Charlemagne's black marble: the origin of the epitaph of Pope Hadrian I

IL MARMO NERO DI CARLO MAGNO: L'ORIGINE DELL'EPITAFFIO DI PAPA ADRIANO I Questo articolo presenta nuove evidenze che identificano l'origine del marmo nero dell'epitaffio di papa Adriano I, conservato attualmente nel portico di San Pietro in Vaticano. L'epitaffio fu commissionato da Carlo Magno dopo la morte di Adriano (avvenuta il 26 dicembre del 795) ed è il più eminente esempio attualmente esistente di epigrafia carolingia. È un pezzo maestro del rinascimento carolingio. Questo articolo dimostra attraverso analisi paleontologiche, petrologiche e geochimiche che la pietra per l'epitaffio fu cavata nei pressi del flume Meuse, vicino Namur in Belgio, in un'area situata all'interno delle estese proprietà della famiglia Carolingia. Inoltre, si deduce che la pietra nera fu scelta con il consapevole intento d'imitare le espressioni classiche delle risorse imperiali, e che unitamente all'indubbia qualita dell'epigrafe e del testo poetico, la scelta del marmo può essere interpretata come una testimonianza di eredità culturale e ambizione imperiale da parte del suo patrono, che fu incoronato Imperatore in Roma nel natale dell'800, in un luogo dal quale era possibile vedere l'iscrizione, cinque anni dopo la morte di Adriano.


158
STORY ETAL. Pope Hadrian's monumental Epitaph is a masterpiece of the Carolingian Renaissance, and its superb quality is displayed in every aspect of its production. 5 The form of the poem and the form of the script owe much to classical examples, and it is evident that both the poet and the epigrapher of Hadrian's memorial were learned in the epigraphic legacy of late antiquity, as transmitted through the pages of books as well as extant inscriptions cut or painted onto walls. 6 Its classical aspect, combined with a distinctive Carolingian aesthetic, signified a renewal of epigraphic standards not seen, in Rome at least, for several centuries. 7 Hadrian's Epitaph proved to be a turning-point for epigraphy in Rome in the ninth century; subsequent papal inscriptions up to about 850 are noticeably more regular and precise in letterform and layout than their eighth-century predecessors. 8 The form of 'square' capitals (capitalis quadrata) used by the Hadrian Epigrapher was also influential in the development of display scripts in de luxe Carolingian books, especially those produced at Charlemagne's court. 9 It has been argued convincingly that the author of the Epitaph was the Anglo-Saxon scholar, Alcuin of York, who was one of the most influential thinkers at Charlemagne's court in the 790s. 10 His epigrammatic style was heavily influenced by syllogae of late antique inscriptions, especially those 160 STORY ETAL.
from the catacombs and basilicas of Rome, which had been circulating in England and Francia since the later seventh century. '' In contrast, the quality and significance of the materials used to manufacture Hadrian's Epitaph have been discussed much less. Contemporary chronicles claim that the inscription had been made in Francia, so this aspect of the Epitaph may have something to reveal about Carolingian resources and cultural ambitions. 12 Particularly unusual was the choice of a dark, almost black, stone for the inscription. Given what we know of the high quality of the lettering and the poetry of the inscription, it is reasonable to suppose that the stone was carefully selected and that, like the epigraphy and poetry, it carried significant messages of cultural authority and legacy. We now recognize that coloured stones were often used in antiquity, not just for their aesthetic qualities but also for their symbolic significance, often with quite specific ideological connotations. 13 This paper offers new evidence to identify the origin of the black stone used to make Hadrian's Epitaph. The results of the investigation described here have significant implications for our understanding of Charlemagne's relationship with Rome and the Carolingian notion of renovatio imperil.
Despite its unusual character, the black marble of Hadrian's Epitaph has been subject to much less analysis than other aspects of its manufacture. Two arguments have been proposed hitherto for the origin of the stone, and thus the location of the workshop where the Epitaph was made. In 1888 De Rossi argued, on the basis of contemporary accounts of the commissioning of the Epitaph and from his belief in Alcuin's authorship of the poem, that the stone for the inscription was quarried in western Francia, in the vicinity of Alcuin's own monastery at Tours, which had a famous scriptorium and maintained a late antique tradition for metrical inscriptions. 14 De Rossi acquired a sample of a black marble from a quarry on the river Sarthe, about 100 km northwest of Tours, and, noticing that it was of the same character as the stone of the Hadrian inscription, concluded that the same quarry had been the source for the stone of the Epitaph over a thousand years before. 15 Kohler, however, argued for an East Frankish origin for the stone. His opinion was later discussed by Ramackers, who argued that the stone had been sourced much closer to Aachen, in the region around Dinant in modern Belgium. 16 Kohler's and Ramacker's conclusions were based in part on their knowledge of the exploitation of fine black marble from that region in modern times, but more specifically on comparisons of the epigraphy of the inscription with display scripts used in contemporary manuscripts made in the vicinity of Charlemagne's court.
The conclusions of all these arguments were reached primarily through a comparison of the epigraphy of the monument with the palaeography of de luxe Carolingian manuscripts from the Tours scriptorium and from the court school focused on Aachen. Geological considerations were of secondary significance to the arguments, and were made on the basis of broad probabilities rather than direct comparative petrological analysis of the inscription with possible quarry sources. The lack of precision on the petrology of the inscription is largely a result of its inaccessible location. Its site, more than 30 feet above the ground in the portico of Saint Peter's, does not lend itself easily to close palaeontological or petrological inspection.
An opportunity to investigate the monument arose in September 2002, when the Fabbrica di San Pietro gave permission to examine the Epitaph and agreed to the use of a scaffolding tower to allow close-up petrological and palaeontological observation, and geochemical analysis of the black marble (Fig. 2). These results were then compared with samples taken from eight modern quarry sites across northern Francia. The precise methodology and detailed observations are described below. The broader historical analysis of the results will be dealt with elsewhere, but at the outset it is clear that the use of such a fine black stone for the Epitaph of Pope Hadrian was a deliberate and careful choice. The identification of the likely origin of the quarry permits not just a broader interpretation of the monument than has been possible hitherto, but also encourages a new interpretation of it as a key piece of evidence surrounding the coronation of Charlemagne as emperor before the shrine of Saint Peter on Christmas Day, 800. 17

