This article identifies a military prison (carcer castrensis) in the Roman legionary fortress at Lambaesis (Tazoult, Algeria) and contextualizes the space among North African carceral practices evidenced in epigraphic, papyrological, and literary sources of the first through fourth centuries CE. The identification is made on the basis of architectural comparanda and previously unnoticed inscriptional evidence which demonstrate that the space under the Sanctuary of the Standards in the principia was both built as a prison and used that way in antiquity. The broader discussion highlights the ubiquity of carceral spaces and practices in the ancient and late ancient Mediterranean, and elucidates some of the underlying practices and ideologies of ancient incarceration.
Around 120 CE, Hadrian established Lambaesis (modern day Tazoult, Algeria) as the permanent castra of the Third Legion “Augusta.”2 The emperor personally visited in 128 CE, and except for a short interlude between the years of 238 and 253, the base continued for over two centuries as the center of operations for the Third Augustan Legion.3 The castra resembles many other Roman legionary bases of Hadrianic date, both in its general layout and in many particulars.4 One feature, however, appears to be unique to this site: typically the temple at the center of a castra was intended to hold the standards of the legion, and sometimes it served also as a treasury (aerarium). The temple at the center of the principia in Lambaesis, however, is significantly different from known treasuries under legionary temples (sacella). We argue that, in its initial phase, it was not a holding space for coin. While the upper room served as a repository for the legionary standards, the evidence indicates that the lower room served as a repository for bodies: it was a prison. The space under the temple includes cells for separating prisoners, worn embrasures opening to the street allowing for light and small objects to pass to and from inhabitants below, doors architecturally conducive to keeping people in rather than keeping people out, and over twenty rudimentary graffiti, across every chamber, indicating the presence of people spending ample time in the space with poor tools.
In this article, we identify the space as a Roman military prison (carcer castrensis) and argue that paying it close attention helps us to visualize, contextualize, and better understand other carceral practices from Roman North Africa, which we explore in the second half of this article. Perhaps most importantly, analysis of the military prison at Lambaesis refocuses our gaze on the institution of incarceration—a central facet of both civic and military life in antiquity.
II. Archaeological Description
The Sanctuary of the Standards (sacellum/aedes signorum) at Lambaesis stands on the southern side of the principia complex, opposite the groma and between the praetorium to the west and the barracks of the 1st cohort to the east (Figure 1).5 Under the supervision of six French colonial guards, a group of 95 Algerian prisoners excavated the principia in the spring of 1897 and 1898.6 Friedrich Rakob subsequently conducted stratigraphic soundings across the complex in four campaigns between 1968 and 1973.7 The principia complex exhibits two distinct phases that Rakob assigned to (1) the initial construction of the castra under Hadrian, and (2) its renovation under Gallienus. The sanctuary sits on a small pediment, and its entrance abuts a portico along the south side of the principia (Figure 2).8 As described by Cagnat, west of the sacellum are two apsidal rooms (Figure 3, rooms II and III), and it is flanked on the east by one smaller apsidal room (2) and one rectangular room (3), all of which have been identified as scholae—meeting areas for legionary collegia.9
The first phase is coterminous with the construction of the castra, around 120 CE (Figure 4). The sacellum complex was constructed by excavating 2.5 m below the ground level and building revetments on all sides directly against the native soil. Walls of the initial, Hadrianic phase are built in opus africanum; sandstone ashlars interspersed with lighter, Ain Drinn-Tuff were used above ground, and underground, sandstone field stones and Ain Drinn-Tuff are bound with lime mortar.10 The subterranean space is divided into five chambers of 1.5 m width (east-west), 4.5 m length (north-south), and 2.3 m height, constructed in the same style as the outer walls dating from the initial phase (Figure 5). Cantilevered sandstone slabs span the space between the field-stone walls, and a corridor to the north connected the underground chambers with an 0.8 m wide sandstone block staircase that descends five steps from the portico above, forming the single access point for the underground complex and originally sealed with a “stout locking device” (Figure 6).11 A threshold remains extant on the third stair from the surface, where a post hole for a hinge is visible along with a recessed wall built to accommodate the crossbeam of a large door that opened into the underground space (Figure 7).
The temple pavement rises 0.3 m above the pavement of the portico to its north, and comprises rectangular sandstone blocks with central notches (perhaps related to the storage of legionary standards). It originally held a statue or altar, the 0.9m2 base of which is currently at the center of chamber C. The temple pavement serves also as the ceiling for the lower chambers, with the blocks spanning the distance between the outer walls and the chamber dividers, as well as between chambers (Figure 8). We cannot say anything certain about the upper walls of the first phase because it has been mostly replaced by the second, but it is likely that the above-ground walls were constructed in opus africanum as elsewhere in the principia; the upper-level pavement remains intact from the original Hadrianic construction.
Each underground chamber exhibits a curious feature: an embrasure on the upper corner of the south wall, which remains from the initial phase of construction (Figure 9). Each embrasure was created by hewing roughly conical halves in sandstone blocks before fitting them together (Figure 10). These embrasures connected each underground chamber with the street running behind the sacellum and allowed for the passage of light, air, sound, and small objects from above to the spaces below without needing to open the door inside the principia complex. Curiously, the embrasures on chambers A and E are of the same initial design as those in B–D, but their openings are significantly smaller. (Compare Figures 11 and 12.) By comparison we can see the effects of wear on the embrasures in chambers B–D, and we learn that the inner chambers were apparently more frequently used, or visited, than those on the outer walls of the complex. Cagnat discovered hinges for doors separating chambers A and E from the central access corridor. It is likely that the central chambers (B, C, and D) had doors as well, but the lower pavement does not appear to be intact; no excavator mentions anything about it, and it is currently covered by nearly 50 years of sedimentary deposits.
In a second, Gallienic phase, underground walls on the north and south of chamber E, and on the south of chamber A, were reinforced with modeled sandstone ashlars, presumably to bear the weight of the ashlar walls of the upper level in the second phase (Figures 5 and 13).12 Above ground, the same ashlars are extant in one to three courses, approximately one meter in width. In antiquity these walls were covered with plaster on the inside and decorated with red paint, some of which was discovered during excavations in the late nineteenth century.13 In the second phase, apses were added to the sacellum and the scholae flanking on the east and west, and the street running behind the sacellum was raised some 60 cm, covering the embrasures and rendering them inoperable.
We discovered graffiti in all chambers of the temple’s lower level. Most graffiti comprise simple geometric scratches in the field stone wall, generally between the current ground level and approximately one meter above it. It is likely that more graffiti remain on the upper parts of the walls, but in the years since initial excavation, sinter has obscured the upper courses, rendering visual identification of further graffiti impossible. Many of the visible graffiti resemble tables, comprising lines parallel and perpendicular to the pavement. This fact helped us discern in doubtful cases, where the lineated fabric of the local sandstone caused lines that appear to be anthropogenic. Cases where inscribed lines continue across adjacent stones, however, confirm that these markings cannot be natural. The presence, as well, of an inscription with the Greek letter Π (followed by a fainter marking that may be the Greek letter O) confirms that these are graffiti, and not simply naturally occurring lines in the rock (Figures 14 and 15).14 A Latin inscription is visible on the door frame separating chambers D and E reading “LAM”: likely the first three letters of the ancient city name (Figure 16). Ultimately it is not the content of the graffiti, but their sheer number, that is important for our argument.
In both phases of the principia’s history there were five subterranean chambers directly below the sacellum/aedes signorum, with a walkway connecting the chambers to each other, and ultimately with the stepped entrance leading north to the portico and the central courtyard. The underground complex has previously been interpreted as a treasury (aerarium)—a site for storage of treasures or for coin used to pay soldiers of the Third Legion.15 While strongrooms under aedes signorum are commonly assumed to be treasuries, Lambaesis presents features that resist such an identification. And, as we suggest below, in almost all cases the identification of “strongrooms” as treasuries is made on the basis of partial evidence, lacking contextualization. We argue that the archaeological remains demonstrate that the space was not used as an aerarium, but a place of surveillance where the army detained humans for periods of time with little access to tools. That is to say, it was a military prison. What follows is the evidence pointing to this conclusion.
Graffiti discovered in the chambers suggest something significant about the use of the underground space. Each was inscribed by someone who spent substantial time there, and who had only rudimentary tools with which to mark the dense sandstone walls. Geometric tables are most numerous among the inscriptions that we discovered, and many graffiti bear a close resemblance to prison graffiti from late ancient Corinth. Compare, for instance, Figures 15 and 17. Furthermore, while only a couple of letters, a Greek inscription on the western wall in chamber A reading ΠỌ resembles a similar graffito from Corinth asking the deity to give an awful death to their opponents who caused them to be in prison.16 Poor preservation of the Greek inscription at Lambaesis prevents us from knowing what this person intended, but we can be confident that they had some knowledge of written Greek, poor tools, and ample time to scratch letters into an unforgiving surface.
