THE EMERGENCE OF GLASS BANGLES IN BRITAIN: THE TRADITIONAL INTERPRETATION

Jennifer Price first recognised that type 2 glass bangles were the earliest bangles produced in Britain (fig. 3).Footnote ⁷ She argued that these bangles were first developed in southern Britain in the late Claudio-Neronian period (a.d. 43–65/70) and that in subsequent periods the craft of bangle-making quickly spread further north with the advance of the Roman army.Footnote ⁸ In northern England, new types were introduced; type 3 became popular in north-eastern England and south-eastern Scotland, and type 1 across most of Scotland, never crossing the line of Hadrian's Wall.Footnote ⁹ Since this article focuses on the emergence of the craft, attention will now be limited to the type 2 glass bangles found in southern England. While two other types seem to be descendants of type 2, they follow a different trajectory in terms of their development and will be treated separately in subsequent publications by the author.Footnote ¹⁰ The further development of type 2 bangles in northern England is also a subject worthy of a separate study.Footnote ¹¹

FIG. 3.

An example of a complete type 2Ai glass bangle from Vindolanda, SF5930; image not to scale. (© Vindolanda Trust)

At present, 35 fragments of type 2 glass bangles from 23 sites in southern England are known,Footnote ¹² with two having dubious provenance and imprecise records (appendix 1; fig. 4).Footnote ¹³ A common feature of the bangles is that they are made from translucent blue-green, so-called ‘natural’, and ice-blue glass (22 specimens), or cobalt-blue glass (four specimens).Footnote ¹⁴ Most belong either to type 2Ai or to 2Aii, though three represent types 2Bii, 2Ci and 2Ciii. Another noticeable feature is the large internal diameters (i.e. 70–90 mm) of 17 of the 35 bangles.Footnote ¹⁵

FIG. 4.

Distribution of type 2 glass bangles in southern Britain; dots = confirmed find-spots; crosses = unconfirmed find-spots.

The suggestion that type 2 bangles are the earliest rests upon evidence of bangles on pre-Flavian military sites in southern England. At Fishbourne, two fragments found in period 1 contexts of Barry Cunliffe's excavations in the late 1960s are roughly dated to before a.d. 75 (fig. 5).Footnote ¹⁶ One fragment found outside Britain, at the site of a Roman auxiliary fort in Valkenburg (the Netherlands), was located in a context dated to a.d. 47–69.Footnote ¹⁷ Other fragments have been found on sites with known or suspected Claudio-Neronian or early Flavian phases of occupation, including legionary fortresses at Usk, Kingsholm and Leaholme.Footnote ¹⁸ Fragments are also known from Carmarthen, Baldock and Caersws, with find contexts dated to the early Flavian period.

FIG. 5.

Two fragments of type 2Ai and 2Aii glass bangles from Fishbourne. (Courtesy of the Sussex Archaeological Society)

Because it was thought that glass bangles were mostly confined to military sites in the south, their appearance has usually been viewed in connection with the Roman army, which introduced a large quantity of glass immediately after a.d. 43. The idea was put forward that local artisans began using the new material to create products that would suit the tastes of the newcomers. These new products had, however, a specific local design: the artisans used pre-Roman techniques to decorate the bangles; they made cords from rods of coloured glass and twisted them clockwise.Footnote ¹⁹ Scholars have suggested that such cord decoration is indigenously British.Footnote ²⁰ Twisted, bi-chrome cords in various colours appear on beads of Guido type 9, usually found south of a line from Chester to Lincolnshire. There is a marked concentration in Somerset and the Bristol Channel area, where glass bangles with similar cords appeared in the pre- and early Flavian period (fig. 6).Footnote ²¹ Accordingly, one hypothesis is that the production of type 2 glass bangles is closely related to – or may well have been developed from – the production of Guido type 9 beads; in other words, it was ‘a Roman-inspired development of an existing late Iron Age skill’.Footnote ²²

FIG. 6.

Correlation of find-spots of beads and bangles with cable decoration in southern Britain (beads data based on Guido Reference Guido1978; Foulds Reference Foulds2017).

The traditional interpretation thus views glass bangles in Britain as an isolated local development. Starting near the Bristol Channel area and Gloucestershire, the craft later spread to other southern regions, including Sussex and East Anglia, where, however, only a handful of bangles have been found. Such a haphazard distribution in southern and south-eastern Britain may reflect the rapid movements of the Roman army during the decades after the Claudian invasion.

PROBLEMS AND QUESTIONS

The discovery of two glass bangles in period 1 contexts at Fishbourne has provided scholars with an apparent terminus post quem for the emergence of glass bangles in Britain: a.d. 43. Such a clear watershed suggests not only the absence of type 2 bangles prior to that year but also that their arrival should be equated with the Roman invasion and, in turn, with the Roman army. However, problems with this chronology have arisen as a result of reinterpretations of earlier phases of Fishbourne based on various finds and excavations in the 1990s by the Sussex Archaeological Society.Footnote ²³ For one, various unambiguous features and artefacts found at locations in and around the site provide cumulative evidence of the occupation of the Fishbourne landscape a generation earlier than a.d. 43.Footnote ²⁴ Although the unified phasing of Fishbourne remains incomplete, the site's use during the LPRIA complicates the presumed initial date of bangles and raises the question of whether phases originally dated to a.d. 50–60 to 60–75/80 in Cunliffe's excavation should be moved back on the timeline.

Glass vessels found at Fishbourne present another dilemma. While the site's assemblage of pottery suggests occupation during the LPRIA,Footnote ²⁵ glass has never been mentioned in the contexts of such discussions, though it should have been. The site has mosaic wares dated to the first half of the first century a.d. and gold-band glass vessels in circulation on the continent in the first half of the first century b.c.Footnote ²⁶ The report on the glass finds was written when Fishbourne was still considered to have a post-a.d. 43 foundation; hence the suggestion that early glass pieces ‘were probably brought over as treasured possessions and not as new purchases on the market’.Footnote ²⁷ However, if Fishbourne's earliest phases were indeed LPRIA, then its pre-invasion inhabitants could have acquired mosaic wares and gold-band glass at the height of their fashion, not as cast-offs decades later.