BLACK MARBLE IN ROME
The marble from which the Epitaph is made is fine-grained and a monochrome dark grey, almost black, colour, which to modern eyes appears entirely appropriate 15 De Rossi, 'L'inscription' (above, n. 5), 485. 16  for a funerary inscription. However, in this respect, the colour selected for the inscription was an innovation; prior to 800 the norm for monumental stone inscriptions in the western Roman provinces was a light-coloured or white marble ground. In his study of the monuments of Rome, De Rossi knew of none other, pagan or Christian, carved on black stone. 18 It is hard to find large-scale inscriptions with a coloured background, and where these do occur the material is usually not stone -such as the Lex Vespasiani (c. AD 70), incised in rustic capitals onto a large bronze tablet, or mosaic inscriptions such as that of Pope Celestinus at Santa Sabina, Rome (AD 422-32), or painted tituli (for example at Santa Maria Antiqua, AD 705-7). 19 It may be that Hadrian's Epitaph was intended to recall these other, non-lapidary, types of inscriptions -certainly it is of a similar scale and visual impact to the bronze leges, which may still have been displayed in Rome at this time -but it is possible also that the black stone was chosen because it had other connotations for the patron who commissioned it and the audience that viewed it.
It is clear from studies of marble use in the Roman period that the geographical sources of marbles were remembered in their names, and so in pragmatic as well as symbolic ways the selection of particular marbles resonated with the power of empire to control scarce resources in distant lands. Thus red porphyry was so called because it came from Mons Porphyrites in Egypt; yellow Lapis Numidicum came from the land of the Numidians in North Africa; purple-streaked Phrygium came from the province of Phrygia in western central Turkey; a white marble dramatically veined with black and known as Aquitanicum came from the Pyrenees. 20 Coloured marbles were valued not simply for their aesthetic impact but because they provided a 'material map' of the Roman Empire and, when used in Rome and its provincial capitals, a physical statement of the conquest of distant lands and peoples. The ability to extract and transport monolithic blocks of coloured marbles great distances across sea and land thus symbolized the power of that Empire over the resources of the natural world in lands far from Rome itself. 21 Black marble was scarce in classical Rome. Most significant was the marble known to fifteenth-and sixteenth-century commentators as nero antico. The ancient name of this fine-grained black marble is unknown, but it was valued as a source for small columns and veneer in the first to third centuries AD, and was sourced in 18 De Rossi, 'L'inscription' (above, n. 5), 484. Our thanks to one of the anonymous referees who noted that the eighth-century epitaph of Otto of Brescia, at Monte Cassino, is cut into a polished slab of purple, black and cloudy white africano (from Teos); A. Pantoni, Le iscrizioni medievali di Montecassino (Monte Cassino, forthcoming), no. 25. 19 On the medieval history of the Lex Vespasiani, see Krautheimer, Rome: Profile (above, n. 7), 192-3; for the mosaic inscription at Santa Sabina, see E. Diehl Tunisia. 22 Pliny had known of another 'generally black' marble that was unusual not just for its colour, but also because it had been named after the consul, Lucullus, who had particularly favoured it, rather than the place where it was quarried. 23 Pliny's discussion of this 'generally black' marble followed his account of the importation and use of Lapis Numidicum. This seems to have led to confusion among fifteenth-and sixteenth-century commentators on Old Saint Peter's, who turned to Pliny to discover the source of the black stone of Hadrian's Epitaph; they conflated his description of the (bright yellow) Lapis Numidicum with the black marble of Lucullus, and assumed that this too had an exotic, African origin. For Vegio (c. 1450), Alpharano (c. 1570), and Grimaldi (1619), this reinforced the importance of the stone used for Hadrian's Epitaph; 'it is rare and precious indeed' and was 'without doubt black and Numidian'. 24 Peter Sabinus, writing for Charles VIII of France in 1494, concurred with Vegio's assessment of the rarity and prestige of the stone, referring to the Epitaph as having been incisum in tabula porphyretica litteris elegantissimis. 25 Prominent also, for scholars of classical Rome, is the so-called Lapis Niger, which is an area of paving in the Forum within the Comitium opposite the Curia, made from black limestone streaked with white veins, that is thought to have come from the Tolfa district north of Rome. 26 The Lapis Niger was linked in early imperial texts with the place of the death of Romulus, the legendary founder of Rome; the place (and thus the coloured stone) acquired inauspicious connotations as a locus funestus, a place associated with death. However, it is unlikely that the Lapis Niger was known by the late eighth century; the area of paving would not have been visible by that date, and it could only have been known indirectly through literary sources, namely Pompeius Festus's epitome of the late first-century BC work of Verrius Flaccus. 27  the later eighth century, but did not select this entry for his epitome. It thus seems unlikely that the Lapis Niger was known even as a literary topos in Carolingian court circles, in spite of the fact that Pope Hadrian carried out extensive restoration works in that part of the Forum, restoring the Curia, which was by then a church dedicated to his patronymic saint, the martyr Adrianus. 28

