Slavery on the Northern Frontier: A Stylus Tablet from Vindolanda

Abstract

A Roman stylus tablet discovered at Vindolanda in 2014 preserves the partial text of a deed-of-sale for an enslaved person, only the second such document from Britain. This article presents the results of multiple techniques used to reveal the almost illegible text and proposes a restoration of the format of the document and its lost content, based on more complete examples from Italy and around the Empire. We examine the late first-century archaeological and historical context and suggest that the purchaser is probably the prefect Iulius Verecundus. We consider other possible evidence for the servi of the commanders at Vindolanda, for example in another hard-to-decipher stylus tablet which may be related to their travel. The deed-of-sale provides a new type of testimony for slavery at Vindolanda and adds to knowledge of enslavement in the Roman military.

INTRODUCTION

The first wooden writing tablets at Vindolanda were discovered in 1973. These artefacts, both ink and wax tablets, command around 1,700 inventory numbers in the stores of the British Museum and several more remain, temporarily, in the Vindolanda collection.¹ Around 1200 ink-writing tablets have been published, including texts comprising hundreds of words through to the so-called descripta, namely ‘texts which do not merit fuller editions’,² but excluding those fragmentary tablets on which only indeterminate traces, or no more than isolated letters, could be recovered. The texts have triggered a wide range of scholarship on the economics of the Roman army, life in military communities, the nature of the Roman frontier, military communication, linguistic and palaeographic studies.³ In addition to the ink tablets, around 370 of the inventory numbers record stylus tablets, almost all fragmentary, which makes estimating how many individual tablets are extant problematic, and only a handful have been published to date.⁴ This is not due to lack of interest, but rather the difficulty of recovering their texts. Except in rare cases in which ink has been used to write on these tablets, the texts are preserved only when the stylus used to write on the wax, which initially filled the recess created by the frame of the tablet, penetrated the wax and marked the wood beneath. The reusable nature of stylus tablets also often resulted in palimpsest texts which are very difficult, if not impossible, to differentiate from each other. In the introduction to the second volume of their magisterial Vindolanda Writing Tablets, Bowman and Thomas noted:

It is evident that the now considerable number of stylus tablets requires prolonged study and any attempts to include them in the present volume would have delayed its appearance for a number of years. A significant number of the stylus tablets – perhaps as many as 30 out of a total of well over 100 – contain substantial amounts of writing, although some may be palimpsests. In virtually all cases the incised texts are difficult to see and it is clear that it is essential to find some means of improving their visibility and legibility. Photographic techniques have not so far made significant improvements.⁵

Careful autopsy, using raking light and, at times, a binocular microscope, can help to overcome the challenges faced by the readers of these stylus tablets, but there is a limit to what the naked eye can perceive and Bowman himself has been at the forefront of work to establish digital techniques, particularly Reflectance Transformation Imaging (RTI), to assist the reading process.⁶ Recently, the teams of two projects, ‘LatinNow’, funded by the European Research Council, and ‘Illuminating the Vindolanda Writing Tablets’, funded by the Social Sciences and Humanities Research Council of Canada, have worked together to examine these stylus tablets and new finds, deploying improved versions of existing techniques, e.g. RTI and Polynomial Texture Mapping,⁷ and trialling new methods, including high-resolution 3D scanning, Morphological Residual Modelling and Computer Tomography. These techniques are still being refined for use with the stylus tablets but progress has already been made, and a new stereo photometric system is currently being tested which, early experiments suggest, allows for the most efficient and effective recovery of texts so far achieved.⁸ One major advantage of the digital capturing of extremely detailed imaging is that these data can be reviewed and shared without having to manipulate the tablets after the initial work. The reading of tablets tends to be an iterative process, with the epigraphists often wishing to consult the artefacts repeatedly to double-check traces and verify readings. The objects are extremely fragile and even the most gentle handling can lead to degradation. Digital work has the additional benefit of helping to protect the object itself and to allow access for researchers wherever they are based.

