Multi-Technique Analysis and Digital Reconstruction of Polychromy on a Mithraic Altar from Carrawburgh Roman Fort near Hadrian’s Wall

Abstract

Heritage materials science techniques, including pXRF, FTIR-ATR, XRD, microphotography, and microsampling, have peeled back concealed layers of polychromy on a Roman Mithraic altar near Hadrian’s Wall. The results break new ground by exposing the interplay between light and dark and the transformative impact of colour cast onto cultic carved stone. A powerful pigment palette is revealed, including vibrant cinnabar/vermilion letters overlying an unprecedented purple inscription panel created from a compound of Egyptian blue, ultramarine, cinnabar/vermilion, red ochre, red lead, and realgar framed with purple, Egyptian blue, ultramarine, and orpiment. The panel was covered in a golden surface crafted from a previously unknown recipe of pyrite (fool’s gold) mixed with beeswax. Tantalising traces are also detected on some iconographic features, but conservator intervention and degradation processes combined to leach colour from the sculpted relief. These are paradigm-shifting results. They force a reinterpretation of the symbolism and performance of altars as personifications of dedicators, and we identify, for the first time, high-ranking Mithraic initiates by name and grade. Critically, we present a digital reconstruction of the altar with original polychromy that revolutionises our understanding of relief-sculpted Roman inscriptions—a category that has, until now, remained underexplored by the burgeoning polychromy research community.

1. Introduction

In 1950, three south-facing altars to Mithras were excavated from the north end of a Mithraeum located 73 m south of the southwest corner of Carrawburgh (Brocolitia) Roman fort just to the south of Hadrian’s Wall, northern England [1] (Figure 1). All three were dedicated by prefects in fulfilment of vows and date to around the 3rd century CE, making them hugely significant monuments from antiquity. They overlie a votive deposit containing a Castor ware beaker, sacred pinecone fuel lumps, cervical vertebrae of a domestic fowl, and an upturned tin cup.

The altars have benefited from limited intervention, making them ideal candidates to explore polychromy practices on the Roman frontier. We develop a unique approach combining archaeological investigation and scientific analysis to provide a holistic understanding of the complex colours that once adorned one of the altars and how these contributed to its performance in the darkened spaces of the Mithraeum. The results present exciting and previously completely unknown pigment recipes and, critically, reveal a golden inscription panel that diverges entirely from conventional knowledge about these ancient monuments.

2. The Altar to Mithras in Context

2.1. Historical Context

The mysterious cult of Mithras was particularly attractive to initiates from the Roman military, with a firm, but not exclusive, focus on Mithraea recovered from frontier contexts, including Britain and the German Limes, and more recently Dalmatia, Dacia, Moesia, and wider Danubian provinces [2,3,4,5]. The tiered male-only structure of Mithraic initiates lends itself particularly well to the hierarchical composition of the Roman army since almost 90 per cent of known dedicators are soldiers [6], with more than half being ranking officers [7].

At Carrawburgh, the central altar was dedicated by Lucius Antonius Proculus of the First Cohort of Batavians (RIB 1544; CSIR 1.6.121; CIMRM 294, 845). This was flanked on the east by another dedicated by Aulus Cluentius Habitus, also of the First Cohort of Batavians (RIB 1545; CSIR 1.6.123; CIMRM 294, 846). Based on the epigraphic evidence recording ethnic identities of the dedicators, these two altars probably date to the early 3rd century CE.

On the west, and the subject of this report, was a much more elaborately decorated altar dedicated by Marcus Simplicius Simplex (RIB 1546; CSIR 1.6.122; CIMRM 294, 847), whose ethnicity and legionary association are not recorded but is thought to originate from Lower Germany [8] (p. 50) [9]. The absence of an identifier and comparatively poorer quality of inscription are thought to date this altar slightly later than its companions [8].

All three altars were accessioned into the collections of the Museum of Antiquities at the University of Newcastle, now the Great North Museum: Hancock, in 1956, where they remain on display in the impressive Roman exhibit (Figure 2). Their original placement is altered, with the Marcus Simplicius Simplex altar (Accession No. 1956.10.30) now commanding the central position. Although this was presumably for reasons of aesthetics due to its highly decorated character, the changed arrangement has the unintended consequence of markedly altering its integral symbolism and performative function.

Marcus Simplicius Simplex’s altar is crafted from buff sandstone. Standing at 1.245 m in height with a width of 0.457 m (Figure 3), it has a flat top which may have held a statue or a container for libations [1] and a capital originally decorated with four leaves framing a central triskele. Immediately underneath this is a relief carving (see below for detail) above a framed inscription in abbreviated Latin, which reads

DEO INVICTO MITRAE M(ARCUS) SIM- PLICIUS SIMPLEX PR(A)EF(ECTUS) V(OTUM) S(OLVIT) L(IBENS) M(ERITO)

Translated as, ‘To the Invincible god Mithras Marcus Simplicius Simplex, prefect, willingly and deservedly fulfilled his vow.’

2.2. Sol Symbology

A 25.4 cm wide openwork recess is carved into the rear of the altar to hold a lamp that casts light through the pierced ogival top, framing a relief-sculpted male figure’s head. This feature would have created a dynamic atmosphere during Mithraic ceremonial activities performed, as they were, in deliberately darkened spaces. The head and torso are carved in low relief, facing forward with both arms bent and hands resting on the hips. He is flanked on either side by carved fillets with circular, triangular, crescent, and linear features. A cloak is draped over his right shoulder, held in place by a penannular brooch, and covering most of his naked, muscular torso and his entire left arm. The figure was identified as Mithras [1] (p. 37), and a whip held in his right hand was thought to connect Mithras with Sol or Helios, the Sun god [1] (p. 37) [10].

