Roman Glass from the North of Scotland

Birgitta Hoffmann

As part of a continuing attempt to identify and re-evaluate un/under-published Roman material from Scotland north of the Antonine Wall, the writer was able to reanalyse a quantity of glass from the legionary fortress of Inchtuthil and the Gask system forts of Ardoch and Bertha. All three assemblages came from 19th or early 20th century excavations or surveys and had only previously been published in the most cursory form. In view of modern advances in the study of this material and especially in the light of the forthcoming reports on the glass from Newstead and Cardean (Hoffmann forthcoming a & b), it seemed worthwhile making a full re-examination of the finds and comparing them with the corpus from the rest of Roman Britain. The author wishes to express her gratitude to Mr F. Hunter of the National Museum of Scotland for locating the material and making it available for study.

The Glass from Inchtuthil

31 Fragments of glass have been deposited in the museum from the 19th century excavations at Inchtuthil (Abercromby et al, 1902). No fragments of cast or mould-blown glass survive. The assemblage is divided as follows:

  • colourless: 1
  • fragment greenish: 2
  • fragments blue/green: 12
  • fragments Bottle glass: 10
  • fragments objects: 2
  • fragments window glass: 2

There are also two modern fragments in the assemblage.

Only eight fragments warrant closer attention. One is an indented colourless fragment (1), which probably came from an indented beaker or cup, popular in the second half of the first century. Both types are occasionally decorated with wheel-cut or abraded lines, and tend to be of a cylindrical or slightly convex shape over a flat base or pushed-in base ring (Price and Cottam 1998, 85-88 – Cool & Price 1995, 70). These beakers were extremely popular throughout the Roman world and occur in blue/green and greenish glass, as well as the colourless glass used here. Parallels from the North of Britain include similar blue/green cups from Corbridge (Charlesworth 1959, 166 fig. 22, 1-3 – Allen 1988, 293 fig. 132, 41), York Blake Street (Cool et al. 1995, 1571 fig. 740, 6040), Doncaster (Buckland 1986, 17 fig. 11) and Castleford (Cool and Price 1998, 146; 159 fig. 54, 91). A greenish-colourless indented beaker was found at Strageath (Price 1989, 198 Fig. 100,7). As the latter originated from the Antonine occupation of the fort, it is debatable whether it belongs to this group or a later version of the indented beaker. It shares with the other quoted examples the rim treatment (knocked-off and ground smooth) and the wheel-cut lines, both of which are common in Flavian beakers. That indented beakers were available after the Flavian period in the North of Britain is documented by the Housesteads specimen, which is dated to the later second century (Charlesworth 1971, 35-6 fig. 6). At Vindolanda indented vessels were found in unstratified contexts during the excavation of the third century vicus buildings and none are yet known from the Flavian levels. It cannot be ruled out, therefore, that the single indented vessel from Newstead may belong to the Antonine period rather than to the Flavian occupation. (Cool and Price 1998, 145 on post-Flavian indented vessels).

The other piece of tableware recovered is part of a yellow/green long-necked jug (2). Even though the general shape of the body has not survived, the handle-attachment with its nicked extension of the central rib suggests that it originally had a conical body similar to the Turiff jug from Aberdeenshire (Thorpe 1934). These jugs exist in a number of varieties, differing in colour, shape and decoration. The pale yellow/green example from Turiff shows some diagonal ribs on the body, but nothing suggest that the Inchtuthil specimen was similarly decorated. While the strongly coloured examples of this jug appear to be mainly limited to the second half of the first century, some colours (viz. yellow/brown, blue/green and yellow/green) appear to continue into the third quarter of the second century AD, with the yellow/green example from Felmongers-Harlow (Price 1987, 193-5 fig. 3,20) being the latest closely dateable example (from a context dating c. AD 160-170).