THE TABLET: CONTEXT, FORM AND FABRIC

Perhaps the most significant text so far discovered by the new examination of the Vindolanda stylus tablets undertaken by the authors is on a fragmentary tablet which was recovered from excavations in 2014. The stylus tablet in question (VinTab 2014-5; BM 2019,8012.5) was found in material used to level a terrace or ditch for the construction of foundations for the period II/III buildings approximately 9 m west of the outermost defensive ditch of the period I fort at Vindolanda (c. 85–90 c.e.) (figs 1–2). The excavators have therefore concluded that this filling material was deposited late in period I (85–90 c.e.) of the site's occupation or, possibly, early in period II (90–95 c.e.).⁹ The First Cohort of Tungrians was stationed at Vindolanda during both periods, though they were succeeded by the Ninth Cohort of Batavians later in period II. The find-spot of the tablet was outside the period I fort, but within the foundations of the larger period II fort, and close to a cache of documents relating to Iulius Verecundus, commanding officer of the Tungrians, which includes four ink-written letters by or to Verecundus.¹⁰ The context of this tablet's discovery suggests an association between it and Verecundus, who is known to have owned enslaved people.

FIG. 1. Plan of Vindolanda with the location of VinTab 2014-5 circled in red. (© A.W.M. and the Vindolanda Trust)

FIG. 2. The findspot of VinTab 2014-5 and the line of the feature in which its associated material was deposited. (© the Vindolanda Trust)

The object is a type A1 (Speidel)/type 1 (Tomlin) stylus tablet, with a recess for wax on one face only. Autopsy, digital photographs in visible and infrared spectra, and 3D scanning have not revealed any remnants of ink writing, which are occasionally found on stylus tablets.¹¹ The surviving fragment of the tablet measures 137–139 mm wide by 51 mm high by 5 mm thick at the rim (4 mm in the recess) with a notch half-way along the bottom edge rim about 71 mm from the left side. This notch was designed to receive a binding-cord. It is rare that stylus tablets are found complete, one exception being VinTab 974, which we discuss below. VinTab 974 measures 141.5 mm wide by 106 mm high, and, given that the average size of the complete examples from the excavations at Vindolanda and the Bloomberg site in London seems to be c. 140 mm wide by 110 mm high,¹² we can assume that our tablet is now a little under half its original height.

Analysis at the British Museum by Caroline Cartwright determined that the tablet is made of larch, larix decidua.¹³ Recent research into the fabric of stylus tablets at Vindolanda and elsewhere reveals spruce, larch, fir and maple.¹⁴ The tablets from the Bloomberg excavations were made of silver fir, larch or maple, and the vast majority, 187 of 199 tablets, were of imported silver fir, perhaps reclaimed from used cask staves.¹⁵ Barrel and cask staves are regularly recovered from excavations at Vindolanda, including some made of larch,¹⁶ and their reuse would have been a logical expedient.¹⁷ Sands and Marlière have recently laid out the evidence for the likely reuse of barrel staves on the site for stylus tablets, drop lids, frames and a spatula.¹⁸ They rightly point out that there is no direct evidence to connect these artefacts, including stylus tablets, to barrel staves, but they suggest that the use of the same imported softwoods is unlikely to be a coincidence. Debate continues as to whether writing in the wax was intended to expose the wood beneath.¹⁹ If so, the colour of the pale larch would have stood out from the (probably black) wax that was applied to it as a writing surface.²⁰ VinTab 2014-5 does not retain its original wax; as with most of the Vindolanda tablets this has perished prior to excavation. The only tablet from Vindolanda with significant remnants of wax is VinTab 836, although the wax is now lost. This tablet uniquely illustrated the relationship between the writing in the wax surface and the underlying incisions in the wood.²¹

RECONSTRUCTING THE TEXT AND OBJECT

Initial autopsy of the tablet suggested that some text was preserved, etched into the surface of the tablet, though almost illegible to the naked eye. In order to recover the text, the authors of this paper, in collaboration with Barbara Birley, the curator of the Vindolanda Museum, and with permission of Richard Hobbs, Weston Curator of Roman Britain at the British Museum, subjected the tablet to non-destructive digital imaging using RTI/PTM, raking light high-resolution photography using a Sony Alpha 7Riv camera and high-resolution 3D imaging executed with a Faro Quantum M ScanArm (fig. 3). It was the combination of the results from these techniques, visualised in Photoshop, which allowed us to manipulate, annotate and reconsider sections of the tablet until readings began to form.²² The first word to be revealed was erronem (fig. 4). Examination of the tablet itself at Vindolanda alongside these images yielded the following drawing (fig. 5), text, and commentary:

Edition

– – – – – –

[one line of text now illegible]

[one line of text now illegible]

[6–7 letters now illegible] eum ḥọṃịṇem sanụm

traḍiṭum f̣urṭiṣ noxsiṣque

soluṭum f̣ug̣iṭi(u)ụm erroneṃ

[It is warranted that] this man handed over healthy is free from deceit and wrongdoing, and [is not] a fugitive or a wanderer.²³

FIG. 3. Image of Morphological Residual Model of the tablet by Hugo Pires.

