this may be a reflection of the sampling strategy in this area. The Saxon bead found is notable, as it is one of a very small number of objects from Elms Farm of definite Saxon date (apart from pottery), though it was not from a stratified context. There were ten copper-alloy beads (or possible beads), which is an unusually large number for a Roman site; very few sites produce even a single copper-alloy bead. Only three are from dated contexts, but these span the entire Roman period. Finally, there is a silver 'washer', an unusual form of bead paralleled at Catterick, where it was associated with the magnificent multi-stranded jet necklace from the priest's burial (Mould 2002a; Bell and Thompson 2002, 177). Some of the jet beads could have come from a similar multi-stranded necklace, in particular a group of eight from pit 15042 (Group 650) in Area M. However no connection with the silver washer can be demonstrated, as the latter was in the topsoil, some distance from the pit. Other parts of necklaces, such as metal chains and links, are poorly represented.

The number of bracelets from Elms Farm was not particularly large (Table 58), comprising thirty-six copper-alloy bracelets, one silver, twenty-five shale, and one bead from a segmented bracelet made from jet. Unlike brooches, bracelets were not commonly worn in the Late Iron Age, and there were none at this site from pre-Roman contexts. Indeed, there are only five from Early Roman contexts, and none from mid-Roman. Of the thirty-two bracelets from closely dated contexts, two-thirds come from Period 5-6. The wearing of bracelets can be regarded as a mainly Late Roman practice at this site. Many of the later Roman decorated bracelets can be paralleled at Colchester, and for some at least, a common source is likely. The evidence suggests that there was a votive aspect to the deposition of the later Roman bracelets, though it was probably not a strong tradition at this site. The group certainly does not stand out as numerically significant, as the 300 or so copper-alloy bracelets from Lydney do, for example. However, nearly a third of the bracelets come from the temple area, some from contexts with structured deposits. The three snake bracelets, which have the most overt religious connotations, do not come from the temple area.

The site produced a total of thirty-two finger-rings in various materials, three loose intaglios, and three possible ring fragments (Table 59). Most were 'trinket rings' made from copper alloy or iron, but there were also five silver rings, a gold ring, and a very fine silver-gilt bezel appliqué depicting Leda and the Swan (Figure 447). A range of styles was present, spanning the whole Roman period, though only twelve of the rings were from stratified contexts. One of the intaglios is of interest, as it may provide evidence for the source of some of the jewellery found at Elms Farm (Figure 448). This nicolo intaglio showing a satyr may have derived from a Colchester jeweller's workshop, possibly employing Greek freedmen. If this is indeed the case it is a useful reminder of the cosmopolitan nature of life in the towns of Roman Britain, and also underlines the fact that many of the 'luxury' goods found at Elms Farm would have been made at, or traded through, Colchester.

Evidence for footwear, in common with many other sites, was very limited, as it depends principally on the survival of leather in waterlogged contexts. Fragments of about seventeen shoes came from four late 2nd to 3rd-century wells. Where it was possible to determine, the shoes were of thonged and nailed construction. In only one instance was the evidence present for identification of a calceus, a closed shoe, although it is assumed that this style would have predominated. In total, 788 iron hobnails were recovered (excluding those in situ in shoes), including several groups of nails that must represent discarded footwear rather than individual losses.

There were a few other items from the site that may belong in the personal adornment category, including fragments of several delicate copper-alloy chains and a possible copper-alloy wire bracelet with silver thread wound round it. A small number of perforated Roman coins may represent Early Saxon reuse as pendants; these are occasionally found in Early Saxon graves, and it is thought that the coins may have been used as amulets.

The Brooches by Nina Crummy

Cite this as: Crummy, N. 2015, The Brooches, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.crummy5

Assemblage characteristics

Excluding the brooches from the cremations and pyre features, 246 brooches were recovered from the settlement. Of these over half are unstratified, deriving mainly from machining and site cleaning activities, a few are from contexts that cannot be closely dated, and others are residual. The balance of stratified to unstratified would more closely approach half and half if the brooches from the cremations and pyre features were added, but they have not been included here as the precise number deposited in each case is not always clear.

The total number of brooches recovered, and therefore their statistical value, has also been affected by the unavoidable sampling of areas/features for excavation, and to some extent by illicit metal-detecting. The number of artefacts removed by the latter will inevitably have included many brooches. Legitimate detecting by Mr M. Cuddeford, however, resulted in the recovery of several brooches that have been donated to Essex County Council and are incorporated into this report. They include the only Nertomarus brooch in the assemblage, and the only Aucissa brooch with stamped decoration.

Despite these problems of assemblage integrity, treating the brooches as a unified collection should have some value as an indicator of trade with both the continent and other areas of Britain, and of the social status of the brooch-wearing population. Table 76 presents the assemblage by type, arranged chronologically as far as possible, and by stratigraphic value, that is, by how many were stratified and (reasonably) contemporary with their contexts, how many were residual in very much later contexts, and how many were unstratified. The first column gives the common name of the type, or series of types. Plate brooches have been divided into three groups by date: 1st- and early 2nd-century types, 2nd-century types, and centre-boss brooches. Penannulars have been divided into early and late types. The second column gives the total number of brooches of that type recovered, with only tentatively identified examples given in brackets. Fragments of uncertain type are treated as a separate form. In columns three to five the numbers of stratified, very residual, and unstratified or not phased examples are given. Brooches from contexts given a broad phase-range (i.e. 2-4) are treated as stratified if the first phase is appropriate to the date of the brooch type. The sixth column gives the total in column two expressed as a percentage of the total number of brooches in the assemblage, accurate to the nearest half per cent.

In many cases the number of examples of each type is very small, so any statistical value is slight. It is only when the number exceeds twenty that the balance between stratified and residual and unstratified reflects that of the assemblage as a whole, i.e. is about fifty per cent, and that occurs so rarely that no meaningful variation from this norm can be detected. However, an unexpected result is that few, if any, examples of some of the types current in Period 2, such as Colchesters, Langton Downs and Rosettes, are stratified, in contrast to a good proportion of some later types, such as P-shaped brooches. A more conventional pattern would be that the majority of early brooches would be stratified, with the later ones more likely to have been disturbed by post-Roman activities on the site.

The first column of Table 76 shows that although the majority of brooches are of pre-conquest and Early Roman types, as would be expected in the south-east, a good representative sample of 2nd-century and later forms is also present. It is this number of later forms that provides a contrast to the pattern seen elsewhere in the region and slightly further afield.

There have been several attempts to compare brooch assemblages, all concentrating on the early, well-represented forms, for example Puckeridge-Braughing (Olivier 1988, 52), and King Harry Lane (Stead and Rigby 1989, 100). The Puckeridge-Braughing figures are used here for a simple direct comparison with Elms Farm. Both sites developed in the Late Iron Age, both survived the conquest, and showed continuous occupation through to the end, or close to the end, of the Roman period. Only the major brooch types and series are included, and the percentages given are for the total of the brooches used in Table 77, rather than for the whole assemblage. Three subtotals are given: for forms that chiefly pre-date the conquest; for those that belong to the middle and second half of the 1st century; and for 2nd-century and later types.

It is apparent from Table 77 that only three early brooch types, Nauheim derivative, Simple Gallic/Colchester, and Hod Hill, predominated at Puckeridge-Braughing, representing 21, 26, and 15 per cent of the assemblage respectively. While all three of these types are present at Elms Farm, there are far fewer Hod Hills, forming only 9 per cent of the total, and the proportions of brooch types at Elms Farm are in general much more evenly spread, with only the Nauheim derivative and the Colchester B derivative forming more than 10 per cent of the assemblage.

In the first half of the 1st century AD the number of imported Langton Down and Rosette brooches is fairly evenly matched between the two collections, but Elms Farm relied much more on the Colchester than the Simple Gallic brooch, while the latter slightly outnumbers the former at Puckeridge-Braughing (Olivier 1988, table 7). These figures appear to show that it is unlikely that Elms Farm was a major entrepôt for continental brooches, as it would be expected to show a greater incidence of continental forms than an inland site. However, the wide range of forms present, not all shown on Table 77, such as the Kragenfibeln and the ?Cravat brooch (the latter in cremation pit F2202, Group 317), and the fact that many of the Colchesters at Elms Farm are of the post-conquest small-late variety, shows that before AD 43 the Elms Farm population had access to as wide a range of continental forms as the tribal centre at Camulodunum.

In the middle of the century both sites favoured the cheap Nauheim derivative, matched in popularity at Puckeridge-Braughing by Hod Hill forms. Both types have been taken as indicators of a military presence (Olivier 1988, 52), though this is less certain for the Nauheim derivative, and both were also popular with the native population, as shown by the numbers recovered here. The figures for both types of Colchester derivative are very different between the two sites, both being present in high numbers at Elms Farm but scarce at Puckeridge-Braughing. This may be taken as reinforcing Olivier's interpretation for military activity at the latter.

Brooch use drops at both sites from about AD 70, but Elms Farm shows a steady trickle of brooch loss through to the end of the Roman period. Not all late types are included in Table 77, e.g. the late 4th-century penannular (Figure 442, no. 243), increasing the contrast with Puckeridge-Braughing. Knee and P-shaped brooches occur in particularly good numbers. The latter and many of the former are continental-made, and one reason for their presence may simply be that the site had direct access to continental markets, rather than relying on trade through either London or possibly Colchester. However, Knee, P-shaped and Crossbow brooches are all types associated with the military and perhaps the civilian administration, and it may be that they are indicative of an 'official' presence of some form at Elms Farm.

Evidence for local manufacture

Local manufacture can be identified by the recovery of moulds or unfinished brooches, and there appears to be an unfinished plate brooch from a Period 2B feature on the site (Figure 441, no. 221). It is of an unusual form, being flat and triangular with incurving sides. Its semicircular catchplate has not been bent over to hold the pin; while a continental origin might be expected for this brooch, it seems unlikely that brooches would be exported without being fully finished.

Taken alone, this is not sufficient to demonstrate brooch manufacture as a positive element in the economy of the site, but there is a second oddity on the site which also appears to suggest local manufacture. This is a heavily leaded fragment of crude execution (Figure 441, no. 182), stratified in Period 2, and with no known parallel for its form. It has a spring mechanism similar to Polden Hill brooches and probably a long gently curved bow. It may be testament to a local attempt to create a new brooch type.

Decoration and repairs

Cast decoration, and that essential to the identification of the form, such as the reeding on Langton Downs, or the grooved and incised bow of some Colchester BB brooches, is not included here. Attention is drawn however, to some less standard items, and to the range of forms employing surface plating or enamelling.

Rocker-arm decoration occurs on a Langton Down/Nertomarus brooch (Figure 433, no. 53) and on the triangular plate brooch (Figure 441, no. 221). Another Langton Down may have had niello inlay in punched triangles on the bow (Figure 433, no. 56), and another has decoration punched on either side of a central ridge. One Dolphin has linear punched decoration (Figure 439, no. 157), as does one of the Knee brooches (Figure 440, no. 172). The latter is one of only a small group of similar brooches. One Aucissa brooch (Figure 435, no. 75) has decorative stamps on the head that were applied before the slot for the hinged pin was cut. This ornament links the brooch to a small number of Aucissas from Britain and the continent with similar stamped motifs.

White-metal plating has survived poorly at Elms Farm, but traces remain on a La Tène III brooch, some Langton Downs, Hod Hills, 1st and 2nd-century plate brooches, Knee brooches, and P-shaped brooches. Where identified, the plating is usually tin. One P-shaped brooch has applied gold foil on the bow and white-metal plating on the rest of the brooch (Figure 440, no. 176). Both Centre-boss brooches have traces of mercury-gilding on the front, and one has tin/lead plating on the reverse (Figure 442, nos 235-6).

An unusual brooch is the Trumpet with pelta-shaped bow (Figure 440, no. 166). It has silver wire on the head and a raised border of silver wire around the bow to contain enamel. There is a slight trace of applied silver foil on the side of the foot. Many other brooches were enamelled, in particular the British-made zoomorphic plates, several other plates, the Headstuds, and one Knee brooch, though in many cases the enamel has not survived well and the colours employed are uncertain. Usually the enamel was placed in hollows in the bow or in cells defined by thin walls of metal (cloisonné), but one plate brooch has a field of green and yellow mosaic enamel set in random lines and swirls (Figure 441, no. 229), and another has three small black spots set in a blue field (Figure 442, no. 233). The former was also white-metal plated.

A piece of conical black glass has been inserted into the Centre-boss brooch (Figure 442, no. 235). The boss from the other example is missing. The Gallo-Rhenish plate brooches (Figure 441, nos 216-18) would also originally have been fitted with some form of inset ornament, probably a spot of enamel.

Only one clear example of a repair has been noted. This is the Knee brooch (Figure 440, no. 173), on which the pin and half of the spring have been replaced in iron. There seems to be no purpose in retaining the remainder of the original copper-alloy spring.

Catalogue

Bow brooches

La Tène I brooch

The precise form of this brooch cannot be determined, as little remains apart from the bow. It is similar in size to a Type 1A from Woodcuts, Dorset (Hull and Hawkes 1987, pl. 26, 2600) and a Type IBx from Woodeaton, Oxfordshire (Hull and Hawkes 1987, pl. 27, 2916). Residual in a Period 2B context, it could date to anywhere from the 5th to 3rd century BC.

1. Copper alloy. Only the humped bow, part of the coil, and a small part of the transverse catchplate remain. SF1555, Fill 9231, Pit 9230, Group 288, Area D, Period 2B

Knotenfibeln and various La Tène III brooches

The 1st-century BC occupation on the site is represented by four Knotenfibeln (Figure 431, nos 1-4), three of copper alloy and one of iron. Some of the seven other brooches in this section may also belong to this period.

An unstratified Knotenfibel (Figure 431, no. 2) is delicately made in a copper alloy of unusually fine quality, and may perhaps be of North Italian rather than Gaulish origin (Stead 1976a, 412). The iron Knotenfibel (Figure 431, no. 5) has an undeveloped boss and high bow curve, perhaps more like Feugère's Type 4c than his 8a (1985, 180-1); both types are dated to the last half of the 1st century BC, the former perhaps appearing before the middle of the century (Feugère 1985, 203, 238).

A related brooch is Figure 4311, no. 10, which has a boss set below the bow curve and small side-wings. The low-set boss is matched on a copper-alloy brooch from Grave Y at Aylesford (Hull and Hawkes 1987, pl. S7, 15; Stead 1976a, fig. 4, 1), and on an iron brooch from Presles-Saint-Audebert, Aisne, France, which also has small side wings (Hawkes and Dunning 1931, fig. 11, 1; Hull and Hawkes 1987, pl. S7, 16). However, the external chord and very developed side wings on SF4747 suggest strong affinities with the Colchester series, and a date in the first half of the 1st century AD may perhaps be most likely for it.

The full form of Figure 431, no. 11, is uncertain. The curvature of the bow suggests it is most likely to be a Colchester, but the brooch with low-set boss from Aylesford mentioned above has a similar rib on its upper surface (Hull and Hawkes 1987, pl. S7, 15; Stead 1976a, fig. 4, 1), and this fragment could as easily be accommodated within the wide range of forms represented by other Welwyn period La Tène III brooches.

The lack of a catchplate on the iron filiform brooches nos 6 and 7 means that they cannot be closely dated, but both belong somewhere in the period from the mid-1st-century BC to the mid-1st century AD. The small fragment no. 8 probably also belongs within that date range, but can be placed there with even less certainty.

Figure 431, no. 12 is similar to Feugère's Type 4d, which dates to the first third of the 1st century AD. A 4d came from a Phase 2 grave at King Harry Lane (Stead and Rigby 1989, 98, fig. 50, X3), giving a deposition date of AD 30-55, though Mackreth has suggested a revision to AD 20-40 (1994, 287-8).

2. Copper alloy. The metal is of exceptional quality. Knotenfibel in four pieces, with part of the spring and of the foot missing. The boss is placed just below the curve of the bow and below it the bow is narrow and flat, with flat sides, each with a marginal groove. The foot is squared off, and the catchplate open. The circular hole at the bottom of the catchplate is a result of corrosion. L.74.5mm. SF1387, Cleaning layer 8000, Area E, not dated

3. Copper alloy. Fragment of a Knotenfibel, with most of the bow and part of the spring. The boss is set just below the curve of the bow, which is of rounded flattish section. L. 51mm. SF7844, Fill 23090, Pit 23124, Group 236. Area N, Period 2B.

4. Copper alloy. Two fragments of a small Knotenfibel. One is two coils of the spring, the other the main part of the bow, with a boss below the curve. The section is flat above the boss, round below it. L. 43mm. SF1652, Cleaning layer 6000, Area H, not dated

5. Iron. Complete Knotenfibel, with a small boss just below the high curve of the bow. The spring is of four coils, the chord internal. The round wire bow tapers to a very narrow point. The catchplate is open. L. 67mm. SF4147, Fill 4840, Pit 4786, Group 276, Area K, Period 2-3.

6. Not illustrated. Iron. In five fragments. The upper part of a brooch with four-coil spring and internal chord. The bow section is round. L. about 43mm. Layer 4540, Group 1163, Area K, Not phased

7. Not illustrated. Iron. In two fragments. The upper part of a brooch with the four-coil spring and internal chord. The bow section is round at the curve of the arch, but square below it. L. about 32mm. What appear to be two small mouldings below the arch of the bow are probably the result of corrosion, as they are absent not only on both the reverse of the bow, but also one side. Layer 7103, Group 8015, Area G, not phased

8. Not illustrated. Iron. In three fragments. The upper part of a brooch bow with part of the spring, which has been bent so that it lies at right angles to its original position. The bow is of square section. L. 20mm (main piece only). Fill 13711, Post-hole 13712, Group 2014, Area I, Period 2B-3

9. Iron. The bent and twisted pin of a long brooch with external chord. Most of the four-coil spring remains. L. (bent) 83mm. SF2310, Layer 5441, Group 455, Area J, Period 6

10. Copper-alloy. Distorted brooch with a spring of five coils and an external chord. Most of the pin is missing. There are narrow D-section moulded side wings. The bow is round in section and has a low-set boss. The end of the bow is missing. The catchplate is narrow and has a single perforation. L. 85mm. SF4747, Fill 13538, Pit F13560, Group 175, Area J, Period 2B

11. Tinned copper alloy. Part of a slightly tapering plain bow of circular section with a rib at the centre of both upper and lower surfaces. L. 35mm. SF7811, Cleaning layer 24058, Area M, Not phased

12. Iron. Hinged brooch, complete apart from the catchplate and the ends of the bow and pin. The thick bow section becomes round after the curve. L. (excluding extended pin) 58.5mm. SF1196, Fill 4433, Pit 4517, Group 62, Area K, Period 2

Nauheim brooches and derivatives

The following twenty-nine brooches are the largest group in the assemblage. There are four possible Nauheim brooches, though the identifications of all but one are uncertain. Figure 432, no. 13 has a central groove down the bow, emphasised by rounding the two side elements to give the impression of two adjacent rods. This brooch belongs in the second half of the 1st century BC, or perhaps the early years of the 1st century AD.

Though few of the 1st-century AD derivative forms were stratified in Period 2 contexts, many are likely to be of pre-conquest date, though the form survived into the Roman pre-Flavian period and some may well be post-conquest. Figure 432, no. 17 is unusual, combining the tapering triangular bow and reverse-curve of a Nauheim derivative with a fully-round boss set at the angle of the curve. It belongs in the same tradition that produced the La Tène III brooch above (Figure 432, no. 10), Eye brooches, and Rosettes. A pre-conquest date, possibly early in the century, is appropriate.

The derivative forms have been grouped with those with a reverse-curve bow first (17-18), followed by those with a single curve. The single-curve brooches are then subdivided by bow section.

13. Copper alloy. Two non-joining bow fragments of a Nauheim brooch, and two pin fragments (not illustrated), which may belong to this brooch or the next. The foot and catchplate are missing. The spring is of four coils with an inferior chord. The bow is flat behind, and on the front consists of two rounded mouldings separated by a deep groove, giving the impression of two joined wire rods. What remains of the catchplate is sufficient to show that it was open. L. 58mm. SF1929(i), Layer 5386, Group 455, Area J, Period 6

14. Not illustrated. Copper alloy. A fragment of the spring and upper pin of a different brooch, possibly another Nauheim or a Nauheim derivative. SF1929(ii), Layer 5386, Group 455, Area J, Period 6

15. Copper alloy. Fragment of the upper part of a bow with a heavy four-coil spring and the top of the pin. An iron axial bar runs through the spring. The chord is inferior. The bow is of low D-shaped section. L. 23mm. The heaviness of the spring suggests that this is a Nauheim rather than a derivative form. Nauheim brooches with iron axial bars also occur at Maiden Castle (Wheeler 1943, fig. 83, 11-12). SF1012, Machining layer 4000, Area A, unstratified

16. Iron. The pin and part of the spring of a ?Nauheim brooch with apparently a single coil on one side of the pin and an external chord. L. 60mm. Fill 14466, Pit F14574, Group 259, Area L, Period 2B

17. Copper alloy. Three fragments, most of the bow, part of the spring, and part of the pin. The small triangular bow has marginal grooves and tapers down towards a boss, below which the narrow straight foot kicks forwards. L. 26.5mm. SF2985, Machining layer 11000, Area A, unstratified.

18. Copper alloy. The pin and half the four-coil spring are missing, as is the lower part of the foot and catchplate. The narrow flat reverse-curve bow has knurled marginal grooves and transverse cuts, obscured by corrosion, at the point where the reverse curve begins. L. 41mm. SF3569, Layer 5951, Group 606, Area I, Period 3B

19. Copper-alloy. Lower end of a wide bow tapering to a point. The bow has pairs of marginal grooves. Possibly not of this type as the solid catchplate is cast, not hammered out. L. 30mm. SF3713, Machining layer 11000, Area A, unstratified

20. Copper alloy. Almost complete, only the catch of the catchplate is missing. Spring of four coils. The narrow flat bow is plain apart from two transverse cuts across the lower half. L. 42mm. SF1989, Fill 4661, Pit 14077, Group 276, Area K, Period 2

21. Copper alloy. Two fragments giving most of the bow with one coil of the spring. The narrow flat bow has a deep central knurled groove. L. 38mm. SF7753, Unknown context 22060, J, not phased

22. Copper alloy. Two non-joining fragments, with part of the narrow flat bow missing. The spring is of four coils. Part of the catchplate is missing. Combined L. 30mm. SF2276, Cleaning layer 5543, Area J, Periods 5-6

23. Iron. Simple wire brooch; most of the pin and the end of the bow with the catchplate are missing. The spring has four coils and an internal chord. The bow is flat in section. L. 44mm. Fill 4142, Pit 4141, Group 1147, Area K, not phased

24. Not illustrated. Copper alloy. Three fragments: part of the narrow flat bow, the catchplate, and the upper part of the pin with the curve towards the spring. Combined L. of bow and catchplate fragments 37mm. SF4753, Fill 13546, Pit 13894, Group 3024, Area J, Period 3

25. Not illustrated. Copper alloy. Corroded narrow flat bow and three fragments of the coil with the upper part of the pin. The bow may have had linear grooved decoration. L. 40mm. SF1147, Fill 7029, Post-hole 7027, Group 475, Area G, Period 7

26. Not illustrated. Copper alloy. Crushed flat narrow bow folded in half. Probably of this type. L. 19mm. SF3424, Fill 6727, Ditch 6728, Group 379, Area H, Period 3

27. Not illustrated. Copper alloy. Fragment of the upper part of a narrow flat bow with two coils from the spring. L. 14mm. SF6095, Fill 13639, Pit 13640, Group 594, Area I, Period 3A

28. Not illustrated. Copper alloy. Fragment of a corroded and partly crushed brooch with narrow flat bow and four-coil spring; fragment of the foot with part of the catchplate; small fragment of the pin; and two other small fragments, probably from the bow. L. of largest fragment 20mm. SF7671, Cleaning layer 21615, Group 378, Area J, Period 3B

29. Not illustrated. Copper alloy. Two coils from a four or five coil spring with a short part of a narrow flat bow, twisted at the break. L. 12mm. SF1926, Cleaning layer 5383, Area J, Period 6

30. Not illustrated. Copper alloy. Three fragments together forming a four-coil spring with the pin, bar the very tip, and a small part of a narrow flat bow. L. excluding pin 12mm, with pin 46mm. SF5118, Worn patch 13031, Group 1122, Area J, not phased

31. Not illustrated. Copper alloy. A four-coil spring with the upper parts of a narrow flat bow and the pin. L. 19mm. Possibly a Nauheim, rather than a derivative. SF6222, Cleaning layer 14535, Area L, not phased.

32. Copper alloy. Most of a very narrow D-section bow ribbed down the front. None of the catchplate remains. L. 36mm. SF5404, Layer 4899, Group 749, Area K, Period 3

33. Not illustrated. Copper alloy. Fragment of a D-section bow. L. 30mm. SF1077, Fill 4493, Pit 4463, Group 276, Area K, Period 2

34. Copper alloy. Almost complete, only the end of the pin and most of the foot with the catchplate are missing. Spring of four coils. Round bow. L. 34mm. SF3072, Machining layer 11000, Area A, unstratified

35. Copper alloy. Almost complete, only the end of the pin and the end of the foot with the catchplate are missing. Spring of three coils. Round bow. L. 37mm. SF2994, Machining layer 11000, Area A, unstratified.

36. Copper alloy. Distorted round bow and spring. Spring of four coils. Bow may have been of reverse-curve type. L. 43mm. SF9501, 3999, Recovered from spoil heap by M. Cuddeford.

37. Iron. Fragment of a simple wire brooch with a six coil spring (most of the internal chord is missing) and a short part of the round section bow. The bow has been bent, and its true form obscured. It may have been of continental form, with the bow turning into the head at an angle. L. 30mm. Cleaning layer 10682, Area F, not phased

38. Not illustrated. Copper alloy. Most of the bow of a very corroded brooch, with two coils of a four coil spring. The top of the catchplate remains. Round bow. L. 37mm. SF3311, Machining layer 11000, Area A, unstratified

39. Not illustrated. Copper alloy. Round bow fragment with bend into the spring. L. 26mm. SF4930, Layer 9717, Group 886, Area D, Period 2B

40. Not illustrated. Copper alloy. Upper part of a round bow with one coil of the spring and two other coil fragments. L. 18mm. SF1396, Machining layer 7000, Area G, unstratified

41. Not illustrated. Copper alloy. Upper part of very narrow wire bow, with four coil spring and upper part of pin. The bow has a sharp behind at the head. L. 18mm. SF5963, Fill 17054, Well 17155, Group 86. Area Q, Period 2A

Langton Downs

This group of brooches with cylindrical spring-cover embraces those with the long, usually flat and reeded bow, and the related form with slightly curved triangular-section bow (Figure 432, Figure 433). Other authors separate out those Langton Downs with foliate mouldings on the spring-cover, Nertomarus brooches (Figure 433, no. 53), into a separate group (Stead and Rigby 1989, 95, J), but here Hull's practice of including them among the Langton Downs has been followed (Hawkes and Hull 1947, 317-19). Also included here is a related brooch (Figure 433, no. 58), on which transverse grooves separate differing zones of decoration on bow and foot.

The plain bow of Figure 433, no. 52, is paralleled at Silchester (Cotton 1947, fig. 7, 16) and Bassingbourn, Cambridgeshire (Hull forthcoming, 2406), though the bow of the latter is very wide. An unusual bow form is present on Figure 433, no. 51, which is triangular in section, flat at the front and ridged behind. Several of the brooches are of the Trier type with a square head (Figure 433, nos 47-9; Stead and Rigby 1989, Ee; Olivier 1996, fig. 11.5, 54-5).

A Gallic form originating at the end of the 1st century BC (Feugère 1985, 265-7), Langton Downs are commonly found in south-east British settlements and cemeteries in the first half of the 1st century AD. The distribution centres on the kingdom of Cunobelin, but with a good cluster at Silchester, and extending north to Lincolnshire and west to Dorset and Gloucestershire. Hull believed that the westerly projection followed Cunobelin's iron supply route. The type is so well represented in Britain that the possibility of some being produced here seems high.

In the past, it has been considered that Langton Downs were first brought to Britain by the Roman army (Wheeler and Wheeler 1932, 71-3), an idea followed by Hawkes and Hull (1947, 317), and more recently by Hattatt (1985, 35). The form is, however, well attested as a pre-conquest import from Gaul, for example at the King Harry Lane cemetery (Stead and Rigby 1989, 91-3), and the evidence from Colchester is that the form never found favour with the Roman army or civilian population, as none has been recovered from excavations within the fortress or early colonia, though they were found in some numbers in post-conquest contexts on the native site of Sheepen (Hawkes and Hull 1947, 318). The absence of the form from Londinium confirms this (Hull forthcoming, catalogue). Indeed, how long trade persisted after the conquest is uncertain. At Hod Hill Langton Downs only occur in pre-conquest contexts (Brailsford 1962, 8, fig. 7, C29). Mackreth suggests a terminal date of c. 50/55 (1997, 184) for the form. It is possible that trade may have ceased altogether in AD 43, with those found in post-conquest contexts being those in use at that date. Were this to be the case, their absence from the Colchester colonia and from London could result from economic rather than social factors.