Further yellow-green conical jugs are known from the Pre-Flavian levels at Usk (Price 1995, 179 fig. 46, 100), from a pit dated AD 70-130 from Chichester (Charlesworth 1978, 269 no. 30 fig. 10.23), from Richborough, pit 122 (dated c. AD 130-150: Bushe-Fox 1949, 148 pl.67, 367), two near complete vessels from pit DD II 29A at Alcester (dated AD 150-60), Towcester, Park Str. (Price 1980, 66 fig. 15, 7-8 (dated c. AD 155-165)), and from the Antonine levels at Strageath (Price 1989, 198 no. 9). A further possible Antonine piece comes from Carlisle, The Lanes (Price & Cottam forthcoming I,13) dating to period 8), as well as unstratified handle fragments from Corbridge (Allen 1988, 289 fig. 131,6), Watercrook (Charlesworth 1979, 232 fig. 93, 169), Doncaster (Allen 1986, 103 fig. 24, 1), Shepton Mallet (Price and Cottom forthcoming b) Verulamium, King Harry Lane (Price 1989b, 45 fig. 27,305) and Colchester (Cool & Price 1995, 124 fig. 8.3 no. 895). Three possible pushed-in base rings from Caersws (AD 100-130+): Cool & Price 1989, 39 fig. 21, 34), Baldock (?: Charlesworth 1986, 193 fig. 84, 876) and Puckeridge-Braughing (Price & Cool 1988, 83 fig. 35, 10), as well as a yellow-green ribbed neck and shoulder from Gorhambury (Neal et al. 1990, 203 fig. 163, 22) may all belong to similar vessels.

Fragments 3 and 4 form part of prismatic containers. These storage vessels, most commonly square in section and often decorated with geometric (or far more rarely figural) patterns on the base, became the most common glass vessel on Roman sites. They appear to have been used mainly as storage containers for liquids, although some large examples have large enough openings to allow the storage of small objects (e.g. eggs). Neither rims nor bases of these vessels survive amongst the Inchtuthil material, but the fact that 10 sherds out of 30 probably came from glass containers, either of this type or the slightly earlier circular variety attests to their frequency in Northern Scotland. The majority of such containers are found in contexts dating to the middle of the first to the end of the second century AD (Price and Cottam 1998, 194-202).

Reuse of materials, especially of glass is a common enough feature on Roman sites but the dark blue fragment (5) appears to have been reworked into a squarish shape, which is slightly unusual. The most common form of reuse of glass fragments appears to be their conversion into gaming counters, e.g. from Carlisle (Cool & Price 1991, 176 fig. 155, 684), but the unusual diameter of this piece and its shape makes appear look unlikely for the Inchtuthil ‘counter’.

The material also contained one glass melon bead. Dark blue melon beads are a less common variety of the ubiquitous frit melon beads. Both types appear particularly common on military sites of the first and early second century, and then seem to continue in very much reduced numbers into the later second and third century AD on both military and civilian sites (Guido 1978, 100). They are often associated with horse harness (Böhme 1978, 288-9), but also occur on other forms of military equipment. One of the highest concentrations of melon beads within the province appears to be Newstead with over 129 melon beads (both frit and glass) recorded to date (Hoffmann forthcoming).

The remaining two fragments are Roman window glass, both showing the characteristic combination of matt/glossy finish that comes from casting the material. Recent experiments conducted during the making of the BBC television program ‘What the Romans did for us’, showed that this process most likely involves a bun of hot glass being pulled into shape with tweezers and then being reheated in the furnace. This process could be repeated until the desired shape was achieved. Window glass of this type is first recorded in a Claudio-Neronian context in Colchester (Harden 1947, 306) and appears to have been gradually replaced by the thinner blown window glass during the third century AD (Boon 1966). The occurrence of window glass at the fortress of Inchtuthil (a further piece was recovered by Richmond’s excavations) suggests that even in Scotland glazing in at least some windows formed part of Roman fortress design from the start. The prime candidate for such ‘luxury’ would have been the hot room of the bath block. However, as none of the fragments can be associated with an exact find spot, it is impossible to confirm or deny this.