IN FRANCIA: THE PETROLOGY, PALAEONTOLOGY AND GEOCHEMISTRY OF THE EPITAPH
A contemporary Frankish chronicle known as the Annales Laureshamenses {Lorsch Annals) states that the Epitaph was made in Francia and that it was sent to Rome as a finished object to grace the tomb of the pope. 29 The section of the Lorsch Annals describing the death of Pope Hadrian survives in a manuscript copied by four scribes that is very close to the date of the events it describes (794-803); it is probably not the authorial copy but it is very close to it, and in that respect is unique among Frankish annalistic texts. The contemporaneity of the manuscript, as well as the fact that for these years the Lorsch Annals provide a rather different view of events to the major Frankish chronicles, has encouraged historians to consider the Lorsch Annals a particularly valuable source for events around the year 800. 30 The Lorsch Annals imply that the stone for Hadrian's Carolingian memorial was sourced in Francia. Some basic observations of the stone of the Epitaph help quickly to narrow down possible sources. Firstly, it is not a true marble in the technical geological sense of a recrystallized metamorphosed limestone; the word 'marble' is commonly used today, just as it was in antiquity (marmor), to describe any hard stone that can take a polish and that is suitable for sculpture or high-quality architecture. 31 Secondly, the stone contains several identifiable fossils (described below), some visible by eye and others with a hand-held magnifying glass. The presence of these fossils within a black-coloured matrix confirms that it is a carboniferous limestone. The size, structure and colour of the Epitaph are also 28  important indicators of its petrological qualities; it is a substantial slab measuring 2.23 x 1.16 m. The depth of the inscription is unknown, since it is set tightly in a seventeenth-century frame, but is likely to be 10-20 cm in depth. The colour is uniform across the surface of the slab and is very dark greenish grey rather than jet black; the matrix is fine-grained, showing no lighter-coloured inclusions and no mineral veins, and the fossils are a similar colour to the rest of the stone. These observations of fossil content, size of the slab, and consistency of colour and texture help to characterize the petrology of the stone and to narrow down possible geological sources. A comparative analysis can then be made of the petrological, palaeontological and geochemical characteristics of the stone of the Epitaph and carboniferous limestones sourced within Charlemagne's kingdom. In combination with an historical and archaeological analysis of the exploitation of such stones in antiquity and the Middle Ages, this enables us to determine the source of the marble used for Hadrian's Epitaph within a high degree of probability.

THE PALAEONTOLOGY AND PETROLOGY OF THE EPITAPH
The palaeontological characteristics of the stone of the Epitaph were gauged by assessing the range of fossil species and the density of their occurrence across its surface. A thin section was also made from a small sample, permitting further analysis of the components and matrix of the stone at microscopic level.
Close inspection of the Epitaph by eye and hand-held magnifying glass revealed the presence of a number of types of fossil, including corals, crinoids, brachiopods and bivalve shells (see Fig. 3 and, below, Fig. 5). Irregular lines are also visible in places on the surface of the inscription. These are the same colour as the rest of the matrix and are thought to be pressure-solution bands, known as stylolites. These do not affect the structural integrity of the stone, but in some places they have created sufficient weakness in the surface of the inscription to cause superficial cracking when conditions of preservation have been unfavourable. In some places this has caused damage to the lettering, which was probably precipitated by the removal of the inscription from its original location in the south transept (by 1543 at the latest), and its subsequent storage and display until finally set in the wall of the new portico in 1619. 32 The density of fossils visible across the surface of the Epitaph is also an important diagnostic feature. Fossils do not occur in great numbers or in dense concentrations, and where they are found they are a similar colour to the matrix. Thus the marble was not selected for the visual impact of its fossils, which under favourable depositional conditions can produce striking, bi-coloured patterns. Rather, it is the monochrome consistency of the slab that is striking, since no white or other coloured inclusions are visible. The most prominent geological feature Note also the small pockmarks on the surface, indicating the erosion of ooids. The whiteness around the edge of the letter is white lead, squeezed into each letter to enhance the clarity of the inscription. of the surface of the Epitaph is pitting, a result of the erosion of ooids from the stone in an acid atmosphere, which post-dates the carving of the inscription (Fig. 3). Closeup inspection of the stone also shows that the marble is not true jet black in colour; rather it is a very dark greenish grey. Again, in combination with other physical observations, this is an important diagnostic characteristic of the stone of the Epitaph.
Thin sections of a sample taken from the stone permit microscopic examination of its petrological structure (Fig. 6, below). It is a grain-supported shelly oolite with around 95% grains and 5% dark, fine-grained carbonate mud matrix. The wellrounded and well-sorted grains include fragments of crinoid, brachiopod shell, bivalve shell and some foraminifera. Terrestrial sand grains (for example, quartz and feldspar) were not present in the sample. These observations suggest that the stone was deposited in a shallow, high-energy marine environment, since the production of ooids (small, ~0.5 mm diameter, spherical calcareous grains) depends upon vigorous wave action.

THE GEOCHEMISTRY OF THE EPITAPH
The geochemical analysis of the Epitaph was carried out using a portable energy dispersive X-ray fluorescence (EDXRF) spectrometer, developed in the Laboratory of Archaeometry and Non-Destructive Analysis (LANDA), University of Rome 'La Sapienza'. The technical details of the equipment are given elsewhere. 33 The primary X-ray and the fluorescence beams sample an analytical surface of around 2 mm diameter; several points for each sample have been measured and the results averaged to obtain a 'fingerprint' of the constituent chemical elements of the stone. Each X-ray photon reaching the detector generates a group of electrons, the number of which (that is the total charge) is proportional to the energy of the detected photon. These electrons produce a pulse of electric current, which is amplified and shaped before being sent to the multi-channel analyser. The multi-channel analyser separates the pulses of different amplitudes into 1024 channels, where they are counted for a selected time interval. The multi-channel analyser is connected to a personal computer for data acquisition control and for storing the spectra. The EDXRF spectra were processed later with the AXIL software package released by IAEA. This program allows the assignment of the fluorescence peaks to the chemical elements and yields the area subtended by each peak (counts).
For our purposes it is not necessary to calculate absolute mass concentrations of the elements, but, for each spectrum, it is sufficient to obtain relative percentages, P p that is, the ratios between the counts N t of the peak corresponding to the atomic species i and the total number of counts for all elements: P t = TV./iyV.. In our experiments we used two values for the amplifier gain, one to expand the low-energy region of the spectrum (approximately from 1 keV to 9 keV) and the other one for the high-energy region of the spectrum (approximately from 4 keV to 25 keV).