Only two Langton Down brooches from Elms Farm were from stratified contexts (other than those from cremations). One is a fragment of a square-headed Trier-type brooch from a Period 2 pit, and on that basis it has been illustrated, though poorly preserved. The other is also in poor condition, but has been illustrated as it has been broken just above the catchplate and may be related to the similarly broken Langton Downs from the cremations.

Flat bow

42. Copper alloy, with faint traces of white-metal plating. Part of the spring-cover and the lower part of the pin are missing, and the foot of the bow is damaged. The head is curved below transverse mouldings. The bow is strongly reeded, with tapering ridges between the central and outer elements below the head. The catchplate has a simple cut-out. L. 41mm. SF6243, Fill 14609, Well 14984, Group 710, Area L, Periods 4-5

43. Copper alloy. Small example, with part of the spring-cover and most of the pin and catchplate missing. The foot and one side of the bow are slightly damaged. The head is only slightly curved below a transverse moulding. The reeding on the bow is corroded, but it appears that the two outer double elements were knurled. L. 36mm. SF4696, Machining layer 11000, Area A, unstratified

44. Not illustrated. Copper alloy. Small fragment from the junction of a curved head with the spring-cover. The reeding is typical of the type, with the gaps formed by the splaying head filled by tapering mouldings. L. 14mm. SF1014, Machining layer 4000, Area A, unstratified

45. Not illustrated. Copper alloy. Small fragment of a crushed spring-cover and upper bow of curved type. There is a transverse groove across the top of the spring-cover. The bow is reeded, with the central element of the central ridge probably a wavy line. L. 23.5mm. SF7163, Machining layer 17150, Area A, unstratified

46. Not illustrated. Copper alloy. Small worn fragment from the junction of a curved head with the spring-cover. There are traces of reeding on the bow. L. 13mm. SF1015, Machining layer 4000, Area A, unstratified

47. Copper alloy. Very worn and corroded brooch fragment, consisting of the upper part of the bow with the spring-cover and part of the spring. Part of the top of the catchplate survives on the back. The thin flat bow is square-headed, with at least one transverse moulding at the junction with the spring-cover. There are faint traces of reeding in the corrosion on the bow. L. 31mm. This brooch may originally have been quite long. SF7820, Fill 24179, Pit 24181, Group 246, Area M, Period 2

48. Copper alloy. The front of the spring-cover is framed by double grooves. Part of the cover is missing, revealing the spring and iron axial bar. The pin, the lower part of the bow, and all but a tiny part of the catchplate are missing. The junction of bow and head is square and marked by transverse mouldings, the uppermost knurled. The bow turns sharply downwards below this. The central ridge has three narrow mouldings, the middle one of which may be knurled or wavy. The marginal ridges have two mouldings. The panels flanking the central ridge are decorated with a wavy line of punched squares with semicircles in the curves. L. 37mm. SF948, Machining layer 4000, Area A, unstratified

49. Copper alloy. The pin, part of the spring, and the catchplate are missing. The ends of the spring-cover are grooved, and there are traces of incised lines from a frame at the front. The head is square and turns sharply down. The bow is similar to no. 48. L. 49mm. SF9503, 3999, Recovered from spoil heap by M. Cuddeford

50. Not illustrated. Tinned copper alloy. Small fragment from the head. Only part of the spring-cover remains over the spring, which has an iron axial bar. Part of the pin remains. Strong transverse mouldings suggest the bow may have been square-headed. L. 11mm. SF475, Cleaning layer 6000, Area H, not phased

51. Copper alloy. Brooch with part of the spring-cover, part of the spring with the pin, and the catchplate missing. There are strong transverse mouldings at the junction with the head, which is almost square. The unusual bow is not flat but more-or-less triangular, with a ridge running down the back. It has prominent marginal ridges, and below the crest of the head the central ridge consists of a large wavy line which rises well above the side ridges. The catchplate was cast in one with the brooch. L. 39mm. SF7627, Machining layer 11000, Area A, unstratified

52. Copper alloy. Corroded brooch, with the pin, part of the spring, catchplate, and most of the spring-cover missing. A very plain example, with traces of transverse mouldings at the head. The form of the bow below the head is not original. The edges of the bow are missing, but it was clearly narrow. L. 35mm. SF9505, 3999, Recovered from spoil heap by M. Cuddeford

53. Copper alloy. Nertomarus brooch, with most of the pin and spring missing, and some damage to the catchplate. There are foliate mouldings at the head, and the top of the spring-cover is scored with rocker-arm decoration. The ends of the spring-cover have double grooves. The head is curved. The bow has slight marginal ridges, and a low central ridge with two grooves. The catchplate has a single triangular perforation. L. 45.5mm. SF9504, 3999, Recovered from spoil heap by M. Cuddeford

54. Copper alloy. Fragment of the bow with three stumps of metal on the reverse suggesting the catchplate had two perforations. The bow had pairs of mouldings at the margins and a group of three in the centre. The central moulding may have been knurled. L. 44mm. It is worth noting that this brooch, like those from the cremations, has broken just above the catchplate. SF1145, Fill 7173, Pit 7174, Group 853, Area G, Period 3

Triangular-section bow

55. Copper alloy. Fragment consisting of most of the spring-cover with the spring, and the upper part of the bow. The front of the spring-cover is framed by double grooves. The head is strongly curved, with a distinctive hump in front of a moulding. There is a line of knurling down the centre of the bow, which is narrow and only slightly thickened. This brooch may more correctly belong in the previous group. L. 19.5mm. SF1011, Machining layer 4000, Area A, unstratified

56. Copper alloy. Fragment of the upper bow with part of the spring-cover. This is plain and only a small part of the spring remains in the surviving end. The head is slight, with a small hump below a moulding. Both sides of the bow are decorated with a line of punched triangles, which would originally have held niello. L. 20mm. SF687, Machining layer 4000, Area A, unstratified

57. Copper alloy. Worn brooch missing the back and one side of the spring-cover, the pin, and the catchplate. Like no. 55, the head is strongly curved, with a hump in front of a moulding. The bow is slightly splayed and very worn, but clearly has a moulding down each margin, and may have had decoration on the central ridge. The foot is rounded. L. 40mm. SF4816, Cleaning layer 12220, Area R, not phased

Related form

58. Copper alloy. Brooch with many characteristics of Langton Downs. Crushed spring-cover and bow. Part of the spring and the lower part of the pin are missing. The spring-cover is framed with double-grooves. The head is curved, with a hump in front of a moulding. The bow is decorated with a floral and scroll pattern formed by tiny punch marks. The reeded foot is separated from the bow by double transverse grooves. The bottom of the foot appears to have been clipped off. What remains of the catchplate has been pushed over against the back of the bow. L. 35mm. SF2111, Layer 5453, Group 457, Area J, Periods 5-6

Rosettes

Rosettes (Figure 433, Figure 434) are a pre-conquest form from the continent, with a distribution and date similar to that of Langton Downs. Mackreth suggests various end-dates, c. 40 (1994, 292), and c. 45/50 (1995, 974).

They can be variously subdivided, but are not well represented at Elms Farm and here are all grouped together. The earliest example has the Colchester-type method of fixing the spring (Figure 433, no. 59). The main group with cylindrical spring-cover contains two with the central plaque fitted with an openwork frame (Figure 433, no. 60; Figure 434, no. 61; Hawkes and Hull 1947, Type XB; Stead and Rigby 1989, Type F, 'Thistle'), an example with the bow elements cast in one (Figure 434, no. 62; Hawkes and Hull 1947, Type XC), and three with a lion bow (Figure 434, nos 63-65; Stead and Rigby 1989, Type H). Figure 434, no. 63, with the disc cast in union with the spring-cover, belongs late in the series.

59. Copper alloy. An early form, with the Colchester-type method of attaching the spring. In two fragments. The pin, the end of the foot and the catchplate are missing. Spring of seven coils. The external chord is held by a forward hook. The side-wings have vertical ribs. The disc has a vegetal design. The upper part of the bow has a central groove, the lower, just above the plate, has a central wavy line, which continues on the foot, suggesting that the disc was placed lower down than intended. It has been brazed onto the bow. L. 50mm. SF280, Machining layer 4000, Area A, unstratified

60. Copper alloy. Fragment of a small example, with bow, circular base-plate and part of the foot. The bow is reeded. The lunette at the junction of bow and base-plate has triangular motifs and a knurled groove. The base-plate has traces of concentric circular mouldings. The surface of the foot is corroded. It was probably reeded. L. 31mm. SF4691, Machining layer 11000, Area A, unstratified

61. Copper alloy. Upper part only. Part of the spring remains in the spring-cover, which has a double groove along the top at the front, above traces of corroded engraved radiating lines. The bow is reeded. Fragments of iron packing rods remain between the bow and base-plate. The lunette at the junction of bow and plate has triangular motifs and a deep knurled groove in which tiny fragments of the applied openwork rosette remain attached. The plate has shallow marginal knurled mouldings defined by grooves. The foot was cast in one with the bow and plate, making this typologically later (Hawkes and Hull 1947, 314). L. 39mm. SF587, Machining layer 4000, Area A, unstratified

62. Copper alloy. The pin, the edges and end of the foot, and most of the catchplate are missing. The front of the short spring-cover is squared. There is a pronounced grooved moulding in the centre of the bow, which turns sharply down to a reduced lunette. The disk has regularly spaced concentric circular mouldings. The foot has a central knurled or punched flat ridge. The catchplate has two adjacent circular perforations. L. 50mm. SF5707, Machining layer 11000, Area A, unstratified

63. Copper alloy. A very damaged brooch, with only the solidly cast bow in good condition. L. 51mm. Part of the spring, with an iron axial bar, remains in its cover. The bow is of King Harry Lane Type Hb, with the forequarters of two lions set back-to-back and separated by a moulding (Stead and Rigby 1989, 94). Very little remains of the perforated frame or its underlying rhomboidal plaque, and the sides and end of the reeded foot are missing. A stump of the catchplate survives. It had at least two perforations. SF7142, Machining layer 17150, Area A, unstratified

64. Tinned copper alloy. Damaged brooch lacking the spring mechanism and upper part of the bow. L. 41mm. This bow is probably of King Harry Lane Type Ha, with a complete lion, as SF4688 below (Stead and Rigby 1989, 94). The small lion has a rudimentary face and two incised lines representing the mane. Much of the upper frame remains, but very little of the underlying rhomboidal plaque. The reeded foot is damaged on one corner. The catchplate is small and pierced by a single subcircular hole. SF5381, Machining layer 11000, Area A, unstratified.

65. Copper alloy. The spring-cover, part of the spring, and the bow only of a King Harry Lane Type Ha. L. 23.5mm. The spring-cover is plain. The lion is small, of comparative size to SF5381 above. The details of the forequarters have been lost to corrosion. SF4688, Machining layer 11000, Area A, unstratified.

Kragenfibeln

These brooches are rare in Britain, with only four listed in Hull forthcoming (Type 28): two from Colchester, one from Silchester, and one from Dragonby. Of the two Colchester examples one is unprovenanced (Hawkes and Hull 1947, 313, pl. 93, 67), the other was a casual find from the Sheepen site. The date of the form in Britain is uncertain (Hawkes and Hull 1947; Olivier 1996, 245), but the find spots suggest that they must be somewhere in the first half of the 1st century AD.

The sides of the bow of fragment no. 68 are missing, hampering identification. It is probably from a Kragenfibel as Figure 434, no. 66, but may be from a simple Rosette brooch, as Figure 433, no. 59 .

66. Copper alloy. Upper part only. The lower part of the pin and most of the foot are missing, as are the edges of the spade-shaped bow and button. The spring is of six coils. The ends of the side-wings are grooved. The bow is decorated with grooves radiating down from a cross-groove near the top. The button has edge indents. L. 40mm. SF7845, Cleaning layer 23093, Area N, not phased

67. Copper alloy. Fragment. Only part of the spring and spade-shaped bow survive. The bow is divided down the centre by a moulding. L. 23mm. SF3377, Fill 6408, Gully 6409, Group 545. Area H, Period 4

68. Not illustrated. Copper alloy. Fragment consisting of the lower part of the bow, the disc, and the upper part of the foot. The sides of the bow are missing, and appear to have been clipped. Most of the edge of the disc is also missing. The rim is slightly reduced in thickness. The small fragment of foot is narrow, and marked on the top close to the disc by a pronounced step. L. 21mm. SF3378, Area H, Fill 6408, Gully 6409, Group 545. Area H, Period 4

Fantailed brooch

One of a small native East Anglian type of brooch and named the Reversed Fantail by Hattatt (1987, 61-3; 1989, 47). Three of the ten listed by Hattatt are unprovenanced, the remainder come from Norfolk (four), Suffolk (two), and Essex (one). The type uses the Dolphin method for attaching the spring, in which a rearward-facing hook holds the chord. Another characteristic is the angled groove on the wings present on nearly all the brooches of this type and on many Dolphins (Brown 1986, fig. 15, 76-7, 80, fig. 16, 82, 87). The distribution, together with the rearward-facing hook, indicates that this is an Icenian type, dated to the Claudian period.

Hattatt suggested that the short double humps of the head and bow are reminiscent of lion-bow Rosette brooches, and on the Elms Farm brooch this is confirmed. The bow hump has a clearly defined raised area representing a mane. The clarity of this shape may be an indication that the brooch is an early representative of the type.

69. Copper alloy. The spring, rearward-facing hook, pin, catchplate, and one side-wing are missing. The other side-wing is damaged, but sufficient remains to show the angled groove running out from the head towards the terminal. The bow is in the form of a lion. Below this the lozenge-shaped element is decorated with incised grooves. The tip of the foot is also grooved. L. 36mm. SF6478, Prepared surface 5935, Group 102, Area I, Period 2B (Figure 434)

Eye brooch

Together with one from Colchester (Crummy 1983, fig. 4), this debased Eye brooch is the most easterly of its type, most of which come from the Midlands (Hull forthcoming, Type 43). Eye brooches are a German form (Ritterling 1913, 120-3) and this example is probably of Claudio-Neronian date.

70. Copper alloy. Fragment of a plain bow separated from the foot by two slight transverse mouldings. Both bow and foot taper, so that the bow above the mouldings is narrower than the foot just below them. A small part of the catchplate remains. L. 32mm. SF1647, Cleaning layer 6000, Area H, not phased (Figure 434)

Aucissa and Bagendon

Called after a maker who set his name across the head of his brooches, British writers have generally considered that Aucissa brooches developed in the Augustan period in Gaul (Hawkes and Hull 1947, 321; Hattatt 1982, 83). However, Feugère has established that the main series is of Northern or Central Italian origin (1985, 321), with North Italy suggested by Martella (1996, 120). The majority in Britain came in with the army of conquest, but a few may have been imported before the conquest (Mackreth 1981, 134-5; Olivier 1996, 248). With Hod Hills, they died out in the pre-Flavian period.

The form is hinged and characterised by the strongly curved bow and short knobbed foot. The most ornate example in this assemblage is Figure 435, no. 75, with its stamped palmettes and circles. Palmette punch marks are rare but widespread, occurring in Britain at Islip, Canterbury, Ogbourne St George, and Silchester (Hull forthcoming, 2434, 3917, 5445, 4735), on the continent at Augst (Riha 1979, Tafn 23, 600-601, 25, 675), Loupian, Hérault (Feugère 1985, pl. 119, 1517), and in Pannonia (Kovrig 1937, pl. IV, 36), and also on an unprovenanced brooch in Hattatt 1985 (fig. 22, 304).

A small fragment of a Bagendon brooch (Figure 435, no. 76), an Aucissa derivative with a divided bow, came from an unstratified layer. It is unfortunately too damaged for a close parallel to be found. The large area of the head may have carried some form of decoration or mark (Figure 434, Figure 435).

71. Copper alloy. Well-preserved Aucissa brooch, with only the tip of the pin missing. The axial bar is iron. The decoration is of a common type, transverse mouldings on the head and at the junction of bow and foot, two flutings, marginal mouldings and a central knurled rib on the bow. The foot ends in a large knob. L. 58mm. SF7425, Fill 15903, Pit 15902, Group 899, Area M, Period 3

72. Copper alloy. Distorted Aucissa brooch, the pin, foot knob and part of the catchplate are missing. Parts of the iron axial bar remain in the head. The corroded decoration on the bow consists of double marginal mouldings, probably knurled or beaded, and a prominent central rib with double knurled longitudinal mouldings flanking a slight shallow fluting. Transverse mouldings separated bow and foot. L. (bent) 33.5mm. SF4015, Fill 10376, Foundation trench 10382, Group 3068, Area F, Period 3

73. Copper alloy. Well-preserved Aucissa brooch, most of the pin, part of the head, and part of the catchplate are missing. The head was not rolled over to hold the axial bar, but largely cast in one with the rest of the brooch. The surviving terminal knob is integral, so the knob at the other end, now missing, must have been made separately and fixed on only when the copper-alloy axial bar was slotted in to hold the pin. The mouldings on the head include a panel with a wavy line. The marginal mouldings on the head are beaded, and the two beaded mouldings on the sides of the prominent central rib flank a shallow knurled fluting. Transverse mouldings, also beaded or notched, mark the end of the bow, and the foot has longitudinal mouldings. The foot knob has two pronounced collars. L. 55mm. SF6638, Cleaning layer 15281, Area M, not phased

74. Copper alloy. Distorted Aucissa brooch, the pin is missing. Similar, though smaller and meaner, to Figure 434, no. 66, but the central rib is obscured by corrosion. L. (bent) 24mm. SF840, Machining layer 4000, Area A, unstratified

75. Fragment of the head and upper bow only of an Aucissa brooch. The head has a grooved moulding at top and bottom, and cut-out 'eyes' at each side. The centre is decorated by two punched palmettes, the points set between three punched circles. The slot for the pin was cut after the punch marks were applied. The bow has marginal mouldings, and a rouletted groove down the middle of the raised central rib. L. 29 mm. SF9508, 3999, Recovered from spoil heap by M. Cuddeford

76. Copper alloy. The upper part of a Bagendon brooch with divided bow. The head is rolled forward to take the axial bar for the hinged pin. The bow is divided, with the three vertical bars all small and plain, and with no sign of any horizontal bars or lugs. L. 18mm. SF939, Machining layer 4000, Area A, unstratified

Strip-bow

All these fragments are unstratified (Figure 435). Both nos 77 and 79 are probably from copies of Aucissa brooches. No parallel has been found for the trellis decoration on the former.

77. Copper alloy. Twisted and bent fragment, with two panels of raised trellis decoration between plain mouldings. The head is missing. The fragment tapers towards the foot, the end of which has broken off, and only a small part of the catchplate remains. L. 40mm. SF3303, Layer 9425, Group 1301, Area D, not phased

78. Copper alloy. Poorly preserved brooch, with damaged head, foot, and catchplate. The pin is missing. The bow decoration consists of longitudinal mouldings. The bow, now bent to lie straight, would originally have been slightly curved. It appears to taper, but was probably parallel-sided, the edges are not original. L. 51mm. SF1863, Machining layer 4000, Area A, unstratified

79. Copper alloy. Fragment from the base of a bow, with a small part of the catchplate. The bow is flat and has a grooved and knurled central moulding and marginal mouldings. It tapers sharply towards a plain narrow foot. SF9500, 3999, Recovered from spoil heap by M. Cuddeford

Hod Hill brooches

This large and varied series of hinged brooches (Figure 435, Figure 436) was introduced to Britain in large numbers at the time of the conquest, though there is an early form from Skeleton Green (Mackreth 1981, 134-5). It is unlikely to have survived much beyond c. 60/65 (Crummy 1983, 10). So many examples are found with tinning that it is likely that all were originally.

With parallel-sided bow (Hull forthcoming, Type 60)

A very close parallel for Figure 435, no. 80, comes from London (Guildhall Museum Catalogue, pl. iv, 18). The cross-moulded foot is not common on Hod Hills of this form, but in rather different forms also occurs at Colchester (Hull forthcoming, 0443, 3616), Rushall Down (Hull forthcoming, 2853). and London again (Hull forthcoming, 5395). An unprovenanced example in the Roach Smith collection in the British Museum may also be from London (Hull forthcoming, 6598; BM 56.7-1.978). A brooch from Gestingthorpe has a similar bow to SF7022 but a less busily cross-moulded foot (Butcher 1985, fig 8, 5).

The central linear decoration of Figure 435, 81, is rarely found, but occurs, variously described, on Hod Hill brooches from Marks Tey (Hull forthcoming, 0447, 'short notches'), Kingsdown Camp (Hull forthcoming, 2530, 'rectangular punch marks'; 2528, 'incised lines'), and Norfolk (Hattatt 1985, fig 24, 313, 'fine incised lines').

80. Copper alloy. Blackish patches on the bow may be the remains of white-metal plating. Complete apart from the pin. The head has a transverse ridge. The bow has a central knurled ridge flanked by cavetto mouldings, but no marginal ridges. A series of flutes and mouldings runs across the foot. L. 48mm. SF7022, Layer 13568, Group 600, Area I, Period 3B

81. Copper alloy with white-metal plating. Only part of the bow with the foot remains. The catchplate is damaged. The bow, as SF7022 above, has a single central knurled rib flanked by cavetto mouldings, but its sides are damaged and may have developed into marginal ridges, as Crummy 1992a, fig. 6.1, 7. The foot has slightly lower margins, a line of punched rectangles down the centre, and a small transverse ridge above the terminal knob. L. 41mm. SF7274, Post-hole 18577, Group 3029, Area J, Period 3

82. Tinned copper alloy. A short and poorly preserved example, the pin of which is missing. L. 40mm. A single knurled ridge runs down the centre of the bow, and there is a line of knurling across the base of the head. The foot is very short. There is a similar brooch, but with a longer foot, from the Harlow temple (France and Gobel 1985, fig. 39, 27). SF4929, Prepared surface 9708, Group 765, Area D, Period 3B

83. Not illustrated. Tinned copper alloy. Fragment of a narrow bow, foot, and part of the catchplate from a small brooch, probably of this type. (A second fragment in this number is not related.) The bow appears to have had a central ridge and marginal ridges. There are four transverse mouldings, possibly knurled, at the junction of the bow and the short plain foot. L. 32.5mm. A similar and equally corroded brooch comes from Winchester (Hull forthcoming, 6602; BM 92.9-1.1731). SF5839, Layer 6790, Group 484, Area H, Period 3

84. Not illustrated. Copper alloy. The head and bow only. The head is plain, the bow has a broad central ridge and marginal ridges. L. 22.5mm. SF7695, Fill 21966, Post-hole 21965, Group 401, Area J, Period 3B

85. Not illustrated. Copper alloy. Two pieces of a small and very corroded brooch, probably a Type 60. The pin and catchplate are missing. The bow appears to have had marginal ridges but no other decoration, and the foot was probably plain and lacking a terminal knob. L. approximately 35mm. SF2214, Layer 10179, Group 355, Area F, Period 3

With lateral lugs at the base of the bow (Hull forthcoming, Type 61)

86. Brass with traces of tinning. Well preserved and complete apart from the pin, the lug on the left side of the bow, and part of the catchplate. L. 62mm. A knurled transverse moulding separate the bow from the head, and the five ridges on the bow are also knurled. The lug terminates in a collar and a knob. Three knurled transverse ridges separate bow and foot. The latter has knurled margins, and a transverse ridge above the knob. SF4758, Layer 13576, Group 600, Area I, Period 3B

87. Copper-alloy. The head and bow of a very small poorly preserved brooch, with the upper part of the pin remaining fixed in the head. L. 25mm. There is a prominent transverse moulding on the head. The bow is triangular, with stubs of lateral lugs at the base. It has a central ridge flanked by two shallow grooves, and marginal ridges. SF5422, Machining layer 11000, Area A, unstratified

88. Not illustrated. Copper alloy. Reeded bow fragment, probably of this type, though there is only a hint of a lug at the base of the right-hand side of the bow. L. 23mm. The head appears to have been plain, and the junction of bow and foot narrow. SF4910, Layer 9427, Group 1301, Area D, not phased

With lateral lugs at the top of the bow (Hull forthcoming, Type 63)

Apart from the solidity of Figure 435, no. 89, which is rare but not unknown in the series, the distinguishing features of this brooch are the mechanism for fixing the pin, and the half-moon button. The head was not rolled up and forwards, as is usual with the Hod Hill series, but terminates in a cast cylinder, part of which is now missing. It may have been closed, as Feugère's head form e (1985, fig. 5), or open-sided, with the open side to the rear (Hattatt 1982, fig. 11a 'cylindrical wing'), or to the top, as on an equally stout Hod Hill from the Midi (Feugère 1985, pl. 144, 1808). The latter is most likely, particularly as the Midi brooch is a close companion, being of the same type, Feugère's ribbed and 'full-bodied' Type 23d1 (Feugère 1985, 184).

The half-moon mouldings at the junction of bow and foot are another expression of the cross-fertilisation between the Hod Hill series and Kragenfibeln and Thistle brooches summarised by Olivier (1996, 249). Similar mouldings are found on Thistle brooches, being the visible part of the base-plate, cast in one with the bow and foot, around which the discs were fitted (Stead and Rigby 1989, 93, Fa). They are most noticeable when the discs are loose or missing, e.g. Hattatt 1987, fig. 19, 785-7. They can also form the half-disc seen on some Kragenfibeln, such as examples from Silchester (Hull forthcoming, 4800; Reading Museum 03166) and Springhead (Hull forthcoming, 9363; Gravesend Museum 4/15.6). The latter completes the circle back to Hod Hills by being tinned and having a rolled-over head to take a hinged pin, and supports Hull's suggestion that the Kragenfibel was most likely to be the true progenitor of the Hod Hills.

Two other Hod Hills have similar knurled half-moon mouldings, and both are from Essex: an equally stout Type 63 from Fingringhoe (Hull forthcoming, 0461; CM 462.60), and a Hod Hill not otherwise allocated to a type from Camulodunum. The latter is also linked to Figure 435, no. 89, by having a closed cylinder for the axial bar and pin (Hawkes and Hull 1947, pl. 97, 144).

89. White-metal plated copper alloy. A stout example, complete apart from the pin, part of the cylindrical case for the axial bar and pin, and part of the catchplate. There is a prominent transverse moulding across the head. The bow has four knurled longitudinal ridges separated by deep narrow flutings. The marginal ridges are rather lower than those in the centre. The lateral lugs have a rounded moulding flanked by a knurled ridge on the inner side and a small knob on the end, possibly also knurled. The junction of bow and foot is marked by half-moon mouldings, knurled on the edge. The triangular foot has beaded margins. The cast catchplate has a single round hole. L. 50.5mm. SF5634, Fill 6864, Ditch 6868, Group 379, Area H, Period 3

90. Copper alloy. Two fragments of a very corroded brooch. The pin, right-hand lug and catchplate are missing. The lower part of the foot is missing, the upper encrusted by corrosion. There is a transverse moulding on the head. The long thin bow has a central ridge and marginal ridges, and a ridge on the lug. There is at least one knurled transverse moulding separating bow and foot. No details of the foot can be determined. L. 55.5mm. SF3401, Layer 6418, Group 506, Area H, Period 3

91. Not illustrated. Copper alloy with white-metal plating. The head and part of the bow only of a small brooch. The right-hand lug and the left side of the bow are damaged. There is a transverse moulding on the head. The narrow bow has a deep cavetto moulding down the centre, flanked on each side by a pair of close-set knurled ridges, the inner rather wider and higher. There are two knurled ridges on the short more or less triangular lugs. L. 22mm. SF3411, Cleaning layer 6609, Area H, not phased

92. Not illustrated. Copper alloy. Bow fragment with part of the head, probably of this type, though the lugs at the top of the bow are missing. The head has a transverse moulding. The bow has two central ridges flanking a narrow flute and separated from marginal ridges by wider flutes. There were probably transverse mouldings separating bow and foot. L. 23.5mm. SF5420, Machining layer 11000, Area A, unstratified

With an opening on each side of the bow, formed by a curved limb (Hull forthcoming, Type 65)

A brooch very similar to this was found at Baldock, though its foot appears broader (Stead and Rigby 1986, fig. 48,121), and two from Colchester also have open triangles but with plain side bars. The find spot of one is unknown (BM Acton Collection, 53.8-15.55), the other is from Sheepen (Hawkes and Hull 1947, fig. 59, 13). The limited number and narrow range of these four open-triangle brooches suggests that they may not be far from their port of entry, presumably either Colchester or Heybridge. They appear to be even more infrequent on the continent, but their wide distribution may suggest a centre of manufacture in the Upper Rhine area: one comes from Besançon, France (Lerat 1956, no. 233) and another from Augst, Switzerland (Riha 1979, no. 928), and there is a variant with a fantail foot from Germany (Behrens 1954, Abb 7.7). However, given the imaginative range of variation within both the Hod Hill series as a whole, and within Hull's defined types, the wisdom of separating this open triangle group from others within Type 65 that have solid triangles or other solid bow forms flanked by curved limbs is uncertain. The other Type 65s are much more widespread in Britain, with examples as far west as Cirencester (Hull forthcoming, 1407) and Hod Hill (Brailsford 1962, C78), and the related Types 66 (with triangular limbs flanking various bow shapes) and 67 (with open triangular bow) also range from Ham Hill in the west, northwards to Wroxeter, and Colchester and East Farleigh in the east.