More accurately, though, this representation of a whip likely signifies a connection of Sol with Heliodromus, the Sun-Runner, as a messenger of Sol [11], a rare representation not explicitly referenced in epigraphy across the Empire [12]. Heliodromus is associated with the Mithraic torchbearers, Cautopates and Cautes, who customarily flank scenes of the tauroctony. Statues of both were recovered from the opposite end of the nave.

Heliodromus is usually dressed in bright red and thought to correspond with Cautes, whose torch is traditionally held upward and associated with solar iconography, divine authority, and astral ascent [13]. This is expressed through the symbology of a torch signifying the morning star (Phosphoros), a whip to control fiery horses, and a radiant crown to denote solar power, illumination, and spirituality [14]. All these components are represented on our altar except for a torch, which is instead here physically articulated through the placement of a lit lantern in the rear openwork recess to project light through the radiant crown pierced with seven rays.

This iconography is a striking departure from the more common representations of Mithras sacrificing a bull in the tauroctony. The altar’s original positioning to the left of a central one might, in fact, speak to the hierarchical positioning of the dedicator as the sixth of seven graded initiates in the cult, i.e., the penultimate level of Mithraism. Therefore, instead of depicting Mithras, this figure is here suggested as a manifestation of Heliodromus as the embodiment of Sol in ritual activities. Representations with perforated radiant crowns are uncommon, with known examples predominantly recovered from Mithraic contexts [15], primarily from Britain and Germany [16]. For example, Mithraic altars, such as one dedicated to Sol from a Mithraeum further north in Scotland, to the east of Inveresk Roman fort [17] (Figure 4), and others from Germany at Bingen (CIRM 1241) [6] (p. 125) and Mundelsheim [12,18] (p. 242), support this proposition.

It is here proposed that the physical placement of these dedicated altars represents seating arrangements during banquets where Heliodromi, the sixth-grade Mithraists, are depicted on Mithraic reliefs seated on the right of a Pater (father), such as at Konjic in Bosnia (Figure 5). The Pater is the seventh and highest-graded initiate and the earthly representation of Mithras [14] (pp. 89, 122) [19]. On the Konjic relief, other initiates also adorn various paraphernalia, including headgear and masks, aligned to their grade in the cult, including a raven and a lion.

This iconography supports the proposition that the three Carrawburgh altars represent material manifestations of the dedicators, placed according to their ceremonial role and Mithraic grade and, potentially, their military rank. Thus, the Pater (central altar) is positioned facing south into the nave, overseeing activities, flanked by Heliodromus on his right and Perses on his left.

At Carrawburgh, then, in their original placement and facing into the Mithraeum, the dedicators are physically, symbolically, metaphorically, and materially entangled in the symbology and ritual of Mithraicism. In other words, through the combination of epigraphy, iconography, materiality, and context, we can confidently identify three individuals holding the uppermost tier of Mithraic grades at this frontier fort as follows:

(1)

Central altar—the embodiment of Lucius Antonius Proculus as the Pater, the highest-ranking Mithraist and earthly manifestation of Mithras in Mithraic rituals. This is validated by his later raised rank in service of the emperor and ultimately high office in Roman Egypt [8] (p. 51).

(2)

Right altar—the embodiment of Marcus Simplicius Simplex as Heliodromus, the sixth (and penultimate) graded initiate and earthly manifestation of Sol in rituals, who sits on the right of the Pater.

(3)

Left altar—the embodiment of Aulus Cluentius Habitus as Perses, the fifth graded initiate and earthly manifestation of Luna, corresponding with Cautopates and often represented in Mithraic iconography by a crescent moon, further symbolising the partnership between the torchbearers and Mithras in the tauroctony.

A mosaic in the Mithraeum of Felicissimus, Regio V, Insula IX, at Ostia detailing a seven-stepped ladder symbolising grades of initiates validates this proposal. It depicts a representation of Heliodromus with a radiate crown in the penultimate position of the initiate hierarchy (Figure 6) and Perses with a crescent moon below him on the fifth rung. The excavators at Carrawburgh later reference this as the ‘Sun-god’ relief [1] (p. 43), thereby perhaps unconsciously amending their original incorrect assertion of this relief as a depiction of Mithras. A survey of perforated altars to Sol [16] (p. 248) also picks up this erroneous identification, but both interpretations wrongly identify this as a depiction of Sol, not of Heliodromus.

The Sol and Luna altars recovered from a small Mithraeum near the Roman villa at Mundelsheim, Baden-Württemberg, with similarly pierced sun radiates and crescent moons, respectively, refs. [20,21], lend further credence to this proposal (Figure 7). There, the figure thought to depict Sol is dressed exactly as our Carrawburgh figure. The excavators at Carrawburgh recognise that all seven grades of initiates are not always represented in sculpted reliefs, but they do make the connection between seven altars occurring on many reliefs as potentially representative of each step in the grading process [1] (p. 59). That said, they do not associate the altars with physical manifestations of each named initiate in the cult and their connection to the deities represented.

The common feature of seven rays in depictions of Sol, or Heliodromus, also alludes to these seven grades. The altar to Sol at Inveresk, discussed above (see Figure 4), was discovered face-down to the right of an altar to Mithras [17], lending further support to the symbiotic relationship between these two deities and their physical representation as well as their positioning in ceremonial activities undertaken in the Mithraeum. Indeed, the iconography on these altars clearly adopts a dominant position over and above the inscribed text [16], where it is present, though see Figure 7.