MEASUREMENTS AND ANALYSIS
Initially, measurements were made in situ on the Epitaph itself, measuring EDXRF spectra on several points chosen on the flat regions of the stone as well as inside the carvings of the letters (Figs 1-2). The Epitaph is covered by a layer of polishing wax, which has been rubbed over the surface to clean it and to enhance the patina at infrequent intervals since at least the early seventeenth century, when the inscription was placed in its current position. Because wax is an organic compound composed of light elements only, it should be almost transparent to the X-rays detected in our measurements, and thus should have had little effect on our results. However, the wax contained also a fairly large amount of lead (Pb) that affects significantly the relative intensities of the measured X-ray fluorescence lines, and was thus very likely to distort the results. The lead may have derived from air pollution (some sulphur is also present), but is likely also to have come from the white lead paste that, at some point in the past, was squeezed into the letters of the inscription to enhance their clarity against the dark background (Fig. 3). The white lead in the letters also obscures any possible evidence of gold that contemporary sources suggest may have originally enhanced the inscription. 34 Because of the omnipresence of this distorting layer of lead across the surface of the inscription, permission was given by the Fabbrica di San Pietro to obtain a small sample of the Epitaph stone. This sample was used for two purposes: firstly to measure the EDXRF spectra on the clean, recently broken surfaces on these flakes, and, secondly, to prepare thin samples for the petrographic analysis discussed above.
The results of the EDXRF analysis of the Epitaph stone are discussed below alongside the comparative data gleaned from the samples taken from quarries of known provenance (below, Table 2 and Fig. 7). For each sample we measured at least nine EDXRF spectra on different points in order to check the reproducibility of the measurements and the homogeneity of the samples. Significantly, the results demonstrate that this technique can distinguish between stone from different quarries, some no more than a few kilometres apart. The results reflect geochemical differences between the discrete layers of carboniferous limestone that were laid down under variable depositional conditions. Subsequent folding and faulting of these horizontal beds caused them to break the surface at different places, so that a bed of carboniferous limestone laid down early in the geological sequence may be quarried near the surface not far from strata laid down much later in time, each with different geochemical signatures. These experiments have shown that EDXRF analysis can distinguish between carboniferous limestone beds of different ages, even when the stone produced outwardly appears to be similar.

THE GEOLOGY OF CARBONIFEROUS LIMESTONES IN FRANCIA
Carboniferous limestones are relatively scarce in Europe; where they occur they are found stratigraphically below coal measures. Carboniferous limestones range primarily, northwest-southeast, in a band from Ireland, Wales and northern England, to southern Belgium. Of these, the Belgian sequences are the most substantial and, historically, the most significant.
The European carboniferous limestones were laid down in a rapidly changing shallow sea. The periodic exposure and erosion of land masses in the sea led to the addition of silicate sediment to some of the limestones. This combination of sea-level change and changing sediment source means that the limestones are palaeontologically and chemically variable and distinct from each other. There were three episodes of relatively high sea level, during each of which black limestones 170 STORY ETAL.
were deposited; these are the Frasnian (F2), Tournaisian (Tn2c-Tn3a) and Visean (Vl-V3b). Although the co-joined seas in which the limestones were deposited contained a sequence of microfossils that can be correlated across the region, there are local variations in the macrofossil communities that have made it possible to identify the source of the limestone used for Hadrian's Epitaph. A summary of the main phases of black limestone deposition is given in Table 1.
The fluctuations in the marine environment that produced the black limestones in our area are described briefly since, over geological time, these changes account for the different characteristics of the black marbles sampled. From the early Devonian period, progressive marine transgressions covered the lower reaches of the mountains formed during the Caledonian orogeny, in a process that culminated during the Siegenian stage. In the late Emsian stage, the process was reversed, and by the Frasnian stage marine regression had created an archipelago in the area with numerous coastlines and many well-developed reefs that, in the area of Dinant, constituted a barrier reef. Tectonic changes meant that at some times the waters of the lagoons behind the barrier reef were isolated from the open sea, while at other times they were connected to it. These changes mean that both the composition of the water and the sediment being deposited in the basin were constantly changing; it is these variables that were responsible for the geochemical variations evident in the EDXRF analysis.
Marine transgression and deep-water deposition at the end of the Frasnian stage were reversed again during the Famennian stage and, again, rhythmic sequences of varied sediments were deposited during this period of marine regression, the Golzinne character of which depended upon the varying terriginous inputs and carbonate production of the time. Another period of marine transgression during the lower Carboniferous period (often called the 'Dinantian' stage) led to the production of rhythmic carbonates. This long phase of marine transgression was temporarily interrupted by a minor regression at the end of the Toumaisian that reached its maximum during the lower Visean stage. At this time there were three interconnected basins of 'Dinant', 'Namur' and 'Campine', separated from each other by islands (Fig. 4). After the sea-level high point of the lower Visean stage, the sea again began to retreat until, at the end of the Visean, the carbonate sequence was overwhelmed by siliciclastics flooding in from the south.
The sites from which samples of the black marbles laid down during these various geological phases have been analysed are shown in Figure 4 and Table 1. The famous black stone of Tournai in western Belgium is representative of the lower Carboniferous series. It is well known to scholars of medieval sculpture, especially through the fonts and low-relief tomb monuments that were exported up the river Scheldt into southern and eastern England in the twelfth and thirteenth centuries. 35 Because it is geologically older, Tournaisian marble has lithological and palaeontological characteristics that are different from the younger carboniferous rocks of the Visean stage. It also ages differently from the Visean rocks, since the surface often develops a yellowish tinge as the surface layers oxidize. 36 The geological stages of the lower Carboniferous series, known as the Tournaisian and Visean, are often collectively termed 'Dinantian' because rocks of both stages crop out around the southern Belgian town of Dinant. 37 The Visean is itself stratified; the lower layers produce the marbre noir de Dinant (Via) that was widely quarried in the modern period. It is compact, very fine-grained and thinly bedded, and is found (as its name suggests) in outcrops in northeast-southwest bands in the vicinity of Dinant. Further up the Visean sequence, and further north, are the carboniferous limestones of the Namur syncline. These are found below and around the substantial coalfield of the valley of the rivers Sambre and Meuse between Charleroi and Liege. These middle Visean beds tend to produce black limestone that is more massive and has a different range of fossils to that found around Dinant. The marbre noir de Namur is part of this middle Visean sequence, and is found particularly within the so-called Calcaire de Lives (V2b) along the banks of the river Meuse, between Namur and Liege. The black limestone of the Calcaire de Lives is also commonly known as 'Mosan marble', and it is this stone that provides the closest analogue to the stone of Hadrian's Epitaph. 38 Windows of carboniferous limestone also crop out further east of Namur, particularly around Theux, southwest of Verviers; this stone is from the lower phases of the Visean (VI). 3 9 Black limestones are also found on the eastern edge of the Amorican Massif in western Francia, to the northwest of Tours in the 'Laval syncline' around Sable. These are mainly of the Tournaisian sequence (with thick coal measures interleaved), and to a lesser extent of the lower and middle Visean period. The marble known as noir de Sable was that which De Rossi thought had been used for the Hadrian Epitaph. De Rossi was right to recognize that the sample of noir de  38 Groessens, 'L'industrie' (above, n. 37), 2 2 8 -9 ; Anderson and Groessens, 'Black altars' (above, n. 36), 134. 39 Groessens, 'L'industrie' (above, n. 37), 2 2 1 -8 .
Sable he saw had the same general petrological character as the stone of the Epitaph. The bed outcrops along the banks of the Sarthe between Port Etroit and Juigne for no more than 2 km, and was extracted primarily at the quarry at Port Etroit (the source of De Rossi's sample), where the stone is found in regular, thick beds that are steeply angled. The overgrown remains of smaller quarries are identifiable further along the river-bank; in the nineteenth century the Benedictine abbey at Solesmes on the south side of the river operated a marble factory that worked different marbles from across the region. However, the main commercial feature of noir de Sable is the extensive and prominent white veining running through the black ground; these are entirely absent from the stone of Hadrian's Epitaph. The white veins are very likely to be found throughout the sequence of black limestones at Sable since, geologically, the beds lie in the nose of a fold that would cause extensive veining through pressure and cracking. It is possible that De Rossi saw a small sample in which the white veins were absent, but it would be most improbable that a slab the size of the Epitaph could be produced from these beds without showing any sign of the white veins so characteristic of the noir de Sable. Additionally, the noir de Sable tends to contain prominent fossils that appear bright white against the dark background and are found in a much greater density and variety than in Hadrian's Epitaph. 40