93. White-metal plated copper alloy. Complete apart from the pin, part of the catchplate, and the lug on the right-hand side of the bow. The inner triangle of the bow is open, the bars forming its two sides are finely knurled. The junction of bow and foot has a narrow transverse ridge between cavetto mouldings. The foot is plain, with a small transverse ridge above the knob. L. 49mm. SF7421, Fill 15589, Pit 15594, Group 238, Area M, Period 2B

With multiple strong cross-mouldings (Hull forthcoming, Type 74)

This may be a new form as the almost heart-shaped element between the cross-mouldings is unusual.

94. Tinned copper alloy. Most of the bow of a distorted Hod Hill brooch bent sharply over just below the head. The lower part of the bow with most of the catchplate are missing, and only a short stump of the pin remains, held in place by an iron axial bar. Groups of transverse mouldings at the top and base of the bow suggest that this is an example of Type 74. L. (bent) 23mm. SF5726, Machining layer 11000, Area A, unstratified

Unknown form

Only a small fragment of this brooch survives.

95. Not illustrated. The head and part of the bow of a flattened and very corroded brooch, probably a Hod Hill, but possibly an Aucissa. The head has been rolled forwards to take the axial bar for the hinged pin. There appear to be marginal mouldings. L. 22.5mm. SF2873, Fill 5721, Gully 5722, Group 607, Area I, Period 3B

Simple Gallic and Colchester brooches

The British-made Colchester brooch and its continental forerunner (Stead and Rigby 1989, Type Bb) are grouped together here (Figure 436, Figure 437). Both are made in one piece, have small sidewings, and a forward hook to secure the external chord of the spring. The bows of Simple Gallic brooches (96-9) are of flatter section than the native form. Figure 436, no. 100, may also be of this form and is unusual in having a slight reverse curve. At King Harry Lane the Simple Gallic does not appear in graves of Phase 3, but is well represented in those of Phases 1 and 2.

Large forms of the Colchester brooch (Hull forthcoming, Type 90) have a wide date range from Tiberius to Nero, but small examples (Type 91) are late in the series. Evidence for the manufacture of Colchesters has been found at Baldock (Stead and Rigby 1986, 122-3) and also in Cambridgeshire at Foxton (Stead 1998, 133). At Elms Farm large Colchesters are not very numerous compared to the 'small-late'. This is unexpected on a site where other pre-conquest forms are well represented, particularly the Langton Down, and could be taken as an indication that Elms Farm was so well supplied with continental brooches that it had little need to import native examples until after the conquest.

96. Copper alloy. Well-preserved Simple Gallic brooch, only the end of the pin and the catchplate are missing. The bow is of flat section, with a sharp angle at the head. L. 49.5mm. SF6255, Cleaning layer 14573, G=Area L, not phased

97. Copper alloy. Fragment of a Simple Gallic brooch, consisting of the upper part of the bow, one side-wing and stumps of the forward hook and spring. The bow is of flattened roundish section, with a sharp angle at the head. L. 33mm. SF4810, Cleaning layer 12207, Area R, not phased

98. Copper alloy. Bow only of a Simple Gallic brooch, with one side-wing broken off and the other rolled forwards. The bow is flat at front and back, with rounded sides. L. 80.5mm. SF5533, Fill 14258, Pit 14257, Group 36, Area L, Period 2A-B

99. Not illustrated. Copper alloy. Fragment from the base of a long narrow bow, with traces of a very cut-out catchplate. Probably from a Simple Gallic brooch. L. 27.5mm. SF7373, Fill 8990, Well 8989, Group 662, Area P, Period 3-5

100. Copper alloy. Brooch with Colchester-type spring mechanism and slight reverse curve to the bow. The foot is flattened and rather wider than the rest of the bow. This too is probably continental in origin. L. 39mm. SF5433, Fill 14052, Pit 14098, Group 4019, Area K, Periods 4-5

101. Copper alloy. Large Colchester, complete apart from the pin and catchplate The spring is of eight turns, and has an iron axial bar fixed through it. The side-wings are plain. L. 76mm. SF5706, Machining layer 11000, Area A, unstratified

102. Copper alloy. Colchester, the pin and catchplate are missing. An iron axial bar has been passed through the spring, which is partly obscured by iron corrosion but appears to be of six or seven turns. The side-wings are ribbed. The bow is D-shaped in section. L. 51mm. SF378, Fill 4029, Pit 4026, Group 62. Area K, Period 2

103. Copper alloy. Colchester bow with the hook, one side-wing, and one turn of the spring. The hook is long. The bow is stout, and almost round in section. L. 67.5mm. SF2683, Fill 8065, Ditch 8066, Group 844, Area E, Period 6

104. Copper alloy. Fragment of a small Colchester: upper bow with the hook, one side-wing and part of the spring. The side-wings are ribbed and there are transverse mouldings on the hook where it passes over the head. L. 19mm. SF1009, Machining layer 4000, Area A, unstratified

105. Copper alloy. Small Colchester, with parts of the side-wings and catchplate, the pin, and most of the spring missing. There are three rectangular perforations in the catchplate. L. 37mm. SF3579, Machining layer 11000, Area A, unstratified

106. Copper alloy. A small corroded Colchester, complete apart from the pin and part of the spring, and some damage to the catchplate. The spring is of ten turns, and the side-wings may be ridged. The angle of the bow is very sharp, and the bow has a central groove. The catchplate is solid. L. 34.5mm. This is very similar to a brooch from Camulodunum (Hawkes and Hull 1947, pl. 91, 29). SF7162, Machining layer 17150, Area A, unstratified.

107. Copper alloy. Most of the bow and part of the spring of a small Colchester. The small side-wings are plain. L. 31.5mm. SF2393, Fill 7269, Pit 7415, Group 309. Area G, Period 2

108. Copper alloy. Small distorted Colchester bow with a spring of seven turns. The surviving side-wing is plain. L. 30mm. SF2792, Machining layer 4000, Area A, unstratified

109. Copper alloy. Fragments of a spring and a brooch bow, probably a Colchester, with stumps of side-wings, spring and forward hook. Below the curve the bow has been hammered flat. L. 20mm. SF6989, Fill 11379, Pit 11385, Group 61, Area N, Period 2

110. Copper alloy. Most of a plain bow with a trellis-like catchplate, probably from a Colchester, though no parallel has been found for the catchplate pattern. L. 58.5mm. SF2998, Machining layer 11000, Area A, unstratified

111. Not illustrated. Copper alloy. Upper part of the bow of a small Colchester brooch with a marked angle between head and bow. Only stumps remain of the side-wings and the forward hook. L. 23.5mm. SF8124. 3999, Spoil heap

112. Iron. Colchester fragment, with small side-wings, the stumps of spring and forward hook, and the beginning of an open catchplate. L. 35mm. SF3373, Cleaning layer 5597, Area I, not phased

Colchester B derivatives

These derivatives (Figure 437, Figure 438) of the one-piece Colchester are made in two parts, with the separate spring and its external chord fixed in a pierced double lug behind the head. The forward hook of the Colchester survives as a crest running down the bow on a central ridge. Either side of the ridge are cavetto mouldings. Hull dates this form to AD 50-70 (Hull forthcoming, Type 92), though there is a possibility that the form developed before the conquest (Mackreth 1981, 137). A brooch from Southwark is exactly similar to Figure 436, no. 106, on which the cavetto mouldings give way to a bow of rounded section, and the pair may be late in the date range (Wheeler 1946, fig. 27, 20).

Two sub-groups are well represented here. Figure 437, nos 118-22 and unillustrated fragments 123-4 are small solidly made brooches with an incised zigzag running down the bow, the latter the main feature of Mackreth's 'Harlow' family (1995, 959). Brooches shown in Figure 438, nos 127-30, are similar but of more slender construction. On all, the bow slants to the right (viewed down from the head), though on 130 the foot has been pushed back to the left. This distortion has been noted before on similar brooches, and, as it does not occur on more solid examples, can be presumed to be an effect of wear.

The fragment recovered by M. Cuddeford from the spoil heap is one a small group of Colchester B derivatives that has a small foot.

113. Copper alloy. The centre of the bow has been squashed. In two fragments. The pin is missing. The spring has ten-twelve coils. The side-wings are grooved close to the head and at the ends. The crest is cut by angled grooving, the bow ridge has a central groove and is knurled. The unusual catchplate has three openings, which give the impression of a ring set in an open triangle. L. 61mm. The decorative openings on the catchplate are matched on exactly similarly sized plates on B derivatives from Colchester (Hawkes and Hull 1947, pl. 91, 41), Braughing, and Strood (Hull forthcoming, 2314, 8626). SF6072, Fill 8742, Pit 19176, Group 663, Area P, Period 3

114. Copper alloy. Most of a very corroded and distorted brooch in four fragments. The pin and part of one side-wing are missing. The spring is of eight coils. The side-wings are plain. There is no crest, but a fairly wide flat-topped ridge down the centre of the bow, tapering towards the foot. It is decorated with a engraved zigzag line, doubled at the top of the bow. The catchplate is solid. L. approximately 70mm. The bow form and decoration are matched on a brooch from Stebbing (Hull forthcoming, 3314). SF3354, Layer 5936, Group 606, Area I, Period 3

115. Copper alloy. Complete apart from most of the pin. The spring is of eleven coils. The side-wings have a single groove at the end. The crest and bow-ridge are plain. The cavetto mouldings peter out below the crest to give a bow of rounded section with marginal mouldings (see also no. 132). The catchplate has a small triangular opening. L. 57mm. SF5713, Machining layer 11000, Area A, unstratified

116. Copper alloy. In two fragments. The spring mechanism is very corroded and partly missing. The side-wings are plain. A groove passes to one side of the crest and runs onto the ridge. There are marginal mouldings. The catchplate has a large triangular opening. L. 62mm. SF6225/6232, Cleaning layer 14541, Area L, not phased

117. Copper alloy. Corroded brooch in three fragments. The pin is missing. The spring has five coils on one side and seven on the other. The side-wings are plain. As with Figure 436, no. 105, there is no crest. A narrow ridge with engraved zigzag runs down the full length of the bow. The catchplate has a large triangular opening. L. approximately 70mm. SF7016, Machining layer 17000, Area A, unstratified

118. Copper alloy. Solidly made brooch. Two fragments: upper part from [5941], lower from [5945]. Most of the spring mechanism is missing. There is a single groove at the end of each side-wing. The bow ridge has an engraved zigzag line. The catchplate has a triangular opening. L. 42mm. SF3374/8222. Prepared surfaces 5941/5945, Group 603, Area I, Period 3B

119. Copper alloy. Solidly made brooch. The pin, part of the ten-coil spring, and chord are missing. The side-wings are plain. The bow is decorated with an engraved zigzag. The catchplate is solid. L. 42mm. SF6275, Fill 14985, Well 14984, Group 710. Area L, Periods 4-5

120. Copper alloy. Solidly made brooch. Complete apart from one coil of the spring and the lower end of the pin. The spring has ten coils. The side-wings are plain. An engraved zigzag runs down the bow. The catchplate has a single large opening. L. 45mm. SF2606, Machining layer 11000, Area A, unstratified

121. Copper alloy. Solidly made brooch. The pin, and parts of the spring and chord are missing. The side-wings are plain. The bow has an engraved zigzag line. The catchplate has two perforations. L. 41mm. SF5113, Unknown context 13301, Area I, not phased

122. Copper alloy, probably with white-metal plating. Two corroded fragments of a solidly made brooch. The head with the spring mechanism is missing. An engraved zigzag runs down the bow. The catchplate has two small perforations, probably punched through it, as one side is burred. L. 44mm. SF7472, Cleaning layer 20184, Area L, not phased

123. Not illustrated. Copper alloy. Lower part of the bow with damaged catchplate. The bow has an engraved zigzag line. There is a small triangular perforation in the catchplate. L. 24mm. SF3281, Fill 9370, Pit 9218, Group 768, Area D, Period 3A

124. Not illustrated. Copper alloy. Lower part of the bow with the catchplate. There is an engraved zigzag on the bow. The catchplate has two stepped perforations. L. 22mm. SF7154, Machining layer 17150, Area A, Unstratified

125. Copper alloy. Corroded and slightly squashed brooch. The spring is missing. Part of the pin is separate. The bow ridge is grooved, but details of any further decoration are obscured by corrosion. The catchplate is solid. L. 37mm. SF5699, Cleaning layer 8000, Area E, not phased

126. Copper alloy. The pin is missing. The spring has ten coils. The side-wings and bow are plain. The catchplate has two perforations. L. 42mm. SF3352, Layer 5936, Area I, Period 3

127. Copper alloy. The bow is of slender construction and slants to the left. The pin, most of the spring, part of the doubly-pierced lug, the end of one side-wing, and most of the catchplate are missing. The side-wings are plain. An engraved zigzag line runs down the bow ridge. The catchplate may have been perforated, but this is not certain. L. 42mm. SF1553, Fill 9217, Pit 9218, Group 768, Area D, Period 3A

128. Copper alloy. A distorted brooch of slender construction, with the bow slanting to the left. The pin and most of the spring, chord, and catchplate are missing. The side-wings and bow are plain. There is a groove to one side of the crest. L. 43mm. SF3351, Layer 5883, Group 369, Area I, Period 3B

129. Copper alloy. The slender bow slants to the left. The spring mechanism with the pin is missing. One side-wing, the catchplate and the edges of the bow are damaged. The side-wings are plain. There is an engraved zigzag down the bow. There are two perforations in the catchplate. L. 40mm. SF3390, Cleaning layer 6515, Area H, not phased

130. Copper alloy. The slenderly made bow slants to the left, but the lower part has been pushed back towards the right The pin and spring are missing. The upper part of the doubly-pierced lug has worn through. The side-wings and bow are plain. The catchplate is solid. L. 42mm. SF6951, Machining layer 17000, Area A, unstratified

131. Copper alloy. Complete apart from most of the pin and part of the chord. The axial bar is iron. The spring has fourteen coils. The side-wings have astragaloid mouldings with knurled central elements. Below the crest the bow ridge has a central groove and is knurled. There are marginal mouldings. The catchplate is slightly damaged. It has a thickened upper edge, notched for the pin, and two more or less triangular perforations. L. 35mm. SF6097, Fill 13669, Pit 13717, Group 594, Area I, Period 3A

132. Copper alloy. Complete apart from a small part of the pin and part of the chord. The tip of the pin is corroded to the catchplate. The spring has eight coils. The short crest is decorated with a raised zigzag line. The bow has well-defined marginal mouldings. The cavetto mouldings peter out level with the crest, and the bow below is rounded (see Figure 436, no. 106). Unusually, there is a foot-knob. L. 49mm. This brooch is almost certainly from the same factory as three others from the south-east: London (BM 56.7-1.975; Hull forthcoming, 6854), Maidstone, and Strood (Hull forthcoming, 8624-5). All are much the same size, and share the short plain side-wings, the marginal mouldings, and similarly-formed foot-knob. The reduced importance of the cavetto mouldings and central bow ridge suggest a date late in the range for this group. SF411, Fill 4164, Pit 4163, Group 729, Area K, Period 3

133. Very small, corroded, and heavily lacquered brooch. The spring mechanism and pin are missing, also part of the catchplate. Traces of cavetto mouldings are visible on the bow. The catchplate is solid. L. 29mm. SF9502, 3999, Recovered from spoil heap by M. Cuddeford

134. Copper alloy. Lower part of a bow with a small foot and solid catchplate. The bow has marginal grooves, and at the top a low grooved central ridge. It is separated from the foot by a transverse groove. L. 40mm. SF3735, Machining layer 11000, Area A, unstratified

135. Not illustrated. Copper alloy. Upper part only, broken just below the crest. Part of the spring mechanism is missing. The side-wings are plain. L. 14mm. SF7279, Prepared surface 18839, Area I, Period 2B.

136. Not illustrated. Copper alloy. Corroded fragment consisting of a head with part of the spring mechanism. The side-wings have a groove at the end. L. 18mm. SF7872, Fill 17226, Ditch 17314, Group 330, Area Q, Period 2B

137. Not illustrated. Copper alloy. Corroded fragment consisting of a damaged head with part of the spring. L. 13mm. SF5652, Cleaning layer 16017, Area H, not phased

138. Not illustrated. Copper alloy. Damaged head with part of the spring and chord, and a separate spring section. The side-wings are plain. L. 10mm. SF4913, Layer 9427, Group 1301, Area D, not phased

Colchester BB derivatives

These derivatives (Figure 438, Figure 439) are two-piece brooches with the same spring construction as the B form, but lacking the cavetto mouldings on the bow. This is Hull's Type 93, for which he defined three sub-types, all represented here. Being absent from Camulodunum, they appear to be post-Boudican in origin, with a probable date range of c. 65-80.

Grooved bow crossed by oblique incised lines

139. Copper alloy. Well preserved and complete except for part of the six-turn spring and the pin. L. 47.5mm. The side-wings are plain and slightly facetted. The back of the bow has a filed flange, from the mould, running between the catchplate and the underside of the head, which is badly finished. The catchplate has a perforation of irregular shape. SF564, Layer 5149, Group 3024, Area J, Period 3

140. Copper alloy. The foot of the bow, part of the catchplate and the pin are missing. L. 32mm. The side-wings are long and decorated with mouldings. The spring was of ten turns, and has an iron axial bar. The bow is worn and corroded. The catchplate appears to have been perforated, but the form of the hole is uncertain. SF6305, Fill 8888, Pit 8912, Group 663, Area P, Period 3

141. Copper alloy. Upper part only, with part of the eight-turn spring remaining. L. 32mm. An unusually stout brooch, with the deep groove down the bow lacking incised crossing lines. The side-wings are probably plain. SF6482, Machining layer 12000, Area R, unstratified

142. Not illustrated. Copper alloy. In two pieces. The eight- or ten-turn spring is damaged, the pin and part of the catchplate missing. L. 45mm. The side-wings are short and plain. The catchplate had a single large, probably triangular, perforation. SF2331, Fill 6203, Pit 6201, Group 530, Area H, Period 3

143. Not illustrated. Copper alloy. Fragment of the upper part of the brooch only. Part of the six- or eight-turn spring and the pin are missing. L. 23.5mm. Similar to Figure 438, no. 139, but the back is better finished. SF7817, Fill 24116, Pit 24115, Group 247, Area M, Period 2

144. Not illustrated. Copper alloy. Fragment of the head only, with part of the ?six-turn spring. L. 10mm. The side-wings are slightly facetted. SF3561, Layer 5907, Group 601, Area I, Period 3B

145. Not illustrated. Copper alloy. Upper part only. Part of the pin is missing. L. 23mm. The spring had six turns. The side-wings are plain and slightly facetted. SF 829, Cleaning layer 23002, Area N, not phased

146. Not illustrated. Copper alloy. Upper part only. Only a short length of the external chord remains of the spring mechanism. The side-wings are plain and tapered. Two flutes, vestiges of the cavetto mouldings of the B derivative, flank the crest. SF8125, Machining layer 3999, Area X, unstratified

147. Not illustrated. Copper alloy. A very corroded brooch, lacking the spring, part of one side-wing, and the lower part of the bow with most of the catchplate. L. 35mm. The side-wings are short and probably plain. The catchplate had at least one perforation, probably triangular. SF1008, Machining layer 4000, Area A, unstratified

Plain bow

148. Copper alloy. Complete apart from the pin. L. 58mm. The spring has six turns. The side-wings are plain. There are three transverse nicks across the toe. The catchplate has a large triangular perforation. SF6562, Machining layer 11000, Area A, unstratified

149. Copper alloy. Complete apart from part of one side-wing, part of the ?eight-turn spring, the pin, and the toe. L. 43mm. The side-wings are plain but slightly facetted. The upper part of the brooch is twisted to one side. The bow has a nick across the toe, and there was probably at least one other where the toe is damaged (as on Figure 438, no. 148). There is a single round hole in the catchplate. SF1388, Machining layer 8166, Area E, Period 2

150. Not illustrated. Copper alloy. Fragment of the head and upper bow only. L. 14mm. The bow has been bent backwards beneath the side-wings. SF7738, Prepared surface 21959, Group 381, Area H, Period 3

151. Not illustrated. Copper alloy. Fragment of the head and upper part of the bow only. L. 19mm. Both side-wings are damaged, but were probably plain and slightly facetted. SF2778, Machining layer 4000, Area A, unstratified

152. Not illustrated. Copper alloy. The pin and the lower part of the bow with the catchplate is missing. L. 45mm. The spring, of six turns, has an iron axial bar. SF5749, Machining layer 11000, Area A, unstratified

The crest continues as a ridge down the bow

153. Copper-alloy. Almost complete Colchester BB brooch, only the pin and part of the six-turn spring are missing. The side-wings are plain. The crest continues as a ridge down the bow. The toe juts forward slightly. The catchplate has two large stepped perforations. L. 50mm. SF1482, Fill 10071, Gully F10072, Group 3011, Area I, Period 3B

154. Copper alloy. Complete apart from the spring and pin. A small fragment of the external chord remains in the upper perforation of the double lug. The side-wings are plain. The ridge continues down the bow for only a very short distance. The catchplate is solid. L. 41mm. SF6633, Cleaning layer 15468, Group 8006 Area M, not phased

Dolphins

The Dolphin (Figure 439) is a Colchester derivative which takes its name from the comma-shaped bow and thick rounded head (Hull Type 94). When sprung, rather than hinged, the brooches have a distinctive rearward-facing hook to retain the external chord. The side-wings are usually long and may be plain or decorated, and the pin of hinged examples is held in the centre of a thick, usually rounded, cross-bar. Sprung dolphins are Claudian in date but the hinged variant may be rather later (Crummy 1983, 14). Number 155 is a hybrid, with a single perforated lug replacing the rearward hook.

155. Copper alloy. Corroded brooch with the spring, pin, catchplate, and part of one side-wing missing. A hybrid, with a single perforated lug replacing the usual Dolphin rearward hook to hold the external chord, part of which survives. This arrangement is reminiscent of the double lug of the preceding Colchester derivatives. The side-wings retain traces of mouldings. The bow has a slight groove on the angle, flanked by lines of nicks, which continue down towards the foot. The groove may also continue downwards, but is obscured by corrosion. L. 52mm. SF912, Machining layer 4000, Area A, unstratified

156. Copper alloy. Well-preserved sprung brooch, with the spring, pin, and catchplate missing. The side-wings are moulded, with the penultimate moulding knurled. A slight knurled rib runs down the centre of the bow. A line of small punched dots runs from either side of the rearward hook down to the junction of the underside of the bow and the wings, and then back up to meet the central rib. L. 46mm. SF4295, Machining layer 11000, Area A, unstratified

157. Copper alloy. Sprung brooch, with rearward hook, pin and catchplate missing. The spring has thirteen turns. The side-wings are decorated with incised lines. The bow has a plain central rib, and incised lines. L. 44mm. SF4807, Cleaning layer 12207, Group 8010, Area R, not phased

158. Copper alloy. Corroded sprung brooch. The spring, pin, one side-wing, and most of the catchplate are missing. The side-wings were long. No decoration can be distinguished. The bow retains traces of a small central rib. The catchplate was solid. L. 51mm. SF5970, Layer 17201, Group 325, Area Q, Period 2B

Polden Hill

Polden Hills (Figure 439) are distinguished by the method of passing an axial bar through the spring and fixing it in the returned ends of the semi-cylindrical cross-bar. They can sometimes, as here, be similar to the Dolphin in profile. This example is probably pre-Flavian in date.

159. Copper alloy. Poorly preserved brooch, the spring mechanism and catchplate are missing, and the side-wings are damaged. The side-wings are moulded, and the bow has a grooved central rib flanked by flutings and slight marginal ridges. L. 50.5mm. SF2411, Layer 7000, Area G, not phased

T-shaped

The series of British-made T-shaped brooches is huge and very varied, but not often found in great numbers in the south-east. Enamelled Lamberton Moor brooches are an exception, being more in evidence in the south than the north (Figure 439, no. 161; Figure 440, no. 162). These brooches (Figure 439, Figure 440) probably belong to the later 1st and 2nd century.

160. Copper alloy. Fragment of a hinged brooch, with the pin and lower half of the bow missing. The crossbar is long, thin and moulded. The head is wide and triangular, with a prominent grooved and moulded panel running part way down the bow. L. 29mm. SF5733, Machining layer 12260, Area B, unstratified

161. Copper alloy. Large sprung Lamberton Moor headstud brooch. Most of the loose head-loop, part of the pin and part of the catchplate are missing. The head-loop passes through the spring to fix it in a lug at the back of the head. The short arms are ribbed. The panel below the stud has a central row of enamelled lozenges flanked by triangles. Nearly all the enamel from this panel and from the stud is now missing, though there are traces of red enamel in one cell. L. 54mm. SF6619, Cleaning layer 15044, Area M, not phased

162. Copper alloy. Small hinged Lamberton Moor headstud brooch, with part of the fixed head-loop, pin, and catchplate missing. The short arms are flat at the front and ribbed. The panel below the stud has a central row of enamelled lozenges flanked by triangles. The lozenges were probably red, what remains of the enamel in the triangles is now discoloured green. L. 41mm. SF2321, Fill 6213, Post-hole 6209, Group 561, Area H, Period 4

163. Copper alloy. Small corroded brooch, with most of the hinged pin and catchplate missing. There is a grooved and moulded boss on the upper bow, with an enamelled panel of triangles and lozenges running down to the forward-pointing foot. Nearly all the enamel has decayed, but appears to have been red in one triangle. L. 38mm. SF6476, Fill 18353, Ditch 18358, Group 593, Area I, Period 3A

164. Copper alloy. Fragment of a small hinged brooch with a stud for enamel (now missing) on the head. Below the stud a series of triangular cells runs down either side of the bow. The enamel from these is also missing. L. 15mm. SF8121, Machining layer 3999, Area X, unstratified

Trumpet

Trumpet brooches are also not common in the south-east (Figure 440). Figure 440, no. 165, is an early form, with a wide date-range from Flavian to Antonine, while the pelta-bow form of no. 166 belongs to the second half of the 2nd century.

165. Copper alloy. Corroded fragment, with part of the head-loop, the pin, most of the foot, and the catchplate missing. The bow is flat-backed with a frontal three-leaved acanthus button. The head-loop doubles as an axial bar, fixing the spring in a lug behind the head. There is the stump of a spike on the head. L. 36mm. SF1018, Machining layer 4000, Area A, unstratified

166. Copper alloy, with applied silver wire and silver foil. Brooch with enamelled pelta on a straight bow. Corroded, but complete apart from slight damage to the penannular or annular foot. The headloop is integral. Silver wire runs over the head. The pelta has a central raised dot and circle, flanked by smaller dots, and a raised border of silver wire. The latter no longer follows the edge on the left-hand side. It has presumably become detached and pushed inwards. No enamel survives. There is a trace of silver foil on the side of the foot. L. 40mm. SF5292, Prepared surface 13063, Group 365, Area I, Period 3

Knee brooches

Of the two main groups of Knee brooches (Figure 440), those with the spring hidden by a plate on the head and those with it enclosed in an open-backed closed-ended cylindrical or rectangular head, only the latter is represented here. Most of these are British varieties of the group, though the distinctive no. 172, with punched scroll decoration, is a continental form rarely found in Britain, where it only occurs in the south-east (Hull forthcoming, Type 177).

The form of no. 168 is unusual, approached only by one from Guildford, Surrey (Hattatt 1989, fig. 91, 1660), and two from Silchester, Hampshire (Hull forthcoming, 4879-80, included in Type 173), though these all have a cylindrical rather than rectangular spring-cover, and varying patterns of enamel.

The pin and part of the spring of no. 173 have been replaced in iron.

Knee brooches belong to the 2nd and early 3rd century, and derive predominantly from northern, usually military, sites, with those in the south mostly coming from large towns, of which the closest to Elms Farm is Colchester (Hattatt 1987, 262). They are military brooches (Riha 1979, 85) that occur also in civilian contexts, and so might have been used as 'badges of office' by the civilian administration. If the latter is an accurate interpretation, then the high numbers from Elms Farm (seven compared to Colchester's twelve (Hull forthcoming, catalogue (8); Crummy 1983, 14-15 (3); 1992a, 143 (1)) might be due to the presence of the temple, the market-place, or the port.