FIG. 4. Section of RTI of the tablet by James Miles, focusing on the first word recovered, erronem, without (above) and with (below) Photoshop annotation.

FIG. 5. Drawing by R.S.O.T.

Commentary

1. 1–2. Slight and illegible trace of two lines.

2. 3. eum is clear, despite the diagonal drawn across the u which may be a long descender from above; it possibly relates to the diagonal across the um of traditum below. The next word, hominem, like the rest of the surviving text, can be securely read by reference to other deeds-of-sale (see further below), but its letters are difficult. The second stroke of initial h is an exaggerated diagonal. It is followed by two downstrokes not much like o, but comparable with that of solutum. m and n are both incomplete, and parted from -em by the s which crosses them; this is unexplained, but might be a mistake, or the remains of an earlier text. Then we read sanum, which is reasonably clear, granted that the s is rather straight and close to the a, and that u, now incomplete, resembles e.

3. 4. traditum can be read, although the horizontal strokes tend to disappear into the wood-grain. The initial t seems to have an exaggerated cross-stroke, but the second t has lost its cross-stroke. furtis has little trace of the cross-stroke of t and a rather straight s. noxsisque can then be cautiously reconstructed. nox- is clear (x with a long descender cutting the line below), followed by si, thus forming the sequence xs which is sometimes used to render /ks/ instead of the usual x.²⁴ But i is apparently followed by n or ae, the diagonal of which is crossed by a very long s which descends into the next line, like the s which reinforces x. This s would seem to be a correction, probably of noxsa (singular) for noxsis (plural). The word is completed by -que, the q being guaranteed by its long diagonal descender, even though its angular loop is now incomplete.

4. 5. solutum can be read, with the cross-stroke of t lost in the wood-grain. It is followed by fugiti(u)um, with the first stroke of g not semi-circular, but almost straight; t largely lost; and u now rather like o. The second u was omitted before final -um, by elision (thus u for uu). The second r of erronem is cut by the long descender of s from the line above, but can be separated from the descender of r. Its final m is incomplete.

The appearance of the phrase eum hominem sanum traditum furtis noxsisque solutum fugiti(u)um erronem clearly marks this document as a deed-of-sale for an enslaved person. Deeds-of-sale have appeared in several Latin documentary texts found in the area of Alburnus Maior (Dacia), Herculaneum, Murecine (near Pompeii), London, La Graufesenque (Millau, France) and Arsinoites in the Fayum, all on stylus tablets with the exception of the last two on ceramic and papyrus.²⁵ We can therefore compare more complete documents of this type which allow us to reconstruct the form of the original document and its layout in a codex. For example, the two deed-of-sale tablets written in the canabae of the legionary fortress Apulum at Alburnus Maior in Dacia, T.Dacia 7 and 25, both comparable to VinTab 2014-5, were found with their triptychs preserved.²⁶ Triptychs of this type were usually constructed as follows (fig. 6). An A1 tablet with recess facing inwards began the set of three tablets and contained the beginning of one version of the text, the ‘inner text’. The second tablet would have been a ‘Siegeltafel’, with a simple waxed recess on one side which faced the first A1 tablet and continued the second half of the inner text which began there. The other side of that Siegeltafel might have taken various forms, but usually had a section to receive wax seals, with space to write in ink and/or in wax on either side.²⁷ In some tablets only names of the witnesses appear beside the seals; in others, for example in Dacia, the start of the ‘exterior’, unsealed version of the text can begin here.²⁸ This ‘exterior’, consultable text then either continues onto, or begins on, a final tablet A1, with its waxed face facing the seals, and its unrecessed face ending the codex. These three tablets were hinged together via two small holes in one edge of each tablet. The first A1 and the Siegeltafel were additionally bound together and sealed, protecting the inner text. The three tablets of the triptych were then tied together around the outside. This format allowed the inner text on the confronting faces of the first and middle tablets to be sealed against tampering – adversus falsarios (Suet., Ner. 17) – while the exterior copy of the text, which occupied the third tablet and possibly began on the accessible side of the middle tablet, could be examined freely, without breaking the seal of the inner text. In this format, if the inner text fills two pages of stylus tablets, then the exterior has to be fitted into just one page after the Siegeltafel. This problem is solved to an extent if space beside the seals can be used for the start of the exterior text (though this has not yet been attested in British examples), but clearly even then the same amount of space is not available. Based on the evidence of texts on triptychs in TPSulp, it seems that the exterior copy was simply squeezed in, since there does not seem to be abridgement of the exterior text (e.g. TPSulp 25, 45, 46, 48). We have not yet been able to find any examples of abridged exterior copies in a triptych.