Another intriguing feature of the Sol altars discussed here is the absence of pivotal components of the dedicators’ identities, specifically their ethnicity. The Mundelsheim altars bear no inscription; the dedicators at Bingen do not declare their rank, legionary affiliation, or ethnicity; the altar at Carrawburgh contains only the dedicator’s rank; and the altar at Inveresk records a dedication to Sol and the dedicator’s name and rank (Centurion—Gaius Cassius? Flavianus?), but not his legionary affiliation. At Inveresk, the same individual dedicated the associated altar to Mithras, though on that, he does not declare his rank.

Is it then possible that initiates at this sixth grade in Mithraicism are required to keep their identity more closely guarded than others by declaring very limited information, or none, in their inscribed dedications? To take this a step further, at Inveresk, might we be tracking the trajectory of a Centurion and Roman citizen, both through his military rank and elevated grading in the cult of Mithras, through his initial dedication as Heliodromus of the altar to Sol and later altar as Pater to Mithras?

These are rich representations in iconography and material embodiments of central participants in initiate rites of Mithraicism through dedications on monumental inscriptions, votive altars with sculpted depictions of deities, and initiates wearing costumes signifying their respective roles. Taken together with the creative manipulation of light shining through openwork details carved out of altars, this demonstrates the fundamental importance of atmospheric conditions and theatrical performance of all participants and associated material culture woven through ceremonial activities.

It is in this context that the altar of Marcus Simplicius Simplex at Carrawburgh must be understood to fully engage with its pivotal position in the Mithraeum. The interplay of light and darkness, combined with the iconography of the altar and privileged position flanking the right of Mithras at Carrawburgh and at Inveresk and Bingen, strongly supports the prominence they played in Mithraic practice performed in these darkened spaces [16] as well as their physical embodiment of the role played in proceedings by dedicators.

The application of vivid colours would have brought the altars vibrantly to life and fundamentally elevated their performance for all participants. Marcus Simplicius Simplex’s altar is adorned with a rich tapestry of previously unrecognised pigments through which we can now explore these additional layers of meaning.

3. Materials and Methods

3.1. Polychromy on the Altar

The altar was found fragmented into three large sections at the top (see Figure 1) and then underwent reconstruction for display. Aside from cementing these and smaller fragments around the sculpted relief, the altar is reported as not having been subjected to any further intervention from cleaning or conservation (Lindsay Allason-Jones, pers. comm.). It has, therefore, benefited greatly from sensitive and careful curation in recognition of the potential survival of pigments ripe for scientific analysis, making it an ideal candidate for identifying polychromy practices on the Roman frontier. All that said, a dichotomy exists between the survival of pigments on the inscription panel, with numerous areas where tantalising traces of pigments remain visibly extant, and the relief sculpture, which retains no visible colouration. It is likely, therefore, that conservators took the opportunity to ‘clean’ the carved relief during the reconstruction and, in so doing, leached some of the pigment colours. Regrettably, there are no records documenting actions taken during this treatment to validate or refute this (Andrew Parkin, pers. comm.), so we must turn to heritage materials science for answers.

A further complication exists since plaster casts were taken of all three altars to create replicas that were previously installed in the Museum of Antiquities on the University of Newcastle campus. When the museum closed in 2008, the replicas were removed but were in such poor condition that they disintegrated and could not be retained (Andrew Parkin, pers comm.). An additional set of casts, in concrete, now stands in place of the originals in the partially reconstructed Mithraeum at Carrawburgh. The process of creating moulds for these casts could well have lifted surviving pigments from their surface or negatively impacted them through leaching. Again, this hypothesis cannot be tested in the absence of moulds, as their whereabouts are unknown, but the degraded condition of some pigment samples is indisputable; see below.

Fortunately, several areas of colouration were reported during the altar’s excavation [1] (pp. 37–39), including the following:

• The inscribed letters ‘had been painted red’;
• The cloak ‘once painted scarlet’;
• The face contained ‘fragments of gesso in the deepest cutting, showing the whole of his skin had originally been plastered white and painted’;
• A ‘richly decorated scarlet’ fillet hanging down the right side of the figure; and
• A ‘deep scarlet’ stain on the cloak.

The following methods were deployed to identify surviving surface treatments, and the results are presented below, covering firstly the sculpted relief, then the framed inscription panel.

3.2. Methods

Non-invasive microphotography and portable X-ray fluorescence (pXRF) were initially deployed for detailed surface examination of the altar and to characterise the elemental compositions of pigments present. It would have been preferable to undertake exclusively non-destructive analysis, but previous work by the authors on similar sandstone statuary confirmed that some instruments are not useful due to the complex heterogeneous character of the painted surfaces. For example, it is not possible to obtain good contact on uneven carved surfaces, which is essential to produce meaningful FTIR results. Also, the natural fluorescence of sandstone overwhelms the Raman spectrometer, leading to unhelpful results for analysis. It was, therefore, decided to extract microsamples from some areas for further lab-based analysis. The preliminary results prompted deeper exploration with Fourier Transform Infrared Spectroscopy with Attenuated Total Reflection (FTIR-ATR) and X-ray diffraction (XRD) on selected microsamples. Composition tables comprising the full datasets can be supplied on request from the first-named author. The paper takes a novel approach, combining archaeological investigation and scientific analysis insofar as the inherent complexity of the compounds requires explanation, so we address these systematically at the point of reporting before drawing all the data together in a defined ‘Discussion’ section.

3.2.1. Microphotography and Microsampling

Surface examination at the visible and microscopic levels of surviving pigments is fundamental to comprehensively reviewing their condition and revealing painted features. A detailed inspection using a hand-held magnifying glass with in-built LED lights capable of ×20 magnification identified candidates for more detailed analysis. Following this, in-situ digital microphotography was captured using a Dino-Lite Edge Digital Microscope (AM4515ZT), manufactured by AnMo Electronics Corporation, Taiwan, China, to capture high-resolution JPEG images with DinoCapture 2.0 software that controlled illumination and exposure, viewing, export, and measurements. This model has adjustable 20–220× magnification, flexible LED control (FLC), integrated adjustable polariser, Automatic Magnification Reading (AMR), and 1.3-megapixel edge sensor. These instruments have been used to great effect for portable microscopy on a diverse range of archaeological artefacts [22,23,24,25,26,27,28,29].