167. Copper alloy. Three fragments, of which the largest consists of the upper part of a distorted bow with cylindrical head. The corroded spring probably has six coils. The curved bow is of low D-shaped section and tapers slightly below the head. L. 19mm. SF4952, Fill 15058, Post-hole 15057, Group 1195, Area M, not phased

168. Copper alloy. Enamelled brooch, complete apart from the pin. The head is rectangular, with semicircular closed ends in which a four-coil spring (shorter than the available space) is fixed by an axial bar. The head is surmounted by a loop. The bow has a square central panel enamelled in blue and yellow, below which it is triangular in section and terminates in a small foot. The catchplate is vertical and short. L. 35mm. SF1781, Machining layer 4000, Area A, unstratified

169. Not illustrated. Copper alloy. The lower part of a bow with a small round foot and part of a vertical catchplate. L. 20mm. SF3380, Cleaning layer 6000, Area H, not phased

170. Copper alloy with traces of white-metal plating. Complete apart from the pin and the end of the horizontal catchplate. The spring is corroded, but appears to be of eight coils. It is held in the cylindrical head by an iron axial bar. A transverse moulding separates the head and S-shaped bow. There is another transverse moulding at the central bend matched by a step on the back of the bow. The front of the round foot is damaged. L. 30mm. SF4049, Machining layer 12000, Area R, unstratified

171. Tinned copper alloy; the spring is a different alloy to the bow. Complete, apart from the pin and the end of the horizontal catchplate. The central coils from what was probably a ten-coil spring remain on an iron axial bar in the cylindrical head. The foot is rounded. L. 25mm. SF3594, Machining layer 11000, Area A, unstratified

172. Copper alloy. The upper part of a tinned heavy angular brooch, with a small fragment of the spring. The damaged projection on the head was a lug, rather than a head-loop. The ends of the cylindrical head have a small moulding, the central section is slightly raised and marked along the top below the lug with a line of triangular punch marks. The bow below the angle is decorated with a series of scrolls either side of a central line, executed with a round punch. L. 27mm. SF128, Cleaning layer 3501, Area W, Period 3

173. Bronze. Corroded brooch in two fragments. The catchplate, pin and part of the head are missing. Originally of copper alloy, the pin and half the spring were replaced in iron. The plain bow is concave behind. There is a small forward pointing foot. SF4804, Cleaning layer 12207, Area R, not phased

P-shaped brooches

P-shaped brooches (Figure 440, Figure 441) are continental imports of the late 2nd to early 3rd century. The earliest stratified example here comes from a Period 3B context. The overwhelming majority of examples with divided bow come from military sites, especially along Hadrian's Wall, but individual examples also occur at Silchester and Nor'nour (Hull forthcoming, 4882, 7934).

174. Copper alloy and poorly preserved. The spring and pin are missing, as are parts of the semicylindrical spring-cover and the catchplate. The bow is an elongated D-shape. There is a concave moulding just below the flange and above the forward-projecting rounded foot. L. 56mm. SF6981, Fill 11310, Pit 11304, Group 674, Area N, Period 4

175. Copper alloy with applied gold foil on the bow and white-metal plating on the rest of the brooch. Complete, but poorly preserved. The axial bar is iron. The bow is of elongated D-shaped section. The flange is small, and the foot below it is facetted. L. 51mm. SF3554, Layer 5929, Group 3011, Area I, Period 3B

176. Copper alloy with white-metal plating. Spring-cover and upper part of the bow only. The top of the spring-cover is a slightly raised flat panel (see no. 179). There are transverse mouldings at the head of the bow. L. 22mm. SF810, Machining layer 4000, Area A, unstratified

177. Copper alloy. The spring-cover and bow only. The bars of the triangular divided bow are triangular in section. L. 24mm. SF4024, Fill 10362, Ditch 10538, Group 838, Area F, Periods 5-6

178. Bronze with traces of tinning. The foot and a small part of the divided bow only. The foot and the bars of the bow are triangular in section. There is a small flange below the bow. L. 32mm. SF4736, Cleaning layer 5628, Area I, not phased

179. Copper alloy. Head and short stump of the bow only, with thirteen-turn spring. There is a slightly raised flat panel on top of the spring-cover (see no. 176). L. 14mm. SF9506, 3999, Recovered from spoil heap by M. Cuddeford

Crossbow

Crossbow brooches (Figure 441), like the previous P-shaped group, were imported from the continent. The fragment of a light early crossbow (no. 180), belongs to the first half of the 3rd century, while the foot fragment (no. 181) is probably from a brooch of Keller's Type 1 or 2, dated from the late 3rd to the mid 4th century (1971, 32-6). The decoration on the foot is quite roughly executed.

180. Copper alloy. Fragment of the bow with one side of the crossbar. The rounded knobs are collared. The bow is slightly facetted. L. 18mm. SF8150, Machining layer 3999, Area X, unstratified

181. Copper alloy. The foot only. There is a notch flanked by mouldings either side of the top of the foot, and two at either side of the bottom. L. 33.5mm. SF3757, Machining layer 11000, Area A, unstratified

Unclassified

This fragment (Figure 441) has an enclosed spring with iron axis bar similar to that of Polden Hill brooches, with a stump at the centre of the spring-cover that may be a rearward hook (Hattatt 1987, fig. 2, 16). What remains of the bow, however, suggests it was long and gently curved, with strong associations to Langton Downs, which probably had little if any overlap with Polden Hills. No parallel for it has been found. Made of a heavily leaded alloy, and crude in its execution, it might perhaps best be seen as a local experimental product.

182. Copper alloy. Long enclosed spring of sixteen to eighteen coils with an iron axis bar, and a short length of the bow. A stump on the top of the spring-cover may be from a rearward hook. The head is marked by transverse mouldings and the bow has three mouldings either side of a flat central area. L. 16mm. SF3442, Layer 10293, Group 2099, Area F, Period 2

Fragments

These bow brooch fragments, which cannot be assigned to a type, are arranged by period, and within period alphabetically by area.

183. Not illustrated. Copper alloy. Spring fragment with the curve down for the pin. Diameter 6mm. SF1349, Fill 8011, Pit 8012, Group 296, Area E, Period 2

184. Not illustrated. Copper alloy. Spring fragment with large thick coils, probably from a La Tène III brooch of the early to mid-1st century AD, or earlier. Diameter 9mm. SF400, Fill 4025, Pit 4018, Group 275, Area K, Period 2

185. Not illustrated. Large pin from a hinged bow brooch. L. 63mm. SF7941, Fill 24137, Pit 24134, Group 232, Area M, Period 2

186. Not illustrated. Copper alloy. Two fragments, one from the spring. Diameter 4mm. SF3436, Dump 6794, Group 195, Area H, Period 2B

187. Not illustrated. Copper alloy. Fragment from the foot of a reverse-curved bow with part of the catchplate. The catchplate is cast, not hammered out, ruling out the possibility that this is from a Nauheim derivative brooch. L. 14mm. SF7498, Fill 20196, Pit 20195, Group 88, Area L, Period 2B

188. Not illustrated. Copper alloy. Part of a spring and bent pin. L. (bent) 20mm. SF5446, Layer 4878, Group 2030, Area K, Periods 2-3

189. Not illustrated. Copper alloy. Three fragments, two from the pin, one the lower end of a bow with the catchplate. The bow is narrow and rounded and terminates in a very slight knob. The catchplate has an oval perforation. L. 18mm. Possibly from a Colchester derivative. SF2428, Prepared surface 7505, Group 387, Area G, Period 4

190. Not illustrated. Copper alloy. Seven fragments, not necessarily from the same brooch. Three, one part of a fairly large spring with the top of the pin, are very corroded. They may be from a La Tène III brooch. The others are better preserved and consist of three fragments of a long pin, and part of a plain bow. SF3435, Layer 6788, Group 484, Area H, Period 3

191. Not illustrated. Copper alloy. Fragment of a foot with part of a perforated catchplate. Flattened at the lower end, or possibly a miscasting. L. 33mm. SF7737, Prepared surface 21959, Group 381, Area H, Period 3

192. Not illustrated. Copper alloy. Spring fragment. Diameter 4mm. SF2252, Build-up 5807, Group 398, Area J, Period 3B

193. Not illustrated. Two fragments of spring coils, diameter 4mm. Coils this small may be from a plate rather than a bow brooch. SF8364, Layer 13468, Group 600, Area I, Period 3B

194. Not illustrated. Copper alloy. Spring fragment with the curve for the pin. Very narrow wire. Diameter 7mm. SF7731, Prepared surface 21788, Group 377, Area J, Period 3B

195. Not illustrated. Copper alloy. Corroded and soil encrusted fragments of a spring and ?bow. SF2367, Fill 6286, Well 6280, Group 531, Area H, Periods 3-4

196. Not illustrated. Copper alloy. Lower part of a bow with catchplate. The bow is narrow and plain, as Colchester BB derivatives, and the catchplate is solid. L. 18mm. SF1005, Cleaning layer 5001, Area J, Periods 3B-6

197. Not illustrated. Copper alloy. Two fragments. Pin and ten-turn spring with an iron axial bar. L. 42mm. SF348, Fill 5092, Pit 5093, Group 433, Area J, Period 4

198. Not illustrated. Copper alloy. Probably the pin of a large brooch, round in section for most of its length, but flattened towards the turn for the spring, with a slightly groove in the centre. The one surviving spring coil is also thick and flat. L. 52.5mm. Unlikely to be a brooch bow, as it rolls over at the top to form the spring, rather than dips down and then rolls up. SF5948, Fill 15259, Pit 15227, Group 468, Area M, Period 4

199. Not illustrated. Copper alloy. Long spring fragment with axial bar and part of the pin. Diameter 4mm. SF5437, Fill 14052, Pit 14098, Group 4019, Area K, Period 4-5

200. Not illustrated. Copper alloy. Spring fragment with tiny coils. Diameter 3mm. SF3484, Fill 10296, Ditch 10406, Group 838, Area F, Periods 5-6

201. Not illustrated. Copper alloy. End of a bow with the catchplate. The bow is of D-shaped section, plain and narrow and ends in a tiny knob. The catchplate is solid and grooved for the pin. L. 19mm. Possibly from a Polden Hill brooch, as Hattatt 1982, fig. 23b, 25. SF693, Machining layer 4000, Area A, unstratified

202. Not illustrated. Copper alloy. Spring fragment with an iron axial bar. Diameter 4mm. SF1084, Machining layer 4000, Area A, unstratified

203. Not illustrated. Copper alloy. Spring fragment with an iron axial bar and part of the pin. Diameter 7mm. L. of pin 33mm. SF3007, Machining layer 11000, Area A, unstratified

204. Not illustrated. Copper alloy. Small fragment of a flat brooch bow with marginal mouldings and a wide central moulding with a wide channel in which is a raised zigzag. Probably either an Aucissa or Hod Hill brooch. L. 12.5mm. SF6935, Machining layer 11000, Area A, unstratified

205. Not illustrated. Copper alloy. Small fragment, convex in section, with flat-topped longitudinal mouldings. Possibly from a brooch bow, though the section and general form of the mouldings make this unlikely. Maximum L. 11 mm, maximum width 10.5mm. SF7209, Machining layer 11000, Area A, unstratified

206. Not illustrated. Copper alloy. Two fragments, one from a spring cover. Width 11mm. SF6968, Machining layer 17000, Area A, unstratified

207. Not illustrated. Copper alloy. Two fragments from the end of a bow with part of the catchplate. The bow is of D-shaped section, plain and narrow. The catchplate has a rectangular perforation and may have broken across a circular one. Possibly a Colchester brooch. L. 35 mm. SF1520, Layer 9004, Group 8012, Area D, not phased

208. Not illustrated. Copper alloy with traces of white-metal plating. The head and top of the bow only, with an iron spring held in a lug behind the head. There is a short spike on top of the head. The bow is slightly hollow behind. Probably from a Knee or Headstud brooch. L. 15mm. SF1649, Cleaning layer 6000, Area H, not phased

209. Not illustrated. Copper alloy. Three fragments, one a flattened hinged head. Possibly from a Hod Hill brooch. L. 10mm. SF3410, Cleaning layer 6609, Area H, not phased.

210. Not illustrated. Copper alloy. Spring fragment. Diameter 7-8mm. SF5629, Context 6859, Area H, not phased

211. Not illustrated. Copper alloy. Squashed spring fragment. Diameter 3-4mm. SF7588, Cleaning layer 21500, Area J, not phased

212. Not illustrated. Copper alloy. Pin fragment. L. 34mm. SF402, Fill 4142, Pit 4141, Group 1147, Area K, not phased

213. Not illustrated. Copper alloy. Spring fragment with a long pin, similar in size to those of large Colchester brooches. L. 56mm. SF1968, Cleaning layer 4518, Area K, not phased

214. Not illustrated. Copper alloy. Fragment of a strip of copper alloy with three sunken channels, each containing a raised zigzag, between mouldings. Slightly convex in section. Possibly part of a brooch bow, though it has no longitudinal curve, nor does it appear to have been flattened. Moreover, it is very narrow for a brooch, of unusual thickness, the ends appear to have been neatly cut, and there is no real sign of corrosion (Illustrated in Function Category 20, no. 47). L. 16mm, width 8mm. SF6986, Unknown context 11367, Area N, not phased

215. Not illustrated. Copper alloy. Lower part of a bow with catchplate. The bow is narrow and plain, as Colchester BB derivatives. The catchplate has a small triangular perforation. L. 19mm. SF8026, Machining context 3999, unstratified

Plate brooches

Early

Figure 441, nos 216-18 are part of a large Gallo-Rhenish group that also includes cruciform and wheel types (Hawkes and Hull 1947, pl. 98; Simpson 1979, pl. 58; Feugère 1985, 337-44, Type 24). All were tinned and bear the same concentric circular motif, often beaded around the outside, and with an iron rivet, a spot or cup of red enamel, or other similar ornament at the centre. The corrosion at the centre of both sides of the lunular Elms Farm brooch may be all that remains of an enamel-filled disc or cup riveted to the plate. All forms of this group appeared in Britain at the conquest and belong to the Claudian-Neronian period, They are mostly found at Camulodunum, Silchester, Hod Hill, Bagendon, and Dragonby, and in the new Roman settlements of Richborough and the Colchester colonia (Hull forthcoming , catalogue). The lunular form is the most numerous and widespread.

The most unusual brooch in this collection is no. 221. A hinged flat triangular brooch with incurving sides, the edges marked by rocker-arm decoration, it was found stratified in Period 2. No close parallel has been found for it, and the unbent catchplate suggests that it is an unfinished, and therefore locally made, product. However, a continental link is provided by a pair of more elaborate triangular hinged brooches with incurved sides from Bingen, with a third cited at Hungary (Behrens 1920, 162, Abb. 77, 9). These have recessed wavy lines along the sides, for which rocker-arm decoration might be seen as a cheap substitute, but are otherwise very different. They are described as being made similarly to Rosette brooches with applied plates, and the illustrated example is flat at the back but convex in front (rising from all sides), with linear decoration or cut-outs running up to a central inset blue glass roundel.

The incurved sides of no. 221 also suggest links with a small number of early brooches, probably continental again, usually ornamented with punch marks, but occasionally with marginal rocker-arm decoration (Mackreth 1981, fig 72, 59; Hattatt 1987, 160, fig. 53, 1012-1015). Marginal rocker-arm zigzags also occur on an early hare brooch, probably from East Anglia, and a bird brooch from Yugoslavia (Hattatt 1989, fig. 78, 1634, 1637). It is also found on some Colchester B derivatives (nos 118-22) and on a strip-bow brooch from Maiden Castle in a Romano-Belgic context dated c. 25-70 (Wheeler 1943, fig. 84, 25).

Another unparalleled brooch is no. 222. While it appears to be similar to equal-ended brooches of the 2nd century, this brooch with its white-metal plating (tinning?) and ribbed bow reminiscent of the Hod Hill series, must belong to the middle years of the 1st century. It may be a locally made attempt to produce a small and cheap plate version of the Hod Hill.

The applied plate type represented by no. 223, with its embossed triskele, includes examples with designs of animals, human heads, and the 'Verulamium' and 'Adlocutio' brooches based on coin reverses of Hadrian. Those with the Celtic triskele design may be earlier than the 2nd-century fully Roman forms, though good dating is scarce (Hattatt 1989, 132; Mackreth 1986, 66). Kilbride-Jones describes the design as in the Aesica style and some examples do approach the elaboration of the Aesica brooch (1980, 56-7, fig. 4). Others are more restrained, or were perhaps made by craftsmen less capable of executing a fully developed version of the design.

The distribution of embossed triskele brooches is mainly northern, centred on Yorkshire but well represented in Northumberland and Cumbria, but there are also examples from Wales and in the south from Caerleon, Richborough, St Albans, Silchester, and Colchester (Kilbride-Jones 1980, fig. 4; Hull forthcoming, Type 249A). These scattered southerly locations might support an early date, and the mechanism by which the brooches came south could be seen as an example of 'gift exchange' between members of an elite (Peacock and Williams 1986, 55-7). However, the designs on the Caerleon and Colchester brooches, and to some extent on the Elms Farm example also, could be described as debased, and may be locally made copies (Hull forthcoming , 6957, 0546, 7965).

216. Copper alloy. Lunular plate brooch, with one terminal and the pin missing (Feugère 1985, Type 24d1). Part of the edge is damaged. L. 32mm. The pin was hinged between two lugs. The surviving terminal has a double lug split by a groove. There are two tiny lugs on the outer edge, and a third would have lain where the edge is damaged. In the centre of the inner edge is a tiny perforated lug, perhaps for a very fine chain. The circular motif consists of a narrow ring and a slight moulding around a recessed centre. This contains an unusual clump of corrosion, which is matched on the underside. SF7547, Machining layer 11000, Area A, unstratified

217. Copper alloy with some white-metal plating surviving on the surface. Fragment of a circular plate brooch with six small lugs around the rim (Feugère 1985, Type 24a). Diameter 27mm. The pin was fixed by a rivet between two lugs. In the centre is a circular motif similar to that on No. 216. SF2984, Machining layer 11000, Area A, unstratified

218. Copper alloy. Brooch with a lozenge-shaped head and long leg (Feugère 1985, Type 24e). The pin was hinged between two lugs on an iron pin. Most of the catchplate is missing. There are ring-and-dot stamps along the upper two sides of the lozenge. The lower sides are incurving, and between them is a circular motif similar to that on the two preceding brooches, with an iron rivet at the centre. The leg has prominent mouldings at the top, and tapers to a small knob. L. 37mm. The use of ring-and-dot ornament along the upper sides is reminiscent of the eyes on the tails of early Peacock brooches (Hattatt 1987, fig. 124, 1415; 1989, fig. 78, 1636; Hull forthcoming, 5681). SF7144, Machining layer 17150, Area A, unstratified

219. Copper alloy with white-metal plating. Fragment, the upper part with the pin and its mechanism are missing. The foot and catchplate are also damaged. The form consists of a circular element (partly missing) set above a triangular foot with incurving sides. The 'waist' between the two parts is marked by three flat-topped transverse mouldings. The upper part has concentric circular mouldings around a large central hole. There is a suggestion of a lug on the right-hand edge. The central hole seems too large to take a rivet or other insert. L. 29mm. This brooch has marked similarities to 1st-century keyhole-shaped brooches (see no. 220), the final plate form of the Rosette series, and to other early circular brooches with concentric mouldings, in particular Hattatt 1987, fig. 52, 1008 from East Anglia, and Hattatt 1989, fig. 53, 1561, from Lincolnshire. SF8131, Machining layer 3999, Area X, unstratified

220. Copper alloy with traces of tinning. Small keyhole-shaped plate brooch in two fragments. Complete apart from the centre of the circular element, the corners of the foot, and part of the catchplate. L. 23mm. The pin is hinged on an iron bar between two lugs and is hidden behind a prominent bossed lug on the rim of the circular area. Two similar lugs are set on either side of the rim. The circular part is dished and badly corroded. Keyhole-shaped brooches are usually flat and fitted with an embossed plate in imitation of the Rosette form (Hattatt 1985, fig. 72, 633). This was usually riveted on, or held by small inward projections on the lugs. A similar plate may have been soldered onto this brooch, and the missing centre means a riveted attachment cannot be ruled out. However, in neither case would dishing the centre be necessary. It may instead have held enamel or some other decorative feature, though no trace of enamel was found. SF4737, Spread 13418, Group 600, Area I, Period 3B

221. Copper alloy. Complete triangular plate brooch with incurving sides. The three points are very slightly knobbed. The margin is decorated with a wide incised line of rocker-arm (walked-scorper) decoration. The pin, now fully open, is held between two slightly tapering rectangular lugs by a small rivet. The catchplate is more or less semicircular. It has not bent over to hold the pin, nor is there any scar where a section may have broken off. L. 30mm. SF7410, Fill 15589, Pit 15594, Group 238, Area M, Period 2B

222. Copper alloy with traces of white-metal plating. Fragment of a small corroded plate brooch, possibly of flat equal-ended form. The pin is missing, but was hinged on an iron axial bar between two lugs. Both sides of the bow and part of the lower projection with the catchplate are missing. The upper projection is moulded. The bow is ribbed. L. 19mm. SF1668, Floor 6165, Group 496, Area H, Period 2B

223. Copper alloy. Disc brooch with an applied plate ornamented with an embossed triskele and radially incised moulded rim. Diameter 24mm. The pin is missing, but was hinged between two lugs. The stout catchplate is deep and projects forward so that part is visible beyond the rim of the brooch. The triskele design is fairly simple in execution, and the brooch lacks the curvilinear motif that often springs from the rim between the arms of the central design on many of these brooches (Kilbride-Jones 1980, fig. 4). SF3592, Machining layer 11000, Area A, unstratified

224. Not illustrated. Heavily leaded copper alloy. A very corroded disc brooch with a pin hinged between two lugs. Most of the edge is damaged. Diameter 23mm. Possibly the same type as Figure 441, 223, with the 'warty' surface being the remains of solder. However, possible traces of white-metal suggest it may be of some different type, with the rough surface simply the alloy's reaction to its deposit conditions. SF8130, Machining layer 3999, Area X, unstratified

Zoomorphic

Most of these brooches are all of enamelled native types widespread in Britain, and all date to the 2nd century (Figure 441). The exception is the flat duck brooch (no. 227), which only reaches as far as York in the north. They mainly come from eastern Britain, and are particularly well represented at Colchester (Hattatt 1987, 231; Hull forthcoming, 0613-0615).

225. Copper alloy. Flat brooch in the form of a dog. L. 43mm. The tips of the ears and forepaw, and possibly of the tail, are missing, as is the pin. This appears to have been sprung on an axis bar fixed between two lugs. Parts of both lugs remain behind the hindquarters of the animal, with one forced outward and now visible above the rump. The catchplate has been squashed into a tube. The playful crouching stance in which the animal is shown is not matched on other dog brooches, nor can the enamel pattern be paralleled. The original colour of the enamel is uncertain. Individuality among brooches of this native British type was noted by Hattatt (1987, 240). SF3200, Layer 5693, Group 609, Area I, Period 3B

226. Copper alloy. Damaged and worn brooch in the form of a hollow-bodied swimming duck. L. 28mm. The head is missing and the neck pushed to one side. The catchplate has been bent out and back. The missing pin was hinged between two lugs. On either side of the narrow body are six crescentic cells for enamel feathers, and a line of opposed triangular cells runs down the back forming lozenges of bronze and terminating in an oval cell at the tail. All the enamel is missing. SF6795, Machining layer 11000, Area A, unstratified

227. Copper alloy. Flat corroded fragment showing most of the tail of a swimming duck. Only the stumps of the pin mechanism and catchplate remain. The body has four panels of enamel, the inner pair both red, the outer pair now pale green and mostly missing. The metal dividing the innermost cells was fitted with a ribbed strip, part of which remains. L. 22mm. SF18, Machining layer 400, Area W, unstratified

Enamelled discs

Number 228 belongs to a group of 2nd-century sprung flat disc brooches (Figure 441) which are more often ornamented with concentric rings of enamel (sometimes also with a vertical strip), or with a scalloped, star-shaped, dot, or triskele design. They sometimes have, as here, ten or twelve small lugs around the rim (Mackreth 1986, fig. 40, 5-6; Hattatt 1989, figs 55-7).

Though there are few examples similar to no. 230, they are very widespread: two from Nor'nour, Isles of Scilly (Hull 1968, fig. 22, 202-3), one from Castle Nick Milecastle on Hadrian's Wall (Hull 1968), one from Chichester (Butcher 1978, fig. 10.48, 1), one from Stonea (Mackreth 1996, fig. 94, 19), and on the continent one from Trier (Exner 1939, Taf. 17, 2). Though dating evidence is scarce, this small group probably belongs in the 2nd century (Mackreth 1996, 305).

228. Copper alloy with slight traces of white-metal plating. Mackreth has noted a silvery finish to the trim on other examples of this type (1986, 64). Flat disc brooch, complete apart from part of the catchplate. Diameter 16mm. The humped pin is sprung on an axial bar between two lugs. The face is divided by metal strips into six triangular fields, four filled by blue and two by green enamel. Twelve small lugs are set around the rim, which had a beaded appliqué trim, little of which now remains. SF7806, Layer 24080, Group 4027, Area M, Periods 4-5

229. Copper alloy. Disc brooch, with enamel set in a raised field and six lugs set on the rim. The raised rim is damaged and only two lugs survive. The pin is missing, and both the two-lug hinge fitting and the catchplate are damaged. Diameter 18mm. The zone between the rim and the raised disc retains slight traces of white-metal plating. The edge of the raised field is marked by tiny triangular notches or punch marks. The enamel is a green and yellow mosaic, set in random lines and swirls. At the centre of the underside is a small cup set on a moulding. Its purpose is uncertain. SF3740, Machining layer 11000, Area A, unstratified

Lozenge-shaped

Lozenge-shaped brooches (Figure 442) with lugged corners and central enamelled field are a 2nd-century type widespread across the Empire (Exner 1939, Group III). In Britain they are particularly well represented at Nor'nour, Isles of Scilly (Hull 1968, Type 29, 143-55). The more unusual no. 232 is very similar to one from Suffolk (Hattatt 1985, fig. 65, 568A), which also had lathe centre-marks in the hollow underside and the cup. The moulds, rather than the brooches, are presumed to have been lathe-turned to provide the means of fitting a central boss or other surface detail (Hattatt 1985, 148).

230. Copper alloy. Lozenge-shaped brooch with lugged corners. L. 27mm. Two lugs and the pin are missing, the two-lug hinge and catchplate are damaged. The lugs are circular and bear a ring-and-dot motif. The field rises in two steps to a recessed bed for enamel, now missing. The outer rim and the two steps are knurled. SF2362, Prepared surface 6320, Group 547, Area H, Period 4

231. Not illustrated. Copper alloy. Fragment of a lozenge-shaped brooch with lugged corners. L. 16mm. The fragment has a lozenge-shaped lug over the catchplate, and a high step up to the inner field. This is recessed for enamel, traces of which remain, now greenish white in colour. SF2436, Stake-hole 7604, Group 858, Area G, Period 4

232. Copper alloy. Lozenge-shaped brooch with a hollow high centre-boss raised on two steps, cf. Exner 1939, 101, Taf. 12, 15. Three of the corners are missing, the fourth has no lug. The missing pin was hinged between two lugs on an iron axial bar. The lozenge-shaped base has a pronounced marginal groove flanked by knurled ridges. Beneath the boss the underside is hollow and has a small lathe centre-mark and the concentric striations typical of lathe-turning. The boss also has a lathe-centre mark in the bottom and retains traces of the substance, presumably enamel, with which it was filled. SF3580, Machining layer 11000, Area A, unstratified

Equal-ended

Number 233 has a long rectangular enamelled plate set across a humped bow (Figure 442). Few similar are known from either Britain or the continent, where they are widely spread. British examples with three spots in an enamelled field, as here, come from Charterhouse-on-Mendip, Somerset and Nor'nour (Hull forthcoming, 1163, 6244). The former has black spots on bright red, the latter black on an unknown colour. A date within the 2nd century is appropriate.

The cruciform no. 234 matches one from Colchester, found in the backfill of an Anglo-Saxon hut, but probably of late 1st- or 2nd-century date (Crummy 1992, fig. 5.1, 31). The Colchester brooch had a field of red enamel, whereas this example is of white-metal and niello. An early date appears confirmed by a brooch from Augst with a similar central panel but only two terminals, both zoomorphic, to mask hinge and catchplate, which is dated to the second half of the 1st century (Riha 1979, 193, Taf. 63, 1654). Cruciform equal-ended brooches with similar floret decoration but rather more humped and with more developed terminals are among Exner's Group II (1939, Taf. 10, 3). The Colchester and Elms Farm brooches can be seen as contemporary with the Augst brooch and precursors of the more-developed 2nd-century cruciform type.