Comparison with other glass material known from Inchtuthil

The excavations within the legionary fortress in 1952-1968, produced a further 75 fragments of Roman vessel glass and another melon bead, as well as the one fragment of window glass already mentioned. The glass recovered belonged to various forms of vessel glass, including a pillar-moulded bowl, a hemispherical cup, a further five long-necked jugs (including a yellowish green discoid jug), an unguent bottle, one square bottle and six cylindrical bottles. With the addition of the current indented cup, a further jug and another 2 prismatic (square?) bottles. The assemblage now presents a good cross-section of the most popular glass vessels on northern military sites of the late first century. The omission of the facet-cut beaker, known from other contemporary sites, may be due to the comparatively small number of fragments recovered from the fortress overall. Only the hemispherical cup found in the 1952-1968 excavations still forms an unusual marker and is hard to explain on a site that presumably started only in 83-89, as these beakers are usually associated with sites in occupation in Britain in the Pre-Flavian period and possibly extending into the early seventies, although an heirloom should perhaps not be ruled out. In fact in many ways (e.g. the high number of long-necked jugs and the indented beaker) the Inchtuthil glass assemblage would also not cause comment in a fort that had an Antonine as well as a Flavian occupation (compare, e.g. Strageath (Price 1989)), although no indicators for such a reoccupation have been found in the case of Inchtuthil.


1. FY 156/11 1 body fragment, colourless. Indented beaker. One indent surviving. no bubble. No weathering. Dims: 25 x 19 Th: 1.5 EVE: 0.20

2. FY 156/29 1 body fragment, yellow/green. Conical long-necked jug. Lower handle attachment of Handle with central rib, Handle drawn out over body and nicked in regular intervals. Elongated bubbles. No weathering. H(ext): 57 W(handle): 30 EVE: 0.14

3. FY 156/21 1 fragment, body. Blue/green. Straight sided prismatic container. One edge preserved. Small bubbles. No weathering. S: 45 S2: 11 H(ext): 87 EVE: 0.14

4. FY156/24 1 body fragment, blue/green. PrismaticBottle. Dims: 40 x 32 Th: 2 EVE: 0.14

5. FY156/27 1 fragment. Dark blue, transparent. Round glass object, chipped into shape rectangular section. Scratched. One small bubble. No weathering. D: 46 Th: 14.

6. FY156/31 1 Glass melon bead, dark blue. Grey surface. L: 21 D: c.30 D (int): c. 7

7. FY156/13 Window glass. Dims: 21 x 37 x 4

8. FY156/25 Window glass. Dims: 29 x 67 Th: 3

Surface finds from the Roman fort at Bertha

The NMS preserves 14 fragments of Roman glass from the fort at Bertha, near Perth. The majority of these were deposited on March 13, 1781 by Mr Anderson and Mr. Marshall of Perth. As to their find circumstances the following information is given in the ‘Account of the Institution and Progress of the Society of Antiquaries of Scotland’ (Smellie 1782,

‘The remains of some Roman cinereal urns of yellow clay, and some fragments of glass vessels, of a bluish colour, lately dug up on the north bank of the river Almond, near its influx into the Tay, where there had been a bridge built by Agricola over this last river, of which some remains still exist.’

The claims that these glass vessels belong to the remains of Roman burials is hard to verify, especially as the find spot is likely to have been long since eroded by the river, but the large size of the cylindrical bottles would allow for a secondary use as funerary urns.

In all 12 fragments from at least three vessels are preserved and were briefly commented on in the 1892 catalogue of the NMS (NMS 1892, 221).

In addition to these antiquarian finds, 2 further fragments of glass – the rim of a bottle – and a dark blue melon bead were found by R.W.Feachem in Bertha and donated to NMS in 1958 (PSAS 92, 120 (14)).

FR200 is a rim fragment of a blue/green pillar moulded bowl. These vessels were very popular in Flavian forts and parallels are known from Inchtuthil (Price 1985, 308 fig. 93,1), Cardean (Hoffmann, forthcoming b), and Elginhaugh (S.Worrell, pers. comm) whilst Strageath has produced a pillar-moulded bowl of purple glass with yellow marbling (Price 1989, 193;197 fig.100,1). Outside the military installations they are also known from Castlehill Wood Dun, Stirlingshire (Robertson 1970, 202 Table II), albeit also in the millefiori variety (Feachem 1957, 35 fig. 9,1). Further south, blue/green pillar-moulded bowls are a dominant feature of Flavian assemblages and they can be found in large numbers at Newstead (Hoffmann, forthcoming a), and Vindolanda (Hoffmann, in print), to name but two.