Another potential problem is the cord decoration on glass beads and bangles. Margaret Guido dates beads with cable-wave decorations to a period covering the first century b.c. to the end of the first century a.d.Footnote ²⁸ Combining Guido's data with new additions to these classes of beads, however, the date now skews towards the Roman period.Footnote ²⁹ Even though the concentration around Somerset, the Bristol Channel area and Gloucestershire still holds (fig. 6), one may ask which came first: a cord on a bangle or a cord on a bead?

Yet another major complication lies in conceiving cord decoration as a genuinely British invention. Many continental beads, as well as some bangles dated to the late first century b.c., have similar corded decorations.Footnote ³⁰ Twisting the glass rods of multiple colours together is, moreover, a familiar method used in the production of glass stirring rods commonly thought to have been used to stir or mix ointments or cosmetics. For the former, a peculiar technique used in the decoration of beads known on the continent, and called Überfangtechnik, can be considered. It involved an opaque, porous, faience-like glass core being fully covered with glass of another colour, onto which thin rods of glass of a third colour were applied.Footnote ³¹ The process has often been dated to the late first century b.c., or La Tène D1, based on examples from Manching and Stradonice.Footnote ³² Manching and Bibracte have rather interesting beads, the faience-like glass cores of which are covered with cobalt-blue glass, onto which thin, opaque white rods are placed horizontally and topped by light-blue rods laid diagonally.Footnote ³³ Visually, the result is nearly indistinguishable from that of British cord decoration; in fact, one bead from Bibracte uncannily resembles a bead of Guido type 9 from Britain.Footnote ³⁴ A fragment of a glass bangle from the same site, decorated by following the same technique, was at first glance mistaken by the author for a piece of a British bangle.Footnote ³⁵ This complicated continental technique may have been the first step in developing cord decoration; thus cords on British bangles and beads could have been a later, simplified version, whereby one rod of cobalt blue was twisted with two rods of opaque white.Footnote ³⁶

The idea that twisted cords originate in Britain is also weakened by the glass stirring rods produced by twisting glass rods of multiple colours together, which were in use from the second century b.c. to the fourth century a.d. These stirring rods were usually manufactured by winding a thin glass trail around a thick glass rod – i.e., opaque white around a cobalt-blue thick rod;Footnote ³⁷ this is a different method from the manufacture of cords on glass bangles discussed above. They are also much thicker than bangle cords, with measurements of c. 0.6 cm versus 0.1 cm for the cords. However, a very brief survey conducted by the author has revealed some continental stirring rods that were produced by a technique similar to that used for the bangle cords.Footnote ³⁸ One may therefore hypothesise that these objects could have served as an inspiration for the development of bangle cords, although an in-depth study, which includes spatio-chronological analysis of glass stirring rods found in Britain, is required in order to resolve this question.

Such evidence prompts a final problem as well as a question: how did continental knowledge of making seamless bangles and decorative elements for them reach Britain, especially when seamless glass bangles, or bead Überfangtechnik for that matter, were not of Roman origin? Seamless glass bangles are strictly phenomena of LIA continental Europe, ‘a cultural and temporal landmark of Celtic civilization’.Footnote ³⁹ Approximately 11,000 fragments of glass bangles have been found on the continent, dating to between the fifth and the first centuries b.c. (fig. 7).Footnote ⁴⁰ This stands in stark contrast to the number of continental bangles found in Britain: the current count is 20 (appendix 2; fig. 7).Footnote ⁴¹ The utter lack of interest in continental glass bangles in LIA Britain and their ‘apparent’ rather abrupt re-emergence in new designs from a.d. 43 onwards in southern Britain require more profound analysis, especially if the craft was neither indigenous to Britain nor Roman inspired.

FIG. 7.

Distribution of glass bangles of continental origin across the European mainland and Britain, fifth to first century b.c. (continental data from Rolland Reference Rolland2019, 37; British data: appendix 2 this paper). (Image courtesy of and © Joëlle Rolland)

The suggested association of glass bangles with the Roman army is similarly problematic. The traditional view sees a marked association between the distribution of glass bangles and the military in south-western England, because some of the bangles were found on military sites and legionary vici. However, that association now looks rather ambiguous, as only 13 of the 35 known examples come from military sites or legionary vici (appendix 1).Footnote ⁴²

Given all these conflicts, the traditional view is certainly open for revision, especially when viewing British bangles in the light of their continental counterparts and contextualising them within the political, social and cultural processes of southern Britain at the time. To quote John Manley and David Rudkin's thinking about Fishbourne, ‘If we hold our nerve and stay this side of AD 43, what else can we see?’Footnote ⁴³

TECHNOLOGICAL AND STYLISTIC TRANSFER

The question of how glass bangles in the LIA and Roman period were made to be seamless has consistently puzzled scholars, albeit those on the continent more than those in Britain.Footnote ⁴⁴ Because no written sources are available, apart from a medieval treatise (On Divers Arts) describing a method of making glass rings without visible soldered joints,Footnote ⁴⁵ continental scholars have turned their attention to alternative sources. Ethnographic evidence available from other countries where seamless glass bangles are known to have been produced has provided clues about the technology,Footnote ⁴⁶ as have analyses of how complex beads were made and detailed visual as well as microscopic analyses of bead and bangle fragments.Footnote ⁴⁷ Piecing various strands of evidence together and replicating some of the techniques in experimental work enables support of some hypotheses and dismissal of others.Footnote ⁴⁸