Prison graffiti of tables have been interpreted elsewhere as game boards (tabulae lusoriae), or perhaps instruments aiding captives in keeping time (Figure 15).17 We discovered over 20 graffiti of this type in the five chambers, and each has a distinctly unprofessional feel about it. That is to say, they are not the kinds of things one might scratch into the walls of a treasury, where there is little reason to believe that someone would linger for extended periods of time. These graffiti, however, fit well in a site of incarceration, where people with various levels of literacy spent ample time with little to do.
We discovered a brief Latin inscription, as well, in the corridor that connects the five chambers with the stairwell leading to the surface. This inscription is neater than those found in the cells, but nevertheless its execution does not imply that the inscriber was wholly literate. It reads “LAM,” the first three letters of the city’s name, and was scratched into the lintel separating chamber E from the central corridor (Figure 16). Again the brevity of this inscription precludes us from saying anything significant about its historical context, but the presence of serifs on the vertical stroke and horizontal crossbar of the L suggest some familiarity with epigraphic habits found elsewhere on the site, and the letter resembles the typical “L” seen across the site announcing the name of the Third Legion.
Second, the presence of embrasures opening to the street presents problems for an identification of this space as a treasury (Figure 10). The embrasures are a critical component of the original design of the subterranean structure; two adjoining stones were hewn to form a connection between the north side of the apse above to the chambers below, and the embrasure widens conically as it extends into the underground space (Figure 10). These openings would have allowed air, light, and sound to enter, and could be used for the passage of small items directly from the street running behind the aedes signorum to the space underneath the standards. It is hard to imagine a treasury being rendered more effective with the addition of such holes, and we have no evidence for anything similar in securely identified strongrooms elsewhere.18 The conical opening is especially puzzling if this space were intended as a treasury. Such a design does not allow appreciably more light or air to enter the space; the only clear benefit is that people on the inside can more easily receive objects passed through the opening. It seems that all five embrasures initially had fairly small, rectangular openings of approximately 10 cm, as we see in Figure 11: the opening between chamber A and the street which shows minimal wear on its outer surface. On the other hand, the openings in chambers B–D are irregular, and each is roughly 20 cm in diameter. These were apparently dilated through wear to a point where they were wide enough to allow an arm to pass through, but not large enough for a body. (See Figure 12)
The presence of such access points, and their apparent wear through use, militates against the identification of this space as a treasury. If the structure were intended as a prison, however, the function of such embrasures becomes immediately apparent: they would allow easy surveillance of and care for captives without the need to unlock the door or enter the underground space, or even to enter the principia complex. Feeding prisoners was a constant source of anxiety in ancient Roman prisons, and embrasures would also have allowed the passage of sustenance to prisoners below.19 An inscription from Lambaesis’s closest neighbor, Thamugadi (Timgad, Algeria), stipulates duties incumbent upon the community, including providing food to locals awaiting trial: those “in carcere…ab ordine civitatis.”20
The embrasures would also have served as small windows, facing south as Columella recommends for prisons, relatively out of reach, and allowing necessary ventilation for the people stored below.21 In fact, in his generalized depiction of second-century Roman prisons, Calpurnius Flaccus indicates that light reached prisoners through “narrow openings” precisely. “I can see the state prison (carcerem publicum), constructed of huge stone blocks, receiving just a faint trace of light through narrow openings (angustis foraminibus tenuem lucis umbram recipientem).”22 While such access points are unprecedented in legionary strongrooms, the municipal carcer at Cosa boasts a similar feature, with what must have been an identical function: providing the jailer or members of the public secure access to prisoners, and allowing the faintest semblance of light to reach deep underground (Figure 18).23 We can situate the use of these embrasures directly in the context of Roman carceral practices, and even the carceral practices of Roman North Africa as attested by local inscriptions. Their presence in an aerarium would be counterintuitive, to say the least. No known treasury has embrasures which resemble those under the aedes signorum at Lambaesis, nor is there reason to think that a treasury’s security or functionality would be increased by the introduction of such passageways apart from the door. Coins do not need to breathe, after all, and windows do not render treasuries more secure.
Third, we know that a number of events took place in Roman legionary bases apud signa—“with the standards”—including incarceration. As at all legionary castra, the forum adjacent to the aedes signorum in Lambaesis was used as a mustering ground; even today a monumental altar stands in the center of the complex, reading ara disciplinae.24 Anne Johnson rightly notes that, according to Vegetius, wages for the soldiery were regularly held “with the standards” (apud signa) or “with standard-bearers” (apud signiferos)—presumably in strong rooms adjacent to or underneath the Sanctuary of the Standards, as we have in the principia at Chester’s on Hadrian’s wall.25 Vegetius’ predecessor Ammianus Marcellinus, however, tells us twice in his History that soldiers are typically incarcerated apud signa. In book 21, he mentions that the Roman general Vadomarius was “handed over to the commander of the soldiers, to be held at the standards” (rectori militum arte custodiendum apud signa commisit), and in book 26, Ammianus indicates that “common soldiers” (gregarii milites) were typically “locked up at the standards” (coericiti sunt apud signa).26 It is true that legionary treasure was often kept near the standards, as Vegetius claims. But lots of things happened in legionary camps apud signa. We argue here that, if scholars allow that Vegetius’s comment is not the sole and unimpeachable guide to interpreting archaeological sites, the intention and function of nearly all structures previously identified as treasuries are open to reconsideration. We contend that scholars should return to all such subterranean “strongrooms” with fresh eyes and a wider range of comparanda.
Fourth, treasuries are rarely divided into multiple chambers. For instance, the treasury of Messene is c. 2.5 m square without distinct holding areas, and only large enough to secure a small lock box or a few stacks of treasure.27 Moreover, the doors securing chambers from the walkway to the north create individual cells, not simply underground arcades. In fact, literary and documentary sources indicate that prisons were often designed precisely to allow separation of prisoners into cells. Consider, for instance, Suetonius’s Life of Tiberius 61, which assumes a prison with the capability to isolate prisoners in a condition similar to what we might call solitary confinement today; namely, without books to read or the ability to converse with one another. Additionally, the Acts of Paul and Thecla 17–20 assumes that there is one guard outside the prison structure itself, and another guard inside the prison overseeing its individual cells, in which different people were kept; in this case, Thecla is depicted bribing her way past both sets of guards to enter the cell where Paul was being held. Literary sources suggest that different types of prisoners were expected to be kept separate—an issue which may have been particularly relevant in a military context, where prisons would hold political, military, civil, and criminal captives.