The remaining fragments all come from bottles, i.e. storage vessels of varying sizes. Only one, No. 4, can be identified as to its original shape, in this case cylindrical. Other shapes that have been identified from other Flavian and Antonine sites north of the Antonine Wall include square (Strageath, Inchtuthil, Cardean) and hexagonal bottles (Cardean). All varieties circulated in large numbers from about AD 43 until the end of the second century. The cylindrical bottles vanish from the assemblages around the beginning of the second century AD. (Price & Cottam 1998, 191-194). It is therefore, unlikely that these ‘funerals’ related to the recently recognised Antonine period of the fort (Woolliscroft forthcoming, 38).

While the majority of these containers held up to 2 pints of liquid, a number of larger containers, particularly cylindrical vessels existed. These are occasionally encountered on settlement sites. Cardean produced vessels with likely minimum volumes of up to 8 l (Hoffmann forthcoming b), and the largest of Vindolanda’s cylindrical bottles may have contained about 12 l of liquid. The surviving body fragments from Bertha have an extant volume of 3.2 l, if reconstructed according to the most likely diameter/height proportion the bottle would have contained c. 15 l as a tall bottle, or c. 6.5 l for a low bottle (for likely proportions see, Cool & Price 1995, 182 fig. 11.4 and Hoffmann forthcoming b). Containers of this size provide storage capacity for largish amounts of material and it is therefore hardly surprising that they were frequently reused as funerary urns (Price & Cottam 1998, 191-3). Thus, although no cremated remains survive in the case of the Bertha examples, a secondary use for these vessels cannot be ruled out.

The dark blue melon bead is a standard find on Roman military sites of the first and second centuries AD (Guido 1978, 100), and one of its more common counterparts – the frit melon bead was encountered in the 1973 excavations at the fort (Adamson & Gallagher 1986, 200 fig. 3,1).

Overall, this very small assemblage contains the most common types of Flavian glassware. Unlike the pottery finds, however, there is nothing in it to suggest an Antonine occupation, which serves as a reminder that such small collections rarely reflect the full range of materials available from the full occupation history of a Roman site.


1. FR200. 1 rim fragment, blue/green. Pillar moulded bowl. 2 ribs surviving. Polishing marks on inside. No bubbles. No weathering. D: c. 200. H(ext): 42 Th: 5 EVE: 0.4

2. FR200. 3 fragments, 2 joining, probably all three from the same vessel. Blue/green. Container/bottle. Rim, Neck, shoulder and handle of large container. Folded rim, turned out, up and in. Trail of upper handle attachment surviving. Cylindrical neck and reeded handle. Bubbly glass. No weathering. D(rim): 64 H(ext: rim): 23 W(handle): 58 H(handle): 69 EVE: 0.42

3. FR201. 1 fragment. blue/green. Container. Complete reeded handle. Reeding does not apply to upper level of handle. Cylindrical body? A lot of elongated bubbles as well as clouds of tiny bubbles. No weathering. H: 106 W: 95 L(upper handle): 71 Th: 11 EVE: 0.14

4. FR202. 7 fragments, joining. Bluegreen. Container. Straight cylindrical upper body. A lot of vertical scratches on the shoulder turn. Few small bubbles. No weathering. D: c. 220 H(ext): 93 Th: 4 EVE: 0.28

5. FR664. Fragment of pale green glass from protruding flat rim of bottle (1958-86) Fragment, rim. Blue/green. Folded rim. Fire-rounded edge bent out, up and in. Very many small bubbles. No weathering. D: 46 H: 6 Th: 2 EVE: 0.14

6. FR663. Half of melon bead of deep blue transparent glass (1958-86) Bead, fragment. Dark blue translucent glass. No weathering. D: 26 H: 12.5 Th: 6