233. Copper alloy. Equal-ended brooch lacking the pin and the lower projection above the catchplate. There is a long rectangular enamelled plate across the slightly humped bow. The pin was hinged on an iron axial bar between two lugs. The innermost mouldings on the upper and lower projections are knurled. The edge of the rectangular plate is also knurled, and the frame for the enamel, raised very slightly from the edge, is marked along its long sides with a line of tiny punched triangles. The field of the enamel is a bright blue. The three spots are partly damaged, partly missing, and the original colour is uncertain, though the best preserved appears very dark. It is probably black like those on the brooches cited above. L. 31mm. SF5727, Cleaning layer 5601, Group 8002, Area I, Not phased

234. Copper alloy. Cruciform brooch lacking the two side bars, the pin and the catchplate. The rectangular central panel has a beaded frame around a raised central panel set with a white-metal four-petal floret in a field of niello. The one complete arm has a knobbed end. L. 20mm. SF9507, 3999, Recovered from spoil heap by M. Cuddeford

Glass centre-boss

Centre-boss brooches (Figure 442) are a native form with a wide distribution from Hadrian's Wall across to the Scilly Isles (Allason-Jones and Miket 1984, 3.136; Hull 1968, fig. 24, 237). There are usually one or two decorated zones around the boss, as here, but a brooch from near Dereham, Norfolk, has three (Hattatt 1985, fig. 73, 643). The punch marks used to decorate the zones are very varied, concentric circles, ring-and-dots, crosses, squares, rectangles, zigzags, I-, J-, L-, and S- shapes (the latter occurs most frequently), and some designs are highly idiosyncratic (Hattatt 1985). The boss is usually a very dark green, brown, or purple iron-rich glass, appearing black. The front surface of these brooches is usually mercury-gilded, the back white-metal plated. Traces of mercury-gilding occurs on both the Elms Farm brooches, but plating on the reverse has only been confirmed on Figure 442, 236, The type occurs most frequently in 4th-century contexts, but seems to make its appearance in the 3rd, with one from Fishbourne in a late 3rd- to 4th-century context (Hull 1971, 106).

235. Copper alloy. Very corroded oval centre-boss brooch. Part of the edge is missing. The pin was sprung on an iron axial bar through a single lug. The conical boss is of black glass and surrounded by a single zone of decoration. The front and sides are gilded, the back is so corroded it is impossible to detect any coating. L. 29mm. SF1536, Layer 9004, Group 8012, Area D, not phased

236. Copper/lead/tin alloy, with mercury-gilding on the front, and white-metal (tin/lead) plating on the back. An oval centre-boss brooch. The central, usually glass, boss is missing, as is the sprung pin and part of the single-lug hinge fitting, and the edge is slightly damaged. The ridge dividing the field into two zones is made of wire. The stamped decoration in the two zones around the missing boss is not clear. The inner zone may have consisted of S-shapes or radial lines. L. 33mm. SF6813, Machining layer 17000, Area A, unstratified

Fragments

Arranged by period and within period by area.

237. Not illustrated. Copper alloy. Large pin. The angle at which the lower edge of the flattened perforated end leaves the shaft suggests this is perhaps more likely to be a buckle tongue than a brooch pin. L. 38mm. SF5111, Fill 13238, Pit 13239, Group 642, Area I, Period 4

238. Not illustrated. Brooch pin or buckle tongue, the upper end damaged. L. 21mm. SF7476, Cleaning layer 20184, Area L, not phased

239. Not illustrated. Small fragment of very corroded copper alloy. Possibly part of a plate brooch with the stump of the catchplate, though this is very far from certain. Diameter 12mm. SF2521, Machining layer 11000, Area A, unstratified

Penannular brooches

The typology of penannular brooches (Figure 442) established by Fowler in 1960 is followed here, with reference to Hull forthcoming where appropriate.

While never very numerous as site finds, the absence of penannular brooches of a particular type or in a particular metal can be intriguing. For example, Type A penannulars, as Figure 442, 240, are absent from Camulodunum, while only iron penannulars were found at King Harry Lane.

Type C brooches are, in general, the best represented form. In terms of date they fall into two groups, both represented here by stratified examples. Early examples, broadly dated to the 1st century BC and 1st century AD, usually have a round-section hoop (no. 242), while later examples, belonging in the 4th century AD (Fowler 1983, 18-19) usually have a flat-section hoop decorated with incised lines or notches (no. 243). The latter is a particularly well-made example of the late form. Made of iron coated with a white-metal alloy, and both grooved and notched to appear ribbed and segmented, when new it would have appeared as a brooch of considerable quality, equal to silver brooches of the period such as a small segmented Type C from a late 4th-century grave at Colchester (Crummy 1983, fig. 16, 103; Crummy 1992, 135), and two Type Aas from probable 4th-century hoards (Fowler 1964, 110, note 6).

The flat grooved hoop of the Type D brooch (Figure 442, 244) is matched at Camulodunum on a Type C brooch (Hawkes and Hull 1947, fig. 59, 4, Class A). Number 244 belongs to Type P4 in Hull forthcoming, several of which have flat decorated hoops. One from Cirencester has an identical hoop and almost smooth terminals (Hull forthcoming, 5877; Corinium Museum C.234).

The thin wire brooch with S-shaped terminals and pin attachment is very unusual (no. 246). It can best be compared with continental wire Omega brooches, where the terminal is S-shaped but lies in the same plane as the ring (Riha 1979, Type 8.1.5, Taf. 69, 1823-5), rather than at right-angles to it as here. A well-stratified example at Augst sets the date in the middle of the 1st-century AD, though as it is not an exact parallel a similar date for the Elms Farm brooch should be treated with caution.

240. Copper alloy. Fowler Type A1 brooch, complete but distorted. The terminals are plain and bun-shaped. The junctions of terminals and hoop are marked by slight constructions. The flat pin appears to have been humped. Internal diameter probably about 20-25mm, circular section hoop, 2mm diameter. SF5094, Prepared surface 13063, Group 365, Area I, Period 3

241. Not illustrated. Fragmentary copper-alloy brooch or finger-ring embedded in soil. Possibly a Fowler Type A brooch. Internal diameter about 16mm, circular section hoop, 1.5 to 2mm diameter. SF5223, Layer 10407, Group 2097, Area F, Period 2

242. Copper alloy. Fowler Type C brooch, with coiled terminals at right-angles to the plane of the ring. The pin and one terminal are missing. A large part of the outer surface of the hoop has been destroyed by corrosion. The remaining terminal shows that the metal was flattened before being wound in three very tight coils. Maximum internal diameter 24mm, slightly flattened circular section hoop, 3mm diameter. SF2403, Fill 7267, Pit 7266, Group 2114, Area G, Period 2

243. Complete iron brooch coated with a lead-tin alloy. The form is the late variety of Fowler's Type C, with flat decorated hoop. The upper face has a central groove, and the edges to either side are notched, giving both a ribbed and segmented appearance. The terminals are of a single coil, and the upper part is marked similarly to the hoop. The coil of the pin is also decorated in this way. Internal diameter 29mm, hoop 4 by 2.5mm. SF4158, Fill 4825, Pit 14098, Group 4019, Area K, Periods 4-5

244. Copper alloy. Fowler Type D brooch, in fragments (partly joined) but complete apart from the tip of the pin. Two grooves divide the upper face of the flat section hoop into three equal strips. The terminals are corroded, but may have been grooved to match the hoop. Internal diameter 21mm, hoop 3mm wide. SF484, Fill 6001, Pit 6006, Group 553, Area H, Period 4

245. Not illustrated. Fragment of a very corroded Fowler Type D brooch, probably with a hoop of circular section. The terminals are too damaged to show any detail. Internal diameter approximately 16mm, hoop at least 2.5 m in diameter. SF6453, Fill 13825, Pit 13771, Group 594, Area I, Period 3A

246. Complete copper-alloy brooch of thin wire, with S-shaped terminals bent back to lie above the hoop, The coil of the slightly humped pin around the ring is similarly S-shaped. Internal diameter 22mm, circular section hoop, 1.5mm diameter (the surface has been almost entirely removed by corrosion, but the diameter cannot have been much greater than it is now). SF7528, Fill 20826, Pit 20825, Group 1173, Area L, Not phased

The Saxon brooch by Susan Tyler

Cite this as: Tyler, S. 2015 The Saxon brooch, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.tyler

247. Copper-alloy/iron cruciform brooch terminal (see Figure 442). Detached cruciform brooch terminal of copper alloy with central iron attachment rod (the latter runs internally throughout the whole length of the terminal protruding some 6mm out of the terminal base). The terminal is a variant of Aberg's Group 1 (Mortimer's Group A) with 'fully rounded knob' as it has facets rather than a simple spherical appearance. The terminal may have been the only attached knob on the head plate of the cruciform brooch or it may have been one of three equidistant around a square or rectangular headplate. In good condition. L. 18mm. Max. width 8mm. SF6238, Fill 14558, Well 14529, Group 722, Area L, Period 6

The brooch terminal is most likely to be one of an original full complement of three head plate terminals; one on the top and two on the sides of the rectangular headplate . The two visible projections are the iron attachment rod and the decorative ridged copper-alloy arm; the former would have secured the knob to the head plate, the latter would have helped to keep the knob attached by running down the front of the brooch and joining up with those attached to the other two knobs. It is unusual in that it exhibits facetting rather than a simple spherical appearance. It can be seen as a variant of Aberg's Group 1 (Mortimer's Group A) and as such belongs to a brooch that would be placed early on in both typologies (Aberg 1926; Mortimer 1990). With only one terminal extant, more precise dating is impossible; however, a date of manufacture in the first half of the 5th century rather than later (when half rounded knobs became the norm) can be postulated. It is not unusual for cruciform brooch terminals of this type to become detached from the rest of the brooch and repairs can frequently be observed in complete examples. At Springfield Lyons, Essex, grave 4988, for example, a 5th-century cruciform brooch, with fully rounded knobs, has a 6th-century repair comprising a half-rounded knob (Tyler and Major 2005, 21).

Crook-headed pin

248. Iron. Rod fragment, with a crooked terminal. The rod has a fresh break. This is the terminal of a crook-headed pin, a type used principally in the Late Iron Age. There is a parallel from a Late Iron Age context at Slough House Farm, only a few kilometres from Heybridge (Major 1998b, 123, no. 1). Although the type probably continues at least into the Early Roman period, the Elms Farm pin must be residual in its context. L. 47mm, W. across loop 17mm, diameter of rod 5mm. Layer 6742, Group 586, Area H, Period 5-6

Hairpins

Bone hairpins by Ros Tyrrell

Cite this as: Tyrrell, R. 2015, Bone hairpins, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.tyrrell2

Four complete bone hairpins and twenty-nine fragments were found. Some of the headless fragments may be needle shafts but are included here. Four of the pins are somewhat roughly finished (Figure 443, nos 249, 256 and 263) but, in the absence of manufacturing waste, it is suggested that these are merely examples of poorer quality workmanship or lack of wear.

The hairpins have been classified using the Colchester type series (Crummy 1983, 19-27). Only types 2, 3 and 5 are identified as present at Elms Farm.

In Colchester, Type 2 hairpins (as no. 249) tend to come from contexts dating to the pre-Flavian period onwards (Crummy 1983, 21). Type 3A (no. 252) and 3B (nos 255-6) pins commonly come from deposits dated to around the late 3rd to 4th century. The Type 3 pins from Ivy Chimneys, Witham (Crummy 1999, 198) however are dated from the early 3rd century onwards. Type 5 (nos 262-3) seems to be popular in the 4th century. The Type 2 pins from Elms Farm are from pits with pottery dating from the late 2nd to mid-4th century AD, which underlines the longish life of this simple style. Types 3A, 3B and 5 (both examples of Type 5 are from the same context) conform to the Colchester pattern and come from Late Roman contexts with one exception. A pin of Type 3A (no. 252) came from a Period 4 pit, and was found with early 3rd-century pottery, as at Witham.

A damaged pin with traces of a decorative head (no. 264), from an unstratified context in Area J, is possibly similar to a 1st to 4th century AD find from Balkerne Lane, Colchester (Crummy 1983, 25, no. 443). The head of the Elms Farm pin is broken off above the first groove, and it may have had a complex top like the Colchester example, though the evidence is slender.

Only one pin (no. 265) did not fit into the Crummy type-series. It came from a Period 4-5 pit (14098) in Area K. The head is damaged but was possibly symmetrical, forming two outward curving 'horns' similar to the shape of a lyre. The caducei from shrines at Uley (Henig 1993, 102-3) are a similar crescentic shape but the two 'horns' point inwards and incorporate Hercules knots. If the Elms Farm pin is a caduceus there are possible votive implications. A more convincing parallel was recovered from the main building of the Gadebridge Park villa site (Neal 1974, 154, fig. 67), in the form of a cockerel-headed pin. The 'horn' could be the curled tail of a cockerel. Both caducei and cockerels were regarded as symbols of Mercury.

Crummy Type 2

249. Tip broken off. Two shallowly cut grooves round a conical top. The shaft is squarish in section. SF385, Fill 4064, Pit 4016, Group 739, Area K, Period 4

250. Not illustrated. Tip broken off. Two deeply cut grooves around a spherical head. The shaft is polished, or worn to a gloss. SF1185, Fill 4392, Pit 4429, Group 741, Area K, Period 5

251. Not illustrated. Two grooves, shallowly cut round a conical top. Tip broken off. L. 41mm, D. 4mm. SF6272, Layer 14608, Group 5016, Area L, Period 5-6

Crummy Type 3A

252. Tip broken off. Rather roughly cut. SF4155, Fill 4870, Pit 4913, Group 4016, Area K, Period 4

253. Not illustrated. Shaft broken off. L. 11mm, D. 3mm. SF7871 Spread 15893, Group 751, Area M, Period 5-6

254. Not illustrated. Tip broken off. L. 54mm, D. 3mm. SF7598, Layer 21619, Group 446, Area J, Period 6

Crummy Type 3B

255. Complete. SF1192, Fill 4380, Pit 4379, Group 741, Area K, Period 5

256. Complete. Rather roughly cut. SF5440, Fill 4994, Pit 4989, Group 1147, Area K, not phased

257. Not illustrated. Shaft broken off. L. 22mm, D. 2mm. SF1903, Fill 5358, Pit 5359, Group 443, Area J, Period 6

258. Not illustrated. Complete. Worn? smooth. L. 97mm, D. 2mm. SF5776, Fill 10891, Pit 10910, Group 676, Area N, Period 5

259. Not illustrated. Part of shaft broken off. Roughly cut. L. 43mm, D. 3mm. SF5790, Fill 10987, Pit 10953, Group 673, Area N, Period 4

260. Not illustrated. Complete. Shaft roughly cut. L.91mm, D. 2mm. SF7490, Spread 20186, Group 5014, Area L, not phased

261. Not illustrated. Tip broken off. L. 59mm, D. 2mm. SF7676, Fill 21692, Pit 21618, Group 439, Area J, Period 5

Crummy Type 5

262. Complete. Two reels under a biconical head. SF416, Fill 4129, Pit 4128, Group 744, Area K, Period 6

263. Tip broken and repointed. One ring. Conical head. Somewhat roughly finished. SF428, Fill 4129, Pit 4128, Group 744, Area K, Period 6

Crummy type miscellaneous

264. Head damaged above the first groove. Tip broken off and shaft worn or polished? glossy. SF7775, unstratified context 21116, Area J

Other types

265. An incomplete pin-head, perhaps a cockerel, lyre or caduceus, decorated with incised diagonal lines. Tip and head damaged. SF5444, Fill 14022, Pit 14098, Group 4019, Area K, Period 4-5

Shaft fragments Not illustrated

266. No tip. L. 51mm, D. 4mm. SF559, Fill 5135, Pit 5093, Group 433, Area J, Period 4

267. No tip. L. 25mm, D. 2mm. SF6843, Fill 5392, Pit 5394, Group 432, Area J, Period 4

268. No tip. L. 72mm, D. 4mm. SF8214, Fill 5864, Pit 5805, Group 444, Area J, Period 6

269. Three fragments of a shaft. L. 41mm, D. 4mm. SF468, Cleaning layer 6000, Area H, not phased

270. Two fragments of different shafts. L. 60mm, D. 3mm; L. 24mm, D. 3mm. SF5772, Fill 10891, Pit 10910, Group 676, Area N, Period 5

271. A fragment of shaft with tip. L. 51mm, D. 3mm. SF5809, Fill 11139, Pit 10910, Group 676, Area N, Period 5

272. A fragment of tip. L. 12mm, D. 2mm. SF7874, Fill 11571, Kiln construction 11423, Group 693, Area N, Period 4

273. No tip. L. 47mm, D. 4mm. SF6452, Fill 13816, Gully 13737, Group 615, Area I, Period 3B

274. No tip. L. 63mm, D. 2mm. SF4192, Fill 14022, Pit 14089, Group 4019, Area K, Period 4-5

275. No tip. Signs of wear. L. 54mm, D. 3mm. SF5443, Fill 14022, Pit 14089, Group 4019, Area K, Period 4-5

276. No tip. L. 51mm, D. 3mm. SF7493, Fill 20203, Pit 20193, Group 718, Area L, Period 5

277. Two fragments of shafts with no tips. L. 83mm, D. 3mm. L. 40mm, D. 4mm. SF4155, Fill 4870, Pit 4913, Group 4016, Area K, Period 4

Jet hairpin by Ros Tyrrell

278. Not illustrated. Part of the shaft and point of a hairpin. L. 30mm, Diam. 3mm. SF5768, Fill 10877, Pit 10910, Group 676, Area N, Period 5

Glass hairpins by Sally Worrell

Glass hairpins normally have globular heads and either a plain shaft with circular section, or a twisted 'barley-sugar' shaft. Both pins from Elms Farm are broken beneath the head, but there is evidence that at least one of them once had a twisted shaft. Glass hairpins are considered to have been in use during the Late Roman period (Cool and Price 1998, 194; Crummy 1983, 28). Hairpins are commonly found near the skull in burials; four glass pins were found in a grave in the Butt Road cemetery (Crummy 1983, 28). The twisted shaft from a hairpin was found in previous excavations at Crescent Road, Heybridge (Drury and Wickenden 1982, 28, fig. 12.23) in a Late Roman/Early Saxon context. The examples from Elms Farm are unstratified (Figure 443).

279. Fragment from a hairpin. Blue/green. Plain globular head. Plain shaft with circular section. Lightly worn. L. 16.5mm, head diam. 9.5mm, shaft diam. 2.5mm. Machining layer 11000, Area A2, unstratified

280. Fragment from a hairpin. Blue/green. Plain globular head, uneven and pitted surfaces. Bubbly. L. 14mm, head diam. 11-12mm, shaft diam. 4.5mm. Machining layer 12000, Area R, unstratified

Copper-alloy hairpins

There were thirty-seven copper-alloy hairpins (Figure 443), eight of them fragments not definitely identified as hairpins. Many of the hairpins were in very poor condition, and not all could be drawn. The group numbers are those given in Cool 1990.

Group 2 Knob on cordon heads

Five examples (Figure 443). Pins in this group can date from anywhere in the Roman period. Two of the examples from datable contexts are from Period 3 (Early Roman), and one from Period 4 (mid-Roman). SF4170 appears to have affinities with group 21, the dated examples of which from Elms Farm are Period 2-3.

281. No surviving surface, bent. It has a swollen shaft, pointed at one end, with a thickened terminal at the other. The terminal is relatively long and slender, and comprises a small block element, with a groove above, and an elongated cone-shaped head, possibly with a square section. As with SF6967, it is probably a variant of Cool group 2. A hairpin from Colchester (Crummy 1983, 30, no. 504; context dated c. 250-300) has a similar cone terminal, with multiple reels below. The swollen shaft is relatively rare on copper-alloy hairpins, but is a feature of most Late Roman bone pins. L. 92mm. SF1682, Fill 6210, Pit 6209, Group 561, Area H, Period 4

282. Incomplete and bent. It has a narrow, elongated head with traces of either cross-hatching or pine-cone decoration, with two collars below. Variant of Cool group 2. The form is paralleled on a silver pin from Baldock (Stead and Rigby 1986, 128, no. 206) which is very similar bar a small knob at the top of the head, and less pronounced collars. L. 65mm. SF6967, Machining layer 17000, Area Q, unstratified

283. Point missing; allied to Cool group 2, with one cordon, but very similar in profile to members of group 21 (Cool 1990, fig. 11.9), with a similar incised circumferential line. It is perhaps best seen as a form of group 21 without the cup. Surviving L. 59mm, head diam. 8mm. SF4170, Layer 4899, Group 749, Area K, Period 3

284. Not illustrated. Complete and bent. It has a sub-globular head, and the cordons are sharply moulded, with a bulge below the cordon. Cool group 2A. L. 122mm, head diam. 5.5mm. SF3562, Layer 5927, Group 3024, Area J, Period 3

285. Not illustrated. Bent and incomplete. Sub-globular head with two cordons. L. 59mm, head diam. 6mm. SF3578, Machining layer 11000, Area A, unstratified

Group 3 Curved units between cordon heads

Five examples (Figure 443). This is a disparate group, with many different variations. Three of the Elms Farm hairpins are very similar, and probably belong to sub-group A, to which Cool assigned an Early Roman date based on the dating evidence from a single site. However, here they are all from mid- to later Roman contexts, suggesting that the type was not confined to the Early Roman period.

286. Complete and bent. It has a button head narrower than the shaft, flaring below the head to a cordon. Very similar to SFs 3552 and 6263, and cf. Crummy 1983, 30 no. 108, from a later 3rd-century context. L. 92mm. SF328, Machining layer 5000, Area J, unstratified

287. Bent, with the point and head missing. The moulding at the top of the shaft is unusual, comprising a square-sectioned moulding between two square cordons, rather than usual rounded profiles. L. 97mm. SF5115, Fill 13332, Pit 13433, Group 637, Area I, Period 4

288. Not illustrated. Bent and in poor condition, with little detail surviving. The head is missing. This is almost certainly Cool group 3, sub-group A, with a curved unit at the top of the shaft with ?two grooves below. It was probably similar to Cool 1990, fig. 2.5. L. 96mm. SF1291, Fill 5229, Post-hole 5232, Group 427, Area J, Period 4-6

289. Not illustrated. Complete and bent; very similar to SF328, except that the neck below the button head is more slender. Cool group 3A. L. c. 97mm. SF3552, Layer 5453, Group 457, Area J, Period 5-6

290. Not illustrated. Point missing. The shape of the head is similar to SF328, but is slightly more elongated. Cool group 3A. L. 48mm. Stoke-hole 14744, Group 714, Area L, Period 5

Group 5, sub-group D Simple grooved heads)

One example. The bulk of the dated examples are 2nd century, although they can occur earlier, and this example is from a mid-1st-century to mid-2nd-century context.

291. Small, low conical head, slightly facetted on top. There are two grooves, with a shallow groove on the cordon between them. In good condition, with the point missing. L. 94mm, head diam. 4mm. SF1272, Fill 5157, Pit 5158, Group 409, Area J, Period 3

Group 6 Button and cordon heads

Three examples (Figure 444). Hairpins of this group are common in eastern England, and were produced in the late 1st century, into the 2nd century. Both the dated examples from Elms Farm are from contexts of this date.

292. Two cordons below a flattened spherical head. In poor condition and incomplete. Surviving L. 75mm, head diam. 8mm. SF6499, Cleaning layer 5602, Area I, not dated

293. Not illustrated. Two cordons below a sub-globular head, similar to Cool 1990, fig. 5.6. In poor condition, with the point missing. Surviving L. 90mm, head diam. 9mm. SF4167, Layer 4899, Group 749, Area K, Period 3

294. Not illustrated. Sub-globular head with one cordon. In poor condition, point missing. L. 62mm, head diam. 6mm. SF1070, Layer 10174, Group 797, Area F, Period 3

Group 9 Triple unit standard heads

Two examples (Figure 444). This is a small group of hairpins, which may have been made at a single workshop. The distribution is limited, with most examples having been found in south-east England, and it is possible to say that the group was in use by AD 125. Neither of the Elms Farm hairpins was from a dated context.

295. Head only; a flat ring with the top missing, with a rectangular collar below, a variation on the examples illustrated by Cool, which have a rectangular section at this point, but no collar as such. Below this is a cross-hatched drum. Surviving L. 28mm, head W. 9mm. There is a very similar hairpin from Kelvedon (Rodwell 1988, fig. 49, no. 59). SF3218, Cleaning layer 5703, Area J, not dated

296. Not illustrated. In poor condition and incomplete, with most of the head missing. It has a rectangular collar, as SF3218. L. 48mm. SF3392, Cleaning layer 6515, Area H, not dated

Group 10, sub-group A Knob head with grooved triangle decoration

One example. Sub-group A of this group is common in Essex. The type was in use by AD 125.

297. Not illustrated. Head only. Bi-conical, with three pairs of grooves forming a triangle on the top, and a circumferential groove on the base of the head. This is similar to a pin from Chignall Roman Villa, though with a flatter head (Major 1998c, 74, no. 37). L. 34mm, head diam. 8mm. SF6646, Fill 15516, Slot 15517, Group 240, Area M, Period 3

Group 12 Knob heads with incised cross decoration

One example (Figure 444). The type was in use by AD 125.

298. Complete but bent. The top of the head has a cross formed by intersecting double lines, with a single line round the edge. Any incised decoration that may have been present on the underside of the head is masked by corrosion. Cool notes that the cross pattern is unusual, and that the pins show a high degree of similarity. While the head on this one is similar to her illustrated examples, the top of the shaft is treated somewhat differently, comprising a rather delicate baluster moulding between two small cordons. The top of the shaft expands slightly below the cordon. The distribution is in North Kent and London, and this may be the only example from Essex. The shaft is in good condition, but the head is poorly preserved. L. 100mm, head diam. 8mm. SF408, Cleaning layer 4149, Area K, not dated

Group 14, sub-group B Heads with glass insets

Two examples (Figure 444). There were two dissimilar examples of Cool group 14, sub-group B, which is of late 3rd to 4th century date. One of the examples is from a context earlier than the date given by Cool for the sub-type, layer 6790, which is later 1st-mid-2nd century. However, the other example is from layer 6025, which physically overlay 6790, and it is possible that the hairpin from the earlier context was intrusive from the overlying layer.

299. Head only, in poor condition, with the top of the cup incomplete and damaged. The head is square sectioned, with a sub-globular green glass inset (now detached) and grooves below the head. The sides of the head have incised decoration, probably Vs. Surviving L. 38mm, head W. 6mm, glass inset diam. 4mm. SF892, Layer 6025, Group 573, Area H, Period 5-6

300. Incomplete and in fairly poor condition, with the glass inset missing. Circular sectioned head with moulded cross-hatching, as Cool 1990, fig. 9.5, but with a cordon below, above a single reel at the top of the shank. Surviving L. 70mm, head diam. 6mm. SF3439, Layer 6790, Group 484, Area H, Period 3

Group 15 Faceted cuboid heads

One example (Figure 444). This is one of the predominant forms of Late Roman pin (late 3rd-4th century), occurring in materials such as bone and jet as well as copper alloy. The group is very variable.

301. Complete but in three pieces, with an applied faceted head, now detached. Cool does not specifically mention two-piece pins, but fig. 9.9 and 9.14 appear to be possibly made from two pieces. The facets on the head are not very regular, and in this aspect it resembles fig. 9.9. The shaft is slender, with a slight groove down one side. The colours of the head and shaft are slightly different, and may be different alloys. AS notes that the corrosion products at the head of the shaft indicated the possibility of a slightly different composition, or some kind of solder, and the detached head had similar corrosion products. A similar pin with an applied head came from Colchester, although in that case the head was a glass bead (Crummy 1983, 29, no. 486). L. 52mm, head W. 5mm. SF6229, Fill 14528, Well 14529, Group 722, Area L, Period 6

Group 21 Hollow spherical heads

Four examples (Figure 444). The four hairpins in this group are of some importance, as they are the only ones of this type to have been found outside Colchester, and included the first ones from dated contexts. Cool suggested a 1st-2nd century date on stylistic grounds, and the dates for contexts 11113 and 13639 bear this out. Pit 11221 (fill 11113) is possibly no later than mid-1st century AD, giving a very early Roman date for the start of the type.

Three of the pins are extremely similar, with diamond cross-hatching on the top of the head, and incised or moulded circumferential lines; the fourth differs only in that the cross-hatching is absent. In one hairpin, decayed glass survives in the cup.