GEOLOGICAL COMPARISONS
Samples were taken from seven Belgian quarries: Golzinne (middle Frasnian); Tournai (upper Tournaisian); Salet, Dinant (BE003) (lower Visean); Anhee, Dinant (BE004) (middle Visean); St Anne, Dinant (BE002) (middle Visean); Sclayn (BE001) (middle Visean); Theux (upper Visean); and an eighth sample from the French quarry at Port Etroit (Sable, BE005) (middle Visean). It is necessary to draw attention to the variables that affect this sampling strategy. The samples provide a keyhole-view of the geology of the region; the palaeontology is sufficiently similar within rocks of the same geological age to render a comparison meaningful, but the local circumstances of deposition mean that geochemical readings from a small sample may differ from a bed not far above or below it, even within the same geological sub-phase. This variability can be minimized by taking a number of readings from each sample; here, at least nine readings were taken from each. It is exacerbated, however, by the very nature of quarrying; over time and use, a quarry destroys the evidence of its own production, as beds are extracted sequentially. Thus, even if a sample were to have come from exactly the same quarry as the Epitaph, the geochemistry and palaeontology of strata at the front and back of the 174 STORY ETAL.
quarry may differ sufficiently to make a match impossible to achieve with 100% certainty; we must thus work within a range of probabilities and use a combination of scientific techniques and historical observations to reduce the variables.

P E T R O L O G I C A L A N D P A L A E O N T O L O G I C A L C O M P A R I S O N S
The combination of analytical techniques comparing the palaeontology and the petrological and geochemical qualities of Hadrian's Epitaph with samples from quarries producing black marble of varying geological ages in Belgium and France strongly suggests that Mosan marble of the middle Visean period is the best match. The modern quarry at Sclayn, 14 km downstream from the confluence of the Sambre and the Meuse at Namur, produces Mosan marble of the middle Visean stage that displays a similar range of fossil species, in similar density, to the stone of Hadrian's Epitaph. These are compared in Figure 5. The Sclayn marble is also very dark greenish grey, like the Epitaph. In these two respects, colour and palaeontology, the samples from Sclayn are quite unlike the marble of different geological ages produced in other quarries. Mosan marble is also more massive than that of other period quarries and would be more likely to produce a slab of the size of the Epitaph. The quarries around Dinant produce marble of the lower Visean sequence that is much blacker, shows fewer fossils of a more restricted fossil assemblage, and tends to be thin-bedded and compact. In addition to this, the most distinctive fossil of the Epitaph, the coral zaphrentis, is absent from the French middle Visean limestone at Sable. The petrological differences between the samples are best illustrated by the thin sections taken from each sample and from the Epitaph (Fig. 6a-b).
The thin sections show that the limestones form two contrasting groups. The first includes the rocks of Sclayn (BE001) and the Epitaph, which have a large number of well-rounded and well-sorted carbonate grains (including shells and ooids) with a small amount of matrix (5-15%). The other quarries form a group that is matrix dominated (70-97%), containing varying amounts of shells, sand and ooids. While the components of each of these rocks will change from bed to bed within a quarry, the features of the Mosan marble (Calcaire de Lives, sampled at Sclayn) and the Epitaph suggest that the rocks were laid down in high-energy, shallow water, while the preponderance of matrix in samples from the other quarries suggests deposition in deeper quiet water.