Roman Glass from Ardoch

During the 1897 excavation (Christison and Anderson 1898) a number of glass fragments were recovered which were later deposited in the NMS by Colonel H.S. Home- Drummond. The original report of the excavations included a rather brief glass report (Christison and Anderson 1898, 453), but a more detailed treatment is still required. In total 28 fragments of glass and 4 glass objects are preserved in the NMS in Edinburgh (Inv No. FQ 203-217), of which two are most likely modern. The only item of tableware is a conical facet-cut beaker, which has disintegrated into 21 small fragments and a small amount of decomposed glass. This form of decay is very common among this type of glass ware. It occurs on examples from a wide range of locations and is probably due to a problem with the glass mixture, although to the writer’s knowledge no scientific analysis has been attempted. The surviving fragments show the original vessel to belong to Oliver’s Group 1, which is characterised by the ‘zone of faceting having been left raised above the level of the undecorated zone bordering the rim and foot. Ridges and ledges above and below the faceting are lacking.’ (Oliver 1984, 36). This puts the fragments into the same group as the tall beaker from Cardean (Wilson 1969, 202 pl.14,1 & Hoffmann forthcoming b) and sets it apart from Group II, which may not have this raised area, but characteristically has a series or ridges or ledges to separate the decorated and undecorated area of the beaker (Oliver 1984, 36). It is impossible to judge from the fragments, whether the Ardoch beaker was a tall or a squat beaker.

Facet-cut beakers have been dated to the last third of the first century AD and can still be found in Hadrianic, and sometimes early Antonine, contexts (e.g. the undecorated Group II beaker from Rough Castle (Charlesworth 1978-80, 268-9. fig. 12 – Cool et al 1995, 1566 – Price and Cottam 1998, 80-83.), although Cool and Price have recently argued that these later beaker may well be residual (Cool and Price 1998, 146). Further Scottish examples of these beakers include finds from Newstead (Hoffmann, forthcoming a), Lyne, Castlecary and Cadder (Charlesworth 1959, 41-2, fig. 4), along with Strageath and the beaker already mentioned from Cardean. On Hadrian’s Wall and its vicinity they have now been recovered from most of the sites on which large scale excavations have been conducted e.g. Carlisle (Price 1990, 168 MF 66 No. 11 and 13 fig. 160), Birdoswald (Price and Cottam 1997, 348 fig. 248,11), Vindolanda (Hoffmann in print) and Corbridge (Corbridge Mus.), while the fort at Castleford in Southern Yorkshire with 11 facet cut beakers stands out as yielding a particularly large number of such vessels (Cool & Price 1998, 145f.).’s. 2-3 are the upper bodies of cylindrical bottles. These are a standard find on sites north of the Antonine wall (cf. Bertha, Inchtuthil (Price 1985, 312 fig. 94,11), Cardean (Hoffmann forthcoming b), and were popular from the middle of the first century AD to the beginning of the second century (Price & Cottam 1998, 191-194). Prismatic, often square, containers like no’s 4 and 5 began at about the same time but continued in popularity until the end of the second century (Price and Cottam 1998, 194-202). A further 2 fragments (6 and 7) cannot be attributed to any specific type of bottle.

The three frit melon beads (8-10) can be considered as standard finds on Roman military sites this far north (cf. Inchtuthil and Bertha above).

No. 5 was, after the breakage of the original vessel, reworked into a gaming counter. Such counters have been occasionally found on other Roman sites and examples from the north of Britain include a piece from Carlisle (Cool & Price 1991, 176 fig. 155, 684), and a further four examples from Vindolanda (Hoffmann in print).

Like Inchtuthil the excavations at Ardoch have produced one piece of matt/glossy window glass, which would fit equally well in an Antonine or a Flavian context.

The yellow bead originally reported by Anderson (Christison and Anderson 1898, 453) was on closer inspection identified as amber and has, therefore, been excluded from this report, as have two 18/19th century fragments.