302. Head only, in poor condition, surface flaked. Cupped head with no glass or enamel inset surviving. The top of the head has incised cross-hatched decoration, with two moulded circumferential lines. There are traces of a cordon round the top of the shank. Similar to Cool 1990, fig. 11.9. Surviving L. 16mm, head diam. 8mm. SF8134, 3999, spoil heap

303. Head only, in poor condition, very similar to SF8134, but with one incised and one moulded circumferential line on the head. It is broken above the cordon, if there was one present, and there is no trace of glass or enamel in the setting. Surviving L. 9mm, head diam. 8mm. SF970, Machining layer 4000

304. Bent and incomplete. It has cross-hatched decoration on the head, as SFs 970 and 8134, one incised and one moulded circumferential line on the head, and a cordon below the head. There may be decayed glass or enamel in the central cup. Surviving L. 68mm, head diam. 7.5mm. SF5803, Fill 11133, Pit 11221, Group 226, Area N, Period 2

305. Shank incomplete, with decayed glass in the top. It is similar to the other hairpins in this group from the site, but lacks the incised cross-hatched decoration. It has one incised and one moulded circumferential line on the head, and a cordon at the top of the shank, which is almost square-sectioned. Surviving L. 59mm, head diam. 7.5mm. SF6094, Fill 13639, Pit 13640, Group 594, Area I, Period 3A

Group 22/13?

306. In poor condition, distorted and incomplete (Figure 444). The top of the head has vertical grooved lines, with a circumferential groove round the middle. The bottom of the head appears plain. There is possibly a groove round the top of the shaft, but this is unclear on the X-ray. This pin is closely similar to Cool groups 13 and 22, both with distributions principally in the West Country. The top of the pin is less of a 'cap' than on the illustrated examples of group 22, but the underside of the head is convex rather than concave, a principal feature of group 13. A source in the West Country is therefore very likely, with a date possibly in the later 2nd to 3rd century. L. 66mm. SF5632, Cleaning layer 6000, Area H, not dated

Other fragments

307. Tapering rod, bent towards the tip. The other end is flat topped, and complete as buried, though it could have been broken prior to burial. This is possibly a simple hairpin of Cool's group 24. L. 80mm. SF794, Machining layer 4000, Area A, unstratified (Figure 444)

308. Not illustrated. Probable sub-globular hairpin head, possibly with a cordon below. Alternatively it could be a tack head; the shank is rather irregular. Analysis by AS indicated an ordinary copper-tin alloy with a small amount of lead and the considerable presence of iron, which had caused an unusually thick dark patinated surface to occur. Surviving L. 12mm, head diam. 7mm. SF8192, 3999, spoil heap

309. Not illustrated. Point of hairpin or needle, probably the former. Slightly bent. L. 44mm. SF576, Fill 5146, Pit 5147, Group 409, Area J, Period 3

310. Not illustrated. Tapering rod, bent, both ends missing. Possibly a hairpin shaft. L. 87mm. SF2277, Fill 5539, Ditch 5540, Group 4013, Area J, Period 4

311. Not illustrated. Hairpin shaft, head and point missing, but probably nearly complete. Not enough is present to assign it to a group. The shaft is slender, and in poor condition, with one small collar surviving. L. 64mm. SF2279, Fill 5539, Ditch 5540, Group 4013, Area J, Period 4

312. Not illustrated. Rod fragment, possibly a hairpin shaft. Broken at both ends. L. 25mm. SF5865, 5601

313. Not illustrated. Hairpin in poor condition, head damaged and point missing. The top of the shaft had at least one square-sectioned block element. Possibly Cool group 11. L. 57mm. SF3278, Layer 9427, Group 1301, Area D, not phased

314. Not illustrated. Rod fragment in very poor condition, possibly not a hairpin. Both ends are broken, and there is little or no apparent taper. One end has four grooves, with rounded elements between. Such rounded elements are rare on hairpins, but occasionally occur, for example on Cool 1990 fig. 4.5, which is similarly robust. L. 103mm. SF2355, Machining layer 11000, Area A, unstratified

315. Not illustrated. Hairpin terminal? Very corroded, not X-rayed. It may have a sub-globular or onion shaped head with two cordons below. Possibly Cool group 2. L. 14mm. SF5941, Natural 15002, Area M, not phased

316. Not illustrated. Hairpin or tack head, similar to SF8192. Roughly globular, with a poorly finished shank. L. 17mm, head diam. 5mm. SF7176, Layer 16166, Group 573, Area H, Period 5-6

317. Not illustrated. A slender rod, complete as buried, with a bent point. It is probably a hairpin with the head snapped off. L. 104mm. SF7836, Fill 23025, Pit 23024, Group 906, Area N, Period 3

Necklaces and beads

Glass Beads by Sally Worrell

Cite this as: Worrell, S. 2015, Glass beads, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.worrell

Sixty Roman beads and bead fragments, in a broad variety of commonly found types, were recovered, plus half a Saxon bead. Short cylindrical and square-sectioned beads are the most numerous, accounting for nearly half of the total found. Of note is a complete oval, translucent purple bead with facets (no. 369). There are just two complete melon beads and a further two fragments. Melon beads are generally more numerous on sites with military associations. A scatter of beads and objects was found across the site, with higher incidences in Areas F and L and a concentration in Area J. It is notable that more than twice the average number of bead finds per area came from the vicinity of the temple (Area J). Almost a third of the total beads present were recovered from bulk soil samples. The highest proportion of beads and objects came from mid-Roman contexts, although many of the bead types present date to the 1st and 2nd centuries. Two complete beads were recovered from funerary features; a green cylindrical bead came from the fill of cremation burial 12203 and a tiny globular bead from pyre feature 2189. Small numbers of beads have been found on most Roman sites in Essex. The highest number, apart from Colchester, came from Kelvedon (Rodwell 1988, 76), although, as with Colchester, many of these were retrieved as necklaces and bracelets from burials.

The beads from the cremation burials are described with the burial groups.

Annular beads

A range of annular beads is illustrated in Crummy (1983, fig. 33). This type of bead, especially when decorated, is more common in the Late Iron Age and again in the Late Roman and post-Roman periods, when native influences were stronger (Guido 1978, 101). Number 318 (Figure 445) is purple, an unusual colour for glass beads as these decorated annular beads are commonly dark blue. There are two purple annular beads from Usk (Price 1995, fig. 31.3; fig. 31.6); the former is similar to the Elms Farm bead, but the latter was probably formed from a reheated fragment of polychrome mosaic vessel.

318. Fragment, c. 50% annular bead. Purple with irregular, swirling, opaque white trails. Perforation slightly irregular and lightly chipped. Diam. 25mm, perf. diam. 11mm, height 8.7-9.7mm. SF1380, Fill 8141, Pit 8142, Group 836, Area E, Period 5

319. Complete annular bead. Mid-blue. Irregular circumference. Bubbly. Diam. 13mm, perf. diam. 8mm, height 2-4.5mm. SF7371, Layer 13468, Group 600, Area I, Period 3B

320. Complete small annular bead. Mid-blue with black specks. Irregular circumference. Diam. 9.5-10.5mm, perf. diam. 6mm, height 2-4mm. SF5237, Fill 10625, Unknown 10650, Area F, Period 3

321. Not illustrated. Fragment, c. 50% annular bead. Dark blue. Irregular section. Diam. 10mm, perf. diam. 6mm, height 3-4.5mm. SF5238, Cleaning layer 10513, Area F

Globular beads

322. Complete small globular bead. Dark blue. Diam. 4mm, perf. diam. 1.5mm, height 4.5mm. SF6220, Cleaning layer 14516, Area L

323. Complete small globular bead. Dark green. Light iridescence. Diam. 3.5mm, perf. diam. 0.5mm, height 3.5mm. SF5157, Cleaning layer 5603, Area I

324. Complete small globular bead. Green. Diam. 2.5mm, perf. diam. 0.5mm, height 1mm. SF6567, Fill 14058, Pit 14006, Group 753, Area K, Period 3

325. Not illustrated. Two joining fragments, small globular bead. Blue/green. Diam. 4.5mm, perf. diam. 0.12mm, height 4mm. SF6839, Fill 10381, Hearth construction 10501, Group 793, Area F, Period 3

326. Not illustrated. Two joining fragments, ?small globular bead. Turquoise with opaque yellow interior. Diam. 3.5mm, perf. diam. 0.5mm, height 4.5mm. SF3295, Fill 9491, Ditch 9520, Group 776, Area D, Period 3-4

Segmented beads

A common Roman bead type made by crimping a rod of glass along its length and breaking the resulting crimps into single beads or into segments (Guido 1978, types 1 and 2; Figure 445).

327. Not illustrated. Fragment, ?segmented bead. Blue/green. Part of one segment. Diam. 3.5mm, perf. diam. 0.7mm, length 4mm. SF7369, Fill 13317, Pit 13358, Group 645, Area I, Period 5

328. Not illustrated. Fragment, ?segmented bead. Dark green. Part of one segment. Diam. 3mm. SF8469, Fill 14621, Kiln construction 14858, Group 714, Area L, Period 5

329. Complete long wound cylindrical bead. Mid-blue. Tapering at one end. Circular section. Black specks. Diam. 2.5-3mm, perf. diam. 0.5-1.2mm, length 16.9mm. SF7370, Fill 14525, Kiln construction 14858, Group 714, Area L, Period 5

330. Fragment, wound and twisted bead. Dark appearing brown. Diam. 1.3mm, perf. diam. 0.2mm, length 5.7mm. Fill 5430, Pit 5431, Group 24, Area J, Period 2B

Long cylindrical beads

Guido's cylinder beads (1978, types 4 and 5), made almost always from either blue or green semi-opaque glass (Figure 445).

331. Complete long cylindrical bead. Dark blue. Faint horizontal striations. Ends at perforations chipped. Diam. 4.1mm, perf. diam. 0.5mm, length 5.7mm. SF560, Spread 5148, Group 457, Area J, Period 5-6

332. Complete long cylindrical bead. Dark blue/green. Horizontal striations. Diam. 4.5mm, perf. diam. 0.8mm, length 15mm. SF6995, Cleaning layer 11413, Area N, not phased

333. Not illustrated. Complete long cylindrical bead. Dark green. Horizontal striations. Slightly rounded ends. Diam. 3mm, perf. diam. 0.7mm, length 5.5mm. SF6746, 17000, unstratified, Area A4

334. Not illustrated. Complete long cylindrical bead. Pale blue/green. Slightly flattened at one end. Horizontal striations. Diam. 2.5-3mm, perf. diam. 0.5mm, length 8.5mm. SF6570, Fill 4692, Slot 4695, Group 1161, Area K, not dated

335. Not illustrated. Fragment, long cylindrical bead. Green. Horizontal striations. Diam. c. 3mm, perf. diam. c. 0.7mm. SF8415, Fill 11478, Pit 11731, Group 665, Area N, Period 3

336. Not illustrated. Fragment, long cylindrical bead. Blue/green. Surface pitted. Diam. 2mm, perf. diam. 0.5mm, length 7mm. SF1290, Cleaning layer 5228, Area J, Period 5-6

Short cylindrical beads

Figure 445. Guido's types 4 and 5; the shorter version of long cylindrical beads (see above).

337. Complete short cylindrical bead. Dark blue. Ends flattened and worn. Surfaces scratched. Diam. 7.5mm, perf. diam. 3mm, height 3mm. SF2228, Cleaning layer 10237, Area F, not phased

338. Not illustrated. Complete small cylindrical bead. Dark blue. Diam. 2mm, perf. diam. 0.5mm, height 1mm. SF8209, Layer 20451, Group 4021, Area L, Period 4-5

339. Complete short cylindrical bead. Opaque mid-green. Perforation off-centre. Diam. 5.5mm, perf. diam. 1.5-2mm, height 2-3.1mm. SF568, Spread 5148, Group 457, Area J, Period 5-6

340. Not illustrated. Complete short cylindrical bead. Opaque dark green. One end chipped. Diam. 5.7mm, perf. diam. 1.3-1.5mm, height 3.5mm. SF1920, Cleaning layer 5383, Area J, Period 6

341. Complete short cylindrical bead. Opaque mid-green. Diam. 5.4mm, perf. diam. 1mm, height 2.3mm. SF4031, Fill 10396, Ditch 10538, Group 838, Area F, Period 5-6

342. Not illustrated. Complete short cylindrical bead. Opaque light green. Slightly worn at one perforation. Diam. 5mm, perf. diam. 1.5-2mm, height 3-3.5mm. SF7277, Machining layer 5000, Area J

343. Not illustrated. Complete short cylindrical bead. Opaque green. End at one perforation worn. Diam. 5.5mm, perf. diam. 1.5-2mm, height 2.5-4mm. SF6241, Fill 14558, Pit 14529, Group 722, Area L, Period 6

344. Not illustrated. Complete short cylindrical bead. Opaque green. Diam. 5.5mm, perf. diam. 2-3mm, height 3mm. Diam. 5.5mm, perf. diam. 2-3mm, height 3mm. SF949, Machining layer 4000, Area A1

345. Not illustrated. Complete short cylindrical bead. Semi-opaque mid-green. Slightly truncated. Diam. 5.5mm, perf. dia 2mm, height 3mm. SF6945, Machining layer 17000, Area A4

346. Not illustrated. Complete short cylindrical bead. Opaque mid-green. Diam. 4-4.5mm, perf. diam. 1.5-2mm, height 2.5-3mm. SF6569, Fill 4819, Pit 4780, Group 4016, Area K, Period 4

347. Not illustrated. Complete short cylindrical bead. Opaque green. Diam. 3mm, perf. diam. 0.3mm, height 1mm. SF3878, Fill 8076, Pit 8152, Area E, Period 6

348. Complete short cylindrical bead. Opaque green. Diam. 3.4mm, perf. diam. 0.2mm, height 1mm. SF3880, Fill 5214, Pit 5209, Group 442, Area J, Period 5-6

349. Not illustrated. Fragment, c. 50% short cylindrical bead. Green. Diam. 3mm, perf. diam. 0.4mm, height 2mm. SF7877, Fill 20031, Pit 20030, Group 248, Area L, Period 2B

350. Not illustrated. Complete short cylindrical bead. ?Blue/green. Heavy iridescence. Diam. 2.5mm, perf. diam. 0.12mm, height 1.5mm. SF7367, Fill 10891, Pit 10910, Group 676, Area N, Period 5

Square-sectioned beads

These beads are also normally made from opaque blue or green glass (Guido 1978, types 6 and 7) and are also very common (Figure 445).

351. Complete square-sectioned bead. Semi-opaque mid-blue. Length 3.3mm, perf. diam. 0.6-1mm, height 2.5mm. SF4942, Fill 15059, Post-hole 15090, Group 1223, Area M, not dated

352. Complete square-sectioned bead. Opaque dark blue. Tapering slightly at one end. One perforation worn. Length 4mm, perf. diam. 0.5-1.5mm, height 2.5-3.5mm. SF1459, Fill 10001, Pit 10018, Group 837, Area E, Period 5

353. Not illustrated. Almost complete, square-sectioned bead. Opaque mid-blue. Worn at both ends. Length 5mm, perf. diam. 1mm, height 2.5mm. SF7935, Layer 24210, Group 9010, Area M, not dated

354. Not illustrated. Complete square-sectioned bead. Opaque mid-blue. Slightly irregular section. Worn at one perforation. Length 4mm, perf. diam. 1.5mm, height 3mm. SF7518, Cleaning layer 20334, Area L, not phased

355. Not illustrated. Fragment, c.50% square-sectioned bead. Dark blue. Length 7mm, height 3mm. SF2870, Fill 4392, Pit 4429, Group 741, Area K, Period 5

356. Not illustrated. Fragment, square-sectioned bead. Opaque mid-blue. Length 5mm, perf. diam. 0.5mm, height 2.5mm. SF4770, Layer 12041, Group 966, Area R, Period 4

357. Not illustrated. Two fragments from square-sectioned beads, one semi-opaque dark blue, one opaque mid-blue. SF7181, Fill 16182, Robber trench 16197, Area H, Period 5; SF8412, Fill 14093, Pit 14099, Group 4019, Area K, Period 4-5

358. Complete square-sectioned bead. Dark green. Worn at both perforations. Length 6mm, perf. diam. 0.5-1mm, height 4mm. SF5107, Cleaning layer 5628, Area I

359. Complete square-sectioned bead. Blue/green. Irregular; tapering at one end. Some black specks. Length 14.5mm, perf. diam. 1.5-2.5mm, height 4.5-6.5mm. SF4766, Fill 12033, Pit 12034, Group 967, Area R, Period 4

Polygonal beads

Beads that are normally hexagonal in cross-section (Guido 1978, types 8-10; Figure 445).

360. Complete hexagonal-sectioned bead. Dark green. One perforation serrated. Length 10.7mm, perf. diam. 1.5mm, height 5.5mm. SF4768, Layer 12056, Group 966, Area R, Period 4

361. Complete hexagonal-sectioned bead. Dark green. Two sides flattened. Length 5.2mm, perf. diam. 0.7mm, height 2.8-4mm. SF514, 7000, unstratified, Area G

Conical bead

This bead shape is uncommon and could be the result of faulty manufacture, or was, perhaps, a custom-made bead (Figure 445).

362. Complete small conical bead. Translucent mid-blue. Length 4mm, width 2.5-4mm, perf. diam. 0.5-0.8mm, height 3mm. SF6577, Fill 10396, Ditch 10538, Group 838, Area F, Period 5-6

Bi-conical beads

These beads can be either short or long (Guido 1978, types 12-14). Only one long bi-conical bead was recovered (Figure 445, no. 368).

363. Complete short biconical bead. Translucent mid-blue. Diam. 4mm, perf. diam. 0.4mm, height 1.5mm. SF6568, Cleaning layer 4534, Area K

364. Not illustrated. Complete short biconical bead. Dark green. Diam. 5mm, perf. diam. 1-1.2mm, height 2mm. SF7875, Fill 14591, Kiln construction 14858, Group 714, Area L, Period 5

365. Not illustrated. Complete short biconical bead. Dark green. Slightly chipped. Light iridescence. Diam. 4.3mm, perf. diam. 0.5mm, height 1.5mm. SF6463, 5000, unstratified, Area J

366. Complete short biconical bead. Opaque green. Diam. 3.5mm, perf. diam. 0.2-0.6mm, height 1.5mm. SF7740, Fill 21974, Pit 21972, Group 23, Area J, Period 2A

367. Not illustrated. Complete small biconical bead. Opaque green. Diam. 3.3mm, perf. diam. 0.7mm, height 1mm. SF7192, Fill 16230, Ditch 16231, Group 584, Area H, Period 6

368. Almost complete, long biconical bead. Dark blue. One end chipped. Diam. 1.8-3.8mm, perf. diam. 0.5mm, length 10.7mm. SF1458, Fill 10001, Pit 10018, Group 837, Area E, Period 5

Miscellaneous small beads

The facetted bead, no. 369, is very unusual and there is some doubt as to whether the bead is made of glass. There are no bubbles present in the metal and the bead has been shaped in the same way as for a gemstone. The bead might be a semi-precious stone such as amethyst, but has been drilled along the long axis to produce a bead, rather than made for use as a ring setting. A circular, flat-sectioned bead of translucent purple glass was found at the Fortress Baths in Caerleon (Brewer 1986, fig. 48.66), where it was noted that beads in purple glass were not recorded in Guido (1978). The diamond-sectioned bead (Figure 445, 370) is also an unusual shape, cf. no. 362.

369. Complete oval bead with facets. Translucent purple. Heavily scratched surfaces. Length 16mm, width 5mm, perf. diam. 0.5mm. SF2207, Fill 10165, Pit 10166, Group 802, Area F, Period 3

370. Complete diamond-sectioned bead. Opaque mid-blue. White specks and streaks. Length 5mm, width 3.5mm, perf. diam. 0.5-1mm, height 6mm. SF7679, Layer 21613, Group 424, Area J, Period 4

Faience melon beads

These are common finds on most Roman sites, although normally their occurrence is restricted to the 1st and 2nd centuries AD. They are unknown in the Late Iron Age period. A description is provided in Guido (1978, 100, types 21 and 22) and also in Crummy (1983, fig. 32.520-1). Only beads in turquoise faience were found at Elms Farm, although melon beads in dark blue glass are equally common (Figure 445).

371. Complete faience melon bead. Bright blue. Seventeen slightly curving vertical grooves. Small area chipped before glazing. Slightly worn at perforations and light wear around centre of ribs. Diam. 19.5mm, perf. diam. 9mm, height 15-17mm. SF504, 7000, unstratified, Area G

372. Complete faience melon bead. Turquoise. Twenty-one slightly curving vertical grooves. Surface of ribs and edges of perforations worn. Diam. 21.5mm, perf. diam. 8mm, height 15-18mm. SF3269, Fill 9284, Ditch 9283, Group 3063, Area D, Period 3-4

373. Not illustrated. Melon bead fragment. Pale blue/green with grey/brown core. Five ribs extant. Height 14mm. SF6840, Fill 4150, Pit 4139, Group 744, Area K, Period 6

374. Not illustrated. Melon bead fragment. Bright blue with grey core. Chipped and worn. Five ribs extant. Height 13mm. SF3400, Layer 6418, Group 506, Area H, Period 3

Post-Roman bead

Beads made from opaque yellow glass, and with applied opaque blobs, are common in Saxon contexts. The Elms Farm example is unstratified (Figure 445).

375. Fragment, c. 60% annular bead. Opaque yellow and red. Three circular applied blobs running around the centre of the circumference of the bead. Diam. 11mm, perf. diam. 3mm, height 6.8mm. SF6811, 17000, unstratified, Area A4

Jet and shale beads by Ros Tyrrell and Hilary Major

Cite this as: Tyrrell, R. and Major, H. 2015, Jet and shale beads, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.major3

The site produced fifteen jet and shale beads; a group of eight and seven individual finds. The beads are made from jet, apart from SF8468, which is shale. All but one is cylindrical in shape and nine of these are transversely drilled with a pair of holes, rather than the usual single longitudinal hole. Whether these were used in segmented bracelets or necklaces is uncertain. The only non-cylindrical bead is a semicircular, wedge-shaped bead. Apart from one very small cylindrical example, which is from a 2nd-century layer in Area I, all the stratified beads were found in later contexts, dated between the mid-3rd and the late 4th centuries. The group of eight transversely drilled beads were found in a pit with late 3rd to mid-4th-century pottery.

The single shale bead was turned, as is evident from the chuck mark at one end. Presumably, the bead was cut from a longer lathe-turned rod, and only the two end beads from this rod would have chuck marks. In some cases, beads were turned individually, as can be seen on a jet bead of the same form from Colchester (Crummy 1983, 33, no. 974), which has a chuck mark at either end.

The following are jet unless otherwise specified.

376. Cylindrical bead with a circumferential groove. SF3879, Fill 6152, Pit 6151, Area H, Period 4

377. Eight cylindrical beads, transversely drilled with two holes. Two are incomplete, and the others range in length from 10mm to 15mm, though the diameter is constant (4mm). SF7372, Fill 15043, Pit 15042, Group 650, Area M, Period 6 (two illustrated, one of each size)

378. A thin, semi-circular bead, wedge-shaped in section, pierced twice, from a segmented bracelet. The edges of these beads are often chip-carved, but this one is worn to a gloss and the pattern is obscure. A similar example was found in a late 3rd to 4th-century context in Colchester (Crummy 1983, 35). SF1950, Fill 4692, Slot 4695, Group 1161, Area K, not phased

379. Not illustrated. Cylindrical bead with a circumferential groove. One end is damaged. If the groove was set centrally, the original length would have been 6mm. L. 5mm, diam. 2mm. SF7366, 16102, Hearth construction 16108, Group 559, Area H, Period 4

380. Not illustrated. Part of a cylindrical bead, with two circumferential grooves,1.5mm apart. It is broken across the second groove. L. 3mm, diam. 3mm. SF6578, Fill 10396, Ditch 10538, Group 838, Area F, Period 5-6

381. Shale. Cylindrical, transversely drilled with two holes. Despite being such a small object, the bead was turned rather than carved, as it has the chuck mark on one end. This end is flat, while the other is chamfered round the edge. L. 8mm, diam. 6.5mm. SF8468, Fill 7597, Beam slot 7601, Group 9024, Area G, not phased

382. Not illustrated. A short, plain, cylindrical bead. L. 2mm, diam. 4mm. SF7365, Layer 13568, Group 600, Area I, Period 3B

Amber

Three small fragments of amber were recovered from a Period 4 spread in Area F (SF4675, context 10330). Amber is very rare on Roman sites, but, where it occurs, it is generally in the form of beads. The amber from Elms Farm was in poor condition, with no indication of utilisation, but is most likely to be the weathered remains of beads.

Copper alloy and silver

There were ten copper-alloy beads (or possible beads), and one silver bead, which is an unusually large number for a Roman site (Figure 446). Very few sites produce even a single copper-alloy bead. Only three are from dated contexts, but these span the entire Roman period. Most are relatively large annular or bun-shaped beads, occasionally with simple circumferential lines, but two are barrel-shaped. The silver 'washer' is an unusual form of bead paralleled at Catterick, where it was associated with a magnificent multi-stranded jet necklace.

383. Silver. A flat, hexagonal bead or 'washer' with a relatively large central hole, slightly irregular in shape. A silver bead with a similar large hole, but with a wavy edge, was found at Catterick, associated with a multi-stranded jet necklace, dated to c. 275-350 (Mould 2002a). External W. 13mm, internal diam. 10mm, Th. 1.5mm. SF2356, Machining layer 1100, Area A, unstratified

384. Annular bead. It is not a perfect circle, but is slightly flattened on one side, and the section here is almost triangular, rather than rounded, as with the remainder of the bead. This is presumably a casting flaw, as it is difficult to see how it could have worn to this shape if used as a necklace bead. In good condition, with a dark grey-green patina. XRF analysis showed that it was a leaded copper-alloy with a trace of nickel. External diam. 13mm, internal diam. 6mm, Ht 5mm. SF818, Machining layer 4000, Area A, unstratified

385. Flattened biconical, not very symmetrical, with a circumferential line either side of the middle. There are two similar, though slightly smaller, beads from the Airport Catering Site, Stansted, one from a very early Roman context (Major 2004c, 130, nos 12 and 13). In good condition. Diam. 14.5mm, Ht 7.5mm. SF839, Machining layer 4000, Area A, unstratified

386. Barrel-shaped, with a groove round either end. Complete bar damage to the ends. Condition fair. L. 8mm, diam. 4mm. SF1285, Spread 5148, Group 457, Area J, Period 5-6

387. Bead? Ring-shaped, with one side slightly constricted. The central hole is perhaps a bit big for a bead. Surface flaking. External diam. 21mm, hole diam. 12mm, Ht 12mm. SF1126, Fill 7123, Pit 7122, Group 868, Area G, Period 4

388. Not illustrated. Short barrel-shaped with large central hole. Diam. 11mm, Ht 7.5mm, hole diam. 7mm. SF2423, Layer 7463, Group 871, Area G, Period 5

389. Not illustrated. Slightly oval bun-shape, with a round hole. 10x12mm, Ht 7mm, hole diam. 6mm. SF5700, Cleaning layer 8000, Area E, not dated.

390. Bun-shaped, with a line round the hole at one end, but probably not the other. The surface is in poor condition. The patina is very grey in places, and this may be coated with white metal. Diam. 14.5mm, Ht 9.5mm. SF6083, Machining layer 11000, Area A, unstratified

391. Annular bead, with two lines round the middle. It is fairly roughly finished, but in good condition. Diam. 16mm, internal diam. 6mm, Ht 6mm. SF7194, Machining layer 11000, Area A, unstratified

392. Not illustrated. Large annular bead or small ring weight. Diam. 16mm, Ht 8mm, hole diam. 4mm. SF6819, Machining layer 17000, Area A, unstratified.

393. Perforated flattened sphere, in poor condition. Either a large bead or a small weight. Diam. 17mm, Ht 13mm. SF7809, Cleaning layer 24058, Area M, not dated

Bracelets

Copper alloy and silver

There were thirty-seven copper-alloy and silver bracelets, and bracelet fragments, mostly later Roman types. A possible child's bracelet, made from a cut-down cable bracelet, has been included with the finger-rings.

This cannot be considered a large group of bracelets, and although a third of them come from Area J and could therefore be seen as possible votives, the votive deposition of copper-alloy bracelets was probably not a strong tradition at this site. The group certainly does not stand out as numerically significant, as the 300 or so bracelets from Lydney do, for example. The three bracelets with the most obvious religious connotations, the snake bracelets, do not come from the temple area.

Many of the later Roman decorated bracelets can be paralleled at Colchester, and for some at least, a common source is likely. Number 400, for example, has closely spaced punched ring-and-dots very similar to those on a bracelet from the Butt Road cemetery (Crummy 1983, 44, no. 1708). In her study of the distribution of Late Roman bracelets, Swift (2000, 136) notes that this motif (b1) is more common in Britannia than on the continent, and that it has a south-western bias. Their variability suggests that they not all were made in the same place, and the two known Essex examples may be part of a sub-group. The spacing of the ring-and-dots may be relevant; those on a bracelet from Frocester, for example, are widely spaced (E.G. Price 2000, 45, no. 187).