The combination of digital photography and microscopy identified previously undetected polychromy traces, validating a more detailed programme of analysis incorporating pXRF supported by selective microsampling. Consequently, with permission from the museum curatorial team, microsamples were collected from 16 areas by scraping with a scalpel. These were sealed in labelled glass vials and then studied in the laboratory under a Leica M80 microscope manufactured by Leica Microsystems in Heerbrugg, Switzerland with incident LED light, and photographs were captured with the integrated digital camera (

Table 1. Log of Microsamples extracted from the altar.

No.	Location	Visible Colour
M1	Left inscription frame	Blue
M2	Left inscription frame	Yellow
M3	Inscription background left of D in DEO—1st row	Purple
M4	Gold beside M—2nd row	Gold
M5	Bottom groove of D in DEO—1st row	White
M6	Band above PREF—last line	Dark blue
M7	Left of reversed E—3rd row	Red
M8	Underside of top inscription frame	Purplish blue
M9	Bottom frame	Gold overlying red
M10	Background above N—1st row	Red
M11	Crustation at bottom of M—3rd row	Brownish
M12	Right cheek of deity	Cream
M13	Discoloured area above crack over the sculpted relief (in top altar frame)	Brown
M14	Upper undecorated part of fillet left of deity	Brownish black
M15	Head wreath (left)	None visible
M16	Fillet left of deity	None visible

). Whilst it is preferable to take a minimum number of samples from curated artefacts, the complexity of compounds and layering evident across the altar, including different colours occurring in the same feature, justified the extraction of 16 microsamples, all less than 1 mm in size. These are now available for future analysis as new techniques emerge.

3.2.2. Portable X-Ray Fluorescence (pXRF)

Following exploratory work along with Dr Richard Jones using a Niton XL3t 900 SHE GOLDD Alloy Analyser manufactured by Thermo Fisher Scientific, Waltham, MA, USA, which hinted at possible surface treatment on some features, a comprehensive programme of pXRF analysis was undertaken in 2023. The pXRF instrument used was an Olympus Vanta M Series (VMR-CCC-G2-K) hand-held analyser manufactured by Evident in Waltham, MA, USA, with rhodium anode in a 4-W X-ray tube capable of voltage up to 50 kV. The instrument operates with two beams, one at 10 kV and one at 50 kV. Analyses were undertaken in the GeoChem (G2) mode, where the X-ray tube operated at 40 kV and ∼70 μA to measure heavier elements and at 10 kV and ~90 μA to measure lighter elements. Measurement time was 30 s:10 s for the heavier elements and 20 s for the lighter elements, and the area of analysis was 7.069 mm². Several of the forty elements from Mg to U that the instrument can detect were present below the limit of detection (LoD), and light elements with fluorescent peaks at low energies were poorly resolved at low concentrations. Because the instrument emits X-rays, standard protocols were in place, including ensuring no staff or visitors were in the vicinity during the capture of data.

A total of forty analysis spots were captured across the relief-sculpted upper portion of the altar (Nos. 1–40), three of which were ground in areas considered unlikely to have been painted (Nos. 38–40). A further thirty-four analysis spots were captured across the framed inscription panel (Nos. 1i–34i). Concentrations of each main element associated with pigments are provided in

Table 2. Locations of pXRF analysis spots on the sculpted relief and results expressed as PPM (parts per million).