FIG. 6. Schematic reconstruction of a triptych used for legal documents (Tomlin 2016, fig. 17). (© MOLA/Bloomberg)

Considering the usual format of the triptych and texts of other deeds-of-sale, it seems likely that the new Vindolanda tablet is the first of a triptych and that our recoverable text begins part way through the first page of the inner text. Certainly the final line that we have been able to read ends mid-formula, so it cannot be the final tablet of the codex (unless it was completed in ink on the back, but there is no evidence to support this). We cannot completely exclude the possibility that the scribe ran out of room for the exterior text on the third (and standardly final) tablet (A1) and chose to add an extra A1 at the end of the codex to complete it, meaning that we could be dealing with the first of two A1s for the exterior text, but to our knowledge this ‘extra A1’ solution is unattested.

If we are to assume that our tablet was originally part of a triptych, the orientation of the text poses a problem. Normally, one would expect the binding holes which held the triptych together to be below the text of the first page of the inner text, but there is no sign of binding holes on the bottom edge of VinTab 2014-5.²⁹ Various explanations are possible. We know that stylus tablets were regularly reused,³⁰ and entire sets could be repurposed: diptychs, triptychs and polyptychs could be disassembled, and their constituent parts used as single tablets, for example. Sometimes pre-drilling of tablets occurred,³¹ and it is possible that they were written on before they were assembled in ‘books’. It may be that the first (and only surviving) tablet of this codex was written before the assembly of the triptych and the scribe simply made a mistake in its orientation and chose not to correct it. Or, if not pre-drilled, that the driller made the mistake. Alternatively, we could consider the possibility that this text was drawn up with the intention of making a codex, but that, for whatever reason, this was not completed, or even that this is a draft copy.

INTERPRETATION OF THE TEXT: DEEDS-OF-SALE

This tablet is undoubtedly part of the deed-of-sale of an enslaved man. The clauses preserved provided a guarantee that the man was handed over in healthy condition and not liable for any theft or wrongdoing, nor prone to run away or wander. At least 11 other documents of this type in Latin survive from around the Empire and in larger numbers in Greek among the papyri from Egypt.³² One of the most complete of the Latin comparanda is T.Dacia 6.³³ The text which includes the witnesses is as follows:³⁴

Maximus Batonis puellam / nomine Passiam siue ea / quo alio nomine est anno/rum circiter p(lus) m(inus) sex emp/ta sportellaria emit man/cipioque accepit de / Dasio Verzonis Pirusta / ex Kavieretio / (denariis) ducen/tis et quinque / sanam esse / eam puellam furtis no/xaque solutam fugi/tiuam erronem non / esse praestari quot / si quis eam puellam / partemue quam ex eo ³⁵ / quis euicerit quo / minus Maximum Ba/tonis quoue ea res [pertinebit habere possidereque recte liceat tum quanti ea puella empta est tam pecuniam et alterum tantum dari fide rogauit Maximus Batonis fide promisit Dasius Verzonis Pirusta ex Kaviereti proque ea puella quae s(upra) s(cripta) est (denarios) ducentos quinque accepisse et habere se dixit Dasius Verzonis a Maximo Batonis actum Karto XVI K(alendas) Apriles Tito Aelio Caesare Antonino Pio II et Bruttio Praesente II co(n)s(ulibus)] // Maximi Ve/neti princi/pis / Masuri Messi / dec(urionis) / Anneses An/dunocnetis / Plani Verzo/nis Sclaietis / Liccai Epicadi / Marciniesi / Epicadi Plaren/tis qui et Mico / Dasi Verzonis / ipsius uendi/toris