In fact, this is not even the only North African military site with a carceral space purpose-built to keep captives detained in separate chambers or blocks. Simitthus (modern-day Chemtou, Tunisia) lies 300 kilometers northeast of Lambaesis and was a younger contemporary of the castra.28 The site is home to a Roman military outpost that oversaw the extraction of giallo antico, the famed yellow Numidian marble. As an army establishment, the site is both remarkable and atypical. According to Friedrich Rakob, who excavated there in the 1960s and 70s, the site was meant simply to hold land and extract resources, not to muster troops or to project military might.29 It had a forum, a theater, a modest bathing facility for the soldiers stationed there, and one more thing: quarters for captives who worked in the quarries—enslaved people and prisoners condemned ad metalla.30 While the ergastulum (workers’ quarters) at Simitthus is significantly larger than our carcer castrensis at Lambaesis, both have the same overall design and, apparently, the same intention. Similar to our carcer, Simitthus has six long, hermetically sealed chambers overseen by guard towers and accessible through a single entrance (Figure 19).31 In the first two phases, from the late second century to the end of the third, stout doorframes opened to long benches which served as sleeping spaces for those inside—perhaps as many as 180 people in each chamber.32 While the earliest phase included no obvious latrine, in a second phase each chamber was given its own combination water source/bath/toilet, proving that the space was purpose-built to house people kept under guard.33 At Simitthus, like at Lambaesis, the principle of hermetically sealing captives from one another was so ingrained that, even when the space was renovated in the late Severan period, bringing running water and sewer to those living inside, six individual latrines were constructed, with six separate sewers, rather than a communal latrine or individual latrines which emptied into a common sewer. Further, each sewer has a deliberately minuscule opening: small enough that it would be impossible to escape, even if someone confined inside was willing to crawl through the muck toward freedom. At Simitthus, the ideology of carceral separation determined the architecture of the space and its renovation, even at the added cost of separate facilities for the different types of prisoners. As Rakob notes, the difference between the open baths provided for the soldiers nearby and the hermetically sealed toilet-cum-water source in each chamber of the ergastulum offers an “indication of the differing quality of their users.”34 Or, in the words of Michael Mackensen, who re-excavated the site in the 90s, “Apparently these are prison-like, rather than barracks-like, accommodations.”35
The carceral spaces at Lambaesis and Simitthus are different: one is a small carcer, while the other is a sizable ergastulum.36 But, both are second-century North African military installations with similar forms and similar functions. The design of the spaces appears to reflect a similar ideological motivation, which held that prisoners of different types were not supposed to mix. While the archaeology itself does not speak to the particular ideological distinctions at play, the architects at Simitthus and Lambaesis went to great lengths to separate prisoners into cells. Theodosian Code 9.3.3, a later constitution, but one that may shed some light on typical practice in this earlier period, indicates that prisoners were separated according to the type of crime committed and legislates further that prisoners should be separated by sex. Writing while the Simitthus ergastulum was in use, Ulpian indicates that women condemned ad metalla were sent to be aids to the (male) miners (in ministerium metallicorum).37 Were the prisoners at Lambaesis, Simitthus, and similar North African establishments in the Principate separated based on sex?38 Were they separated based on class or freedom/enslavement? Based on the charging authority, or the reason for their confinement? We do not know. But what is clear in both cases is that an ideology of separation was at play in the initial design of both of these spaces, even if we do not know how each particular space worked in practice.
Finally, the remains of the threshold at Lambaesis suggest an entry point more suited for a site of incarceration than a treasury. The difference between the two is subtle. Treasuries and prisons were both criminalized spaces: their architecture responds to perceived criminality. The difference is this: treasuries were designed to keep people outside of an area; prisons were designed to keep people in. The entrance to the subterranean chambers in Lambaesis is a stout structure, with a door opening to the inside and resting against a significant rock cut backstop (Figure 6). It would have been impossible to push the door open from the inside. By itself, the structure and orientation of a threshold is indeterminate. But in this context, the threshold’s construction suggests an intention to keep people in, and thus lends further credence to the identification of the space as a carcer.
In sum, the structure underneath the aedes signorum at Lambaesis does not conform to the typical features of a treasury, but it bears all the marks of a prison. There are rudimentary graffiti in all of the chambers, indicating the presence of individuals with significant time and poor tools. Embrasures which show signs of significant wear allowed for the passage of air, light, and small objects from the street behind the sacellum directly into the underground complex—features that would be counterproductive in a treasury, but which are attested in numerous literary sources for prisons, where friends and relatives are tasked with delivering provisions to incarcerated people. There are multiple chambers that communicate only with a common hallway, a feature expected from sources describing prisons and paralleled in another military carceral context of second-century North Africa, but without parallel in a securely identified treasury. Finally, the space features a threshold that is more secure against being pushed open from the inside than from the outside, suggesting that the space was meant to contain bodies rather than to keep them out. The architectural features of this space point to its function as a place of incarceration. The inscriptional evidence points to its use as a space of incarceration. Together the data allow for a highly probable identification.
It is not an accident, either, that the structure lies directly underneath the legion’s Sanctuary of the Standards. The placement of the prison itself signifies the might of the Roman army, and invokes cultic duties to the preservation of order. The space demonstrates materially the power of the Roman military over captive bodies, an oppositional connection well established in other North African carceral spaces. Consider, for instance, the case of Roman Cuicul (modern-day Djemila, 100 km northwest of Lambaesis): the city’s main prison lies directly underneath the civic basilica, which is to say that the seat of judgment was physically constructed directly over the site of incarceration.39
In Roman North Africa, even when carceral sites were not located physically underneath the space of judicial deliberation, the conceptual connection remained. At the Roman colonia of Tipasa, for instance, on the Algerian coast, the apse of the civic basilica has a mosaic pavement of the second half of the second century CE, contemporary to our prison at Lambaesis, called “The Mosaic of the Captives.” Its central panel depicts a family of prisoners: a man, woman, and child, each with their hands bound, with helpless, pathos-inflected faces (Figures 20 and 21). Twelve small portraits surround the central image. The identity of these individuals is obscure, but the effect of their depiction is clear. Judicial decisions were made from the apse of the civic basilica at Tipasa; the power to bind and loose was exercised on a pavement depicting bodies quite literally under the power of the state apparatus. The point stands no matter what kind of trials were happening in the civic basilica in Tipasa, or in Djemila, or near the Sanctuary of the Standards at Lambaesis. At issue here is not legal procedure, but ideological optics.40 When we look across known North African sites and depictions of incarceration from the imperial period, one theme surfaces again and again: the Roman state is displayed artistically and architecturally as having power literally over captive bodies—the bodies of deviant sectors of society.
In some senses, the identification of a Roman military prison raises more questions than it answers. The deeper question—namely, who was incarcerated there, under the legionary standards?—is something that the archaeology cannot answer. Perhaps some evidence about second- and third-century prisoners was present in the small finds unearthed by Algerian prisoners in the spring of 1897 and 1898, but as is common for nineteenth-century excavations, these items were not catalogued, likely because they were not collected. However, the sense in which the Lambaesis site is a Roman military prison is not that it was used solely for military prisoners. It is a military prison insofar as the military used it for imprisonment. In this case, the evidence does not tell us who was held there, but there are some hints. The wear on embrasures in chambers B–D suggests something: either that chambers A and E were less used, that prisoners of the type kept in the central cells were more likely to receive visitors, or that those in the outer cells were typically bound in such a way as to preclude their access to the windows themselves. The presence of a Greek graffito suggests some amount of linguistic diversity among prisoners.41 Ammianus Marcellinus, quoted above, claims that common soldiers were often confined “at the standards,” while P. London 6 1914, a fourth-century letter about drunken soldiers who mistreated and incarcerated some Christians, presents civilians as being incarcerated ἐν τοῖς σίγνοις. It is unlikely that we will ever know the identity of any particular person incarcerated in the principia of the castra at Lambaesis—we just do not have the evidence to say. But, seen together, the evidence that we do have suggests that in Roman antiquity, there were not distinct prisons for different types of prisoners. Rather, there were distinct prisons for different types of imprisoners, and cells within prisons to keep different classes of prisoners separate.
IV. The Lambaesis carcer among ancient carceral practices
Our findings allow us to offer the first tenable identification of a Roman military prison.42 While it is widely acknowledged that Roman castra routinely contained prison spaces, no such structure has been plausibly identified before now.43 With the remainder of this article, we attempt to contextualize some of the features of the Lambaesis prison among carceral practices in North Africa as evidenced in literary sources, and among other archaeologically attested examples of ancient prisons. Specifically, we will discuss evidence for military prisons in Roman North Africa, prison graffiti, carceral spaces divided into cells, and the seemingly peculiar placement of a site of incarceration in close proximity to a place of imperial cult.
Prisons in Roman castra are mentioned in literary sources such as Tacitus, Annals and Juvenal’s Satires.44 Geographically and temporally closer to the military prison in Lambaesis is the Passion of Perpetua, a literary account set in Roman North Africa of the early third-century CE—and thus, likely while the Lambaesis carcer was in use.45 Perpetua and her companions were said to have been transferred from the public prison (carcer) near the forum to a military prison (carcer castrensis) in or near Carthage, where they were held in anticipation of fighting in the military games.46 Epigraphic evidence likewise confirms the presence of carceres within Roman castra, as well as those tasked with their oversight. Typically, a military prison would fall under the responsibility of the optio carceris, who was charged with guarding the structure and surveilling its captives.47 The Passion of Perpetua mentions someone named Pudens, a soldier with the rank of optio who had been put in charge of the prison (Pudens miles optio, praepositus carceris).48 Other epigraphic evidence indicates that such responsibilities could also fall to a beneficarius or a frumentarius.49 Thus, while no military prisons had yet been securely identified in the archaeological record, it is hardly surprising, based on other literary and documentary evidence, to find one in the castra at Lambaesis, in or near the aedes signorum. In fact, what is surprising is that this is only the first military prison to be credibly identified. The prison in Lambaesis gives us an example with which to think spatially and archaeologically about other data concerning prisons in Roman castra, and Imperial era prisons in general.
