A recent examination of some unpublished data in my archives identified the likely presence of a glass kohl bottle in York: a form specifically designed as an eye make-up container. They were very common in Egypt but have not, to my knowledge, been identified elsewhere amongst the vessel glass of Roman Britain. The use of kohl can rarely be traced outside the Egypt/Nubia area and so this article explores the implications of this find for York.

The bottle came from excavations conducted by the York Archaeological Trust in 1983–4 at 24–30 Tanner Row (NGR SE 59965176). At that time it was also known as the General Accident Extension site. This is an area in the Colonia lying to the south of the River Ouse. As was the custom of that unit, evidence from their excavations was published in fascicules as the reports on different aspects were completed. For this site there are reports on the animal bones, the environmental remains and the pottery.Footnote ¹ The site narrative is now available and shows that this area of York did not start to be developed until the mid-second century.Footnote ² Other aspects of the material culture from the site remain unpublished. I catalogued the glass about 30 years ago whilst working for the Trust.

The bottle (sf1719: Fig. 1, Appendix 1) was recovered from context 2262 which was one of the deposits associated with Building 5 in Period 5 Phase 2.Footnote ³ In the archive report these deposits are described as mixed, consisting of 16 separately recorded contexts, overlying and in between the timbers of the building. The possibility that they were an accumulation below a raised floor was suggested, and it was noted that the structure had not had any recognisable floor levels.Footnote ⁴ In the published report Period 5 is assigned to the late second century. Perrin noted that pottery was very mixed but included material with late second- and third-century characteristics.Footnote ⁵ In Period 6 the area was occupied by a series of ephemeral buildings occupied for a short period. A stone building followed this in Period 7. Both periods are assigned to the mid-third century. The pottery evidence for Period 7 suggested an early to mid-third century date.Footnote ⁶ The bottle is thus securely dated by its context to the later second century at the earliest, and possibly extending into the early third century.

Fig. 1.

The kohl bottle sf1719 from 24–30 Tanner Row. The central views show the marked asymmetry of the vessel. Scale 1:1. (© Author).

Sf1719 retains the reservoir and base of a small blue/green flask with an approximately conical profile and is broken at the neck/body junction. It is asymmetrical with one side forming a thickened bulge at the base. Its defining features are that the lower wall and base become very thick and the central reservoir is near cylindrical, retaining the same diameter within the body as it does in the neck. This sets it apart from the vast majority of glass unguent bottles across the empire. Normally these bottles had thinner walls and the shape of the reservoir mirrors the form of the external profile. This is well demonstrated by comparing sf1719 to the types of unguent bottles in common use in late second- to third-century York. Fig. 2.4 depicts a profile of sf1719. Fig. 2.1–2 are examples of indented unguent bottles of Isings Form 83 which were found in grave S47750 at Hungate.Footnote ⁷ Fig. 2.3 is a candlestick unguent bottle of Ising Form 82 from the Railway Station cemetery.Footnote ⁸

Another difference is the quality of the glass. Although both sf1719 and the indented unguent bottles from Hungate are blue/green, sf1719 has thick silvery iridescence on all surfaces. There are also small patches of much darker weathering which overlies it in places, especially on the base. As can be seen in the photographs in Fig. 2, the Hungate bottles lack this thick iridescence and it is possible to see through them. The bubbles in Fig. 2.1, for example, are clearly visible, as are the black impurities within the glass on the upper body. Roman glass in the north-western provinces is generally very chemically stable. It does not decay as, for example, medieval potash glass does. It may develop a light iridescence as it moves from the state of being a super-cooled liquid to being a solid, but surfaces which are as heavily ‘weathered’ as those of sf1719 are, in my experience of Romano-British glass assemblages, uncommon. The blue/green glass found in York, for example, more commonly resembles the Hungate bottle state than it does that of sf1719.

Fig. 2.