The bracelets are copper alloy unless otherwise specified (Figure 446).

Twisted single-strand bracelets

394. Bracelet, made from a single twisted strand, Lankhills type B2. One end has the original hook (as Clarke 1979, fig. 69, nos 30 and 31). The other end was cut in antiquity, and the bracelet bent so that the ends overlapped, forming an almost square bracelet. The current size would fit a child (internal diam. 25mm). The original size (assuming that the curve nearest the hook is original) would have been c. 40mm. SF1928, Unknown context 5385, Area J, not dated

395. Twisted single strand of square-sectioned wire. The rounded terminal is probably complete, though damaged, with a small reel below, somewhat like an acorn. One face only, adjacent to the terminal, has shallow transverse lines. The twisting is uneven, and tighter towards the broken end. This is probably a distorted bracelet or armlet, though the terminal is unusual that of a bracelet, and this may be a distorted drop handle, or even a toilet implement handle. L. 107mm. SF3405, Layer 6575, Group 1000, Area H, not dated

Decorated strip-bracelets with a rectangular cross-section

396. Not illustrated. Bracelet, tapering fragment from near the clasp. In fairly good condition, surface partly obscured by earth. The clasp end has a recent break, the other end an old break. It has a central line with notches either side, similar to SF2315, above, but this bracelet is thinner. The decoration is as the end element of Crummy 1983, 45, no. 1725, which is from a grave deposit, dated c. 320-450. L. 31mm, W. 4-7mm, Th. 1mm. SF1935, Layer 4689, Group 4019, Area K, Period 4-5

397. Bracelet fragment. Invariable rectangular section, with ring-and-dot decoration. Flattened, with both ends cut, and one end bent. In very good condition. L. 42mm, section 4x1.5mm. SF2115, Cleaning layer 5306, Area J, Period 5-6

398. Bracelet fragment with tapering rectangular section. It has crisply executed notched decoration and ring-and-dots. The fragment has been flattened, and the tip is turned up, as if to form a hook, although the end has clearly been cut through the decoration. In good condition, with slight damage to the edge. L. 54mm. SF2259, Layer 5416, Group 453, Area J, Period 6

399. Not illustrated. Bracelet fragment. Distorted, and with a fresh break across the terminal loop. The decoration consists of two transverse lines above the loop, then a central longitudinal line with angled notches either side. The other end was broken in antiquity. In good condition. L. (flat) 51mm, W. 3mm. SF2315, Layer 5434, Group 1000, Area J, not dated

400. Strip-bracelet with complex ring-and-dot and linear decoration. Nearly complete, but flattened. The decoration is similar in layout to a bracelet from Colchester, though this example is more elaborate; certainly, a common source can be postulated (Crummy 1983, 45, no. 1731). The Colchester bracelet is from a grave, dated c. 320-450. There is a further close parallel from Lydney Park (Wheeler and Wheeler 1932, fig. 17F). This is in fair condition where the surface survives, but the edges are badly damaged. L. 147mm, W. 5.5mm. SF2875, Machining layer 11000, Area A, unstratified

Snake bracelets

There are two definite snake bracelets from the site, and one probable one, all with the snake motif very debased. None is from dated contexts.

401. Complete, bar damage to each terminal, the shape of which is uncertain. A strip-bracelet with D-shaped section of almost constant width. There are transverse lines across the terminals, and probably longitudinal moulding, but the surface is in too poor condition to be certain of the pattern. However, this was probably a snake bracelet, similar to the other two from the site. W. 4mm, oval c. 50x41mm. SF431, Cleaning layer 6000, Area H, not dated

402. Terminal. D-shaped section with a flared, squared end with a band of transverse lines and three raised converging lines. The terminal may have been shortened; it is similar to 20334, SF7514, but the latter bracelet has a longer terminal. They are possibly parts of the same bracelet, and were found fairly close together. L. 22mm, W. 4-5mm. SF7513, Cleaning layer 20184, Area L, not dated

403. Terminal, oval sectioned, with a flattened terminal. It has a pattern of raised lines and dots below a transverse ribbed band. In good condition. 20184, SF7513 has a very similar motif. SF7514, Cleaning layer 20334, Area L, not dated

Decorated, simple section

404. Bracelet fragment, flattened, with a lentoid section, decorated with bands of transverse grooves. In fairly good condition, with some recent damage. L. 55mm, W. 3mm. SF577, Fill 5161, Pit 5178, Group 437, Area J, Period 5

405. Bracelet fragment. A strip with an oval section, and two groups of sharply defined transverse grooves on the outer face, separated by a plain panel. The plain panel has traces of white-metal coating, but too little survives for XRF analysis. It is comparable to a bracelet from an inhumation at Colchester, also with white-metal coating, dated c. 320-450 (Crummy 1983, 40, no. 1688). L. 25mm, W. 2.5mm, Th. 1.5mm. SF3464, Layer 10255, Group 840, Area F, Period 5

406. Not illustrated. Bracelet? fragment. Distorted, tapering strip, surface mostly missing, with three transverse grooves across one end. Possibly a bracelet terminal, cf. Crummy 1983, 45, no. 1717, although rather chunky. L. 68mm, max. section 5x3.5mm. SF5116, Cleaning layer 13316, Area I, not dated

407. Strip bracelet of almost constant width; two joining fragments, with cast longitudinal grooves, and stamped transverse grooves across the central ridge. The moulding is crisply executed. The punch used for the transverse grooves produced a line with a small dot at each end; in places not all of the impression has come out, giving the appearance of separate lines of dots flanking the central ridge. Only the smaller fragment is illustrated. In good condition. L. 81mm, section 4x1mm. SF6258, Cleaning layer 14737, Area L, not dated

408. Not illustrated. Two joining fragments of a bracelet. The section is circular, with deep grooves across the outer face at 2mm intervals, giving a beaded effect. In poor condition. L. 16mm, diam. 2.5mm. SS 472. SF8375, unstratified in Pit 15232, Group 471, Area M, Period 6

Plain, simple section

409. Not illustrated. Four fragments of flat strip, in very poor condition. Probably fragments of a strip bracelet. 20x5mm; 20x5mm; 15x4mm; 14x4mm. SF4907, Fill 9582, Ditch 9581, Group 770, Area D, Period 3B

410. Not illustrated. Bracelet fragment, distorted, but in fairly good condition. Plain D-shaped section with one slightly bulbous terminal, changing to a circular section towards the middle of the bracelet. The two right-angled bends appear to be deliberate. Original diam. c. 80mm, max. section 5x4mm. SF4741, Fill 13387, Pit 18019, Group 3024, Area J, Period 3

411. Not illustrated. Half a ring, with a plain D-shaped section, surface partly eroded. Probably a bracelet. External diam. 48mm, internal diam. 40mm. SF9528, Context 3999, Spoil heap

412. Not illustrated. c. 75% of a bracelet, with a D-shaped section of almost constant width. The hook end was broken in antiquity, the other end has a recent break, and there is a crack across the middle. Original diam. c. 45mm, section 3x1.5mm. SF1931, Cleaning layer 4683, Area K, not phased

413. Bracelet, c. 50%, now distorted; plain, with variable section, circular to oval. In fair condition, surface flaking. Original internal diam. c. 50mm. It was probably similar to either of two bracelets from Colchester with variable sections, one with a hook and eye fastening, the other with an expanding clasp, both from Late Roman graves (Crummy 1983, 38, nos 1650-1651). SF1121, Layer 7073, Group 1319, Area G, not dated

414. Not illustrated. Probable bracelet fragment, with a plain, circular section. A child's bracelet? In poor condition, with very little of the surface surviving. Internal diam. c. 38mm, diam. 3mm. SF4897, Cleaning layer 9539, Area D, not phased

415. Not illustrated. Silver. Slightly curved strip, probably a bracelet fragment. The section is probably square, and there is no sign of decoration. The surface is now somewhat pitted. L. 23mm, section 3x2mm. SF5228, Cleaning layer 10569, Area F, not phased

416. Not illustrated. Plain bracelet fragment, with an oval section. In poor condition. Section 5x4mm, external diam. 60mm. SF2643, Machining layer 11000, Area A, unstratified

417. Not illustrated. Ring fragment, c. 25%, probably from a plain bracelet with a circular section. External diam. 58mm, internal diam. 48mm. SF4669, Machining layer 11000, Area A, unstratified

Wire armlets

418. Not illustrated. Probable bracelet fragment, comprising two pieces of wire, one wrapped round the other. This could be a sliding fastening, cf. Crummy 1983, 38, no. 1601, or part of a two-strand cable bracelet. In poor condition. L. 42mm, wire diam. 2.5mm. SF6500, Cleaning layer 5602, Area I, not dated

419. Not illustrated. Wire armlet, with twisted expanding clasp. Distorted, and in poor condition, but probably complete bar part of the clasp. Similar to an armlet from a 3rd to 4th century inhumation at Colchester (Crummy 1983, 38, no. 1601). Original diam. c. 90mm, max. diam. of section 3mm. SF2233, Layer 10255, Group 840, Area F, Period 5

Cable bracelets

Fragments from eleven cable bracelets were found. The type is well illustrated elsewhere, and none of the Elms Farm examples is of particular interest.

420. Not illustrated. c. 50% of a three-strand cable bracelet, broken at both ends. Internal diam. c. 46mm, Th. 3mm. SF3532, Build-up layer 5807, Area J, Period 3B

421. Not illustrated. Small fragment of three-strand cable, probably from a bracelet, although it could be a finger-ring. In fairly poor condition. Diam. 3mm. SF5445, Fill 14043, Pit 14098, Group 4019, Area K, Period 4-5

422. Not illustrated. Fragment; three-strand cable with loop fastening as Crummy 1983, 38, no. 1628. Broken in antiquity, in fair condition. Original diam. c. 60mm, Th. 3mm. SF2019, Cleaning layer 5307, Area J, Period 5-6

423. Not illustrated. Cable bracelet fragment, probably two-strand; in very poor condition, surface powdery. L. 20mm, diam. 4mm. SF3490, Fill 10296, Ditch 10406, Group 838, Area F, Period 5-6

424. Not illustrated. Four fragments of a four-strand cable bracelet. The largest is 20mm long. SF3502, Fill 10296, Ditch 10406, Group 838, Area F, Period 5-6

425. Not illustrated. Cable bracelet fragment. This has completely unravelled, but was not a very large fragment. The original number of strands is unclear. SF7175, Layer 16166, Group 573, Area H, Period 5-6

426. Not illustrated. Fragment of a two-strand cable bracelet, in very poor condition. L. 17mm. SF5942, Fill 15233, Pit 15232, Group 471, Area M, Period 6

427. Not illustrated. Small fragment of a three-strand cable bracelet fragment, made from unusually thin wire. In good condition. L. 23mm, diam. 2.5mm. SF6662, Fill 15639, Gully 15527, Group 146, Area M, Period 6

428. Not illustrated. Three-strand cable bracelet fragment. Broken both ends. L. 15mm, diam. 4mm. SF1016, Machining layer 4000, Area A, unstratified

429. Not illustrated. Three-strand cable bracelet; two joining fragments comprising c. 60% of the bracelet, with the looped fastener surviving at one end. Similar to Crummy 1983, 38, no. 1628. The condition is fair, the surface variable. External diam. 72mm, Th. 3mm, loop L. 4mm. SF2743, Machining layer 4000, Area A, unstratified

430. Not illustrated. Three joining fragments, forming c. 50% of a three-strand cable bracelet. (Two from 13565, SF6087 and one from 13691, SF6102; the break is fresh.) External diam. 70mm, diam. 3mm. Both contexts are poorly recorded, and this must be regarded as unstratified. SF6087/6102, Spread 13565 and cleaning layer 13691, both Area J, not phased

Shale bracelets by Ros Tyrrell

Twenty-six fragments of shale bangle or armlet were found, two of which joined. None of the pieces represented more than a quarter of a complete bracelet, and most were considerably less. The bracelets were found mostly in later Roman contexts, as shown in Table 79.

The bracelets were concentrated in the areas round the temple (Table 80), and Area J by itself had 38% of the total. However, most of the fragments from Area J came from only two features, a late 4th-century well [22210], which produced four fragments, and a nearby post-hole [22062] of a similar date that contained three.

Plain

431. Damaged. The section shape is uncertain but was probably plain and round. There is a transversely drilled hole, suggesting that this may be a hinged bracelet. L 70mm. Estimated internal diam. 80mm. SF3413, Cleaning layer 6609, Area H, not phased

Not illustrated

432. D-shaped section. L. 45mm, diam 11mm, estimated internal diam. 50mm. SF8522, Layer 5159, Group 457, Area J, Period 5-6

433. Fan-shaped section, rounded on the outer face and angled on the inner face. L. 30mm, diam. 6mm, estimated internal diam. 60mm. SF7759, Fill 22061, Post-hole 22062, Group 5007, Area J, Period 5-6

434. Oval in section. L. 30mm, diam 5mm, estimated internal diam. 60mm. SF4665, Cleaning layer 5662, Area I, not phased

435. Round sectioned. L. 47mm, diam. 9mm, estimated internal diam. 80mm. SF7453, Fill 15745, Pit 15744, Group 691, Area M, Period 3

436. D-shaped section, internal rough edge from turning. L. 24mm, diam. 4mm, estimated internal diam: 60mm. SF1840, Fill 4129, Pit 4128, Group 744, Area K, Period 6

437. D-shaped section, internal rough edge from turning. L 41mm, diam. 6mm, estimated internal diam. 60mm. SF7456, Fill 16182, Robber trench 16197, Group 567, Area H, Period 5

438. D-shaped section. L. 32mm, diam. 5mm, estimated internal diam. 50mm. SF549, Machining layer 7000, Area G

439. D-shaped section. L. 47mm, diam. 10mm, estimated internal diam. 80mm. SF8523, Fill 20009, Pit 2008, Group 892, Area L, Period 3

440. D-shaped section; not part of the same bracelet as SF8049 from the same context. L 60mm, diam. 7mm, estimated internal diam. 50mm. SF7953, Fill 22051, Well 22210, Group 448, Area J, Period 5-6

441. D-shaped section. L. 47mm, diam. 6mm, estimated internal diam. 50mm. SF8049, Fill 22051, Well 22210, Group 448, Area J, Period 5-6

442. Fan-shaped section, rough edge inside. L. 25mm, diam. 5mm, estimated internal diam. 60mm. SF7491, Fill 20194, Pit 20193, Group 718, Area L, Period 5

443. Fan-shaped section, rough edge inside. L. 40mm, diam. 7mm, estimated internal diam. 60mm. SF1013, Layer 6008, Group 590, Area H, Period 5-6

444. Oval in section, internal rough edges from turning. L. 29mm, diam. 4mm; Estimated internal diam. 40mm. SF4171, Fill 4910, Post-hole 4911, Group 1157, Area K, not phased

445. Oval in section, internal rough edges from turning. L. 52mm, diam. 8mm, estimated internal diam. 60mm. SF7640, Fill 14022, Pit 14098, Group 4019, Area K, Period 4-5

446. Oval in section. L 36mm, diam. 5mm, estimated internal diam. 60mm. SF7699, Cleaning layer 4881, Area K, not phased

447. Oval in section. L 31mm, diam. 7mm, estimated internal diam. 50mm. SF8076, Fill 24398, Well 22210, Group 448, Area J, Period 5-6

448. Oval in section. L 65mm, diam. 9mm, estimated internal diam. 60mm. SF7495, Fill 20203, Pit 20193, Group 718, Area L, Period 5

449. Oval in section. L 38mm, diam. 6mm, estimated internal diam. 60mm. SF7641, Fill 16182, Robber trench 16197, Group 567, Area H, Period 5

Decorated

450. Not illustrated. Trefoil-shaped in section. L 29mm, diam. 7mm, estimated internal diam. 50mm. SF7494, 20203, Pit 20193, Group 718, Area L, Period 5

451. Trefoil shaped in section. L 31mm, diam. 7mm, estimated internal diam. 70mm. SF574, Fill 5160, Pit 5179, Group 442, Area J, Period 5-6

452. Moulded and notched, with a rough inner edge. Estimated internal diam. 40mm. SF7968, Fill 22051, Well 22210, Group 448, Area J, Period 5-6

453. Moulded with ellipses and notches cut round the outer edge. Estimated internal diam. 60mm. SF8027, Fill 24071, Ditch 24070, Group 146, Area M, Period 6

454. Rectangular in section with 'V' shaped notches round the outer edge. Two non-joining pieces, a and b. Estimated internal diam. 60mm. SF7761 and SF7762, 22061, Post-hole 22062, Group 5007, Area J, Period 5-6

455. Not illustrated. Trefoil shaped in section. L 25mm, diam. 4mm, estimated internal diam. 70mm. SF7758, Fill 22061, Post-hole 22062, Group 5007, Area J, Period 5-6

Finger-rings and intaglios

Thirty-two finger-rings were found, comprising twenty-four in copper alloy, four silver rings, one iron, one glass, and a single gold ring. There was also a loose silver-gilt bezel appliqué, three loose intaglios, and three possible finger-ring fragments. The one ring-key found is described with the locks and keys. Only twelve of the rings were stratified, so their dating is very dependent on parallels. The main text cited for the metal finger-rings is Guiraud (1989), which, although specifically a typology of Romano-Gallic finger-rings, is equally useful for Romano-British rings.

The following are copper alloy unless otherwise specified (Figure 447).

Plain bands

456. Glass. Fragment from a finger-ring. Dark yellow/green appearing black. D-shaped section. Thickened in part of circumference. Max. thickness 4mm, internal diam. 15mm. SF1297, Cleaning layer 5228, Area J, Period 5-6

Glass finger-rings have been found on several Romano-British settlements, including Exeter (Charlesworth 1979, 230) and Birdoswald (Price and Cottam 1997, 283). Almost all are recorded from Late Roman contexts and appear to have been in circulation in the late 4th century, possibly continuing in use into the 5th century (Cool 2000, 50-6). The Elms Farm finger-ring fragment was recovered from a late 4th-century layer in the temple area.

457. Gold. Plain finger-ring. Internal diam. 19mm. SF5739, Machining layer 12257, Area R, unstratified

458. Silver. Plain, thin, circular section. The surface is now rather uneven owing to corrosion, but the ring was probably never very well finished. XRF analysis showed this to be silver, with copper and lead traces. External diam. 21mm, internal diam. 18mm. SF1841, Machining layer 4000, Area A, unstratified

459. Not illustrated. Finger-ring; a hoop with a plain, oval section. Complete, but in fairly poor condition, with the surface flaking. Internal diam. 18mm, section 3x2mm. SF4697, Machining layer 11000, Area A, unstratified

460. Not illustrated. 50% of a ring, probably a finger-ring, with a plain oval section. External diam. 19mm, section 3x2mm. SF4740, Layer 13419, Group 634, Area I, Period 2-6

461. Ring. A band with a flat section, possibly a large finger-ring, with two incised circumferential lines. Surface poor. Diam. 23mm, W. of band 6mm. SF2492, Machining layer 4000, Area A, unstratified

462. Not illustrated. Ring, probably a simple finger-ring. Plain thin hoop with an oval section. In good condition, surface obscured. External diam. 23mm, internal diam. 18mm. SF5718, Cleaning layer 5603, Area I, not dated

Penannular rings

463. A slightly tapering strip bent into a penannular ring. The wider end is probably complete, with a rounded terminal, the other end is probably broken. There appears to be a single punched dot in the middle of the terminal, although this may be an artefact of corrosion. Although not definitely a finger-ring, this would fit a child, and is possibly cut down from a bracelet. Internal diam. 12mm, W. 5-9mm, Th. 0.5mm. SF5684, Machining layer 11000, Area A, unstratified (Figure 447)

Spiral rings

Crummy (1983, 47) suggests an early date range for the type because of the high numbers from Sheepen, and it does seem to be a particularly common Late Iron Age form. However, they also occur in Roman and Early Saxon contexts (e.g. West Stow: S.E. West 1985, figs 225 and 263), and since neither of the Elms Farm examples is from a dated context, it cannot be assumed that they are early (Figure 447).

464. Spiral finger-ring, in good condition, with a dark greyish-green patina reminiscent of that often found on Late Roman bracelets. Two-and-a-half loose coils, made from oval-sectioned wire pointed at both ends. Internal diam. 15mm. SF1987, Cleaning layer 4604, Area K, not dated

465. Not illustrated. Spiral finger-ring, with one-and-a-half turns. In very poor condition, with little original surface left, and damaged. The core is a void, perhaps due to the metal having completely corroded away. Internal diam. 17mm, external diam. 22mm. SF5697, Machining layer 11000, Area A, unstratified

Polygonal rings (Guiraud type 9)

In France, the type occurs from the mid-2nd century to the 4th century, with the peak of deposition around AD 250. Neither of the Elms Farm examples is from a dated context (Figure 447).

466. Silver. An octagonal band with a transverse line either side of each angle. External diam. 20mm, internal diam. 16mm. SF9510, Context 3999, Spoilheap

467. Not illustrated. c. 25% of a polygonal finger-ring made from a thin strip of variable width. It was probably octagonal. The condition is fair, although the edges are damaged. Original diam. c. 15mm. SF5848, Cleaning layer 5603, Area I, not dated

Cut-down bracelets

The practice of making finger-rings from cut-down bracelets has been noted at other sites such as Colchester (e.g. Crummy 1983, 49, no. 1774) and Great Holts Farm, Boreham (Major 2003, 75, no. 15). Some of the larger examples were possibly bracelets for children, rather than finger-rings. The practice seems to be largely confined to the later Roman period, as most of the bracelets used are Late Roman types, often decorated strip-bracelets. Both of the fragments below have the decoration on the outside, but often the original decoration of the bracelet is on the inner surface of the ring, and the visibility or nature of the decoration does not seem to be an important consideration on such rings. The use of a stranded cable bracelet rather than a strip-bracelet is unusual, and paralleled at only a few sites such as Shakenoak (Brodribb et al. 1978, 96, no. 235; although this may be an earring) and from a late 3rd-century context at Latimer Villa (Branigan 1971, 156, no. 156). The Elms Farm example could also possibly be an earring, of Allason-Jones' Type 6, but is a complete ring rather than penannular. It could have been suspended by the loop, although none of Allason-Jones illustrated examples appears to have been used in this way (Figure 447).

468. Bracelet fragment, formed into a small finger-ring with overlapped ends. It was cut or broken across the fastening hole in antiquity, and there is a band of transverse lines adjacent to this end. The other end was cut. The ring is now slightly squashed. In fair condition. Diam. 14mm. SF3201, Cleaning layer 5703, Area J, not dated

469. Not illustrated. Finger-ring, made from a cut-down bracelet of constant section, with notches along the edges. One end was cut in antiquity, the other has a fresh break. From the size, it should be a child's ring. Diam. 12mm, section 3x1mm. SF3506, Fill 10296, Ditch 10406, Group 838, Area F, Period 5-6

470. Not illustrated. Large finger-ring (or a baby's bracelet, or an earring), cut down from a larger three-strand cable bracelet. One end has the original loop fastener, with a small sleeve above it, the other end was cut. The ring is not very circular, and the ends are overlapped by 12mm. Condition fair, surface obscured by earth. External. diam. c. 28mm, internal diam. 18-22mm. SF3236, Cleaning layer 5603, Area I, not dated

Coiled bezel

471. Not illustrated. Finger-ring fragment. A wire finger-ring with a circular, coiled bezel, the ends of the wire wrapped round the hoop. The type is most commonly 2nd to 3rd century (Henig 1995a, 1001), but usually with a more elaborate knotted or coiled bezel. Diam. of bezel 8mm, diam. of ring not determinable. SF4000, Fill 10360, Pit 10349, Group 791, Area F, Period 3

Rings with plain or moulded bezels

472. A hoop of variable width, with a slightly convex outer face where the surface survives. It has a plain, solid, oval bezel with no trace of decoration. The shoulders may have had incised lines down each edge. The bezel is inserted, with the joins clearly visible on the inside of the ring. In poor condition, with little surface surviving. This simple form of ring does not seem to be directly paralleled in Guiraud (1989). Diam. 19mm. SF4140, Fill 4705, Pit 4780, Group 4016, Period 4 (Figure 447)

473. Iron. Finger-ring, now distorted by corrosion, and sprung at the join. A hoop of variable width, with a small flat oval bezel, with three or four transverse lines on the hoop either side of the shoulders. The form is paralleled by a white-metal ring from a 4th-century grave in Colchester (Crummy 1983, 50, no. 1790). Diam. 22mm. SF7873, Fill 15668, Oven construction 15984, Group 695, Period 4-5

474. Fragment; a large hoop with a plain, flat, oval bezel, damaged. The hoop is constricted at the junction with the bezel, and has linear decoration. In fairly good condition. SF5340, Machining layer 11000, Area A, unstratified

475. Not illustrated. Fragment, broken in antiquity. Part of a tapering hoop with a plain, flat, incomplete bezel, either circular or oval. There is no sign of anything having been attached to the bezel. Bezel W. 16mm, hoop min. W. 5mm. Possibly post-Roman. SF8152, Context 3999, Spoil-heap

476. Not illustrated. Fragment; a narrow band with transverse moulding either side of a small, flat, oval bezel. It is very similar to a white-metal finger-ring from Colchester (Crummy 1983, 49, no. 1790), from a grave dated c. 320-450. In poor condition. Internal diam. c. 16mm. SF3397, Cleaning layer 6552, Area H, not dated

477. Not illustrated. Fragment, probably an oval bezel from a finger-ring, with part of the narrower hoop. The bezel has relief decoration, probably an animal, but is in such poor condition that it was not possible to clean the object. The bezel is c. 10x8mm. SF4760, Layer 13696, Group 1116, Area J, not dated

478. Silver gilt. Oval finger-ring bezel appliqué depicting Leda and the Swan in relief. The modelling is good, with variation in the height of the relief giving a three-dimensional effect. The back has traces of a white substance, presumably the cement used to fix the bezel to the ring. This is a rare subject for finger-rings in Britain. Henig (1974, 103) notes only one gem from Britain with the device (from Springhead Temple), plus a lead sealing, perhaps dating to the 3rd century. 11x8mm. SF9515, Context 3999, Spoil-heap

Rings with inset stones

479. Finger-ring with a moulded glass stone. The ring has a hexagonal raised bezel, and broad, flat, truncated triangular shoulders with an incomplete sheet hoop springing from the underside. The metal is brittle, with a very hard concretion on the surface, not all of which has been cleaned. The glass stone is probably light blue, and has five irregular depressions, probably a very degenerate figure. It is quite similar to a stone from London of 3rd century date (Henig 1974, pl. XVIII, no. 574), and there is a close parallel for the form of the ring from Colchester (Crummy 1983, 49, no. 1780), with a better moulded glass inset, from a context dated AD c. 250-300. Diam. c. 18mm. SF7486, Fill 20194, Pit 20193, Group 718, Area L, Period 5 (Figure 447)

480. Silver. A plain tapering band, now rather distorted, with slight shoulders and an oval bezel, contents missing. Margaret Brooks notes that the setting has a smooth whitish surface, but that this may only be silver corrosion below a lost glass stone or intaglio. Diam. 21mm. SF4279, Machining layer 11000, Area A, unstratified (Figure 447)

481. Base silver. A small, plain tapering band with an empty oval socket, containing a whitish deposit. Guiraud type 2c. Complete and in good condition. 23x21mm. SF4694, Machining layer 11000, Area A, unstratified (Figure 448)

482. Finger-ring bezel and part of the band, now flattened. It has incised longitudinal lines on the shoulders, with a petal-shaped setting for the stone, which is missing. There is a separate small strip fragment in the corrosion on the back of the object, almost certainly part of the ring, as it also has incised lines. SF5864, Cleaning layer 5601, Area I, not dated

483. Not illustrated. Finger-ring fragment. A circular, cupped bezel with a plano-convex pale blue glass stone, now detached. A small part of the narrow hoop is present. Too little survives to determine the diameter. It is similar to Henig 1995a, fig. 419, no. 199, which is from a 2nd-3rd century context. Stone diam. 6mm, hoop W. 3mm. SF7576, Cleaning layer 21500, Area J, not dated

484. Not illustrated. Finger-ring fragment. A small part of a plain hoop with an oval bezel, stone missing. The type is as Crummy 1983, 49, no. 1786. There were possible traces of glass on the bezel, but Margaret Brooks notes that this may be just a corrosion film formed in wet conditions before the stone dropped out. Max. W. of bezel 9mm. SF5459, Fill 10688, Pit 10763, Group 842, Area F, Period 5

485. Not illustrated. Finger-ring fragment. Oval bezel, with no shoulders to the hoop. There are small fragments of brownish glass along one edge of the bezel, presumably the remains of an inset. Guiraud type 2a, dated to the 1st century in Gaul. Diam. c. 22mm, bezel 12x8mm. SF9517, Context 3999, Spoil-heap

Intaglios by Martin Henig

Cite this as: Henig, M. 2015, Intaglios, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.henig2

486. Translucent gemstone, dull green in colour with some dark inclusions. Probably a chalcedony. The conventional jewellery name of such a stone is plasma or prase. It is a very small, slightly elongated oval, convex on both surfaces. The intaglio device is a parrot in profile to the left (to the right on an impression), pecking at some food. 8x5x1.5mm. Compare the parrot on a citrine from Colchester (Henig 1978, no. 685) and for the diminutive size and the material used, no. 671, a plasma depicting a bird set in a gold ring from the Roman palace at Fishbourne, West Sussex. SF5405, Layer 5709, Group 600, Area I, Period 3B (Figure 448)

487. Intaglio of onyx with a blue upper face on a dark ground (nicolo). The gem has a flat upper face and bevelled sides. The front face shows a satyr walking to the right holding a bunch of grapes in his left hand and a pedum in his right hand (these of course would be reversed in impression). There is a short ground line. The reverse exceptionally bears a graffito, apparently in Greek letters EYTY for Eutyches. 11x9mm. SF5109, Fill 13238, Pit 13239, Group 642, Area I, Period 4

The subject is a relatively common type; compare Henig 1978, nos 161, 162, App. 37, App. 41. The only parallel to the graffito known to me is the reverse of a nicolo found in Colchester and showing on its obverse a seated satyr. Here the name is EYCEBI for Eusebius (Britannia XVII (1986), 442, no. 44, pl. xxxiii = RIB II, fasc. 3, no. 2423.9). Was this a practice in a Colchester jeweller's workshop? Both names are Greek, which might point to freedmen artisans cutting or setting the gems rather than to the owners of the signets. 2nd century.