Furthermore, we have ancient evidence testifying to the presence of a prison in Lambaesis in particular. The Passion of Marian and James is a hagiographic text set in ancient Numidia during the reign of the emperor Valerian (253–260), though it was redacted to its current form somewhat later. The text records details of the travels, incarcerations, and ultimately the deaths of the saints for whom it is named, and it mentions specifically that they were moved, under guard, from a prison in Cirta to a prison in Lambaesis (Lambesitanus carcer).50 Vitruvius tells us that every Roman forum was intended to have three things: a curia, a treasury, and a prison.51 While the text of the Passion of Marian and James does not give reason to think that the martyrs were held in a military prison, it speaks to the notion that sites of incarceration dotted the ancient landscape, and that we should be attuned to this evidence when interpreting—or reinterpreting—archaeological remains.
The combination of non-literary graffiti and a graffito with Greek lettering, which we have found in the Lambaesis prison, fits well with other data regarding incarceration in Antiquity. The carceraria (prison graffiti) from late ancient Corinth, recently published by Erkki Sironen, are a clear parallel to what we have in the space underneath the Sanctuary of the Standards.52 One of the more common features of the Corinthian prison graffiti is the presence of tables (perhaps used to keep time or as game boards) similar to what we have found in Lambaesis, as well as other geometric markings, like patterns of rings.53
The literary graffiti in the Lambaesis carcer, including the inscription reading ΠỌ, indicate that at least some prisoners spoke Greek and possessed rudimentary literacy, though it is hard to say anything more specific with any security. As a provincial capital active well into the fourth century CE, it is doubtless that some people in Lambaesis spoke Greek. It is equally doubtless that Greek speakers were a minority in the city and in the castra, and would be associated with foreigners and visitors passing through Numidia more readily than with the Latin speaking core of the army. In the Imperial period in general, locals and foreigners tended to be subject to different carceral practices. For instance, when Paul is imprisoned at Iconium in the Acts of Paul and Thecla, the account depicts him as a foreigner (ὁ ξενος) as a way of justifying the fact that he was incarcerated differently from his companion Thecla, who was a local citizen.54 Likewise, the inscription from Thamugadi mentioned above offers a different protocol for feeding prisoners who were “ab ordine civitatis,” and prisoners who were not.55
As mentioned above, literary, documentary, and archaeological sources suggest that prison architecture allowed some manner of separating prisoners, and later imperial constitutions in fact mandated it.56 Even though one can find references to what we might call “solitary confinement” in ancient sources, until now nothing like cells has been found in a securely identified Roman carcer. For example, in his Life of Tiberius, Suetonius claims that Tiberius had “denied those who escaped a prison sentence not only the solace of reading books, but also the privilege of talking to their fellow-prisoners.”57 While such a claim is rhetorically charged, portraying Tiberius as an unreasonable tyrant, it nevertheless offers valuable information into the Roman cultural imagination of incarceration. Until now, it would have been difficult to imagine what kind of prison architecture could have accomplished such an action, as the most well-known ancient Roman prisons are simply rooms or pairs of chambers, and it is unclear how one would go about separating incarcerated persons from each other.58 Cells in the military prison at Lambaesis, and their parallel in the ergastulum at Simitthus, allow us to begin to think architecturally about what kind of a prison structure could have kept incarcerated persons separate.
From the perspective of the history of Roman military camps, and the history of Roman prisons, one of the most interesting aspects of this site is that it shows the presence of a prison underneath a temple—and not just any temple, but an aedes signorum. This military prison in Lambaesis is the first known archaeological example of a purpose-built carceral site situated directly underneath a temple structure. This connection may seem strange at first glance, but we argue that it is not surprising when seen in its broader archaeological and ideological context. Roman incarceration carried both a religious and a military valance, and temples themselves were occasionally used to confine prisoners.59 For instance, P. Oxy. 17. 2154 (fourth century CE) mentions an instance of incarceration inside the Temple of Hadrian in Oxyrhynchus.60 In fact, civic prisons were frequently located in the forum, directly abutting civic and imperial temples or in their immediate vicinity; given the enduring spatial and ideological connection between sites of cult and sites of incarceration, the idea of having a prison underneath a temple is perhaps less surprising than a modern reader might initially expect.61
Roman law, too, operated in the ideological shadow of the gods. In fact, in the city of Rome, law was practiced literally in their shadows. The Forum of Augustus, where the urban praetor heard civil cases, boasted at its center an imposing Temple of Mars Ultor housing recaptured military signa, a colossal bronze quadriga statue of the deified Augustus, and two sets of porticoes housing hundreds of statues depicting gods, kings, mythical people, and the summi viri; the statues of gods themselves often served as meeting points in vadamonia—written commitments to meet for a court procedure.62 So close was the connection between legal jurisdiction and cultic oversight that Richard Neudecker argues that both in the Republic and in the Principate, “divine presence—be it by way of statues or by altars—was necessary to give validity to legal acts.”63 High profile prisoners of war and enemies of the state were detained in the oldest and most famous prison in the Roman world: the Mamertine Prison, sandwiched between the Republican Temple of Concord and Temple of Venus Genetrix in the Forum of Caesar, and looking out over the Comitium, with its sacred tress, statues of famous augurs, and the Lapis Niger.64 But according to recent archaeological work, prior to its conversion to a carceral space, it was apparently a site of chthonic worship, and small finds from recent excavations suggest that cultic observance at the site continued even after it began to function as a prison.65 The connection between sacral and carceral places appears at Rome ab initio. It is no surprise to see the connection enduring through time and transferred from the center at Rome to the imperial periphery at Lambaesis.
Power over bodies
As discussed above, the civic basilica in Tipasa is the space where judicial decisions would be made regarding the binding and loosing of prisoners in the region. Specifically, decisions would be made from the apse of the complex, the pavement of which is decorated with the faces of prisoners. At Tipasa, judgment was pronounced literally over the bodies of captives incarcerated in the prison below. Similarly, for instance, the so-called “Gemma Augustea” invokes an extraordinarily common trope in the Imperial period: Roman military dominance ideologically and physically over the bodies of captives.66 In the upper register of this renowned onyx cameo, the emperor Augustus is seated flanked by Jupiter and Roma, and crowned by Oikoumene, while the lower register depicts the bodies of foreign captives directly under the seat of the emperor, begging their captors for mercy. These captives exist politically under the power of Roman dominance, and their physical submission is depicted spatially. While there are currently no other examples of the phenomenon of prisons under temples known from Roman archaeology, the combination is not surprising, as Heinrich Nissen mused on the connection already in 1904: “It is worth remembering the parallel connection between the state prisons and the high temples of Rome. The carcer and the executioner’s block were closely connected with one-another: not only for the purposes of convenience, but because penal law is a religious act; because the blood of the guilty is offered to the gods as an expiatory sacrifice.”67
Roman religion and Roman incarceration were always connected conceptually. Such a temple-cum-prison, with the sacral space superimposed over the carceral, further expressed Roman military power over captive bodies, and reinforced a conceptual framework visible as early as we have evidence for Roman incarceration.68
V. The persistence of signa over a long Late Antiquity
The Legio III Augusta receives a brief mention in the late fourth- or early fifth-century Notitia dignitatum, and thereafter it withdraws from the historical record, as the Roman military presence faded from North Africa. As mentioned above, there is ample evidence for people being incarcerated “at the signa” during the period of Roman occupation. Long after the Vandals had come and gone from North Africa, however, the memory of military occupation and the connection between signa and incarceration remained in the language that locals used to describe prisons and the people staffing them. Coptic documents from across North Africa attest to the fact that by the sixth century, the term signa had become disaggregated from the actual standards at the center of legionary castra and was used simply to mean “prison.” See, for instance, a sixth-century ostracon from Thebes now at the Metropolitan Museum of Art in which a prisoner complains that “they imprisoned me in the signa (ⲥⲓⲕⲛⲉ).”69 By the sixth century, as well, Jesus’s admonition to care for those in prison (Matt. 25:36) had been institutionalized in the Coptic church: in a papyrus now held at the British Museum we meet a certain “Paul, son of Azarias, the Deacon of the Prisons.”70 The Coptic, however, does not read “Deacon of the Prisons,” but rather “Deacon of the Standards (ⲇⲓⲁⲕⲟⲛⲟⲥ ⲛⲛⲥⲓⲅⲛⲱⲛ).” Another Coptic ostracon from the seventh century attests a prisoner referring to their jailer as a ⲥⲓⲅⲛⲟⲫⲩⲗⲁⲝ, literally the “Standards Guard.”71 The persistence outlived even the lifespan of Coptic as the linga franca of North Africa; while Coptic was only briefly used as a chancery language in the first half-century after the Muslim conquest, the word for standards (signa; ⲥⲓⲕⲛⲉ) came to be used in Arabic as the general word for prison: sijn سجن)72).