Sf1719 (no. 4) compared to typical unguent bottles in York of the later second to third centuries (nos 1–3) and kohl bottles (nos 5–7). Scale: photographs nos 1–2 1:1, line drawings 1:2. (Nos 1–5 © author; nos 6–7 after Harden Reference Harden1936, pl. XX nos 815 and 819).

The Tanner Row site was a heavily waterlogged one,Footnote ⁹ and it is possible that this might have had an effect on the glass. To check this, I recently re-inspected sf1719 and other blue/green and colourless fragments from contemporary contexts at the site. The latter showed the normal pattern of preservation with, at most, only a light iridescence. It does not appear that the burial conditions alone can account for the appearance of sf1719. The bottle was clearly a finished vessel and not a waster from manufacture as the pontil scar on the base indicates the glass blower had finished the rim. It is always possible that it was just an extremely poorly made, standard conical unguent bottle. That said, it can be observed that in 40 years of studying Roman glass, I have not encountered that level of incompetence in other vessels. It seems useful to look for alternative explanations.

When blown glass vessels started to become common in the early first century a.d., small flasks used for perfumes and oils were among the most frequently encountered forms. This reflected the high demand for the contents which were used in a variety of contexts such as dining, religious rites and funerals as well as the care of the body.Footnote ¹⁰ Most of these followed the pattern of having reservoirs that mirrored the outer profile, but the glass industries in Egypt developed an additional form catering for a local need and it is these that had the effectively cylindrical internal profile.

Kohl is a black eye make-up long used in Egypt and Nubia. The lead ore galena was a major ingredient, though most of the analytical work to explore the full composition has focused on third- to early first-millennium b.c. material rather than that of the Roman period.Footnote ¹¹ The containers could be made in a variety of materials, and there is a well-published group of wooden examples from the Kushite cemetery at Ballana.Footnote ¹² The site is in lower Nubia between the first and second cataracts of the Nile. The cemetery was in use from the first to fourth centuries a.d. and so is of the Roman period, though beyond the boundaries of the empire. These show the cylindrical reservoirs very well. The kohl was stored in a solid form and the containers were used with kohl sticks which could be double-ended like a Roman spatula probe.Footnote ¹³ These were used both to prepare the make-up and to apply it. A cylindrical reservoir is thus appropriate, as extracting the kohl using a rigid stick from a vessel with an expanded one would be more difficult.

The earliest blown glass kohl bottles could either have a conical profile like the Tanner Row bottle (Figs 2.5 and 2.7) or they could have expanded bases where the cylindrical reservoir was positioned mainly within the neck and only extended a short way into the solid body (Fig. 2.6). As can be seen in Fig. 2.4, the profile of sf1719 has more in common with the Egyptian kohl bottles than it does with the contemporary unguent bottles from York. The kohl bottles are normally described as being made of dark green glass, but they are frequently found in a very weathered state where the surface has decomposed so much it can be difficult to see the original colour. A group in the Musée du Louvre has been carefully catalogued, and the colours recorded include green, strong green, blue/green and blue-tinged.Footnote ¹⁴ An example from Meroe (Sudan), which was also made of blue/green glass, is also shown here as Fig. 2.5 and demonstrates the degree of ‘weathering’ that can occur.Footnote ¹⁵ The colour of sf1719 would thus be acceptable in a kohl bottle. A similar deterioration of the surface is probably not to be expected given the very different burial environments encountered in water-logged York and arid Egypt.

Both the conical and expanded base kohl bottles were in use in the later first and second centuries. An example of the conical profiled type was in use by the a.d. 90s as it was found in a context of that date at Didymoi.Footnote ¹⁶ Another from Umm Balad was found in a mid-second-century context.Footnote ¹⁷ Both of these are military sites in Egypt’s Eastern desert. A rubbish dump at Quseir al-Qadim (Myos Hormos, Red Sea coast) produced both forms.Footnote ¹⁸ This had been heavily used in the mid-second century with the final dumping taking place in the early third century. In the rubbish dumps of the fort of Wadi Abu Ma’amel at Mons Porphyrites a total of 36 of the expanded base form came from second-century rubbish dumps.Footnote ¹⁹ Kohl bottles like the Tanner Row example would certainly have been common in Egypt in the later second century, and the regular recovery of them within military contexts may be noted.