Maximus, the son of Bato, has purchased and received by mancipium the girl with the name Passia, or known by any other name, who is more or less six years old, having been purchased as a foundling, from Dasius, the son of Verzo, the Pirustian from Kavieretium, for two hundred and five denarii. It is warranted that the girl is healthy at this time and free from deceit and wrongdoing, and that she is not a fugitive (or a) wanderer. Maximus, the son of Bato, had demanded in good faith that, if anyone lays legal claim to this girl, or any part of her for any reason, so that it would not be right for Maximus, the son of Bato, or whomever this matter pertains to, to have and possess her justly, then as much as the girl was bought for, so much and so much again be given, and Dasius, the son of Verso, the Pirustian from Kavieretium, has promised to do so, in good faith, and for the girl who is mentioned above Dasius, son of Verzo, declares that he has received two hundred and five denarii from Maximus, the son of Bato. This was done at Kartum sixteen days before the kalends of April, when Titus Aelius Caesar Antoninus Pius and Bruttius Praesens were consuls, both for the second time. [The seals of] Maximus Venetus, princeps; Masurus Messus, decurion; Anneses, son of Andunocnis; Planus, son of Verzo; Sclaies, son of Liccia; Epicadus, son of Marciniesus; Epicadus, son of Plarens, who is also known as Mico; and Dasus, son of Verzo, the seller himself.

These documents are formulaic, but vary in their detail and the specific terms and stipulations of the exchanges they record. This Dacian tablet, however, contains almost all the elements that are found in the corpus as a whole. table 1 shows the distribution of common elements found in the known Latin deeds-of-sale, with the exception of TH² 59 which is of a different type. The numbering indicates the order in which the surviving elements appear in each text.

Table 1 elements found in the known latin deeds-of-sales, excluding th² 59.

* For this phrase, see Watson 1963.

Unfortunately, the top of the Vindolanda tablet is lost and it is impossible to be sure how much text is missing. However, comparison with complete similar tablets allows us to suggest that five or six lines have been lost. We may be relatively confident that our text began with the name of the purchaser and contained the verb emit. A short description of the enslaved person being sold including his name, and the phrase siue eum quo alio nomine est, would probably have followed. It may have provided the enslaved person's origin. Next would have come the name of the seller and the agreed-upon price. The formula we have recovered would then have appeared.

The close connection between this formula and the Edict of the Curule Aediles has long been recognised. The edict, as preserved in Justinian's Digest, includes the following passage:³⁶

qui mancipia uendunt certiores faciant emptores, quid morbi uitiiue cuique sit, quis fugitiuus erroue sit noxaue solutus non sit: eademque omnia, cum ea mancipia uenibunt, palam recte pronuntianto. quodsi mancipium aduersus ea uenisset, siue aduersus quod dictum promissumue fuerit cum ueniret, fuisset, quod eius praestari oportere dicetur: emptori omnibusque ad quos ea res pertinet iudicium dabimus, ut id mancipium redhibeatur.

Those who sell slaves are to apprise purchasers of any disease or defect in their wares and whether a given slave is a runaway, a loiterer on errands, or still subject to noxal liability; all these matters they must proclaim in due manner when the slaves are sold. If a slave be sold without compliance with this regulation or contrary to what has been said of or promised in respect of him at the time of his sale, it is for us to declare what is due in respect of him; we will grant to the purchaser and to all other interested parties an action for rescission in respect of the slave.

The Roman jurists Ulpian and Gaius both interpreted this passage as a directive that vendors should declare any physical or legal conditions relating to the enslaved person they intended to sell.³⁷ This was clearly a well-known regulation since it is cited or paraphrased by Cicero (Off. 3.17.71), Seneca (Controv. 7.6.23) and Aulus Gellius (NA 4.2.1).³⁸ Such a guarantee need not have been included in a deed-of-sale, however. Ulpian himself also wrote pacisci contra edictum aedilium omnimodo licet, siue in ipso negotio uenditionis gerendo conuenisset siue postea ‘It is quite lawful to make a pact contrary to the edict of the aediles, whether the agreement is made in the course of arranging the sale or afterward’ (Dig. 2.14.31, translation Watson 1998). The edicts provided assurance that such statements about the conditions of the individuals being sold would be made, honoured and enforceable, not that they must appear in a contract.³⁹ The text preserved on VinTab 2014-5 is just such a contractual statement, expressed in language strikingly similar to the Edict of the Curule Aediles and other documents; compare quis fugitiuus erroue sit noxaue solutus non sit in the edict and the reconstructed furtis noxsisque solutum fugitiuum erronem in VinTab 2014-5. The Edict of the Curule Aediles is even mentioned explicitly in a number of deeds-of-sale.⁴⁰ Focusing on the Greek papyri, however, Arzt-Grabner notes that ‘comparably few slave dealers include such a guarantee [about flight or truancy] in the contracts’.⁴¹ He was able to cite only seven examples, including just two Greek texts on papyri.⁴² Now, with the addition of the examples from London and Vindolanda that total rises to nine, and seven of these are on wax stylus tablets from the West. Indeed, all ten of the wax tablets in table 1 contain some form of guarantee, if not always about flight, of the enslaved person's condition.