The technologies in question are the winding and stretching of glass with two wrought-iron rods, or ferrets, or the stretching of it with a cone-shaped mandrel.Footnote ⁴⁹ La Tène glass bangles in continental Europe were produced in various types, and their outer surfaces were modified while the glass was hot by adding peculiar shapes, such as knots, horizontal and diagonal lines, squares and waves, which were incised with a simple knife or a small blade. To achieve the desired shape, craftspeople plausibly followed different operational sequences in each case and used different tools to support the glass while applying the decoration.Footnote ⁵⁰ However, not all glass bangles from the LIA have incised decorations. In the later periods of La Tène (i.e. LT D1), simple monochrome glass bangles, either triangular or D-shaped in cross section and in a variety of colours, but mostly in purple and blue, were produced,Footnote ⁵¹ and some were additionally decorated with zigzag patterns, mostly in opaque yellow.Footnote ⁵² Known as series 36–9 in Rupert Gebhard's typology, these objects were created by following the simple technique of winding and stretching the glass with two wrought-iron rods.

The first stage in the production of this type of monochrome glass bangle is to pick up a gather of glass from its crucible and wind it around an iron rod (fig. 8).Footnote ⁵³ At this stage, the bangle resembles a rounded bead. Gravity and the constant spinning of the rod create a small opening in the middle of the ‘bead’, into which the second rod, or ferret, is inserted. By carefully pulling the glass in different directions with two rods, the opening is enlarged and widened, and the bead is transformed into an annular seamless object: a bangle.Footnote ⁵⁴ Despite the difficulty of producing a perfectly round bangle, experiments have shown that, with enough skill and excellent hand-eye coordination, symmetrical bangles can be created. After the basic bangle is created, decoration is applied onto its outer surface as a final step.Footnote ⁵⁵

FIG. 8.

Artistic impression of the winding and stretching technique used in the production of glass bangles (based on Rolland and Clesse Reference Rolland and Clesse2014, fig. 1). (Drawing by Vicki Herring)

A visual comparison of approximately 500 Romano-British fragments with a few La Tène D1 continental bangles and four present-day Nigerian glass bangles produced by the said method suggests that this technique was also used to manufacture the type 2 glass bangles found in southern and northern England, as well as the type 3 glass bangles that are widespread across northern England and southern Scotland.Footnote ⁵⁶ Most Romano-British glass bangles of these types exhibit the following four features associated with the winding and stretching method (fig. 9).Footnote ⁵⁷

FIG. 9.

Features applied to evaluate the technique used to produce British bangles (images not to scale): (a) elongated gas bubble as seen on a type 2Ai specimen from Seamills (Bristol Museum and Art Gallery, register no. F945); (b) matted inner surface with greyish residue as seen on a fragment of a type 2Ai bangle from Usk (National Roman Legion Museum, Caerleon, acc. no. 82.10H); (c) internal folds as seen on a type 3G specimen from Corbridge (English Heritage Trust, acc. no. CO11357); (d) cross-sectional asymmetry as seen on a type 3A fragment from Corbridge (English Heritage Trust, acc. no. CO32989). (Image (b) courtesy of Amgueddfa Cymru/National Museum Wales)

(1) Elongated gas bubbles. Gas bubbles appear in the glass body matrices of some bangles due to low temperatures being used in their production.Footnote ⁵⁸ Such bubbles are circular before the glass is spun, but stretch because of the constant spinning motion.Footnote ⁵⁹ This feature is visible with the naked eye in bangles made from translucent blue-green and cobalt-blue glasses (fig. 9a). Oblong channels, each once a gas bubble, can be observed in the broken cross sections of opaque white and opaque yellow bangles.

(2) Matted inner surfaces, sometimes with embedded iron scales. The inner surfaces of many type 2 and type 3 Romano-British bangles are matted with tiny greyish or brownish dots, and show horizontal striations, sometimes appearing as wavy lines (fig. 9b). These features do not occur on the outer surfaces of bangles. Similarly matted inner surfaces are observable on the continentalFootnote ⁶⁰ and today's Nigerian bangles (fig. 10).Footnote ⁶¹ The Nigerian craftsmen cover the tips of their wrought-iron rods with clay to prevent objects from adhering to the rods and to allow their easy release.Footnote ⁶² This procedure is akin to modern bead-making, in which steel rods are covered with a material known as bead release, roughly half of which is clay. Many modern beads thus have the same matted inner surfaces as modern Nigerian bangles due to their contact with rod tips covered in bead release. Although many scholars have assumed the dots on Romano-British glass bangles to be dirt accumulated over centuries as pieces have lain in the soil, it is much more likely that this effect is a residue from the parting layer used to cover the tips of the rods to prevent bangles sticking to them.Footnote ⁶³ Another feature of the inner surfaces of some bangles are flakes of iron fused onto the surfaces that now appear corroded. These are scales that form on the surface of hot iron tools, which fuse with the softened surfaces of hot glass masses during the stretching process.Footnote ⁶⁴ As a brown Nigerian glass bangle reveals, scales also form when too little clay is applied to the rod (fig. 10).

FIG. 10.

Inner matted surface with some iron scales on two Nigerian glass bangles.