The material recovered from Ardoch mirrors closely the material found at other Roman sites in the area. Only the melon beads and the cylindrical bottles are more likely to be Flavian than Antonine, whilst the prismatic jugs and the facet-cut beakers could equally well be either.


1. FQ 207-208 21 fragments and many tiny fragments (some joining). Body of facet cut beaker. Facets on raised ledge. Facets in quinqunx. Cracked throughout and devitrifying. Dims: a) 20 x 11 Th: 5 b) 18 x 18 Th: 5 c) 18 x 13 Th: 5 d) 19 x 12 Th: 5 e) 15 x 16 Th: 5 f) 13 x 14 Th: 4 g) 13 x 11 Th: 4 h) 12 x 13 Th: 4 i) 13 x 10 Th: 4 EVE: 0.4

2. FQ 211 Fragment, body. Blue/green. Shoulder of cylindrical bottle. No bubbles. Scratched. No weathering. D: c. 130 H(ext): 25 Th: 5 EVE: 0.14

3. FQ 215 Fragment, body. Blue/green. Shoulder of cylindrical bottle. No bubbles. Scratched. No weathering. D: c. 160 H(ext): 44 Th: 4 EVE: 0.14

4. FQ 216 Fragment, body. Blue/green. Fragment of straight sided prismatic container. Small bubbles. No weathering. Dims 56 x 45 Th: 6 EVE: 0.14.

5. FQ 212 Fragment, body. Greenish. Fragment of prismatic container. No bubbles. No weathering. also: reworked into glass gaming counter. D: 24 Th: 4 EVE: 0.14

6. FQ 214 Fragment, body. Blue/green. Neck of bottle. Turn of shoulder surviving. Elongated bubbles. No weathering. D: 68 H(ext): 32 Th: 4 EVE: 0.14

7. FQ 213 Fragment, body. Blue/green. Shoulder of bottle. Elongated bubbles. No weathering. Dims: 24 x 40 Th: 7 EVE : 0.14

8. FQ 203 Frit melon bead: D: 25 H:20 D(int): 9

9. FQ 204 Frit melon bead: D: 15 H: 10 D(int): 7

10. FQ 205 Frit melon bead D: 15 H:10 D(int): 4

11. FQ 217 Glass Window pane, roughly rectangular. Edge Dims: 62 x 84 Th: 3.5

A comparison of the glass material recovered from Roman sites north of the Antonine Wall

To date there have been reports on glass or glass objects from eight fort sites north of the Antonine Wall. The quantities of material differ widely. Bochastle, Fendoch and Cargill have produced 1-3 items each, whilst Bertha and Ardoch have produced groups of less than 50 fragments, with the Inchtuthil corpus being slightly larger. Only Cardean and Strageath have so far produced samples that may be fully representative of the types originally present on the site.

As Price (1985, 193) has already remarked, there is a marked decline in the use of glass between the Flavian and Antonine periods. All the glass assemblages are, therefore, not surprisingly dominated by material available in the Flavian period, and especially bottles. The fact that a lot of the vessel glass types available in the Flavian period continued in use in the Antonine period compounds the problem of chronological differentiation on the sites and it is perhaps symptomatic that only the largest assemblage, Strageath, has so far produced glass that has to be Antonine, i.e. a colourless cup or beaker with wheel-cut lines.

It is also interesting that the range of glass object types is rather small on these sites. Frit melon beads are ubiquitous, but other types have only been found at Strageath and Cardean. Moreover, it is perhaps worth commenting on the fact that only the square sectioned bead from Strageath has to be associated with Romano-British female jewellery, while the other bead types either have strong local Iron-Age traditions (e.g. the opaque yellow annular beads, which are similar to the ones recovered from the Culbin Sands near Inverness (PSAS 9, 1870-72, 80f.)) or are not attributable either to gender or culture. Also of interest is the fact, that whilst glass bangles are a common find on military sites around Hadrian’s Wall and have been found occasionally as far north as the Antonine Wall. North of this area they seem to be confined to Iron Age sites for, to date they seem to be missing from the Roman military installations.

A long term research project to study the Romans north of the Antonine Wall