No.	Sample Location	Mg	Al	Si	P	S	K	Ca	Ti	Mn	Fe	Cu	As	Sr	Pb	Colour	Pigment
38	ground—right side	<LOD	23,361	229,702	1071	5693	417	2356	1517	113	4546	19	<LOD	18	22
39	ground—left side	<LOD	26,078	180,391	935	29,585	3147	25,305	831	401	5322	58	<LOD	39	28
40	ground—front top	<LOD	13,585	263,953	296	8488	1148	4406	1459	87	3059	33	<LOD	21	24
	AVERAGE		21,008	224,682	767	14,589	1571	10,689	1269	200	4309	37		26	25
	STD DEV		6570	42,006.6	414	13,062	1413	12,699	380	174	1150	20		11	3
																	high Ti and Al = conservation material
1	silvery background left of deity	<LOD	31,534	210,527	632	13,132	3470	9734	2276	170	4770	17	<LOD	25	64		conservation material
2	silvery background left of deity	<LOD	25,960	204,447	191	16,642	4305	13,854	1842	174	3973	30	8	30	35		conservation material
3	silvery background left of deity head	<LOD	28,536	168,596	419	23,208	7399	29,829	1475	132	4635	19	9	33	34		conservation material
4	right of deity head (less silvery)	<LOD	31,729	265,783	988	5554	2530	2277	1626	143	5272	29	13	16	28	green	green earth
5	deity forehead	<LOD	25,332	225,718	1655	8890	6233	4575	1687	<LOD	4278	70	<LOD	23	50		calcium phosphate, green earth
6	deity head wreath	<LOD	24,833	188,319	1685	21,925	8460	15,145	2282	106	4554	195	<LOD	29	73		calcium phosphate, greenearth, gypsum, Egyptian blue
7	deity right cheek	<LOD	30,075	196,088	1428	7352	5498	4722	3978	<LOD	4007	74	7	19	47		calcium phosphate, green earth
8	deity eye socket	<LOD	34,399	182,522	918	5679	4632	8483	3330	184	8139	113	17	40	108	blue–green	green earth and Egyptian blue
9	deity hair	<LOD	20,328	212,500	1569	10,737	5821	6090	1941	66	4829	78	7	17	43		calcium phosphate and green earth
10	deity forearm	<LOD	22,743	231,760	442	11,265	1369	6337	821	38	2775	73	<LOD	23	67
11	deity whip in left hand	<LOD	22,938	219,143	480	11,154	1955	7768	1146	84	3341	46	6	25	39
12	deity chest	<LOD	23,737	243,186	1382	11,930	5574	5779	1765	102	2931	109	<LOD	18	87
13	deity abdomen	<LOD	27,936	251,012	1438	10,340	3456	6430	3103	89	3446	55	9	18	58
14	bottom right V of cloak (silvery shine)	<LOD	18,407	194,881	682	7710	1353	5679	3209	185	5005	73	<LOD	27	79	red	Cinnabar/vermilion and possible ochre
15	chest on fold of cloak	<LOD	15,274	225,911	653	7812	2963	3838	1318	52	3403	38	14	19	64	red	Cinnabar/vermilion
16	cloak below shoulder	<LOD	21,935	220,580	1363	12,077	6286	6690	1119	<LOD	3953	92	<LOD	31	96	red	Cinnabar/vermilion
17	crease at v of cloak	<LOD	18,605	202,569	877	10,965	2782	7466	2624	<LOD	3404	86	<LOD	27	65	red	Cinnabar/vermilion
18	neck of cloak	<LOD	19,972	230,544	1170	8153	5372	2893	2488	77	3064	77	17	30	63	red	Cinnabar/vermilion
19	deity neck	<LOD	21,728	179,653	404	5957	1190	5963	1954	79	3967	40	12	18	71
21	brooch outer circle (repeat test)	<LOD	11,940	90,153	401	15,615	2962	26,125	2653	310	8435	167	55	46	245	purple	gypsum, red ochre, realgar, red lead, Egyptian blue
22	background below left elbow	<LOD	18,755	192,037	<LOD	15,267	772	12,367	1403	184	3586	17	9	25	39
23	brown area above top break	<LOD	9074	151,468	2555	14,659	1112	32,674	1417	177	3868	<LOD	<LOD	128	39		conservation material
24	left column near decorated features	<LOD	12,825	92,596	183	37,800	2841	99,983	2022	351	12,157	40	11	725	35	red	gypsum, red ochre, realgar
25	left column décor— horizontal line	<LOD	10,567	69,441	854	26,412	4248	81,444	1878	401	17,458	73	6	670	29	red	gypsum, red ochre, realgar
26	left column décor— circular feature	<LOD	13,634	82,151	278	36,968	2693	105,628	1631	384	17,798	59	14	714	78	red	gypsum, red ochre, realgar
27	left column décor— left vertical line	<LOD	13,006	93,816	455	33,594	4177	86,247	1882	238	18,958	56	7	653	42	red	gypsum, red ochre, realgar
28	left column décor— crescent moon	11,180	14,003	101,692	146	39,664	2210	94,519	1954	177	13,855	52	16	714	16	yellow– green	gypsum, green earth, orpiment
29	left column décor— centre vertical line	<LOD	13,325	81,715	464	38,728	2406	98,786	1187	338	15,113	54	<LOD	639	24	red	gypsum, red ochre
30	left column décor—frame	12,414	13,347	95,088	556	36,698	2427	95,653	1539	280	14,436	96	21	686	38	greyish–brown	green earth, gypsum, red ochre, Egyptian blue
31	right column décor— bottom left triangle	<LOD	25,678	205,022	1469	11,631	5533	6980	2195	100	4939	110	8	16	57	dark purple	Calcium phosphate, Egyptian blue, cinnabar/vermilion
32	right column décor— middle line bottom feature	<LOD	38,851	203,471	861	7421	3277	5601	2100	200	6677	26	15	23	49	Vibrant green	green earth, possible with trace of orpiment
33	right column décor—centre of circular feature	<LOD	18,134	212,874	1037	10,854	4177	6167	1259	64	4547	61	8	15	35	black	calcium phosphate
34	deity carved outer sunbeam (right)	<LOD	32,140	243,007	665	7027	1424	3128	1331	81	3164	31	<LOD	19	29
35	deity crown/wreath (right)	<LOD	24,084	214,961	645	6074	1324	3076	1543	66	3871	38	7	15	42
36	deity carved outer sunbeam (left)	<LOD	10,121	63,897	235	18,397	6097	116,042	1360	178	11,802	49	13	921	17	bright yellow	gypsum, yellow ochre, orpiment
37	deity lips	<LOD	19,345	209,273	1923	9235	6501	4953	1016	90	3036	83	<LOD	14	57

and

Table 3. Locations of pXRF analysis spots on the inscription panel and results expressed as PPM (parts per million).