In the past these little vessels have puzzled glass specialists when they have tried to interpret them within the context of being a standard unguent bottle. Meyer, writing up another assemblage of them from Quseir al-Qadim, noted that, if they contained perfume, the buyer would have been cheated.Footnote ²⁰ Others, given the numbers found in desert settings, have suggested they were used for eye salves.Footnote ²¹ This is an understandable suggestion. Treating patients suffering from eye complaints must have been an ever-present part of a doctor’s work in the Roman world, given the number of prescriptions known for them from sources such as medical treatises and collyria stamps.Footnote ²² Some of the ingredients in kohl may have reduced or prevented bacterial eye infections,Footnote ²³ but it is unlikely that medical eye salves would have been stored in this way. Ancient authorities said metal boxes should be used if long-term storage was required for them. Scribonius Largus, for example, wrote a manual on the compounding of drugs in the mid-first century a.d. which later authors drew on. He recommended boxes of Corinthian bronze, silver and stagnum, which is a lead/silver alloy.Footnote ²⁴ Fortunately the use of these bottles for kohl is confirmed by the example from the fort at Umm Balad. This retained not only part of its wooden stick wrapped in cloth, but also part of its content, which was identified as galena.

Attempting to trace the distribution of kohl bottles beyond Egypt and Nubia produces few results. This is not the result of the absence of well-published glass assemblages from the provinces to the west of this core distribution in north Africa or to the north-east in the Levant and Syria. For the former there are, for example, large and well-documented groups from Carthage. Many of these deal predominantly with late Roman material, but that from the ‘Quartier Magon’ has an extensive range of early to mid-Roman material and kohl bottles are not present.Footnote ²⁵ For the Levant and Syria, there are detailed typologies and again they are absent.Footnote ²⁶ The scarcity of kohl bottles outside their core area was observed when the examples in the Musée du Louvre were published.Footnote ²⁷ The authors pointed to an example with an expanded base from Cyrene, Libya.Footnote ²⁸ Others come from Dura-Europos, SyriaFootnote ²⁹ and from Beirut, where it was noted there are other unpublished examples from Lebanon.Footnote ³⁰ A comparison between the examples present on these sites with those from Egyptian sites is striking. In Egyptian assemblages they will occur in large numbers. Elsewhere, when they do occur, it tends to be as single examples in large glass assemblages.

Widening the search area suggests they were virtually unknown further north and west. For Italy, de Tommaso produced a very detailed typology of small flasks and unguent bottles, but it lacks the conical and expanded base kohl types.Footnote ³¹ That those did occasionally reach Italy is suggested by two bottles recovered from graves at Tor Vergata and Via Paribeni, Rome.Footnote ³² That those were a very unusual occurrence is certain, given the numerous Italian museum catalogues which include large collections of unguent bottles of different types, but where kohl bottles are conspicuous by their absence. They are also missing from the typology for Greek glass vessels,Footnote ³³ and from the glass assemblage from Augusta Raurica (Basel, Switzerland).Footnote ³⁴ The latter is so large and comprehensive that the typology that has been produced to deal with it is frequently used as a point of reference for assemblages from elsewhere in the north-western provinces and beyond. Against this background, the presence of sf1719 at York verges on the bizarre, and one has to ask what mechanisms might have led to its presence.

One could consider a normal trading environment. We know that there was a great deal of trade between provinces within the Roman Empire. Commodities such as olive oil leave a very obvious footprint in the archaeological record in the form of the huge amount of discarded amphora fragments that are found. Other commodities can be almost invisible. An example is the trade in building materials such as the pozzolana from the Bay of Naples, transported around the Mediterranean to make the concrete for harbour installations. It was most likely shipped as ballast and did not have a container.Footnote ³⁵ Other mechanisms include personal agency. Individuals moved between provinces taking their possessions with them. This could have been for personal reasons, or it could have been involuntary, which must have been the case when a military unit moved. Again, though, individuals and quartermasters would have been moving objects from one place to another.