THE INDIVIDUALS IN THE TRANSACTION

The inclusion of a guarantee for the enslaved man who was subject to this transaction allows us to assume that he was probably healthy and not likely to cause trouble. The use of homo in line 3 of our text identifies him as a grown man. homo now appears twice in Latin deeds-of-sale; the other example occurring in the highly fragmentary text on pottery from La Graufesenque (Marichal 1988, no. 211). The sherd preserves even fewer words than on VinTab 2014-5, but given the content, it is very likely to be a deed-of-sale:

].uanto oppa[

]ẹmit homineṃ [nomine

s]iue is quo alio no[mine est

… has bought a man [by name …] or by whatever other name he is known

In the eight other Latin deeds-of-sale where a such a qualifying description is used, six of the individuals are female (five puella, one mulier), and only two pueri occur.⁴³ Pueri probably refer to enslaved males in several ink tablets at Vindolanda, including Tab. Vindol. 642, in which the pueri are labourers, and unlikely to be children.⁴⁴ Similarly, at La Graufesenque an account of the tasks of a list of enslaved males over a period of days is headed Ateliae puerorum, ‘the pueri of Atelia’, and again it seems improbable that these workers, collecting wood, going to market and making the pottery, were all boys (Marichal 1988, no. 169). However, in the deeds-of-sale more precision was necessary for the description of the person, with the maturity of the servus being relevant, so in these legal documents we can be more certain that a puer refers to a young male (up to around 14),⁴⁵ and a homo to one who had reached physical maturity.

The precise age of our adult-male servus, however, and indeed his origins, are unknown.⁴⁶ The guarantee and the mere existence of this document suggest that he was of perhaps more than average value. Whereas ‘leaf’ tablets in the North were the everyday medium, arguably stylus tablets tended to be associated more often with official and legal texts.⁴⁷ They seem to have been a less widespread medium at Vindolanda; despite their increased chance of preservation compared to the thinner leaf tablets, fewer have been found over time, though as climate change is adversely affecting the preservation of organic materials at the site, the proportions of leaf compared to stylus tablet seem to be dropping.⁴⁸ Having one of these documents drawn up on a triptych, with the appropriate clauses to protect the buyer, and witnessed and sealed, was not a minor undertaking, and would probably not have been an option for less well-educated or connected buyers.

The price that our enslaved man fetched has not been preserved.⁴⁹ The surviving Latin documents that record guarantees state prices ranging from around 200 to 1000 denarii (see table 1), and other document types from the Bay of Naples record similar prices of 600 sesterces (150 denarii) for a girl (TH ² 65) and 1300 (325 denarii) for a woman (TH ² 74). Several factors, including age, appearance, abilities and character would have affected the price, as did local market forces,⁵⁰ and there must have been sometimes significant variation which is not demonstrated in this small pool of Latin-language deeds. It is notable that the prices for the puella at London and the puer in T.Dacia 7 come to 600 denarii, and that the six-year-old puella and the seven-year-old puer attested in the documents written in Kartum (T. Dacia 6) and Seleucia Pieria (P.Lond. II 299) fetch only around 200. We might expect that the enslaved man recorded at Vindolanda fetched a price in the region of 1000 denarii.