(3) Internal folds. Folds are clearly observable in the cross sections of bangles made of translucent glass (fig. 9c). Appearing when individually gathered chunks of glass are wound around an iron rod during the first stage of bangle production, the folds represent badly fused glass worked without constant heat.Footnote ⁶⁵

(4) Visible asymmetry in cross sections. A final feature deriving from the manipulation and constant rotation of glass around a rod is the asymmetrical height of the sides, as clearly observable in the cross sections of broken bangles (fig. 9d). On some fragments, one edge is slightly flatter and thicker than the other, which is conversely slightly elongated and thinner. Such asymmetry may derive from rotating and turning the hot, liquefied glass on a rod; the glass shifts from one side to another, but cools and stiffens too quickly to allow the surface to regularise itself into a semicircular shape. The asymmetry could also stem from annealing, as one side of a bangle is liable to slump slightly under the weight of the glass when placed in an annealing oven.Footnote ⁶⁶

These four features are also noticeable in the continental glass bangles from the LIA, as well as in today's Nigerian glass bangles and glass bangles produced during experiments undertaken by a French team supervised by Joëlle Rolland.Footnote ⁶⁷ Many of these features are the result of the low temperature in the small, wood-fired furnaces in which the Nigerian and experimental bangles were produced. During their experiments, Rolland and her colleagues noticed that objects made in modern gas furnaces, where temperatures easily exceed 1,000 degrees celsius, lack these features.Footnote ⁶⁸

Without a doubt, the winding and stretching method was used to produce both continental LIA and Romano-British type 2 and type 3 glass bangles. Closer inspection suggests that they also share elements of design. Many scholars who have addressed the appearance of, and inspiration for, Romano-British bangles have emphasised the differences in their decoration and quality from those found on the continent. Indeed, continental bangles are more spectacular in their design, given their incised knots, their multiple horizontal and vertical lines that form a chequered pattern and their wavy lines that sometimes resemble running dogs. However, scholars have generally compared Romano-British bangles to continental ones from a different period, since the more spectacular bangles are all from a much earlier era: La Tène C2 (170–110 b.c.), or what is sometimes called the ‘baroque period’ in the development of glass bangles.Footnote ⁶⁹ The continental bangles that are geographically, chronologically and stylistically most similar to the British ones are Gebhard's series 36–9, which are dated to La Tène D1 (c. 110–50 b.c.; fig. 11). The production of series 36 (i.e. undecorated, D-shaped, purple bangles), series 38 (i.e. undecorated, D-shaped, blue bangles), series 38b (i.e. blue bangles with opaque yellow zigzag decoration) and Haevernick type 3b (i.e. purple bangles with opaque yellow zigzag decoration) is strongly suspected to have been undertaken in the Netherlands and the southern German Rhineland.Footnote ⁷⁰ These simple, monochrome, D-shaped bangles have a pronounced regional distribution: roughly 2,600 fragments have been found in the Netherlands, accounting for nearly 60% of all glass bangles recorded there.Footnote ⁷¹

FIG. 11.

Schematic comparison of continental and British bangles: (a) Haevernick 3b: purple with opaque yellow (PAN-00040944, Maasdriel, the Netherlands) and type 2Ai from Usk (National Roman Legion Museum, Caerleon, acc. no. 82.10H); (b) Gebhard 38b: blue with opaque yellow (PAN-00024950, Geldermalsen, the Netherlands) and type 2Ai from Usk (National Roman Legion Museum, Caerleon, acc. no. 82.10H); (c) Gebhard series 39 (PAN-00022824, Zaltbommel, the Netherlands) and type 2Ai from Whittington Court (The Museum in the Park, Stroud, acc. no. CM.3560). (Images (a) bottom and (b) bottom courtesy of Amgueddfa Cymru/National Museum Wales)

Although cross-Channel connectivity is implied by the shared technology of the production and style of continental and British bangles, a chronological conundrum persists. If the supposedly Romano-British glass bangles indeed descended from continental ones, then a gap of two or three generations occurred between the disappearance of glass bangles from the continent in the Augustan period and their appearance in Britain in the Claudio-Neronian period. This chronological discrepancy and other problems raised can be overcome, as long as the initial date of the glass-bangle revolution in Britain is shifted. The technological and stylistic borrowing proposed in this section brings a LPRIA start date for British bangles within the bounds of possibility. What other arguments might also support this hypothesis?

SOCIOCULTURAL TRANSFER

A chief premise of this article is that the development of type 2 bangles cannot be studied in isolation from processes in southern England before a.d. 43. While a complete review of the changes within the historiographical tradition of interpreting the transition from the LIA to the Roman period is beyond the scope of this article, principles central to the argument will be highlighted in the following subsections in order to showcase strands of evidence supporting an earlier date for the beginning of glass bangles.

REDATING GLASS BANGLES?

Given the discussion above, the appearance of British glass bangles combining a continental production method and visual elements but delineated by their final style cannot be coincidental; it must be the result of convoluted processes and the amalgamation of various cultures during the LPRIA.Footnote ¹¹³ Put differently, the bangles are products of their time, born from the intermingling of various elements of an interconnected southern England and European mainland. Such objects of personal adornment were made by using technical expertise from the continent, and, though their development may have been inspired by contacts with societies from across the English Channel and the North Sea, they were not copies or antecedents of continental glass bangles or beads. Rather their design was likely a reinterpretation of those objects in the light of the interplay of the selective borrowing of visual aspects of continental bangles and beads that were more consistent with the tastes of local consumers. In this sense, British bangles represent a clear case of ‘a British particularisation of a universal category of circulating object’.Footnote ¹¹⁴

Glass and glass artefacts are too often studied in isolation from other contemporary objects of material culture. This approach neglects connections already suggested for objects made in other materials – in this case, the appearance of new dress accessories and changes in visual appearance. Bangles may have been developed in parallel with other objects of personal adornment to signify new lifestyles and new forms of outward presentation. Responding to common stimuli, they emerged as a result of selection and customisation instead of in a singular, typological sequence of development.Footnote ¹¹⁵

However, it is of course not enough simply to state that the emergence of glass bangles in Britain should be redated to the LPRIA, from the period ‘after a.d. 43’. A few issues complicate the hypothesis. The first is the absence of direct chronological evidence. No bangles found in southern England date to the pre-conquest period; however, the dating of the genesis of particular types of material culture in southern England during the hiatus of the mid-first century a.d. to either the pre- or post-conquest period is fraught with major problems.Footnote ¹¹⁶ What is sometimes assumed to be LIA material has been shown later to have been persistently used in the late first century a.d. and vice versa, while a few objects once presumed to be in the Roman style have been recognised later as stemming from the LPRIA.Footnote ¹¹⁷ Many sites in southern England have yielded artefactual assemblages that show clear signs of the continued use of materials from the LPRIA into the Roman period.Footnote ¹¹⁸ There is also of course the problem of residuality, with some LPRIA-to-Roman transition sites exhibiting a pattern of pre-conquest artefacts finding their way into post-conquest contexts.Footnote ¹¹⁹