It is worthwhile to pause and consider a further historical twist that emerges from this site. This building we have identified as a Roman military prison in Lambaesis was initially excavated in the spring of 1897 and 1898 by a group of Algerians who were incarcerated by French colonists in a prison that sits directly behind the archaeological site to this day. Unknowingly, these Arabic-speaking prisoners were forced to excavate a site which would eventually help explain their own word for the place of their incarceration. The fact that the local Arabic-speaking community in Tazoult, Algeria still refers to this building as a sijn (سجن) exemplifies how, then as now, one often finds oneself bound up unaware in larger histories, not always visible, yet nonetheless operative in subtle, formative ways.
Prisons were a staple of the Roman urban landscape, and incarceration was an ongoing duty of the Roman army. We should expect that every castra had a carceral space, even if it did not have a dedicated prison like the one at Lambaesis. The identification alone suggests the possibility that other prisons lurk unremarked upon in the mass of Roman legionary archaeology that has been undertaken over the last two hundred years—especially in spaces previously identified as “strongrooms” or “treasuries.” It also adds material evidence to contextualize the copious literary attestations of incarceration from across the Roman world. With our identification, we take a step towards establishing a typology of Roman prisons in general, and military prisons in particular.
The identification of a carcer castrensis at Lambaesis also presents us with an opportunity: it refocuses our gaze on this central practice of legionary forces. In this remarkable space, nearly every facet of Roman carceral practice is visible in nuce. Under the Sanctuary of the Standards at Lambaesis we find cells for separating prisoners; embrasures for surveillance, subsistence, ventilation, and visitation; graffiti attesting to bored prisoners far from home; and we find a dark, damp, odoriferous subterranean complex that brings texture to the affective experience of incarcerated people that lived and died in these spaces long ago. We get to ponder for a moment a Roman military apparatus that placed captive bodies quite literally underneath the divinely assured authority of an army commander. And we get to return to our present day, viewing it in new light, and perhaps to be haunted by how—despite its differences—incarceration remains a ubiquitous yet obscure, creative yet destructive tool of social control over bodies deemed deviant.
We would like to thank Elizabeth Fentress and Matthew J. Adams for their expert advice at various stages of this project, and the anonymous reviewers for SLA, who performed a selfless duty with a rare blend of generosity and critical acumen; we hope to have done justice to their questions and concerns. Thank you, as well, is due to Gina Tibbott for her work on the illustrations for this project, and to Mhamed Gueraini for his advice and joyful companionship during our research trip in Algeria. The order of authorship was revealed in a dream. Research for this article was generously supported by an Innovation Grant from the David A. Gardner ‘69 Magic Project, and a Mellon Digitization Grant through the University of Southern California. All translations are our own.
For a broad history of the site, from initial settlement through the castra construction to the history of its excavation, see Michel Janon, Lambèse: capitale militaire de l’Afrique romaine (Ollioules: Nerthe, 2005).
See ILS 2487, 9133–35, a monumental inscription detailing Hadrian’s oversight of military exercises at the site and his personal addresses to different cohorts of the legion. On the continued use of the base at least into the reign of Diocletian, see René Cagnat, L’armée romaine d’Afrique et l’occupation militaire de l’Afrique sous les empereurs, 2nd ed. (Paris: Léroux, 1913), 474.
The closest parallel is the castra of the Sixth Legion Ferrata at Legio, Israel, currently being excavated under the direction of Matthew J. Adams and Yotam Tepper. See Yotam Tepper, “Lajjun-Legio in Israel: Results of a Survey in and around the Military Camp Area,” in Limes XVIII: Proceedings of the XVIIIth International Congress of Roman Frontier Studies Held in Amman, Jordan (September 2000), ed. Philip Freeman et al., vol. 1, BAR International 1084 (Oxford: Archaeopress, 2002), 231–42. A more thorough description of the principia complex is available in Matthew J. Adams et al., “A Betyl with a Decorated Base from the Principia of the Roman VIth Ferrata Legionary Base, Legio, Israel,” Israel Museum Studies in Archaeology 9 (2019): 72–75.
See Yann Le Bohec, La troisième légion Auguste (Paris: Editions du Centre national de la recherche scientifique, 1989), 417 fig. 54. The identification of this upper structure as an aedes signorum is secure. Not only is the central, apsidal room of military castra always identified as a Temple of the Standards, Cagnat reports that during the excavations of this site in the spring of 1897 and 1989, “fragments of a small altar and the torso of a seated divine figure were discovered” in the space. René Cagnat, “Les deux camps de la légion IIIe Auguste à Lambèse d’après les fouilles récentes,” Mémoires de l’Institut de France 38, no. 1 (1909): 247. On typical features of aedes signorum, see Yann Le Bohec, The Imperial Roman Army (London: Batsford, 1994), 157. For a recent identification of a legionary Temple of Standards on the basis of mosaic inscriptions, see Benjamin Arubas et al., “Capricorno Alae VII Phrygum… (i) Interim Report on the Fort near Tel Shalem,” Journal of Roman Archaeology 32 (2019): 201–13.
Lambaesis, and particularly the French prison there, was an infamous penal colony in Algeria of the late nineteenth century, and the name itself invoked carcerality much the same way that “Rikers Island” today evokes images of carceral hell. The colonial prison there was famous enough to become proverbial: “Ce n’est pas un palace, mais ce n’est pas non plus Lambèse.” Yasmina Khadra, L’écrivain: roman (Paris: Pocket, 2001): 230. The Khadra quotation is mentioned in Hélène Ménard, review of Lambèse, capitale militaire de l’Afrique romaine, by Michel Janon and Hippone, edited by Xavier Delestre, Histoire urbaine 19, no. 2 (2007): 166–68.
Friedrich Rakob and Sebastian Storz, “Die Principia des römischen Legionslagers in Lambaesis: Vorbericht über Bauaufnahme und Grabungen,” Mitteilungen des deutschen archäologischen Instituts, Römische Abteilung 81 (1974): 253–80. It appears that the underground portion of the complex was excavated (or uncovered, in any event) in 1885, twelve years prior to the formal excavations in 1897/8. Maurice Besnier, “Les scholae de sous-officiers dans le camp romain de Lambèse,” Mélanges de l’école française de Rome 19, no. 1 (1899): 231. The rationale for digging only the underground portion of the sacellum is unclear, but one might reasonably speculate that the “excavators” were looking for treasure in the most likely place: in the underground room of an apsidal building in the principia, directly opposite the groma. For a full accounting of the site’s nineteenth-century excavation history, see Rakob and Storz, “Die Principia des römischen Legionslagers in Lambaesis,” 253–59.
The sanctuary at Lambaesis is similar to the aedes signorum in the castra at Castellum Dimmidi, which lies to the southwest of Lambaesis on the southern limes of the province of Numidia. See Gilbert Charles-Picard, Castellum Dimmidi (Paris: Boccard, 1944), 127–31.
The identification of these rooms as scholae is secure; Cagnat offers significant epigraphic evidence, and rooms in comparable legionary castra were outfitted with inscriptions explicitly naming them as such; Cagnat, “Les deux camps de la légion IIIe Auguste à Lambèse,” 249–51. See, for instance, CIL 3.3524, an inscription dated to 228 from the principia of the castra at Aquincum. It reads: “The schola of the watchmen, Legions I and II Adiutrix: pious, faithful, Severianiae” (Scola speculatorum Legionum I et II Adiutricium piarum fidelium Severianae). A number of inscriptions that appear to be from legionary collegia were discovered at Lambaesis as well, though none in situ. See, for instance, CIL 8.2554 and 2557.
Rakob, “Die Principia des römischen Legionslagers in Lambaesis,” 275–76.
“…der Basilika durch ein Schlupfpforte mit schwerer Verriegelung gescholossen war” (Rakob and Storz, “Die Principia des römischen Legionslagers in Lambaesis,” 276).
This technique reduces construction cost and is known from other buildings at Lambaesis. Lower portions of the Capitoline temple, for instance, are built in field stones bound with mortar except for places where full ashlar construction was necessary to support ashlar walls above.
Cagnat, L’armée romaine d’Afrique, 474–75.