The possibility that sf1719 moved as part of trade in kohl across the empire seems unlikely. We have seen that kohl containers had a quite specific internal cylindrical form so they should be recognisable in the archaeological record. Obviously it has not been possible to conduct a comprehensive, empire-wide search, but the picture that emerges for the first- and second-century glass kohl bottles is that they rarely travelled beyond Egypt and Nubia. Make-up and skincare products were used across the empire, but there were clearly regional preferences. Within Roman Britain, for example, it seems highly likely that distinctive pestle-and-mortar sets were used in the preparation of small quantities of pigments for use as cosmetics.Footnote ³⁶ Their use has only very rarely been seen outside Britain, mirroring the very occasional recovery of kohl bottles outside Egypt and Sudan. For something to be a traded item there needs to be both a supplier and a market for the product. For neither kohl, nor the pestle-and-mortar sets, does there appear to have been a market outside the home one. This suggests that an explanation for the presence of sf1719 would be better sought within the realms of someone bringing it with them as a personal possession.

By the late second century, York was a busy centre. A large part of the Legio VI Victrix would have been back in garrison in the fortress after the posting to the Antonine Wall, and the civilian town was expanding. It was the home of a population that included those from distant parts of the empire. At the upper end of the social scale there was Claudius Hieronymianus, a senator and the commanding officer of the legion. In York he is known from the dedication-slab of the temple he built for Serapis.Footnote ³⁷ He went on to be governor of Cappodocia, placing his time in York in the a.d. 190s.Footnote ³⁸ Where he grew up is unknown, but his cognomen suggests an eastern origin. His dedication to Serapis, an Egyptian deity, might support this, though, as the consort of Isis, this was a god worshipped across the empire by people of different ethnicities.

Lower down the social scale there were the potters manufacturing a distinctive north African style of cooking vessels in the local Ebor ware. This activity has been recognised since 1981,Footnote ³⁹ and has given rise to an extensive literature exploring what the implications are: usefully summarised in an article in this journal by Pitts.Footnote ⁴⁰ The date of this phenomenon has been established by Monaghan to lie within the first quarter of the third century, and associated kilns in the Aldwark/Bedern area show it was associated with the military.Footnote ⁴¹ The impetus for this new style of pottery could well have been new drafts of soldiers from other legions who joined the Legio VI Victrix when it returned to York after Clodius Albinus’s defeat at the battle of Lugdunum in a.d. 197, and possibly by those who accompanied Septimius Severus to York during his campaigns of a.d. 208–11. It seems unlikely, however, that the potters making the north African style pottery would have been wearing kohl make-up. The comparanda for the vessel forms are to be found in Tunisia.Footnote ⁴² As has already been noted, that does not appear to be an area where kohl, and its bottles, were in use.

Among the finds recovered from Roman Britain there are, very occasionally, items that suggest that their owner had connections with Egypt at some point in their lives. Of particular interest is the ivory panel from a composite box found associated with a large courtyard house at Vine Street, Leicester occupied in the early third century (Fig. 3 no. 1).Footnote ⁴³ This is a very rare example of a Romano-Egyptian product which, unusually, depicts Egyptian motifs rather than the more normal deities of the Roman pantheon. The front depicts lotus flowers and Anubis, who holds linear objects that may have been intended to be lances. Roman soldiers in Egypt favoured the worship of Anubis, god of embalming,Footnote ⁴⁴ and so the recovery of military equipment within the house is intriguing. The link to the military becomes even stronger when it is realised that three legionary seals were found near the house. An unstratified one from the preliminary evaluation mentions both an individual from Legio VI Victrix on the obverse and Legio III Cyrenaicae on the reverse.Footnote ⁴⁵ The latter was stationed in Egypt until a.d. 106, when it was transferred to the newly created, neighbouring province of Arabia Petraea to the north-east.Footnote ⁴⁶ The find places of lead seals need not be directly associated with the named legion as, after all, they were intended to be fastened to objects that would travel and could get lost en route. Here, though, the recovery of two unique items with connections to the military in the broad Egyptian area does appear to suggest that someone, perhaps an officer or veteran who had been stationed there, was present at Vine Street.