488. Not illustrated. Moulded glass intaglio, imitative of nicolo (upper surface blue on a black ground). Only a fragment remains but this is sufficient to show that the subject is a lion in profile to the left (to the right on an impression). The animal's back and the base of its neck are clearly visible. Greatest L. of fragment 7mm, W. 4mm. Compare Henig 1978, nos 631, 632. Late 2nd/3rd century. SF6573, Fill 9565, Gully 9564, Group 804, Area D, Period 4

489. Blue glass intaglio in a copper-alloy ring. The lower part of the ring is lost. The ring has angled shoulders and a raised bezel, typical of the 3rd century. Diam. 23mm, W. across shoulder 11mm. The intaglio belongs to a type common in southern Britain which I have designated as 'Romano-British imitations'. The subject is a very debased rendering of a human figure. 8x6mm. Compare Henig 1978, 132-3, fig. 2 type 3, no. 561 from Gestingthorpe, Essex. SF2889, Machining layer 11000, Area A, unstratified

Fragments

490. Not illustrated. Curved strip of variable width, in very poor condition with little surface surviving, and damaged. Probably part of a finger-ring. External diam. 22mm, internal diam. 16mm. SF4803, Cleaning layer 12207, Area R, not dated

491. Not illustrated. Ring fragment, probably a finger-ring broken across the edge of the bezel. A slightly tapering strip with a lenticular section. In fair condition. Max. section 12x4mm, original internal diam. c. 18mm. SF5565, Machining layer 11000, Area A, Unstratified

492. Not illustrated. Ring fragment, probably part of a finger-ring (c. 30%). Plain D-shaped section, tapering. Diam. c. 20mm. In good condition. SF3571, Prepared surface 5440, Group 97, Period 2B

Earrings

Only one definite earring was noted, but they are probably under-represented. Broken earrings are very difficult to identify, particularly the simpler forms. Other objects from the site that could be earrings include the probable finger-ring cut down from a three-strand cable bracelet (SF3236, no. 470 above), and a small copper-alloy rod fragment with very thin silver wire wrapped round it, from an undated cleaning layer in Area I. Any broken rings of suitable size are also candidates for earrings. In addition, fragments of chain may be parts of earrings rather than necklaces, as may pendants such as no. 499, below, or the small clapperless bells included in category 20.

The typology for the earrings is taken from Allason-Jones 1989.

493. Type 1 Earring. Butted terminals, possibly broken. In fairly good condition. Internal diam. 14mm. SF1329, Cleaning layer 5306, Area J, Period 5-6 (Figure 448)

494. Not illustrated. Distorted, curved strip, tapering at either end, with no surviving surface, and with the ends of the strip probably missing. It is probably an earring, either Type 1 (penannular) or Type 3 (hoop with intertwined ends). L. (straight) 57mm, max. W. 4mm. SF4757, Layer 13576, Group 600, Area I, Period 3B

Miscellaneous jewellery

495. Not illustrated. Oval chain link made from thin wire, with the ends overlapped. In good condition. It is very delicate, and probably from jewellery. 5x4mm. SF6575, Fill 4692, Slot 4695, Group 1161, Area K, not phased

496. Not illustrated. Chain fragment, in poor condition. It appears to be made from links interwoven in two directions, to create a rope with a square cross-section, although the details are not completely clear on the X-rays. This is probably from a necklace, such as that found on the floor of one of the temples at Springhead, Kent (Penn 1960, pl. IIIB), although it could be part of, for example, a suspension chain. Johns (1982, 71, pl. 54) illustrates a tintinabulum from Pompeii that uses copper-alloy ropes to link the various components. L. 28mm, section 6x6mm. SF2371, Fill 6367, Well 6280, Group 531, Area H, Period 3-4

497. Not illustrated. S-shaped wire, probably a link from a necklace (comparable to, for example, Clarke 1979, fig. 87, no. 405). The surface is mostly flaked, and the ends damaged. There are three beads from the context, two made from glass and one from copper alloy, though none was found in close proximity to this piece. Made from a rectangular-sectioned rod, 2x1mm. L. 17mm, W. 17mm. SF561, Spread 5148, Group 457, Area J, Period 5-6

498. Pendant with central loop. The surface has almost completely disappeared, and there is little detail surviving. The metal is golden, presumably an alloy that that has remained gold coloured through a chemical reaction with the soil of the context. The pendant is curved and asymmetrical, with a shallow V-shaped section. It is uncertain whether either end is complete. The narrower end has part of a transverse rib surviving. The subject is obscure, and this does not seem to be any of the most popular pendant motifs, such as lunulae or phalluses, although the narrower end could possibly be a fist. L. 42mm, max. W. 16mm. SF8430, Fill 16082, Well 6280, Group 531, H, Period 3-4 (Figure 448)

499. Terminal, probably from a pendant, in fairly good condition, with slight damage to the surface. It comprises a broken loop with two moulded reels below; below this are eight small lozenge-shaped insets of blue glass or enamel, set round the circumference. Most are damaged, but one retains its original surface, showing that it stood slightly proud of the metal. Below this are two incised lines. The end has a hole in it, 3mm in diameter, and apparently not very deep. L. 19mm, max. Diam. 8mm. SF8158, Context 3999, Spoil-heap

500. Not illustrated. Perforated denarius, with a silver wash. In poor condition, with almost no detail surviving bar the outline of the head. It appears very worn. This was probably worn as a pendant, and it is possible that it is an Early Saxon artefact. Perforated Roman coins are occasionally found in Early Saxon graves; at the Early Saxon cemetery at Springfield Lyons, Essex (Tyler and Major 2005, 73), two coins were found in association with a glass amulet or spindlewhorl. It is possible that the coins were also considered to have amuletic properties. Diam. 20mm. SF4284, Machining layer 11000, Area A, unstratified

501. Pendant. An arc from a cast ring, both ends broken in antiquity, with a cast-in loop. The ring has an L-shaped section, forming a setting for a separate central part, which is now missing. The flat of the ring has two concentric grooves. This pendant frame may have held a coin. It is similar in size to a pendant from Colchester which featured an as of Hadrian (Crummy 1983, 51, no. 1805. W. 22mm, L. 16mm. SF7527, Fill 20483, Pit 20199, Group 336, Area L, Period 2B

502. Not illustrated. Small copper-alloy rod fragment in very poor condition, with a single strand of very thin silver wire wrapped round it (XRF by Margaret Brooks). A small fragment of loose silver wire was also present. There were also four small fragments of sheet in very poor condition, with no certain original edges. This was probably part of an item of jewellery, perhaps a bracelet. A similar fragment, but with copper alloy rather than silver wire, came from Chignall St. James villa (Major 1998c, 74, no. 27). The largest piece was 32x14mm. Rod L. 14mm, diam. 1.5mm. SF5149, Cleaning layer 5610, Area I, not phased

Footwear

Leather shoes by D. Friendship-Taylor

Cite this as: Friendship-Taylor, D. 2015, Leather shoes, in M. Atkinson and S.J. Preston Heybridge: A Late Iron Age and Roman Settlement, Excavations at Elms Farm 1993-5, Internet Archaeology 40. http://dx.doi.org/10.11141/ia.40.1.shoes

Context

All the leather from Elms Farm was preserved in waterlogged conditions in four wells. Of the eighteen small groups, fourteen came from a large timber-lined well in context 16083 (Area H), the 'market place', interpreted as an area with a domestic function.

Two groups, one the finest shoe in the assemblage, both stylistically and in terms of the state of preservation, came from contexts 9879 and 9895 (Area D), respectively, an area with a road, ditches and a well. Insignificant fragments came respectively from wells in context 18236 (Area J), the temple precinct and context 20034 (Area L), with industrial activity; the well contained a wealth of artefactual material.

Condition

Much of the leather was in a very poor condition, fragile, crumbling, with some delamination, which restricted handling and impaired analysis. There is very little surviving evidence for shoe uppers, the more substantial bottom parts having survived better, as is common in waterlogged assemblages. The leather having lost much of its structure, identification of the leather was extremely difficult and somewhat tentative. Some of the analysis was carried out before full cleaning and conservation were undertaken, due to the possibility of its further deterioration. Shoes with hobnails were X-rayed prior to any analysis or conservation. They did not reveal any additional evidence of construction, but ascertained the degree of iron mineralisation in the nails.

Construction and styles

Where it is possible to determine, the shoes are of thonged and nailed construction. The thonging was used to hold the insole and middle sole together, to which the sole was attached by nailing. The classification of nailing patterns, as used by Quita Mould at Birdoswald (Mould 1997, 335, fig. 243) has been adopted here, as most appropriate for reference for the Elms Farm material.

There is one instance of an 'A' type decorative nailing pattern (SF6630, 518); four of 'B', a lightly nailed variety, all from context 16083; there are no instances of the heavily nailed 'C' type.

In only one instance (SF5479, 503a) is the evidence present for identification of a calceus, a closed shoe, although it is assumed that this style would have predominated. By analogy, it is suggested that SF6630 (518) may be a closed, latchet front-fastening boot. Two sandals are present (SF5661, 509 and SF7026, 515), of late 2nd to early 3rd century types. A child's shoe (SF7023, 513) is the only shoe that may be totally stitched.

Toe shapes have been difficult to determine: two shoes are certainly pointed, a man's or adolescent's (SF5480, 504) and a man's shoe (SF6630, 518). There are two rounded toes: SF5483 (507) and SF7023 (513), the child's shoe.

Heel stiffeners are present in two groups, two in context 16083 and one in context 9895.

Interpretation, further evidence and significance

Only one shoe can certainly be identified as male (SF6630, 518), although SF5480 (504) may be a man's or adolescent's. No women's shoes can be positively identified, although Rhodes (Rhodes 1980, 117) associates the sandals from Billingsgate as being worn by women and children, on the grounds of sizes. The only certain child's shoe is SF7023 (513).

There is no convincing evidence for repairs and the condition of the leather makes it difficult to determine the state of wear.

Evidence of leatherworking comes from five groups, four in context 16083 (Area H) and one in context 9879 (Area D), consisting of cutting waste and an unusable fragment from the edge of a hide.

There are possible dyes present in two pieces: SF5666 (511) exuded a brown colouring, unlike other pieces from that context, or any other. SF7027 (519) produced a madder-like pink.

The feasibility of analysis for tannins, waxes, and fats was considered. There can be difficulties in distinguishing between dyes and tannins colourwise (Jim Spriggs, pers. comm.). No tests were carried out to identify the presence of waxes and fats.

Other classes of artefact from the site may indicate other evidence for leatherworking: three possible iron and two bone awls, eleven iron chisel/punch type tools, three copper-alloy and seven bone knife/tool handles, nineteen iron knives and five iron punches, some of which may have leather associations.

Few iron hobnails were identified from the bulk and small find assemblages, but many were extracted from soil samples. The excavators appreciated the difficulty of distinguishing partially mineralised hobnails from small lumps of iron pan which occurred in many contexts.

The small size of the leather assemblage should not be taken as any indication of how much leather was prepared, made into artefacts, worn/used and discarded during occupation of the site. Only items deposited in suitable waterlogged environments would have stood any chance of survival. Those that have survived are mostly very degraded and have only survived for the period of the late 2nd to the 3rd centuries.

Leather from Well 6280, Fill 16083, Group 531, (Period 4)

503. a. Section of the bottom parts and fragment of upper of a man's thonged and nailed closed shoe, presumably a calceus, comprising an insole and middle sole thonged, and the sole attached by hobnails, securing the three layers. The middle sole and upper are in one piece. From what remains of the upper, it appears to be undecorated. The toe shape is tapered and squared off at the tip. X-rays clearly show the remaining nails in position, in quite a good state of preservation.
b. Section of shoe bottom parts, difficult to match to (a) above and apparently from a broader shoe; the insole of a number of components, thonged together and nailed with the sole and middle sole, which may also have been extended to form the upper. The toe is tapered, but the end has been cut off at an angle (not the original shape as would appear to be the case with (a)).

These shoe parts are too fragmentary to determine the nailing pattern, but both probably fall within the 'B' group of nailing pattern classification, as defined by Rhodes (Rhodes 1980, 105-7) and Mould (Mould 1997, 334-5). Date: probably later part of 2nd, or possibly early 3rd century. SF5479

504. Two of the bottom units of a man's or adolescent's left foot pointed toe shoe, with a pronounced curve at the outside joint, of thonged and nailed construction. A narrow middle sole was thonged to the insole along the medial axis. The sole (absent) was attached to the insole by marginal hobnails (now mostly missing, at an average spacing of 25mm. There are groups of three nails at the toe tip and heel seat respectively and occasional other nails, probably forming a nailing pattern of group 'B' type. Not all the nails penetrated the insole margin. X-rays indicate that the extant hobnails still contain some unmineralised iron. Insole length: 260mm x 85 mm at the tread x 60mm at the waist: approximate modern size 6. A fragment of upper survives at the toe tip. Presumably cattle hide. Date: late 2nd or early third century (on analogy with Vindolanda) (van Driel-Murray 1993, 31-47). SF5480

505. a. Partial bottom units of a thonged and nailed shoe, from the heel seat and waist areas, comprising an insole, with a narrow middle sole, made up with irregularly shaped components, attached by a thong along the medial axis. The sole was attached by large peripheral hobnails (only one survives) and along the medial axis; four nails form a diamond pattern in the heel seat (probably a 'B' type nailing pattern). The sole would have been over 10mm in substance (disregarding the hobnails).
b. Fragments of bottom units from a thonged and nailed shoe, probably part of the forepart. It may belong with (a), as it appears to be of similar construction and substance. Probably cattle hide.
c. Indeterminate fragile fragments, probably belonging with (a) and/or (b).

Date: heel seat nailing pattern and construction similar to 504 (5480) above; therefore, probably late 2nd or 3rd century. SF5481

506. Heel seat of insole and middle sole and heel stiffener, of a shoe of thonged and nailed construction. The insole and narrower middle sole were thonged together along the medial axis. Where discernible, the hobnails just penetrated the upper side of the insole. The peaked heel stiffener margin fitted up to the middle sole margin and nails were placed around the sole margin and in the heel seat. The top edge of the stiffener was not stitched in. The quarters were presumably placed between the stiffener and the sole. Nailing pattern and size of shoe indeterminate. The substance of the leather suggests cattle hide. Date: indeterminate, but consistent with other shoes in this context, i.e. late 2nd to early/mid-3rd century. SF5482

507. a. Possible fragment of shoe upper. Creasing suggests the toe of a rounded toe shoe, but it is thick for an upper fragment . There are stitch or nail holes around the margin. Probably calf leather.
b. Possible waste piece from cutting out, with all cut edges, or, an object cut up for reuse; delaminating; no features.
Other fragments are also possible cutting waste.

Date: indeterminate. SF5483

508. a. Fragments of the sole and middle sole of a nailed shoe. There is a single row of nails (none surviving) around the wide sole lasting margin. There is no evidence of thonging from the surviving parts.
b. Small fragments of the bottom parts of a nailed shoe, possibly belonging to (a). One piece has a group of four small nail holes arranged in a dice '4'. Nails also occur away from the margins. A fragment of insole and sole are thonged together.
c. Probable heel stiffener, with a rounded peak, with vestiges of a damaged lasting margin. The leather is probably cattle hide.

Date: compatible with other material from this context: late 2nd to early/mid-3rd century. SF5659

509. a. Fragment of three bottom unit layers, insole, middle sole and sole of a sandal. Pairs of thonging slits, 6mm apart, run parallel to and 10mm. from the edge. More thonging slits, about 35mm in from the margin, run along the longitudinal axis, not appearing on the upper side of the insole, but tunnel-stitched to it. The fragment is now 8mm in substance. There is no evidence of nailing. More of this sandal occurs as 514 (SF7026). Leather identification is indeterminate. Date: late 2nd to early 3rd century.
b. Delaminated fragment of a sole forepart of an oval or pointed toe shoe, with peripheral nail holes and, apparently, two further longitudinal rows (probable 'B' type nailing pattern). Date: probably late 2nd or 3rd century.
c. Delaminated fragment of the sole of a nailed shoe.
d. Tiny fragment, with three very fine marginal stitch holes, perhaps from a shoe upper or a non-shoe artefact.
e. Fragments of probable leatherworking waste, including a possible unusable piece from the edge of a hide; some may be calf or kid.

Dates for (c), (d) and (e): indeterminate. SF5661

510. Not illustrated. Offcut or trimming from cutting out; possibly cattle. Date: indeterminate. SF5662

511. Fragmented and delaminating piece of leather, with at least two sides cut straight and one cut on the curve: either a waste piece from cutting out, or part of an indeterminate artefact. There are no stitch holes or evidence of seams, or that it has been part of any construction; probably calf leather. A possible brown or red/brown dye exuded from the wet leather before conservation. Date: indeterminate. SF5666

512. Not illustrated. Fragment of ?insole and a little adhering middle sole, thonged together; also fragments of leather with a nail hole and a piece of ?upper. Other fragments may be leatherworking waste. Possibly sheep leather. Date: indeterminate, but compatible with other material from this context. SF6831

513. Bottom units of a child's shoe. The toe is missing, but was probably rounded. It was made more or less as a 'straight'; estimated length: 168mm x 55mm at the tread x 33mm at the waist: modern child's size 8. The shoe is apparently stitched, but some intermittent irregularly placed small holes may be nail holes but this is difficult to determine in the fragile leather. Associated is a probable piece of lasting margin at the ear of the heel seat and a fragment of upper. Possibly made of calf leather. Date: compatible with other material from this context. SF7023

514. Waist portion and part of forepart and heel seat of an adult shoe, the insole and sole thonged together. There is one extant nail. Length and toe shape are indeterminate; approximate measurement at the waist: 43mm; substance: 12mm. Possibly goatskin leather. Date: compatible with other material from this context. SF7024

515. Delaminated fragments of sandal sole, including some from sandal 508(a). The largest piece has pairs of thonging holes, 5-7mm wide and 5mm apart, set 7-11mm. from the margin and comes from a different sandal. Some calf leather possibly present. Date: late 2nd to early 3rd century. SF7026

516. Sixteen fragments of leather, representing part of a nailed shoe. Part of the sole forepart survives, with five extant hobnails, X-rays indicating partial mineralisation. Peripheral nails occur at 10-15mm spacing and other nails may form two further rows, more widely spaced, with more of a concentration near the waist, probably a 'B' type of nailing. The main piece and no. 509b (SF5661) may be delaminates of the same forepart. Probably cattle hide. Date: late 2nd to early/middle 3rd century. SF7029

Leather from other contexts

517. Not illustrated. Delaminated offcut, of indeterminate leather. Date: indeterminate. SF6616, fill 9879, Well 9421, Group 772, Area D, Period 3B

518. Bottom units: sole, middle sole and insole, heel stiffener and fragments of upper of a large man's left foot pointed toe nailed shoe, a calceus. Length: 290mmx 95mm across the tread x 64mm at the waist (insole measurements): modern size: adult 9-10 (Figure 451).

The insole and middle sole were thonged together by a single thong along the medial axis, which then forks into two across the widest part of the forepart. The toe has quite a pronounced point. The insole shows wear creases and there is also wear at the inside toe and outside heel seat, on both sole and insole. The middle sole is fragmentary and there is a wide lasting margin.

The nailing pattern is very distinctive, of A1 type, as illustrated in Mould 1997, 335. There is a single row of peripheral nails, at approximately 8mm spacing. The forepart decorative element is a leaf-tendril pattern, with the curving stem finishing at the waist. A small grouping below the waist leads to a subcircular arrangement of nails around a diamond pattern of four nails in the heel seat. Eight nails, or parts of, survive, of an estimated seventy-two around the margin and a total of 137. X-rays of the shoe indicate that the few remaining nails are partially mineralised.

There are a number of parallels for this A1 nailing design. The closest parallel comes from Castle Street, Carlisle (Padley 1991, fig. 210, no. 887), dated to AD 165-200, on pottery evidence. At least one very similar example comes from Vindolanda, period 6, AD 160-c. 180 (Metcalfe and Longmore 1973, 38, fig. 1, and van Driel-Murray 1993, 36), when nailed soles tended to be natural to pointed, some very pointed, often with the leaf-tendril decoration. They became more common in the 3rd/4th century inner ditch fill, from AD 213. Seven examples occurred at Birdoswald, e.g. 2021, with an oval pointed toe (Mould 1997, 328, fig. 238, no. 1), given a 3rd to 4th century date. At St Magnus Quay, London (MacConnoran 1986, 218), the leaf-tendril design occurs only in adults' shoes, dated early/mid-3rd century. Other instances are New Fresh Wharf (MacConnoran 1986, 218), and Queen Street, London (MacConnoran 1982, fig. 37, 116), where there is an instance of an inverted leaf-tendril at the tread, with a slightly tapered rounded toe, attributed to the '4th century or possibly earlier'; Brayford Wharf East, Lincoln (A2); Old Penrith, Cumbria (Thornton 1991, fig. 114, no. 923); The Saalburg (Busch 1965, Taf. 15, nos 223 and 224); Zugmantel (Busch 1965, Taf. 34, nos 752 and 754). At Birdoswald, the leaf-tendril pattern was found associated with a distinctive latchet fastening boot (e.g. Mould 1997, 2185, fig. 23a, no. 12, fig. 244:2.5), of a type dating to the 3rd century. Upper remains at Queen Street (MacConnoran 1982, no. 116) suggest a closed type of ankle boot, laced up the front. This is also noted at Vindolanda (van Driel-Murray 2001, 188, no. 23, fig. 1 and pp.190-1), where the type begins c. AD 180 and peaks from c. AD 230.

Continental assemblages, too, have produced this association (van Driel-Murray 1987, 38), although latchet boots do occur with other patterns of nailing. The universality of styles underlines the fluidity of people and ideas in the Roman Empire.

The shoe is made of cattle hide. Date: late 2nd-early/mid-3rd century. SF6630, Fill 9895, Well 9421, Group 772, Area D, Period 3B

519. Not illustrated. Small fragment of leather, with no definite edges, possibly delaminated. The conservator noted madder-like pink in the storage bag, prior to conservation. It is a possibility that this is due to metal salts in the waterlogged environment. There was also lead-casting activity associated with the well, which may be relevant. The leather may be from cattle or ?sheep. Date: indeterminate. SF7027, Fill 18236, Pit 13883, Group 595, Area J, Period 3B

520. Not illustrated. Indeterminate fragment of featureless and delaminating leather, all edges torn. Date: indeterminate. SF8063, Fill 20034, Well 14984, Group 710, Area L, Period 5

Iron hobnails

In total, 788 iron hobnails were recovered (excluding those in situ in shoes), comprising 13% of the total nail assemblage. Retrieval of hobnails on site appears to have been patchy, as 62% of those found came from soil samples, which only produced 14% of the total number of nails of all types.

It is possible that the types of context (e.g. rubbish pits and waterlogged contexts) selected for bulk sampling may be more likely to contain hobnails, but it is also likely that, in general, there was poor recovery of hobnails by standard digging techniques. This is no doubt due in part to their similarity to the small lumps of iron pan that occurred in many contexts, and which were themselves often mistakenly collected as iron objects.

Three types of hobnail were noted. The majority, where the head shape was discernible, had rounded heads, but some had conical heads, and some almost flat heads. Types could be mixed within a single context; for example, those from context 10847 included all types.

The assemblage included several groups which must represent discarded footwear rather than individual losses. In a few cases, some details of the shoe could be discerned in the corrosion products. The larger groups of hobnails are detailed below; none is illustrated.

521. A group of hobnails in good condition, including three groups of two corroded together. A few have traces of (presumably) mineralised leather and it seems likely that this is the remains of a shoe sole. Some have almost hemispherical heads, others are much flatter. There are twelve with hemispherical heads, L. 15mm, Diam. 9mm, and ten flatter ones (L. 12-15mm, Diam. 8-11mm), plus four of indeterminate shape and eleven shafts, i.e. a minimum of twenty-six hobnails. A second group of twenty-seven hobnails with domed heads from the same context (SF5762) may be part of a second shoe. Context 10847, Kiln construction 10906, Group 672, Area N, Period 4-5

522. Two groups of hobnails in situ in mineralised leather, probably from the same shoe. The hobnails are close-set, with seventeen on one piece and five on the other. Fill 10009, Pit 10010, Group 3064, Area E, Period 3-6

523. 62 hobnails were retrieved from a soil sample. Many have mineralised leather on them, so this is presumably the remains of a shoe or shoes. Fill 10891, Pit 10910, Group 676, Area N, Period 5

524. Fragment from the heel of a shoe sole, with hobnail impressions in mineralised leather. There is a close-set row of four along the edge, and one inside this. Head diams c. 9mm. SF7860, Cleaning layer 6001, Area H, not phased

525. Two groups of hobnails, in situ in soles. One group consists of eighteen hobnails, the other of about twenty. The latter group may have the edge of the sole present, but not enough survives to be certain of the nailing pattern. Head diameters are c. 10mm. SF3509, Prepared surface 13045, Group 381, Area H, Period 3

526. A group of nine hobnails, closely spaced, and corroded in situ in mineralised leather. The nails are heavily mineralised, some surviving as impressions only. The edge of the shoe appears to be present, with a small part of the seam. Head diams. c. 10mm. SF6652, 15515, Pit 15514, Group 696, Area M, Period 4

Iron Cleats

527. Cleat; an oval plate with one short clenched tang, the other tang broken. Although this is rather on the large side for a boot cleat, the short clenched tang suggests that it was attached to leather, so it is probably from a boot. 56x15mm, tang L. c. 14mm. SF8215, Fill 5214, Pit 5209, Group 442, Period 6 (Figure 452)

528. Not illustrated. Large cleat, probably from a boot. Sub-rectangular plate with a tang at each end, one broken. Possibly Roman (cf. Manning 1985a, R54-64). 46x30mm, tang L. 20mm. SF2843, Machining layer 4000, Area A, unstratified

529. Not illustrated. Probable cleat fragment, one arm missing. L. 25mm, W. c. 8mm, arm L. 17mm. From bulk sample. Fill 5376, Ditch 5359, Group 443, Area J, Period 6

Other items

See Figure 452

530. Not illustrated. Iron. Two sheet strip fragments corroded together, both ends broken, possibly a broken sheet buckle plate. There is a gap between the sheets of c. 1.5mm, consistent with attachment to leather, but no obvious perforations. 35x16mm. SF8057, Context 11306, Pit 11316, Group 227, Period 2A

531. Probable strap end. Two heart-shaped lugs with double perforations, and a tapering square-sectioned shank with two mouldings, and a terminal ring. In good condition, partly obscured by concretion. L. 26mm, 5mm. gap between the lugs, diam. of ring 10mm. SF9522, Context 3999, Spoil-heap

532. Sheet buckle plate, with one rivet, edges damaged, and incomplete at both ends. It is decorated with an incised line along each edge, flanked by two rows of very worn punched dots, a transverse line and row of dots adjacent to the rivet, and two longitudinal lines with dots. This could conceivably be medieval rather than Roman, but the decoration suggests a Roman date; the decoration on medieval buckle plates is usually linear, or walked scorper, and very rarely punched dots. L. 28mm, W. 17mm. SF5679, Machining layer 11000, unstratified

 PREVIOUS   NEXT   CONTENTS   ISSUE   HOME