No.	Sample Location	Al	Si	P	S	K	Ca	Ti	V	Cr	Mn	Fe	Cu	Zn	As	Cd	Sn	Sb	W	Hg	Pb	Colour
38	ground—right side	23,361	229,702	1071	5693	417	2356	1517	<LOD	<LOD	113	4546	19	66	<LOD	<LOD	85	<LOD	<LOD	<LOD	22
39	ground—left side	26,078	180,391	935	29,585	3147	25,305	831	79	<LOD	401	5322	58	80	<LOD	<LOD	<LOD	<LOD	<LOD	<LOD	28
40	ground—front top	13,585	263,953	296	8488	1148	4406	1459	32	<LOD	87	3059	33	63	<LOD	<LOD	<LOD	<LOD	<LOD	6	24
	Average	21,008	224,682	767	14,589	1571	10,689	1269	56		200	4309	37	70			85			6	25
	Std Dev	6570	42,007	414	13,062	1413	12,699	380	33		174	1150	20	9							3
	TOP FRAME
1i	blue spots on left panel frame	22,009	189,368	569	31,598	2874	18,644	2866	35	61	<LOD	4482	89	87	14	<LOD	<LOD	63	<LOD	16	48	blue
2i	yellow spots on inner left panel frame	25,435	171,433	385	7840	2026	8621	2281	<LOD	<LOD	191	5752	88	77	<LOD	<LOD	<LOD	<LOD	<LOD	36	58	yellow
	BOTTOM FRAME
32i	red flecks and gild—bottom frame	24,777	101,029	528	7959	3130	7790	1094	24	55	117	4124	18,456	2753	<LOD	<LOD	<LOD	<LOD	<LOD	<LOD	57	Red and gold
33i	bottom outer frame —crusted area	25,909	212,433	994	10,794	2507	12,092	3606	44	<LOD	252	8445	38	141	16	<LOD	34	<LOD	<LOD	6	83
34i	red flecks and gild—bottom frame	29,129	185,484	887	12,084	2844	18,460	2803	54	130	306	8393	73	157	28	<LOD	<LOD	53	<LOD	<LOD	73	red and gold
	INSCRIPTION BACKGROUND
3i	purple spots on background next to D in DEO	24,109	187,058	212	6851	2706	7173	2102	<LOD	164	138	5199	266	71	31	<LOD	<LOD	<LOD	24	46	45	purple
24i	blue line above R on 4th row	32,976	207,952	786	19,179	1859	5753	2082	90	<LOD	74	4851	175	93	11	<LOD	<LOD	<LOD	<LOD	585	41	blue
25i	blue line above R on 4th row	25,951	214,047	622	14,898	1493	3801	1157	<LOD	<LOD	236	5814	484	113	9	<LOD	<LOD	<LOD	<LOD	13	24	blue
26i	gild? Under final M—4th row	19,102	156,064	226	7481	1627	6733	1745	88	<LOD	115	7333	4627	674	<LOD	<LOD	<LOD	<LOD	67	376	79	gold
28i	dark blue below the top frame	18,120	199,758	<LOD	12,695	615	3545	1063	<LOD	126	386	5680	224	90	25	24	<LOD	49	18	<LOD	55	blue
29i	inscription background under V of VICTOR	29,455	266,674	857	7243	1700	2458	1504	108	<LOD	197	4133	609	127	<LOD	27	<LOD	<LOD	<LOD	19	57
30i	inscription background right of I in SIMPLEX	27,136	253,259	875	8754	977	3569	1635	70	<LOD	305	5805	554	99	7	<LOD	<LOD	<LOD	<LOD	9	58
31i	inscription background under E of PREF	18,750	229,607	1953	13,040	905	7233	1192	45	73	<LOD	10,159	380	85	18	<LOD	<LOD	<LOD	<LOD	47	74
	LETTERS
27i	red dot inside V of VICTOR—1st row	20,364	245,261	592	6868	339	3372	1984	<LOD	<LOD	185	3689	9	28	11	<LOD	<LOD	<LOD	<LOD	11	68	red
4i	white spot in groove of D in DEO—1st row	14,439	246,781	<LOD	37,106	165	23,607	1002	44	554	214	4599	85	45	10	<LOD	72	<LOD	<LOD	1198	43	white
5i	white spot in groove of D in DEO—1st row	16,888	230,091	<LOD	18,819	252	11,474	911	43	515	138	5056	82	65	7	<LOD	<LOD	<LOD	<LOD	1105	43	white
6i	O of DEO—1st row	29,270	183,822	650	7949	2906	4591	1422	61	<LOD	117	5938	4740	828	7	<LOD	<LOD	<LOD	<LOD	1369	84	dark tone
7i	separator between DEO and IN—1st row	32,884	231,669	1098	9017	3262	5898	1210	99	<LOD	166	6122	128	80	24	19	<LOD	<LOD	<LOD	1894	84	dark tone
8i	I of IN—1st row	28,852	234,695	941	7072	1991	5357	2292	55	<LOD	221	4552	48	90	<LOD	27	<LOD	<LOD	<LOD	1529	85	dark tone
9i	I of VICTOR—1st row	27,397	263,897	745	8001	1292	3937	1261	175	<LOD	154	5189	26	86	<LOD	<LOD	40	<LOD	<LOD	1067	104	dark tone
10i	O of VICTOR—1st row	28,409	226,919	1619	6713	1396	3029	1701	78	<LOD	1241	12,803	45	82	20	<LOD	<LOD	<LOD	<LOD	688	108	dark tone
12i	M of MITHRA—2nd row	30,234	232,611	699	9652	1665	7194	1686	<LOD	<LOD	63	4112	269	102	5	<LOD	<LOD	<LOD	<LOD	646	40	dark tone
13i	T of MITHRA—2nd row	33,406	205,278	568	8209	2838	6523	2954	101	<LOD	264	5189	1099	218	7	<LOD	<LOD	<LOD	<LOD	1693	58	dark tone
14i	I of SIM—2nd row	27,944	237,070	1263	8306	1439	4657	1874	77	<LOD	669	7528	80	79	7	<LOD	95	<LOD	<LOD	1334	75	dark tone
15i	L of PLIC—3rd row	32,173	196,619	1720	10,800	2374	9241	1174	102	<LOD	929	9840	285	104	26	<LOD	58	<LOD	<LOD	1675	51	dark tone
16i	C of PLIC—3rd row	32,275	177,857	965	10,063	2320	9558	2087	66	<LOD	934	8476	49	110	17	64	<LOD	<LOD	<LOD	1792	79	dark tone
17i	M of SIMP—3rd row	26,114	222,067	409	15,966	1401	10,595	2156	89	<LOD	160	4298	37	63	9	<LOD	39	<LOD	<LOD	424	37	dark tone
18i	X of SIMPLEX—3rd row	28,484	204,448	1853	7942	2206	7438	2800	<LOD	<LOD	326	12,109	28	135	22	<LOD	<LOD	<LOD	<LOD	998	115	dark tone
19i	P of PREF—4th row	33,937	174,016	1091	9200	2400	11,781	2506	99	<LOD	159	11,409	61	94	16	<LOD	121	<LOD	<LOD	476	62	dark tone
20i	E of PREF—4th row	27,427	158,698	1314	8562	2411	8452	1652	<LOD	<LOD	124	9682	3559	600	14	<LOD	<LOD	65	<LOD	54	46	dark tone
21i	V of V S L M—4th row	27,393	258,223	1097	9699	845	5269	1087	50	<LOD	186	7394	117	76	12	<LOD	<LOD	<LOD	<LOD	891	55	dark tone
22i	L of V S L M—4th row	32,486	236,693	1254	9279	2586	7767	3221	<LOD	<LOD	257	8833	45	88	6	<LOD	94	59	<LOD	157	62	dark tone
23i	M of V S L M—4th row	29,716	196,883	643	7726	1898	6517	1830	79	<LOD	168	8122	52	107	<LOD	<LOD	<LOD	<LOD	<LOD	1787	64	dark tone
11i	flora at end of 1st row	20,397	253,676	774	6244	1142	3161	1274	47	<LOD	432	5630	46	61	<LOD	35	<LOD	147	<LOD	173	37	dark tone