Fig. 3.

Ivory box panel from Leicester (© author) and glass bath flask from Caerleon (after Allen Reference Allen and Zienkiewicz1986, fig. 42 no. 52).

Another example is a glass bath-flask found in the drain deposit dated to c. a.d. 160–230 of the fortress bath at Caerleon (Fig. 3.2). This is colourless with an elaborately cut design of a type that has been regularly found in Egypt and Sudan but which is rare elsewhere. In discussing this, Allen also drew attention to examples thought to have been made in the western empire.Footnote ⁴⁷ A recent reappraisal of examples from Egypt and Sudan has shown that a very strong case can be made for the Caerleon flask coming from the Egyptian/Nubian area because of the linear decoration on the neck. This is a very rare feature but can be directly paralleled by a bath-flask from Semna South in Sudan.Footnote ⁴⁸ If that is correct, it is more likely to have travelled to Britain as a personal possession than being acquired there as an object of trade.

These instances of unusual Egyptian items being found in Britain provide a background against which to place sf1719. It might be suspected that the most likely community of people to bring items like this among their baggage was the military one. There is no secure epigraphic evidence that there were transfers of units or detachments from the area to Britain. The lead sealing linking the Legio VI Victrix with the III Cyrenaicae from Leicester is intriguing, but what it signifies is unknown. There are auxiliary units with the same name attested in Egypt and Britain, but it is normally assumed that they were different ones.Footnote ⁴⁹ That individual soldiers could be seconded far from their unit is shown by the career of C. Iulius Apollinarius, a soldier of the III Cyrenaicae, known from the papyri found at Karanis in the Fayûm.Footnote ⁵⁰ In a.d. 107, when he wrote home celebrating his promotion, he was stationed in Arabia Petraea. By a.d. 119 he was in Rome, where he had been seconded to act as frumentarius, i.e. a member of the imperial supply and intelligence service.

The Tanner Row site lies within the civilian area of York, but there are some grounds for thinking that it had links with the fortress across the river. There is reused stone that seems likely to have been derived from renovations in the fortress after a.d. 160; the leather-working debris includes fragments of military tent panels; and there is also military metal-work.Footnote ⁵¹ Furthermore it is suggested that many of the artefacts like sf1719 recovered in and around the timber buildings of Period 5 were probably deposited after the buildings had been abandoned.Footnote ⁵² The rubbish of a legionary fortress has to go somewhere and it is tempting to wonder if, at the end of the second and into the third century, the site was one of the dumping grounds. This would certainly account for the fact that the material includes a higher proportion of Ebor ware than the other civilian York sites Pitts studied, a pattern that is typical of the military assemblages.Footnote ⁵³ Sf1719 could thus have arrived as part of that deposition.

The contextual dating for sf1719 indicates that it may have arrived during the Severan activities after a.d. 197 or a little before. That the arrival of the Imperial court and its entourage in a.d. 208 would have been accompanied by military personnel from many different units is certain. Egyptians of both sexes used kohl and the quantities of kohl bottles at Egyptian forts suggest Egyptian soldiers did as well. Objects can rarely be used to suggest the ethnicity of their owners, but, with sf1719, it is the contents and the use to which they would be put which is important. Culturally specific practices, such as the use of kohl eyeliner, are much stronger evidence of ethnicity. This is especially the case here where the scarcity of kohl containers outside of Egypt and Sudan indicates it was not a fashion adopted elsewhere. If the identification of sf1719 advanced here is correct, then we may well have evidence of an Egyptian in York in the late second or early third century. We may also have to revise any preconceptions about the appearance of at least some Roman soldiers in the city. After all, wearing eyeliner is not a habit normally associated with the Roman military.