G E O C H E M I C A L C O M P A R I S O N S
The geochemical data from each sample are given in Table 2. The composition of the samples coming from Theux, BE005 (Sable) and BE003 (Salet, Dinant) are rather different from the Epitaph sample, and may thus be disregarded as a possible source for the Epitaph. Theux and BE005 (Sable) are much richer in strontium (Sr)    The other samples have percentage values closer to those of the Epitaph and so it becomes difficult to give any a unique assignment or to discard other samples. For example, the value of calcium (Ca) for the Epitaph is between those of BE001 (Sclayn) and Golzinne; the strontium concentration in the Epitaph is closer to that of Golzinne, but the iron content is closer to that of BE001 (Sclayn). A statistical study, which considered the entire set of measurements instead of their mean values, was used to refine the comparison between the populations of measurements for the quarries and the Epitaph.

Statistical analysis
Principal Component analysis is used as a means of demonstrating the differences between groups of data where each group represents the results from a single sample; it also enables comparisons to be made of the variations within each dataset. The Principal Components (PCs) were calculated from the covariance matrix. The first two PCs provide a good summary of the data, since they account for 99% of the total variance. The two-dimensional scatterplot of the first two PCs (Fig. 7) confirms that there is a significant separation of the Theux and BE005 (Sable) groups from the other classes, and that the two groups are different from each other. Inspection of the Principal Component plot (Fig. 7) shows that the two closest populations to that of the Epitaph are those of BE001 (Sclayn) and BE004 (Anhee, Dinant), both of which are from the middle Visean stage and are parts of the Calcaire de Lives. Together with PCs analysis, a classification algorithm was employed in this study, the so-called Support Vector Machines (SVMs) method, a new generation learning system based on recent advances in statistical learning theory. 41 The learning machine was trained on a training set consisting of eight classes: Golzinne, BE002b, Theux, BE004, BE003, BE005, BE001, Tournai with respectively seventeen, ten, twelve, thirteen, fifteen, thirteen, fifteen and fourteen observations. After the training the machine was asked to classify the twenty observations of the Epitaph sample and it assigned 75% of them to the BE004 class and the remaining 25% to the BE001 one. While the linear trends for the rocks of Golzinne, BE002b (St Anne, Dinant), BE003 (Salet, Dinant) and Tournai are similar to those of the Epitaph, their chemistry is much more variable than that of the Epitaph, BE001 (Sclayn) and BE004 (Anhee, Dinant). This increased variability is illustrated on the chart; each group of data points for the Epitaph, BE001 and BE004 is compact, whereas the groups of data points for the other samples are more stretched. The differences in chemistry between the two groups are probably a reflection of different dates of deposition. The Golzinne marbles were created during the Frasnian stage of the late Devonian series (Table 1). The other two samples are limestones from the lower Visean stage; BE002b (St Anne, Dinant) and BE003 (Salet, Dinant) were deposited at a time of low sea level and increased exposure of the land. This is reflected in the presence of small amounts of sand found in these rocks, as observed in thin section (Fig. 6). These sand grains are probably the cause of the greater variation in geochemistry of these two samples. The distribution of the Tournai samples is compact but has a composition distinct from that of the Epitaph and the other related rocks. This difference probably reflects the difference in chemistry of the sea in which it was deposited, a result of the rock being deposited at a much earlier time.
Chemical analyses of the samples from carboniferous limestone quarries and from the Epitaph do not give a unique result. However, they restrict the provenance to rocks laid down during the Visean stage in the region of the Meuse valley around Dinant and Namur. Note that rocks of this age deposited away from the Namur/ Dinant region (that is Sable) can also be excluded on geochemical grounds. Within the Namur/Dinant region, the closest match to the Epitaph is given by samples BE001 (Sclayn) and BE004 (Anhee, Dinant), which are both outcrops of the middle Visean (V2b) Mosan marble {Calcaire de Lives) and hence have similar chemistry to each other. To further discriminate between these two sources, petrological evidence is required.
The palaeontological analysis supports the geochemical result that the Calcaire de Lives (BE001 (Sclayn) and BE004 (Anhee, Dinant)) is indeed the best match for the stone of the Epitaph as it contains an assemblage of fossils characteristic of the Visean in this area. Furthermore, the thin-section petrography of the stone shows that the environment in which the BE001 (Sclayn) sample was deposited was characterized by high-energy conditions and hence abundant ooids, while the environment of deposition of BE004 (Anhee, Dinant) was generally lower-energy with occasional influxes of shelly material, perhaps storm deposits. Thus, of the two, the sample from Sclayn, BE001, is more like the stone used for the Epitaph.

HISTORICAL EVIDENCE FOR THE USE OF MOSAN MARBLE
This preference for the marble of the middle Visean stage is also supported by the historical and archaeological evidence, which indicates that these rocks were exploited in the Roman period whereas the black marbles from Dinant were exploited only later. There is a considerable history of exploitation of the black limestones in Belgium. Marble known as 'Belgian black' was very highly valued during the seventeenth and eighteenth centuries because of its homogeneous, intense black colour and its very fine grain, which enabled the creation of complex three-dimensional sculptures. 42 The best source of this 'Belgian black' comes from Golzinne (Mazy) to the north of Namur, where it is quarried underground. Geologically, however, the black marble from Golzinne pre-dates the Dinantian stage, belonging to the last stage of the Devonian (middle Frasnian). This is significant since, although the quarry is close to Namur, to the river Meuse and to the younger marbles of the middle Visean stage found along the banks of that river, the geological and geochemical properties of marble from Golzinne are quite different, as we have seen. 43 The first quarry exploiting marble from the Golzinne area was opened in 1645, and there is no evidence that it was used before that time.
The modern term 'Belgian black' was also applied to the lower Visean marbles from Dinant. The stone from this region was also highly prized in the seventeenth and eighteenth centuries, but could only be quarried in blocks of relatively small size; it was used particularly therefore for floor tiles and tomb slabs (and is now quarried at Salet). There is some textual evidence for the use of the stone in the later Middle Ages, and surviving tomb monuments, such as the tomb of John the Fearless (1371-1419) in Dijon, indicate that it was used for high-status commissions. 44 Marble from thicker middle Visean layers was preferred for Romanesque sculpture, especially fonts, in the twelfth and thirteenth centuries. 45 Many survive within the diocese of Liege, some very finely carved, others with less sophistication.