. Elemental concentrations are expressed in parts per million (ppm), and elements related to each painted feature are discussed in-text. Some elements, including Rb and Zr, have been excluded from the discussion as occurring below the instrument’s limit of detection (expressed as <LOD) or unrelated to pigment or other surface treatment, confirmed by the ground analysis spots. The remaining elements provided a level of quantification at various spots in concentrations sufficiently above background levels to confidently identify pigments present, though some only at low trace levels.

3.2.3. Fourier Transform Infrared Spectroscopy with Attenuated Total Reflection (FTIR-ATR)

FTIR–ATR was carried out on microsamples; these were separate from the samples used for XRD and were not embedded in resin. The FTIR–ATR used was a Nicolet iS 10 manufactured by Thermo Electron Scientific Instruments LLC, Madison, WI, USA, with Omnic 9.1 software and fitted with a Universal ATR Sampling Accessory. The ATR crystal used was a diamond/thallium-bromoiodide (C/KRS-5) with a penetration depth up to 2 µm (FTIR–ATR is primarily a surface technique). Thirty-two accumulations were used at a resolution of 4 cm⁻¹. Unless otherwise stated, samples were placed on the ATR with the surface level facing downwards. Omnic software was used to operate the spectrometer, and this also benefited from having a pre-loaded material reference library for spectral comparison with commonly evaluated materials such as polyethylene, which was used as a standard.

3.2.4. X-Ray Diffraction

X-ray diffraction data were collected on a Malvern Panalytical Empyrean diffractometer manufactured in Almelo, The Netherlands, with a copper X-ray source (λ = 1.1.5419 Å) manufactured in Eindhoven, The Netherlands. The samples were placed on a silicon zero-background sample holder. Data were collected between 5 ≤ 2θ ≤ 100° in reflection geometry. Due to the small volume of samples (10–30 flakes < 1 mm in diameter), the collections were taken over extended periods between 17 and 49 h per sample, as required.

The pigment phases present were identified using Highscore (Malvern Pananalytical, Malvern, UK) to search the Crystallographic Open Database (COD) for matching patterns restricted to the elements present as determined using pXRF (X-ray fluorescence) results. The feasibility of these possible pigments being present was then assessed for their accessibility for the period, as well as possible resultant phases from conservation treatments the altar has been subjected to post-excavation.

Using the structural models from the COD for the identified phases, Rietveld refinements were carried out to determine the phase fraction for the crystalline phases in the altar’s fragments.

4. Results

4.1. Visibly Surviving Polychromy

The visual inspection revealed no pigments on the sculpted relief aside from some areas of white in the folds of the deity’s cloak and silvery reflective inclusions restricted only to the left side background (Figure 8i). Paradoxically, a complex matrix of pigments is visibly present on the inscription panel (Figure 8), including

• Bright purplish–blue spots in a line in the overhanging top and exterior left frame (Figure 8a);
• White in D of DEO in the first row (Figure 8a).
• Purple, blue, and gold in the panel background to the right of the D of DEO on the first row (Figure 8a,d).
• Blue line (very thick) overlain with gold above PREF (Figure 8b).
• Yellow spots on the inner groove of the left frame (Figure 8c); Dinolite inspection reveals this overlies black, or this could be the result of degraded pigment (Figure 8f).
• Bright blue on the left panel frame (Figure 8e).
• Gold on several areas, including around the D of DEO (Figure 8a,d,g); the P of PREF (Figure 8b); on the panel background at the final line between the V and after the M; and above the wavy line decoration over the bottom panel frame.
• A crustacean in the final M of the second row (Figure 8h).
• Orange spots on the underside of the top frame (Figure 8j, no microphotograph).
• Red above and below N of IN and above V in the second row (Figure 8k, no microphotograph).

Some of the resulting digital images are striking and entirely unexpected; for example, the gold overlies several areas on the panel (Figure 9). Other samples evidence colour leaching from some pigments where traces remain extant to confirm original pigmentation; see below. This could be the result of degradation processes when the altar was buried and something that has been recognised in other examples of painted Roman sculptures on frontiers from the Antonine Wall [23] as well as the German Limes [30,31] and more widely [32]. More recent interventions could equally have contributed to the process.