In this regard, information about the contexts in which bangles have been located warrants attention. Of the 35 specimens, contextual data have been recorded for 22, with a few discovered in rubbish deposits dated to the Neronian and pre-Flavian phases of occupation at sites of Roman forts.Footnote ¹²⁰ For these few, Neronian and pre-Flavian dates mark their deposition or discard, not their manufacture. The durability of a given bangle is unknown; though abrasions and dull surfaces could indicate wear and use over some time.Footnote ¹²¹ In terms of bangles found at forts founded in the a.d. 50s, such as Kingsholm, Leaholme and Usk,Footnote ¹²² with LPRIA activity detected at Kingsholm,Footnote ¹²³ it is theoretically possible (but not proven) that the bangles arrived at those sites earlier than their foundation dates.

Chronological and contextual evidence for bangles found in rural and smaller urban contexts is also patchy, but permits the hypothesis that they originated before the Roman invasion. Most examples come from contexts broadly dating from the early first to early second century a.d. Such bangles are either unstratified or recorded rather broadly as being associated with ‘occupation secondary deposits’, meaning that the original date of their deposition or discard cannot be established at present. Evidence suggests the uninterrupted occupation of these rural sites from the LIA until after the conquest, and a few likely endured into the later Roman period.Footnote ¹²⁴ For example, a bangle from Littlehampton, a rural site with evidence of continuous occupation, was found in a refuse pit along with domestic and construction rubbish dated from the mid-first to late second century a.d.,Footnote ¹²⁵ which may have been part of the levelling deposit of a farmstead from an ‘earlier date’.Footnote ¹²⁶ As another case in point, an unstratified bangle from Claydon Pike came from an area dated from the early first to the early second century a.d., along with many associated finds dated to either side of a.d. 43 and into the Flavian period.Footnote ¹²⁷ A small industrial Romano-British town, Ariconium (now Weston-under-Penyard), has revealed particularly strong evidence of pre-Roman occupation. The area in which an unusual bangle was found had been continuously used for rubbish disposal with ‘occupation debris of predominantly first and second century’ and ‘significant quantities of Late Iron Age/Transitional period material’; this suggests that the area was already actively in use during the pre-Roman period.Footnote ¹²⁸

Neronian and pre-Flavian deposits indicating the date of discard but not manufacture or actual use, coupled with the occurrence of glass bangles in levelling deposits alongside early material from rural sites, cast doubt on any automatic dating of bangles to after a.d. 43. Such evidence also weakens the association of bangles with the Roman military. Given the possibility of the non-military origin of British bangles, it is appropriate to consider the people who may have propelled the emergence of the craft. Throughout this article, ‘southern England’ has been used as shorthand for their location; however, the bangles’ centre of gravity seems to be in the Severn-Cotswolds region, given the cluster of type 2 bangles there (fig. 4). This territory has been assigned generally to the Dobunni, and though it may be tempting to call type 2 bangles ‘Dobunnian artefacts’, such a label is problematic. It is a somewhat artificial creation, derived from a second-century a.d. map compiled by Ptolemy and adapted to the later Iron Age landscape,Footnote ¹²⁹ and may not mirror tribal or social entities living in the region during the LPRIA or early Roman period.Footnote ¹³⁰ Nevertheless, it is clear that, sometime during the first century a.d., some group or groups living in the region began appropriating glass bangles into their social practices. This certainly fits more generally with societal developments of the LPRIA Severn-Cotswolds groups which Tom Moore has characterised as societies in the process of ‘breaking away from existing well-defined relationships and adopting various attributes of different lifestyles’.Footnote ¹³¹ If the origin of British glass bangles is indeed continental, then continental craftspeople might have arrived in this part of Britain in the LPRIA and begun producing bangles in a new style to accommodate the tastes of local customers. However, evidence points more towards the existence of strong regional exchange networks than to long-distance trade with continental societies.Footnote ¹³² In fact, the relative proximity of the Severn-Cotswolds groups to the middle and late Iron Age lake villages of Glastonbury and Meare in Somerset, with evidence of glass-bead manufacture spanning four centuries, could indicate some sort of regional continuity in glass craftsmanship involving local glass-workers open to external influences in the development of new products.Footnote ¹³³

Although the character of the Glastonbury and Meare lake villages, which were in existence until the late first century b.c., is debatable, in terms of their seasonality, permanence and ceremonial use, it is clear that highly specialised crafts were practised there, including glass-bead manufacture.Footnote ¹³⁴ The existence of a glass workshop at Meare is considered by Julian Henderson to be ‘a virtual certainty’, given abundant evidence of manufacturing occurring on site from the fifth century b.c. onwards.Footnote ¹³⁵ Another site in the south that has provided circumstantial evidence of glass-working is Hengistbury Head in Dorset.Footnote ¹³⁶ Lumps of imported raw glass in purple and opaque yellow showing signs of cutting have been found there, and indicate either trade in raw glass on the site or the working of glass into objects – beads, most likely.Footnote ¹³⁷ Although no glass-working debris has ever been found, the possibility of glass-working is strengthened by the occurrence of various high-temperature industries on the site.Footnote ¹³⁸ The assemblages from both Meare and Hengistbury Head have yielded continental La Tène bangles of series 36, 38 and Haevernick type 3b, all of which morphologically resemble the earliest bangles attested in Britain (fig. 11).Footnote ¹³⁹ One can, therefore, imagine that skilled indigenous glass-workers started catering to the new tastes of consumers who required novel products to satisfy their desire to change their dress and outward appearance. As a result, the type 2 glass bangle was born. Nevertheless, this theory remains plagued by another complication: the absence of evidence for experiments and misfires.