There is little reason to think that Greek would be used as a quarry mark, but Josef Röder’s death in 1974 precluded his publication of the quarries from Lambaesis. Photos of the Lambaesis quarry from Röder’s estate are available in the archives of the Deutsches Archäologisches Institut at http://arachne.uni-koeln.de/item/topographie/8004720 (accessed October 1, 2020).
The fact that this space is subterranean and directly beneath the aedes signorum was context enough for Cagnat to offer an interpretation of the space. In his opinion, in all such cases, “One may conclude that the emperor and the standards of the cohort were worshipped, and if the lower floors were subterranean, it was in order to serve as a reserve for coinage…” (Cagnat, “Les deux camps de la légion IIIe Auguste à Lambèse,” 246, emphasis ours). Cagnat notes that Léon Reiner, an epigrapher and librarian at the Sorbonne in the late nineteenth century, dubbed the space “carceres,” but Reiner does not seem to have ever published his intuition; Cagnat, L'armée romaine d'Afrique, 474. One must imagine that the inscriptions did not escape Reiner’s keen eye, and that his identification of the space as a prison, however informal, was in part based on them. On Renier’s pioneering epigraphic work in Algeria, see Bonnie Effros, Incidental Archaeologists: French Officers and the Rediscovery of Roman North Africa (Ithaca: Cornell University Press, 2018), 178–87.
IG IV2 3 no. 1273: πόησον Κ(ύρι)ε μόρ̣ῳ κακῷ ἀποθάνε τους.
See also similar graffiti on discovered columns near Temple H in Corinth. Corinth drawing 130 036. http://corinth.ascsa.net/id/corinth/drawing/130%20036?q=graffito&t=drawing&v=icons&sort=&s=2 (accessed December 17, 2019).
Anne Johnson claims that two of the subterranean structures in Roman Britain, which she identifies as treasuries, had windows. Johnson, Roman Forts of the 1st and 2nd Centuries AD in Britain and the German Provinces (New York: St. Martin’s Press, 1983), 115–116. However, the windows in question are different in kind from the embrasures in Lambaesis: “at Benwell a window was set low down in the external wall of a room next to the sacellum in which the strongroom lay, and at South Shields a three-barred window admitted light from the cross-hall” (117). Thus, one of the windows in question was not in the sacellum or strong room itself, and the other was barred and led out to the cross-hall, rather than to a public space. Neither allows passage to an unsecured area like the street, as we see at Lambaesis.
See Pilar Pavón Torrejón, La cárcel y el encarcelamiento en el mundo romano, Anejos de Archivo Español de Arqueología 27 (Madrid: Consejo Superior de Investigaciones Científicas, Instituto de Historia, Departamento de Historia Antigua y Arquelogía, 2003), 241–42; Calpurnius Flaccus, Declamations 4; Theodosian Code 9.3.7; see especially Passion of Perpetua 16.2–4. In this passage, the tribune seems to refuse to allow the prisoners food while they were incarcerated in the North African military prison. Starving prisoners to death by refusing deliveries of food seems to have been a common tactic, as attested in the various versions of the Pero and Cimon story. See, for instance, Pliny the Elder, Natural History 7.121 and Valerius Maximus, Memorable Doings and Sayings 5.4.7.
The statute was discovered in the forum on a large pillar (1.17 m tall) immediately north of the curia. CIL 8.17897.
Columella, On Agriculture 1.6.3.
Calpurnius Flaccus, Declamations 4, a second-century CE rhetorical declamatory exercise on incarceration and prison guards.
The Cosa carcer was built as part of the third phase of the town’s forum (241–9 BCE) and remained in use likely through the third century CE, when the town was largely abandoned. An important distinction is that the openings in Cosa and Djemila come from doors, not windows. Both prisons abutted the forum, but only communicated directly with the public street.
CIL 8.18058 (ILS 3810). Cagnat cautions that the statue base is likely not in situ, but it is unlikely to have traveled too far from its initial placement, either (“Les deux camps de la légion IIIe Auguste à Lambèse,” 239).
Vegetius, Epitome of Military Science 2.20. On the strong room at Chester’s, along with other examples of strongrooms under sacella in Roman castra, see Johnson, Roman Forts, 111–20. There are a few problems, however, with her conclusions, and therefore with using her examples as guides to interpreting the subterranean structure under the aedes signorum in Lambaesis. First, all of her examples are from Roman Britain or the German provinces, and thus are not immediately compelling comparanda due to geographical distance from Roman North Africa and significant difference in climate and terrain. As we will show, there are better comparanda from Roman North Africa itself. Second, few identifications of strong rooms/treasuries are, in fact, secure. Nearly all interpretations are predicated directly on the fact that subterranean structures were built under a sacellum, as Cagnat argued above in n. 15. We argue here, on the other hand, that some of these spaces may have been intended or used for a different purpose; the discovery of a stout structure underneath a sacellum is not reason enough to identify a space as a treasury. Third, all of Johnson’s examples are fundamentally dissimilar from our structure in Lambaesis. Their architecture is different, being essentially a pit of ca. 2.5–4 m square. Thus, the buildings in question in Roman Britain are much smaller as a whole than the subterranean complex at Lambaesis, yet much bigger than any of its individual cells.
The context of Ammianus Marcellinus’ comment indicates that the practice was typical. He notes this not as part of his narrative, but while lamenting that some of his contemporaries are sure to unfairly criticize his work if he leaves out what he deems to be trivial details, for instance if he were to “leave out the reason that common soldiers were locked up at the standards.” History 26.1.1 (Loeb Classical Library 315). The first quotation is from 21.4.5.
Interestingly, the subterranean treasury in Messene was also used as a prison. It famously held Philopoemen of Megalopolis, the general of the Achaean League, prior to his poisoning. See Petros G. Themelis, Ancient Messene (Athens: Kapon Editions, 2019), 70–72. Themelis reports similar underground treasuries in Gortyn on the island of Crete, as well as other treasuries at Bouthroton and in the sanctuary of Apollo Karneios (at the Karnasion Grove in Andania). The foundation of the city of Messene began in 369 BCE and Pausanius, who visited and wrote about the city in the mid-second century CE, did not comment on the prominently placed treasury, perhaps because it had been abandoned and filled in by that time. Plutarch, Life of Philopoemen 19; William J. O’Neal, “The Messenian Affair, Philopoemen, and Roman Policy,” International Social Science Review 69, no. 1/2 (1994): 31–35. Similarly, there is evidence of the state treasury in the Roman forum (in the aedes saturni) being used as a site of incarceration. See Dio Cassius VIII, 3, 2 (as summarized in the epitome of Zonaras, Annales VIII, 2 [D]), and Pavón Torrejón, La cárcel y el encarcelamiento en el mundo romano, 109.
The terminus post quem for the construction of the ergastulum is 154 CE, based on a dated inscription found reused in the walls of the initial phase. M. Vegas proposes an initial construction date of c. 170 CE, based on the ceramic evidence. Vegas, “La céramique du ‘camp’ à Simitthus,” in Simitthus II: Der Tempelberg und das römische Lager, ed. Mustapha Khanoussi (Mainz: P. von Zabern, 1994), 143–44.
Friedrich Rakob, “Das römische Steinbruchlager (Praesidium) in Simitthus,” in Simitthus II: Der Tempelberg und das römische Lager, ed. Mustapha Khanoussi (Mainz: P. von Zabern, 1994), 53–54.
It is worth noting that throughout antiquity enslaved people and freedmen were more likely to be condemned ad metalla than higher-class inhabitants of the empire, blurring the distinction between the two in the context of the ergastulum. See for instance, Digest 188.8.131.52; Theodosian Code 7.18.8, 8.5.18. Hermogenian, however, makes clear that both patrons and their clients are subject to damnatio ad metalla, suggesting that class distinctions are present but not inflexibly determinative; Digest 184.108.40.206. It is interesting too that (for Ulpian at least) being condemned to work in the mines renders an otherwise freeborn person akin to a slave: “But, if even a son had been condemned to the mines, or to another punishment which renders him a slave (sed et si filius in metallum damnatus vel alia poena, quae servum efficit)…” (Digest 220.127.116.11); for text, see Theodor Mommsen and Paul Kr?eds., Digesta Iustiniani Augusti (Berlin: Weidmanns, 1870).
On the near impossibility of this space being either horrea or, in its first two phases, a fabrica, see Rakob, “Das römische Steinbruchlager (Praesidium) in Simitthus,” 98–101.
Rakob, “Das römische Steinbruchlager (Praesidium) in Simitthus,” 100.
As above, coins do not need to breathe, and we have no evidence to suggest that items kept in horrea have any use for a toilet.