STORY ETAL.
The colour of the stone is often quite grey, but where features and figures have been polished, a smooth, shiny black patina is created.
The black stones of Belgium were exploited in the Roman period also. It has been argued that the stone of the eastern Belgian site at Theux was used in the Roman period, but the argument has been confused by a misplaced nineteenthcentury attempt to associate this source with Pliny's Lucullan marble. 46 Toumai was an active quarrying centre in the Roman period and stones from that region were transported along the river Scheldt/Escault for use in the construction of the thirdcentury coastal fort at Aardenburg. Black Tournai stone was used in the construction of its walls, the principia and two towers, and it has been suggested that an internal temple at the site may have been made of polished black marble. Black stone was used also at the nearby Saxon Shore fort at Oudenburg, from where it was quarried for reuse in the walls of Bruges in the late eleventh century. 47 Roman quarries were also worked along the banks of the river Meuse, between Namur and Liege, producing middle Visean 'Mosan marble'. Fragments of black marble from this phase are known from the Roman villa at Echternach, Luxembourg, and from six funerary slabs of third-century date from the citadel at Namur. A third-century altar and the octagonal itinerary column from Tongres, now in the Royal Museum in Brussels, are also made from Mosan marble. 48 A fragmentary late fifth-or early sixth-century funerary inscription from the abbey of Saint Servatius in Maastricht was made from a piece of Mosan marble, possibly spolia 49 Mosan marble is also the material of several of the remarkable group of altars dedicated to the goddess Nehalennia; seventeen whole or fragmentary black altars were recovered from the temple site off the Dutch coast at Colijnsplaat, transported there via the Meuse to the mouth of the Rhine. 50