A MISSING LINK: STYLISTIC GENEALOGIES

As products of human actions and individual agency, objects do not develop in a linear, progressive sequence; their development depends very much on the growth of knowledge of a craftsperson involved in their production.Footnote ¹⁴⁰ For type 2 bangles, evidence of a linear typological developmental sequence and progression is nearly non-existent, and that of misfires, failures and trial and error is limited. In fact, all bar two bangles display a high level of conformity in their technical proficiency and design. Indeed, the geographical spread of bangles in the south – from Wales to Kent, with very little difference in design between those from Usk and those from Canterbury – suggests either that knowledge of how to manufacture them was widely shared or that production was carried out in a limited number of centres. However, evidence of the production of bangles in southern England is restricted to a mere 16 mm-long piece of premade twisted cord of opaque white and cobalt blue in the assemblage from the Leaholme legionary fortress.Footnote ¹⁴¹

Two unusual examples come from the small towns of Kenchester and Weston-under-Penyard.Footnote ¹⁴² The bangle with a twisted cord from Weston-under-Penyard is unique insofar as its base colour is opaque yellow. Type 3B bangles with opaque yellow as a base colour are known only from northern Britain, but they are all, save one, entirely undecorated.Footnote ¹⁴³ The Kenchester bangle, while having most features of a typical type 2Aiii bangle, is also exceptional (fig. 12). Not only is it additionally decorated with opaque yellow dots on the apex, but, more importantly, it is extremely large and chunky, with an internal diameter of 90 mm and a height of 12 mm. Experiments with wearing bangles of a similar diameter have shown that such a bangle would have been extremely uncomfortable to wear for any length of time.Footnote ¹⁴⁴

FIG. 12.

Fragment of a glass bangle from Kenchester (acc. no. unknown): outer and inner surfaces and side views. (Courtesy of Herefordshire Museum Service)

Both instances of deviation amid an otherwise uniform collection can be interpreted as attempts by local artisans to recreate a popular design due to demand by using the material at hand, their own knowledge and observation. A similar pattern is noticeable on a set of 25 fragments of possibly seven bangles from a so-called ‘native’ stone-walled fort in Bryn y Castell, Ffestiniog, Gwynedd, Wales.Footnote ¹⁴⁵ The fragments found at this iron-working site, occupied between 300 b.c. and the mid-first century a.d., were most likely produced from recycled glass just before the end of the site's occupation.Footnote ¹⁴⁶ The pieces were produced in translucent blue-green, cobalt-blue and dark-green glass, and decorated either with twisted cords or trails and lines applied directly onto the apex and edges, thus broadly resembling type 2 glass bangles.Footnote ¹⁴⁷ In contrast to standardised bangles circulating in southern England, the cords on these examples meander and curtail on the apex and sides. Instead of being purposeful, the distortion of the cords and manner of their application suggest that whoever applied them did not have sufficient manual dexterity to handle both a bangle on a rod and the application of a cord, or else lacked sufficient knowledge or skill to judge how much cord was needed.Footnote ¹⁴⁸ Though the Kenchester bangle differs from examples found at Bryn y Castell – the cords are laid with mathematical precision – its thickness and large diameter imply that its maker may have misjudged how much glass was necessary.

Altogether, while both products might derive from trial and error, they could be interpreted as the products of artisans with different levels of manual dexterity and training, who produced bangles for local needs and consumption. Such craftspeople probably made popular objects based on their observations of templates received from elsewhere, likely from Gloucestershire or Roman period rural settlements in southern Wales. In either case, bangles could have been produced in a wide range of workshops, from highly specialised ones to smaller, more regional ones specialising in other crafts, to even smaller-scale, possibly peripatetic ones, rather than production having been confined to a limited number of centres.Footnote ¹⁴⁹

Other evidence suggesting the plausibility of manufacture by a local artisan or artisans are the finds of fragments of four enigmatic bangles at four locations in south-western England. These pieces have not been considered previously in discussions on the emergence of glass bangles in Britain. Ascribed to type 3G, they are usually assumed to be exports from the north, where 18 examples of bangles of that type have been recorded.Footnote ¹⁵⁰ Typological group 3G is described as being made of a translucent blue-green glass body and decorated with an opaque yellow inlay, usually in a pot-hook pattern (see fig. 2).Footnote ¹⁵¹ The four bangles from the south stand in contrast to the northern specimens; instead of bearing a pot-hook decoration they have lines applied horizontally on the surface and sides. Base glass colours differ as well. The large bangle found at Shepton Mallet, the Roman period settlement at Fosse Lane, is of emerald-green glass with five such lines; a much smaller but very similar bangle, albeit in greenish-yellow, recorded as ‘possibly from the Meare lake villages’ and now in the collection at Wiltshire Museum, has six lines (fig. 13a and b).Footnote ¹⁵² Two other similar bangles are from the Isles of Scilly's Nornour islet and Meare East lake village. Made in translucent nearly colourless glass, they are decorated with opaque yellow lines; the one from Meare East shows a swirling pattern, while the Nornour fragment is too small to establish the nature of its pattern (fig. 13c and d).Footnote ¹⁵³

FIG. 13.