“Jeder der sechs Raumfluchten A-F war am nördlichen Kopf der Anlage ein Lavacrum-Kompartiment zugeordnet, und aus diesem Grunde kann wenigstens in der ursprünglichen Planung mit der vorgesehenen Unterbringung von Menschen in allen Schiffen gerechnet werden. Doch zeigt der Vergleich mit der nordöstlichen Lagerlatrine und ihrer normalen ‘offenen’ Gliederung in einem Areal…deutlich den Unterschied zu den mit großem Aufwand hermetisch getrennten Lavacra der sechsschiffigen Kaserne; Hinweis auf eine andere Qualität ihrer Benutzer.” Rakob, “Das römische Steinbruchlager (Praesidium) in Simitthus,” 101.
“Offenbar handelt es sich hier um gefängnisartige und nicht um kasernenartige Unterkünfte.” Michael Mackensen, “Erster Bericht über neue archäologische Untersuchungen im sog. Arbeits- und Steinbruchlager von Simitthus/Chemtou (Nordwesttunesien),” Mitteilungen des deutschen archäologischen Instituts, Römische Abteilung 107 (2000): 492–93.
While it is unlikely that an ancient person would call the ergastulum at Simitthus a carcer, it is nevertheless a space intended to hold the bodies of enslaved people and those condemned ad metalla while they undertook forced labor at the quarry. To call it an ergastulum, rather than a carcer, is emically sound, but etically invokes a distinction without a difference. We note, as well, the difficulties more generally of identifying carceral spaces among archaeological remains, on which see Sarah P. Morris and John K. Papadopoulos, “Greek Towers and Slaves: An Archaeology of Exploitation,” American Journal of Archaeology 109, no. 2 (2005): 155–225.
Digest 18.104.22.168. Robinson suggests that “in ministerium metallicorum” indicates sexual service to the male inmates, and there is some evidence that jurists were at least concerned with the possibility that women condemned to the mines might give birth (D 22.214.171.124); but if the implication is present in the text, it is subtle. In all events if women were condemned to work at a brothel overseen by the military, we do not know where that would have been at Simitthus, and we should still expect that prisoners were housed separately; O. F. Robinson, Penal Practice and Penal Policy in Ancient Rome (London: Routledge, 2007), 125. Women at military mines did not only work in ministerium metallicorum, however: in the second century, women worked in the emerald mines themselves (O. Did. 376), and there is evidence for paid prostitution, rather than enslaved sex workers, at the garrison guarding the mines at Mons Claudianus in the first through third centuries. See Adam Bülow-Jacobsen, “Callgirls in the Quarries,” Proceedings of the 29th International Congress of Papyrology, Lecce, 28 August–3 September 2019 (Forthcoming). See also Cuvigny, who shows that soldiers from the Roman garrison at Berenice paid for the services of a prostitute by the month; it would be surprising if male inmates in Numidia were provided with sexual services while soldiers in Egypt paid out of pocket for the same. Hélène Cuvigny, “Femmes tournantes: remarques sur la prostitution dans les garnisons romaines du désert de Bérénice,” Zeitschrift für Papyrologie und Epigraphik 172 (2010): 159–66.
Notwithstanding the fact that, as Krause demonstrates, both in the Republic and in the Empire, although Romans were generally reluctant to incarcerate women, it did happen. See Jens-Uwe Krause, Gefängnisse im Römischen Reich, Heidelberger althistorische Beiträge und epigraphische Studien 23 (Stuttgart: Steiner, 1996), 170–79 and, more recently, Julia Hillner, who demonstrates that “the consensus [in Late Antiquity] was that prison was unsuitable for high-status women,” while “for lower rank women in the Roman provinces…ending up in a public prison was a likely prospect” (“Female Crime and Female Confinement in Late Antiquity,” in Social Control in Late Antiquity: The Violence of Small Worlds, ed. Kate Cooper and Jamie Wood [Cambridge: Cambridge University Press, 2020], 18).
Albert Ballu, Ruines de Djemila (antique Cuicul) (Algiers: J. Carbonel, 1921), 30; Louis Leschi, Djemila, Antique Cuicul (Algiers: Direction de l’intérieur et des beaux-arts, Service des antiquités, 1953), 23.
It is an interesting question: what kind of trials are happening in these civic spaces that stand above sites of incarceration, and who is pronouncing judgment in these spaces? We would be thrilled to know the answer, but these questions are best left to legal historians, and are outside of our purview and the scope of our argument. At issue here is not what is happening procedurally in the spaces above, but what appears to be happening at a cultural or ideological level in placing these courts of judgement and places of incarceration in opposition.
It is interesting that the Greek graffito appears in chamber A, the cell whose embrasure shows the least amount of wear, presumably pointing to relatively fewer visitors when compared with prisoners held in chambers B–E. It is possible, though of course impossible to prove, that this cell was reserved for prisoners far from home, or who could not speak the language of the castra or the city. Prisoners far from home often had trouble receiving the supplies that they needed to survive in prison. See, for instance, P. Cairo Zeno 3.59519, a letter from Phaneisis in which the corn-measurer requests from Zeno “a cloak or some money, as much as you please, to serve until one of my folk sails down” and can offer him assistance. Contrast this with CIL 8.17897, mentioned above, a decree from near Thamugadi, obliging the citizenry to provide only for locals suffering incarceration.
In 1904, Nissen and Koenen identified a building at Novaesium (Bau 53) as an “Arrestlokal,” and on that basis reclassified a similar space at the castra in Bonn that was published in 1819; Constantin Koenen, “Beschreibung von Novaesium,” Bonner Jahrbücher des Vereins von Altertumsfreunden im Rheinlande 111/112 (1904): 176–77. If the two had archaeological reasons to classify these spaces inside the gate on the via principalis as military prisons, they did not offer any evidence in their publication. The space at Bonn, which boasts 68 small cells, has been interpreted variously as small changing rooms “for the beautiful and those of the other gender,” as lockers for bathers, and as small rooms used to store weaponry; Heinrich Nissen, “Geschichte von Novaesium,” Bonner Jahrbücher des Vereins von Altertumsfreunden im Rheinlande 111/112 (1904): 46. For evidence of his reclassification of this space as a prison in particular, Nissen offers only that “A rebuttal to these comforting attempts is unnecessary. The appearance demonstrates that in Bonn, as in [Novaesium], we have a prison.” If these spaces are in fact prisons, they are quite unlike any prison known from antiquity. Hans Lehner published the small finds from the Novaesium site (176 pages worth) and reported no carceral items nor any items at all from the building identified as a prison; Hans Lehner, “Die Einzelfunde von Novaesium,” Bonner Jahrbücher des Vereins von Altertumsfreunden im Rheinlande 111/112 (1904): 243–418. Petrikovits is similarly skeptical about the Novaesium and Bonn identifications, and suggests on comparison with similar spaces in Ostia that these should be understood as “cabinets for otherwise small or valuable objects.” Harald Petrikovits, Die Innenbauten römischer Legionslager während der Prinzipatszeit (Wiesbaden: Rheinisch westfälischen Akademie der Wissenschaften, 1975), 88. To our eyes, Petrikovits’ identification is the most plausible by far.
See, for instance, Cagnat, L'armée romaine d'Afrique, 168–69: “Dans chaque camp, il y avait une prison, dont la surveillance appartenait à un officier nommé optio carceris, analogue à nos capitaines commandants de prisons. Quand la fonction n'était pas confiée à un option, elle pouvait être, et le fait devait se produire assez fréquemment, attribuée à un homme de confiance du légat, un beneficiarius ou un frumentarius.” Interestingly, for our purposes, Cagnat stated at that time that, “La prison du camp de Lambèse n'a pas encore été reconnue.”
Tacitus, Annals 1.16–21; Juvenal 6.561: castrorum in carcere. Cagnat, L'armée romaine d'Afrique, 168.
Whether this text represents an authentic prison diary (so Brent D. Shaw, “The Passion of Perpetua,” Past & Present 139 : 3–45) or a literary text concocted out of whole cloth, it circulated in a Roman world in which the details were intended to be plausible. It is not clear whether Perpetua was in fact moved from a municipal to a military prison, but ancient readers were meant to believe that such a translation was theoretically and materially possible.
Passion of Perpetua 7.9: transivimus in carcerem castrensis. See also 3.5; 6.7. Latin text is from Thomas J. Heffernan, The Passion of Perpetua and Felicity (New York: Oxford University Press, 2012).
On the optio carceris, CIL 6.531; 9.1617; 13.1833; CIPh 2.1.74. Cagnat, L'armée romaine d'Afrique, 168–69.