Given the archaeological and historical context, it is logical to associate this tablet with Iulius Verecundus, the prefect of the First Cohort of Tungrians, which occupied the fort in periods I/II (c. 85–100 c.e.). The commanders of auxiliary units were equestrians and as such had to fulfil a property requirement of 400,000 sesterces, and, therefore, would presumably have had ample funds to make such a purchase. While Verecundus is the most obvious purchaser in this case given the proximity of other documents explicitly related to him, we have to admit that the archaeological context for this tablet cannot definitively single him out, and there were no doubt other more junior officers, veterans or civilians who could have completed this transaction.⁵¹ The tablet does not preserve the circumstances of the sale, and offers no clues as to the legal status of the buyer or seller. But we need not even assume citizen status. The buyer and seller of T.Dacia 6 were both peregrines, as was the buyer, Dasius Breucus, of T.Dacia 7.⁵² Strikingly, Vegetius, the buyer by mancipatio of Fortunata in the deed-of-sale from London, is ‘assistant slave of Montanus the slave of the August Emperor’.⁵³ Wierschowski demonstrated that rank and file soldiers were eager to become slave-holders and Linden-High found that auxiliary soldiers comprised 10 per cent of the slave-owning soldiers in the documentary evidence.⁵⁴ M.P. Speidel has also argued that many, if not all, cavalrymen in the auxilia of the early imperial army had servants, and seems to take for granted that many or all of them were enslaved people.⁵⁵ Though our knowledge of auxiliary soldiers’ pay is incomplete, M.A. Speidel argued that rank and file auxiliary soldiers earned 1000 sesterces per year in the relevant period, while sesquiplicarii and duplicarii earned 1500 and 2000 sesterces.⁵⁶ These salaries were subject to various deductions and expenses, but soldiers’ remaining pay was sufficient to allow slave-ownership.⁵⁷ The lowest cost we have recorded, 200 denarii, represents two to four years of expendable income for an enlisted man, but soldiers are known to have augmented their incomes and enslaved persons had the potential to be profitable investments.⁵⁸ All that being said, however, the very existence of the formal legal document and its find-spot associated with Verecundus suggests this deed-of-sale is more likely to record a servus of the prefect.

THE SERVI OF THE COMMANDERS AT VINDOLANDA

The Vindolanda writing tablets provide a number of glimpses into the experience of enslaved people on the site.⁵⁹ Several commanders’ servi appear in the tablets as authors and recipients of communications. In one letter (Tab. Vindol. 301) a cornicularius named Severus addresses Candidus, who may have been owned by the prefect Genialis.⁶⁰ This letter demonstrates both the education of, and warmth afforded to, certain enslaved men. It appears to be in Severus’ own hand and is addressed to Candidus directly with the possessive adjective suo, suggesting that both men were educated and may have been on friendly terms. The subject matter, payments for preparations for the celebration of the Saturnalia, reinforces the notion that servi had some authority and freedom in the conduct of business and personal matters.⁶¹ The celebration of the Saturnalia implies some limited licence in their relationship with their owners, at least for the few days of the festival itself.⁶² Severus’ use of the friendly and egalitarian epithet frater for a servus also gives a sense of the relative status select enslaved people achieved in some contexts.⁶³ Similarly, Iustinus, who may himself have been a prefect, sent greetings to Cerialis’ pueri, by which he may have meant enslaved people.⁶⁴ Most relevant for our purposes, however, a servus of Verecundus himself is attested in Tab Vindol. 302, recovered from a Period 1 ditch. The name is missing, but it concerns the acquisition of foodstuffs, and seems to represent correspondence between domestic staff.

Letters directly from Vindolanda's commanders to their servi and liberti employ a rather sterner tone with imperatives in place of more polite language. This is most visible in Tab. Vindol. 890 in which Verecundus uses a ‘peremptory tone’ towards his servus, named Audax, for having sent the wrong key to a box.⁶⁵ An Audax appears in another document, difficult to date, alongside several other names in a list of items required by the cook (Tab. Vindol. 590). Whether it is the same enslaved person is unclear. Similarly firm language appears in Tab. Vindol. 616, in which the editors suggest that Cerialis orders his staff of enslaved people and freedmen to make suitable preparations for the arrival of guests. While these letters do not display any particular animus between dominus and servus, they reflect the hierarchies of daily life.⁶⁶

A further stylus tablet from Vindolanda, which was provisionally published in 2005 as part of an earlier project testing imaging techniques, is currently under re-examination and may also be related to the servi of the commanders.⁶⁷ Unusually, the tablet is complete (VinTab 974, BM 1993,1103.4) (fig. 7) and was discovered in a workshop or smithy at the north end of the fabrica that constituted the west wing of the period III praetorium at Vindolanda (c. 100–105 c.e.).⁶⁸ This room also contained three other unpublished stylus tablets and seven fragments of ink tablets, including one tablet which may mention a stolen balteus and servi, but whose interpretation is uncertain (Tab. Vindol. 322).⁶⁹ The discovery of the tablets here suggests that they should be associated with the household of the garrison's commander, Flavius Cerialis. The stylus tablet (VinTab 974) was first thought to record the sale of two servi.⁷⁰ Although it does mention an enslaved person, subsequent examination has revealed that this document is of a different type from VinTab 2014-5. The reading is complicated by the fact that there are traces of more than one text in the upper part of the tablet and some of the traces are faint, but the text that has been relatively securely recovered to date is as follows:

Traces of 4–5 lines

Batauorum ciue meo Bello-

uaco ser(u)um nomine Verecun-

8    du(m) ciue(m) Ambianis et dedi per-

missione(m) et uecturas [written over uecturas] …

triginta quinque et eum

ser(u)um nutriui annos

12  dece(m) quinque

of the Batavians (?) … my fellow-citizen of the Bellovaci [name and verb lost] a slave called Verecundus, citizen (?) of the Ambiani people. And I have given permission and travel expenses (?) … thirty-five; and I have nurtured that slave for fifteen years.