Miscellaneous glass bangle fragments from Somerset/Wiltshire: (a) Shepton Mallet (Somerset Heritage Centre, Taunton, acc. no. TTNCM 99–1996); (b) possibly from the Meare lake villages; (c) Meare East (Somerset Heritage Centre, Taunton, acc. no. 84.AA.13/7); (d) from Nornour (after Dudley Reference Dudley1968, 24, fig. 9, no. 39). (Images (a) and (c) © Somerset County Council and South West Heritage Trust, 2018; (b) courtesy of and © Wiltshire Museum, Devizes)

Instead of attempting to categorise these specimens as type 3G, which is northern in distribution, or creating a new type or subtype when only four examples have been recorded thus far, it seems more reasonable to assign them tentatively to a miscellaneous category. Deriving from contexts with limited dating information, it is uncertain whether the four examples are of a much later date than type 2 or an earlier one. What adds some credence to the former notion is the context, dated a.d. 250–300, of the Shepton Mallet piece; this was found in a pit backfilled with domestic rubbish broadly dated from the late first to the third century a.d.Footnote ¹⁵⁴ And what adds some credence to the latter is the base glass colour. Emerald-green glass was popular in the early to mid-first century a.d., peaking in the decades leading up to the mid-first century, with glass bowls produced in this colour reaching Fishbourne and Colchester.Footnote ¹⁵⁵ However, with the current state of knowledge it is not possible to determine whether the Shepton Mallet glass bangle was produced from the recycled emerald-green glass that found its way to mid-first-century a.d. Somerset. At present, it is simply reasonable to assume that such miscellaneous pieces were created somewhere in the Somerset/Wiltshire area, since no other parallels, neither elsewhere in Britain nor on the continent, have been recorded. We could also consider a range of scenarios: bangles similar to this fragment might have been the predecessors of type 2 bangles – the missing link between continental Augustan and British Claudio-Neronian ones – or an analogous type developed in parallel with type 2 but falling from favour for reasons unknown. Until more examples from stratified contexts are found in southern England, the nature of these enigmatic bangles will remain unclear.

The cases of these four bangles along with the Bryn y Castell pieces emphasise how much remains unknown about the development of the glass-bangle industry in southern England. The existence of possible other types and variability in their production suggest that the development of bangles did not proceed linearly and sequentially, but the current state of the evidence prevents us from drawing a clear picture of this process.Footnote ¹⁵⁶ In short, the stylistic genealogies of the artefacts cannot be established beyond their continental La Tène lineage.Footnote ¹⁵⁷

ISSUES OF AVAILABILITY

There are other pieces of the puzzle missing, and they will be addressed here briefly.

The first issue concerns information about the availability of bangles. Given the paltry number of bangles found and since they are unusual, innovative objects, they may have been restricted to a privileged few and associated with wealth, exclusivity and exoticism. However, there is no strong evidence that these bangles had any special value or ritual significance. Only one fragment has been found in a location associated with ritual activity – debris of a temple at St Albans – but the dating points to the third century a.d., implying a residual context (appendix 1, no. 25). Apart from this specimen, no bangles to date have been reported from conspicuous places within sites associated with religious practice in southern England.Footnote ¹⁵⁸ There is also no indication that bangles had any monetary or similar value – none came from hoards. Furthermore, none have been identified from graves.

Bangles are indeed under-represented in the archaeological record of southern England, especially in the core region of the Severn-Cotswolds. Nonetheless, the quick manufacturing process, with ten minutes being the maximum production time for a single piece,Footnote ¹⁵⁹ and the thin distribution patterns of standardised bangles in southern England suggest that knowledge about the objects spread far and wide. It is thus almost certain that many more existed than are currently known, but it remains unclear whether their low recorded number is a product of a dearth of excavations, a limited number of recorded and published reports on small finds or recycling activity in the past.

Regarding the latter, a high level of recycling has been proposed for continental bangles in Britain during the LIA, and recycling of – and trade in – scrap glass was indeed common in antiquity.Footnote ¹⁶⁰ Even in later periods, when glass-working in Roman Britain was performed on a far larger scale, finds of glass waste or cullet dumps have been rare, as the craftspeople would have recycled everything available to them.Footnote ¹⁶¹ Even broken glass bangles were suitable for recycling. A type 2 glass bangle from Gloucester (appendix 1, no. 17) may have been prepared for recycling of its blue-green glass. The apex of this artefact, where the twisted cord is usually located, exhibits many conchoidal fractures, suggesting preparation of the piece for recycling through deliberate removal of the multicoloured cord.Footnote ¹⁶²

A second factor is the limited availability of glass itself and knowledge of how to handle the material. If, as this article suggests, the origins of glass bangles lie in a pre-a.d. 43 horizon, then the evidence should at least indicate that glass was available, as were craftspeople who were knowledgeable and skilled enough to make bangles. As emphasised previously in this article, a regional connection – from Somerset to Gloucestershire – in the spread of knowledge about producing glass objects is possible, and the same can be said about the availability of glass. The Meare lake villages had access to enough glass to make hundreds of beads, and the possibility of glass-working at Hengistbury Head should not be excluded.Footnote ¹⁶³ Indeed, Strabo mentions that glass vessels were traded with Britain from the continent at the end of the first century b.c.Footnote ¹⁶⁴

Additional evidence of regional continuity, not only in terms of skill and glass availability but also with reference to visual preferences for colours, includes the use of opaque yellow glass. Few glass bangles found in south-western England made use of such glass in their decorative schemes.Footnote ¹⁶⁵ Opaque yellow was a popular, plausibly widely available, colour in south-western Britain during the LIA, as emphasised by the Hengistbury Head lump of imported opaque yellow glass and the opaque yellow decorations on most beads from Meare East.Footnote ¹⁶⁶ However, all the opaque yellows found from this time in Britain differ in their chemical composition, meaning that different opacifiers were used to colour the base glass.Footnote ¹⁶⁷ Since evidence of its production in Britain is non-existent, the colour likely reached the sites at which it has been found from different continental workshops.Footnote ¹⁶⁸ Scientific analysis of the opaque yellow used in British bangles would thus be welcome, in order to determine where the base glass originated and to compare it with other LIA opaque yellows analysed by Henderson.Footnote ¹⁶⁹ A similar conclusion can be reached regarding yet another colour of continental glass bangles from the LIA: cobalt blue.Footnote ¹⁷⁰ This colour appears in four fragments from south-western England, but it is currently not possible to determine from where and when it came.