Passion of Perpetua 9.1: Deinde post dies paucos Pudens miles optio, praepositus carceris. For Latin text and comments on optio carceris, see Heffernan, Passion of Perpetua and Felicity, 52–53, 88, 242, 244. The optio was a temporary position of service underneath an officer, similar to other posts that soldiers could be expected to take up as they worked their way up the cursus honorum. Le Bohec, The Imperial Roman Army, 48: “The optio was the adjutant of someone who had a post, particularly a centurion (this is the ordinary adjutant), but some, called ad spem ordinis or spei, formed an elite who would one day become centurions themselves.” See CIL 8.217. A certain Petronius Fortunatus within four years moved through the ranks of librarius, tessararius, optio, and signifer, before he became a centurion. Graham Webster, The Roman Imperial Army of the First and Second Centuries A.D., 3rd ed. (Totowa, N.J.: Barnes & Noble Books, 1985), 119–20.
On the beneficarius and the military prison, see CIL 3.3412. On frumentarius and the military prison, see CIL 3.433 and Cagnat, L’armée romaine d’Afrique, 168–69.
Passion of Marian and James 9.5: Tunc eos praesidi admotos iterum Lambesitanus carcer accepit. Latin text is from Pio Franchi de’ Cavalieri, ed., La Passio SS. Mariani et Iacobi, Studi e testi 3 (Roma: TipVaticana, 1900).
Vitruvius, On Architecture 5.2.1. The capital city of Lambaesis was quite large, however, and surely would have had more than one prison—even mid-sized cities like Crocodilopolis or Oxyrhynchus had multiple prisons. In late ancient Crocodilopolis, there were enough carceral spaces that in P. Wash. Univ. 1.57, one was simply called “the big prison.”
IG IV2 3 no. 1270–1294. The inscriptions were initially published by Benjamin Dean Meritt, Corinth VIII, Part 1 (Cambridge, MA: Harvard University Press, 1931), though they were neither recognized as belonging together nor as prison graffiti.
Matthew Larsen, Early Christians and Incarceration.
Acts of Paul and Thecla, 17–20. For an early reference to this phenomenon, see Demosthenes 24.163–169. The different treatment of foreigners and local citizens likely had to do with providing sureties; see Antiphon 5.17–18. When Thecla enters the prison to visit Paul, the gatekeeper reports that she has gone “to the foreigner in the prison” (πρὸς τὸν ξένον εἰς τὸ δεσμωτήριον); Acts of Paul and Thecla 19. Greek text is from Richard Adelbert Lipsiu and Max Bonner, Acta apostolorum apocrypha (Hildesheim: Olms, 1959).
Theodosian Code 9.3.3, a law of Constantine indicating that multiple-celled prisons were to be preferred because they allowed for the separation of prisoners by sex and by the nature of their crime.
Suetonius, The Life of Tiberius 61. Translation is from Robert Graves, trans., The Twelve Caesars (London: Penguin, 1989), 144.
See, for example, the Carcer-Tullianum in the Roman Forum, with its carcer above and the Tullianum below, or the carcer at Cosa described above.
See Larsen, Early Christians and Incarceration.
P. Oxy. 17. 2154, lines 14–15: κατάκλειστός εἰμει […] εἰς τὸ Ἀτριανῖον.
See Vitruvius, De arch. 5.2.1. This relationship seems to correlate to archaeological remains of public prisons. See the Carcer-Tullianum (Mamertine prison) in Rome, which was situated at the foot of the Capitoline hill, in between the curia and the Temple of Concord. Similarly, the prison in Cosa was likewise in the corner of the forum and next to the curia and a temple. On the prison in Cosa, see Frank Edward Brown, Emeline Hill Richardson, and L. Richardson, Jr., “Cosa III: The Buildings of the Forum: Colony, Municipium, and Village,” Memoirs of the American Academy in Rome 37 (1993): 38–41. See also Herbert Newell Couch, The Treasuries of the Greeks and the Romans (Menasha, Wisconsin: Banta, 1929), 4–6, 28–29, 47–50. Couch shows a relationship between the prison, the treasury, and the tomb (48–50): “If the language employed to describe the structure in any one case [sc. of a prison, a treasury, or a tomb] casts light on the archaeological peculiarities on construction, it may be transferred at least tentatively to the others” (49).
“The visible sacredness of the forum [of Augustus] was fundamentally important and specifically employed towards functional ends for each of the hundreds of oaths made on a court day. The porticos also visually connote the nature of the site of the court as a sanctuary.” Richard Neudecker, “The Forum of Augustus in Rome: Law and Order in Sacred Spaces,” in Spaces of Justice in the Roman World, ed. Francesco De Angelis, Columbia Studies in the Classical Tradition 35 (Leiden: Brill, 2010), 170. On the ideological dimensions of the Forum of Augustus’s visual program, see Paul Zanker, The Power of Images in the Age of Augustus (Ann Arbor: University of Michigan Press, 1990), 193–238; and, more recently, Joseph Geiger, The First Hall of Fame: A Study of the Statues in the Forum Augustum (Boston: Brill, 2008), on the summi viri.
Neudecker, “The Forum of Augustus in Rome,” 161. Emphasis ours. See, for instance, the Lex irnitana, in which as Neudecker mentions “even an adjournment, a diei difissio, had to be confirmed by making an oath to Jupiter, the Penates, the deified emperors, and the Genius of the reigning emperor” (170)—a formula expanded from Republican oath just to Jupiter and the Penates in, for instance, the Lex latina Tabulae Bantinae 1.24.
E.g., Jugurtha, Vercingetorix, the Catilinarian conspirators, Sejanus, and so forth. See Filippo Coarelli, Rome and Environs: An Archaeological Guide (Berkeley: University of California Press, 2014), 68.
See the recent archaeological work of Patrizia Fortini, Carcer Tullianum: Mamertino prigione di San Pietro (forthcoming), 37: “Prima di essere inglobato nel Carcer, il Tullianum non era una cisterna (o almeno una vera e propria cisterna), ma una struttura a carattere religioso (accoglie offerte, anche se in seconda deposizione) seppur connessa con l’elemento acqua.” See also Fortini, Carcer Tullianum: il carcere Mamertino al Foro romano (Rome: Electa Ministero per i beni culturali e ambientali, Soprintendenza archeologica di Roma, 1998).
Kunsthistorisches Museum Wien, Antikensammlung, IXa 79. https://www.khm.at/objektdb/detail/59171/ (accessed December 1, 2020).
Nissen, “Geschichte von Novaesium,” 47. See also Pavón Torrejón, La cárcel y el encarcelamiento en el mundo romano, 103–105. Ultimately, parsing cleanly between a prison and a treasury might be misguided, both in terms of the purpose for which they were built as well as how such buildings were used; ancient sources do not justify a neat distinction between the two. See Couch, Treasuries of the Greeks and Romans, 5, 28–29, 47–50.
To date the only comprehensive account of evidence for Roman incarceration in English is Julia Hillner’s excellent Prison, Punishment, and Penance in Late Antiquity (New York: Cambridge University Press, 2015). The first chapters offer an ideological overview of incarceration and present a selection of the material, but the evidence is offered largely as prolegomena to a study of late ancient and early medieval monastic self-confinement. Hillner’s work has become a touchstone among anglophone scholarship because of its comprehensive overview and the fact that it is the only resource easily accessible at the moment. It is, nevertheless, hardly the final word on incarceration in Classical and Late Antiquity.
Metropolitan Museum of Art 14.1.33.
British Museum Inv. no. 1745 recto, published by Harold I. Bell, ed., Greek Papyri in the British Museum, vol. 5 (London: British Museum, 1917), papyrus 1709.
O. Mon. Epip. Coptic 177. Metropolitan Museum of Art 12.180.228.
Petra M. Sijpesteijn mentions the brief use of Coptic as an administrative language in an article which presents an excellent overview of the documentary material for incarceration in the early Islamic period: “Policing, Punishing, and Prisons in the Early Islamic Egyptian Countryside (640–850 CE),” in Authority and Control in the Countryside, ed. Alain Delattre, Marie Legendre, and Petra M. Sijpesteijn (Leiden: Brill, 2018), 547–88. The connection between ⲥⲓⲕⲛⲟⲛ and سجن was first suggested by Walter E. Crum, The Monastery of Epiphanius at Thebes Part II: Coptic Ostraca and Papyri, Publications of the Metropolitan Museum of Art Egyptian Expedition 4 (New York: Metropolitan Museum of Art, 1926), 201 n. 8.