This text is somewhat perplexing. It mentions a servus and we might be tempted to assume it may be a receipt or something akin to a deed-of-sale, but the phrase dedi permissionem ‘I have given permission’ seems to suggest something different, and we wonder whether it could be a formal permission note which an enslaved person might carry to ensure safe passage – the sort that the author of Tab. Vindol. 645, Maior, seems to be requesting of his correspondent, Cocceius Maritimus, in the margin of his letter: ‘if you intend to send a puer to me(?), send a hand-written note with him so that I may be more reassured’.⁷¹ If so, the uecturae may be covering the cost of the journey, the details of which we have no knowledge, but thirty-five days might be a reasonable estimate for the time to complete an over-land trip to or from Vindolanda and Beauvais or Amiens, civitas capitals of the Bellovaci and Ambiani respectively.⁷² It proved hard to find parallels for the use of civis to describe a servus, but it is likely that it is simply deployed here in a loose sense to refer to belonging to a specific geographical community (rather than to a specific type of legal status), in the same way that it is regularly used by peregrines (e.g. the civis Trever at Lancaster in RIB III 3185). Another possibility is that the person in question may be an alumnus/nutritus, in which case he may previously have been a free-born civis (given up or orphaned by citizen parents) before becoming a servus or may have attained free-status after being raised as a servus.⁷³ The reference to the period of nurturing may indeed suggest that this may be some kind of letter of recommendation document.⁷⁴ A further problem may be raised by the syntax of the phrase cive(m) Ambianis, in which a genitive plural of the community name or an accusative adjective might be expected. Improved readings may eventually resolve some of these issues, and perhaps more text will be revealed. For example the term uicesima, ‘a twentieth’, sometimes used to refer to the tax levied on the manumission of enslaved people, can now be recovered in the hardest-to-read section at the top of the document. Whether this is a part of the same ‘permission/recommendation’ document or not is unclear, though the divergent size of the lettering suggests it belongs to another. So here is a second stylus tablet from Vindolanda which may well be related to the documentation of the servi of the commanders of the garrison, but which has not yet given up its secrets.

FIG. 7. A high-resolution image of VinTab 974 with raking light. (Taken at the British Museum in 2022 by A.W.M. and A.L.M.)

CONCLUSIONS

The lower section of a single stylus tablet, VinTab 2014-5, is the surviving part of a triptych which contained the deed-of-sale of an enslaved person at Vindolanda in the late first century c.e., only the second known from Roman Britain.⁷⁵ After multiple techniques revealed the text on the fragment, comparison with other known deeds-of-sale led to the reconstruction of the original document. The archaeological and historical context, and the deployment of the legal medium itself, suggest that the purchaser may have been Iulius Verecundus, the commander of the Tungrian garrison. The text can be added to other evidence of the ownership of servi by the commanders at Vindolanda and provides further insight into the role slavery played in provincial society. It demonstrates that a legal formula originating in second-century b.c.e. Rome could be applied to transactions in the furthest reaches of the Empire. It raises the questions of how law worked on the northern frontier and among a military community, analogous to those raised by the Dacian tablets. Others may want to consider how this might fit into the increasingly detailed picture of provincial legal practices.⁷⁶

Stylus tablets were often used for legal documents in the West and therefore have the potential to make significant contributions to our knowledge of their communities and practices. We have attempted to show the benefits of painstaking examination of these difficult-to-read wooden objects. As technologies such as RTI, 3D scanning, photogrammetry, photometric stereo, and handwritten text recognition to detect and classify characters continue to evolve, new and improved techniques will emerge that should aid the recovery of more texts from Roman-period stylus tablets. Meanwhile, careful examination of the unpublished and problematic tablets from Vindolanda using a combination of existing technologies is already yielding meaningful results. The deed-of-sale presented here is just the first fruits of this labour.