BLACK STONES AND THE MEUSE VALLEY IN THE AGE OF CHARLEMAGNE
Our analysis suggests that Hadrian's Epitaph was made from a Mosan marble of the middle Visean stage, cut in the vicinity of Namur. We have good evidence that this material was exploited in the Roman period and again in the Romanesque. Our observations concerning its use for Hadrian's Epitaph provide hard evidence for its use in the intervening centuries, though whether it was freshly cut or reused spolia remains to be seen.
The Meuse is one of the major rivers of Francia, central to the eastern kingdom of Austrasia, providing a communication link northwards and, through its proximity in its upper reaches to the Rhone-Saone axis, to the southern provinces and the Mediterranean. 51 It rises in the plateau of Langres and flows north through the forested lands of the Ardennes, cutting and exposing the carboniferous limestone geology between Dinant and Namur. At Namur it joins the course of the Sambre and turns eastwards, following the line of the geology (exposing the coal and dolomite), before turning north again at Liege, through Maastricht, to Frisia, the Rhine delta and the North Sea.
From the mid-seventh to earlier eighth centuries this region formed a discrete political unit -the kingdom of Austrasia -that was one of the three major units of Merovingian Francia and often in political opposition to Neustria and Burgundy. It was here that the Carolingian (Pippinid) dynasty rose to prominence from the ranks of the Austrasian aristocracy, acquiring political influence and control of the region through marriage into families that had well-established claims to lands and monastic foundations around the central Meuse valley. 52 The charter evidence suggests that the core of the family lands lay to the west of Namur, focused around the monastery of Nivelles, which was founded c. 640 on lands owned by Itta, the wife of Pippin I and Charlemagne's great-great-great-grandmother. Pippin II (d. 714) extended the family's influence further south by marrying Plectrude, whose family had founded the monastery at Echternach and controlled great tracts of land in the Ardennes and the middle Moselle region. He consolidated his influence further east by marrying again into another family based in Maastricht, which possessed lands around Liege. The lower Meuse came under his control through the conquest of Frisia in the 690s and through the consequent division of land, churches and monasteries among men loyal to him and to Plectrude and her sons. This policy was 51  disliked by the aristocracy of Liege-Maastricht, who resented the intrusion of Plectrude's southerners into a region that they felt should have come within the diocese of Tongres/Maastricht/Liege. 53 Charles Martel, Pippin's son by his second wife, was eventually able to overturn the influence of Plectrude and to succeed his father as mayor of the palace in Austrasia by marshalling this unrest among the local aristocracy of Liege-Maastricht, where his own maternal family was based, against the considerable forces of his step-mother, culminating in his victory at Ambleve in April 716. It was from here that Charles was able to gather sufficient strength to take on the Neustrian forces and to become de facto ruler of all Francia, as his father had been before him. 54 The central Meuse valley was thus a crucial power base for the Carolingian dynasty in the late seventh and early eighth centuries, articulated by a network of aristocratic centres that either belonged to the immediate family of Charles Martel or identified with his cause. The creation in the early eighth century of a new cult centre at Liege, focused on the remains of the martyred bishop Lambert, was a demonstration of Pippinid power within the extensive diocese of Tongres/Maastrich/Liege, and served to marginalize the claims of other regional aristocratic families to control of the bishopric and its dispersed estates, and to act as an alternative focus to the royal demesne focused on Maastricht. 55 Namur had been a stronghold of the Pippinid family since the time of Pippin I in the earlier seventh century, and his wife Itta also seems to have owned lands through her family further downstream around Andenne. 56 Their daughter Begga founded a monastery at Andenne, which may have lain within a bigger estate that straddled the river and included Seilles, Landenne and Vezin on the north bank, and Andenne, Sclayn, Bonneville and Coutisse to the south. 57 Begga's monastery at Andenne never rivalled that of her sister Geretrude at Nivelles, which became one of the most influential royal nunneries in Francia, but its location on the banks of the Meuse was an important marker of Pippinid presence that complemented the family's other centres of power along the central Meuse, such as Namur, Huy, Chevremont/Liege, and the palatia publica at Jupille and Herstal. 58  possessed by Pippin's family along the Meuse and its hinterland were amplified by additional estates to the east of the river, including that which became Charlemagne's favoured residence at Aachen, 17 km to the west (Fig. 8). 59 These family Carolingian estates included the areas that produce the black Mosan marbles. The modern quarry at Sclayn lies within the suggested boundaries of the estate on which the monastery of Andenne was founded, and an earlier quarry 186 STORY ETAL.
(possibly Roman in origin) lies between it and Namur. Evidence for the use of these black stones during the Carolingian period is otherwise hard to come by, although local white limestone was certainly used in the early Middle Ages, especially for sarcophagi and monumental sculpture. 60 Perhaps the most intriguing piece of evidence for the use and desirability of black stones at this time comes from the letter written by Charlemagne to King Offa of Mercia in early 796. 61 This letter, which is well known to historians of Anglo-Saxon England, covers a variety of topics of serious concern to the two kings, including Charlemagne's gift of alms to the episcopal sees of Mercia and Northumbria for the intercessions for the soul of Pope Hadrian. Among other topics, his letter discusses 'black stones' that OfFa had asked to be sent to him: As for the black stones which your Reverence pressed earnestly to be sent to you, let a messenger come, who may consider what kind you have in mind, and wherever they are to be found we will willingly order them to be given and to help with their transport. But just as you have intimated your wishes concerning the length of the stones, so our people make a demand about the size of the cloaks, that you may order them to be such as used to come to us in olden times.
The form and function of these 'black stones' is obscure, but they were evidently a topic worthy of negotiation between the two kings. Hincmar of Rheims, writing to his diocesan priests c. 856, told them to make altars de marmore vel nigra petra, which suggests that the material was to be reserved for special use. 62 Levison thought that the reference in Charlemagne's letter was for building stones, but others have favoured a more mundane interpretation, arguing that the petrae nigrae were quern-stones made of dark Mayen lava, commonly imported into Britain in earlier centuries. 63 However, Charlemagne's letter specifies that it is the length of the stones that is important to OfFa. 64 Peacock has argued that OfFa's 'black stones' were columns rather than querns, the length of which would be a crucial feature. Furthermore, he suggested that such columns must have been spolia from Roman sites rather than ones quarried de novo in Francia in the later eighth century.
Peacock's conclusion was prompted by a study of the columns used in Charlemagne's chapel within the new palace complex at Aachen, begun in 792. Nineteen antique columns survive in the building and, of these, two are made of very rare black porphyry. 65 Of the rest, there is a marked preference for grey marbles and granites, which had come originally from the imperial quarries in eastern Turkey and Egypt. 66 The black porphyry columns in Charlemagne's chapel are very significant; the stone is exceedingly scarce and, Peacock has argued, would have been of immense value and prestige in the Roman world. Porphyry was the stone of Imperial YLomspar excellence. It comes only from the quarries at Mons Porphyrites in Egypt and was normally purple in colour. Its exotic origin, the difficulty of extraction and transport, as well as its intense purple colour, meant that it more than any other stone came to symbolize imperial power, and efforts were made to limit its use to imperial projects. Black porphyry came from only a few places on the mountain, and was thus even more rare than the purple variety; it 'may have had a value almost beyond comprehension'. 67 Objects made from black porphyry are few and far between, which makes it difficult to generalize about their significance and primary use; Gnoli knew only of two other columns that are said to have come from an altar commemorating the decapitation of Saint Paul in the church of San Paolo alle Tre Fontane, south of Rome on the Via Laurentina, where the saint was said to have been martyred. 68 Another black porphyry column is to be found -with four red porphyry columns -in the apse of the thirteenth-century cathedral in Magdeburg, perhaps reused from the tenth-century palace of Otto I nearby. 69 188 STORY ET AL. and other specimens' from the floors and walls of a palace in Ravenna. 70 This could refer to objects of varying size and function, such as small coloured tiles, large pieces of marble veneer or squared blocks, as well as columns, capitals or other architectural elements. The letter is undated, but may have been written after Charlemagne's visit to Ravenna on his way home after spending Easter in Rome in 787. The reference to musiva in Hadrian's letter may be to opus tessellatum or to the larger, coloured pieces of opus sectile, the fashion for which had been revived in Rome during Hadrian's pontificate. 71 Opus sectile tiles, as well as column shafts that had been taken from Roman buildings, have been found at Aachen and in Charlemagne's palace at Ingelheim, as well as in several contemporary Carolingian churches; much of this material might have come from Roman buildings closer to home than Ravenna, but its reuse in Francia mirrored the latest fashion in Rome, which was itself derived from late antique interior design. 72 We should note that some of the opus sectile pieces from Aachen, Corvey and Munster are mottled dark grey/black marble (some with fossils), reminiscent of the stone of Hadrian's Epitaph, and are assumed by the excavators to be of Roman manufacture. 73 Hadrian's letter reveals that Carolingian agents were actively collecting stonework from afar for Charlemagne's building projects in Francia, with considerable logistical implications. Peacock's work on the petrology of the Aachen columns implies that these agents were preferentially selecting particular types of worked stone from particular Roman buildings, especially stones that were black or grey in colour. Charlemagne's letter to Offa implies that 'black stones' were widely appreciated, and that they were sufficiently scarce for access to them to be controlled by the king. Hincmar's comment that altars were to be made of marble or black stone suggests that the material retained its value and scarcity into the middle of the ninth century at least. All this has implications for our study of the marble selected for Hadrian's Epitaph. Firstly, it raises the question of whether the Epitaph was also