Continuing on the theme of colour and glass availability, it is worth noting that the most common colour of glass finds from post-a.d. 43 Britain is blue-green. This is the colour of popular tubular, unguent glass bottles used to store cosmetics, oils and ointments.Footnote ¹⁷¹ By way of example, vessels made of bluish-green glass from Colchester form a high proportion of such finds from the first period post-conquest.Footnote ¹⁷² An analysis of glass consumption in Essex during the early Roman period conducted by Angela Wardle reveals that the overall assemblage of glass is dominated by vessels and bottles made of blue-green glass alongside coloured-glass objects.Footnote ¹⁷³ Twenty-two of the type 2 bangles from southern Britain are blue-green. Moreover, Usk and Cirencester both have glass bangles of dark brown, golden brown (amber) and emerald green; these are the typical colours specific to the glass produced in the decades leading to the mid-first and the second half of the first century a.d.Footnote ¹⁷⁴ Yet again, it is not clear when exactly such glass reached Britain. In line with the comment from Strabo, the presence of a very few early Roman continental glass vessels dating from the late first century b.c. to the early first century a.d. at sites dated to the LPRIA in southern England suggests that some glass arrived in Britain prior to the Claudian invasion.Footnote ¹⁷⁵ These vessels tend to be in the polychrome cast tradition, with monochrome blue-green glass vessels beginning to dominate glass assemblages only after the mid-first century a.d.Footnote ¹⁷⁶ Thus, it can be speculated that the opaque yellow and cobalt blue glass used in British bangle-making have their origins in the LIA. In other words, the type 2 glass bangles from the south produced in these colours belong to the pre-a.d. 43 horizon. In contrast, bangles produced in blue-green and strongly coloured glass belong to the Claudio-Neronian period. As speculative as it sounds, this may explain why most early type 2 glass bangles are in these colours: it was a matter of scrap glass availability, not of consumer or symbolic preference.Footnote ¹⁷⁷

SUMMARY

Building upon the work of Price, I concur that type 2 bangles are the earliest glass bangles produced in Britain and that the appearance of the craft of bangle-making should be placed in the context of south-western Britain. The origins of the artefacts should be placed, however, in the LPRIA, rather than after a.d. 43. The viewpoint that bangles were an exclusively Roman period phenomenon confined to Roman military sites can no longer be sustained in the light of new context information as well as greater awareness of the social and cultural developments and transformations in the transitional LPRIA, of which bangles were plausibly an integral part. Current approaches view the artefactual changes in this period in Britain as a result of the construction of new social relations and the reconfiguration of identities of people who were proactive in their choices as consumers.

In his research on various imported vessels found in southern England, Pitts draws an image of integrated societies in which goods and people traversed a globalised network of cross-Channel links.Footnote ¹⁷⁸ Such global networks allowed southern British societies to participate in continental cultural practices and to choose from a range of divergent continental trappings to suit their regional cultural trajectories.Footnote ¹⁷⁹ As a product of these networks, British bangles were not simple copies of continental counterparts but reflect the complex processes of innovation, adaptation, cultural connectivity and glocalisation.Footnote ¹⁸⁰ While simultaneously sharing and maintaining a cultural genealogy from across the Channel, these objects were reinterpreted to correspond better with local cultural practices.Footnote ¹⁸¹ Arguably, British bangles were regional, innovative variations of pan-regional LIA continental glass bangles, on which the British bangles depended technically and stylistically.Footnote ¹⁸²

Who brought the craft of bangle-making to Britain, who first made them and who wore them are all unknown, due to the limited available data and thin distribution patterns that reveal only general tendencies. There are tantalising hints that these bangles developed in response to various stimuli and changing circumstances in Britain in the LPRIA and the early Roman period. In particular, the development of how an individual presented their body may have acted as an impetus in the emergence of this new accessory alongside many other objects associated with personal appearance and grooming, including nail-cleaners, mirrors and brooches.Footnote ¹⁸³

This article advocates for longer chronologies of the development of bangles, suggesting that the idea for them developed in a pre-a.d. 43 horizon. Although some bangles may have been Claudio-Neronian products, the aim of this critical analysis is to show that we should allow for the possibility that other bangles may have appeared earlier than a.d. 43. While the craft may have been continental in origin, the uptake of bangles was strictly regional in south-western England, plausibly due to the history of glass-working in this area, marked by workshops and craftspeople open to the development of new glass products.Footnote ¹⁸⁴ We can speculate that the glass-bangle revolution began in two areas: Wiltshire and Somerset, characterised by failed attempts, and Gloucestershire and Avon, which successfully introduced new designs to the market, probably because artisans here could mobilise faster, cheaper production methods that sparked a rise in demand among consumers. The different regional trajectories in the uptake of bangles might also relate to different cultural and social aspirations within the populations. In his research on Severn-Cotswolds material culture, Moore observes that the adoption of material was variable and locally specific, and this could partly explain why northern Somerset has yielded more beads with cords, whereas the Severn-Cotswolds area has yielded more bangles (fig. 4).Footnote ¹⁸⁵

Thus, the development of glass bangles in Britain had more to it than meets the eye, to borrow Jundi and Hill's phrasing.Footnote ¹⁸⁶ British bangles had become forgotten artefacts, absent from discussions on objects of body adornment from the LPRIA. This article has shown the necessity of adding bangles to analyses of changing practices of personal presentation and the altered identities that were achieved during the period.Footnote ¹⁸⁷ Undoubtedly, wearing a colourful, lustrous, glass bangle made an impact within the changing visual and sensory worlds of Britain's inhabitants during this transitional period.