4.2. XRF Data from the Sculpted Relief

A Greek papyrus from Roman Egypt dating to the 4th century records the opening prayer to Mithras as ‘Thou shalt see a young god, fair of aspect with flaming locks, clad in a white robe and scarlet cloak and having a crown of fire’ [33] (p. 10). Thus, we might anticipate red pigment on the cloak of our deity, and indeed, the excavators do reference a red staining on this feature, but a microsample here produced no useful visible information. Aside from this, the references by the excavators to colouration are somewhat ambiguous on whether they verify the presence of pigment or merely allude to an expectation of a colour on various features, though the mention of possible gypsum in the carved crevices of the deity’s face is of interest. Table 2 records the locations and results of samples from the sculpted relief analysed by pXRF and is followed by a discussion of the results, which are thereafter compared with XRD results on selected microsamples.

The complex compounds evident through visual inspection and confirmed by pXRF mean it is necessary to explain the results holistically at the point of reporting, below, before drawing the evidence together in a formal Discussion section.

4.2.1. Conservation Treatments

Elevated levels of titanium on the relief could indicate titanium dioxide (TiO₂), a polymorphic compound in nanometric form used by conservators since the 1960s to preserve architectural stone due to its photocatalytic properties and for protection against degradation from soil, pollution, or microorganisms [34]. Certainly, the margins of repaired fragments for the upper part of the altar are clearly identifiable, especially the area around the deity’s head and radiate crown, where fragments of the cementing material have fractured. Additionally, the entire area above the head and in the left background has been treated, including the carved fillets flanking him, with the left fillet entirely delaminated when excavated (see Figure 1).

Elevated aluminium aligns with peaks of potassium and could indicate green earth (phyllosilicate mineral: K[(Al,Fe^III)(Fe^II,Mg](AlSi₃,Si₄)O₁₀(OH)₂), but the samples with both are restricted to the top of the relief on the background and the deity’s face. There is also no corresponding peak of magnesium or iron except for sample 4 to the right of the deity’s head (which does not have elevated potassium). The elevated aluminium in samples around the head, therefore, possibly derives from materials used during conservator intervention, possibly as a titania–alumina composite coating since the addition of alumina to TiO₂ optimises photocatalytic properties [35]. This could also explain the highly reflective inclusions visibly present both to the naked eye and at microscopic levels (Figure 10).

That said, it is entirely possible that green earth is also present in the area, which could explain the higher levels of potassium at several of these features (see also the XRD results, below). The low levels of iron argue against this, but at sample 4, there is a clear visible difference in the character of the matrix with no glittering reflective inclusions.

The high potassium on many features corresponds with elevated titanium and could equally derive from conservation materials, or they could alternatively be a product of degradation of the stone through salinisation [36]; see also the XRD results.

4.2.2. Gypsum

Elevated levels of calcium and sulphur confirm a sulphate-rich gypsum (calcium sulphate dihydrate: CaSO₄·2H₂O) used as a substrate for pigments. Elevated levels of strontium point to a Roman origin for mineral gypsum, i.e., not the product of alteration phases or conservation techniques [37]. This combination of elements is restricted to the outer left sunbeam feature and the left fillet column, which contains a variety of linear and curved features.

4.2.3. Relief Background

Taking account of the lack of clarity in the background due to conservation products discussed above, the presence of green earth cannot be discounted, and certainly, this pigment is known to have been used by Roman artists to emulate the more precious chrysocolla [38,39]. From the results here then, green earth mixed with Egyptian blue (an artificial copper calcium silicate—CaCuSi₄O₁₀), red lead (Dilead(II) lead(IV) oxide: Pb₃O₄), and orpiment (arsenic(III) sulfide: As₂S₃) may have been used to create a dark green [40] background for our deity.

4.2.4. Deity’s Hair and Skin

Peaks of phosphorus, potassium, and titanium in all samples of skin and the deity’s hair make identification of pigments challenging, especially in the cheek and eye socket, which, in common with the background, also contain elevated aluminium. The combined peaks of aluminium, potassium, and traces of arsenic here could indicate the mixing in of an organic dye [41], perhaps madder (Rubia tinctorum L.), such as pink lake madder [42]. This would not be easily detectable with pXRF, but it could have been an appropriate and locally available source of pink/red pigment that has been indicated through Raman spectroscopy for unique distance sculptures further north at the Antonine Wall [43]. Elevated phosphorus is predominantly restricted to skin areas (forehead, cheek, chest, abdomen) and hair, so it is likely that a black, such as bone black (calcium phosphate: Ca₃O₈P₂), has been used in the recipe for depicting skin in common with painted faces from Hellenistic, Etruscan, and other Roman contexts [44,45,46]. All samples with elevated phosphorus are mirrored by peaks of potassium, except for one brown area just above a break at the top of the relief (sample 23), which certainly derives from conservation materials.

4.2.5. Deity’s Cloak

Despite the less clear results reported above, there are some features in the sculpted relief that we can say with certainty were adorned with pigments. For example, mercury is elevated in all samples taken of the deity’s cloak, confirming it was painted with cinnabar/vermilion (Mercury(II) sulfide: HgS), a vibrant, deep red.

4.2.6. Pigment Compounds

Intriguing compounds are present on some features, including:

• Deity’s Eye

Aluminium, potassium, iron, and copper at the eye recess might indicate green earth mixed with ultramarine (Na₇Al₆Si₆O₂₄S₃) and Egyptian blue to create a blue–green eye colour for the deity.

• Brooch

The combined elevations of calcium, manganese, iron, copper, arsenic, and lead on the outer circle of the brooch holding the cloak together suggest mixing of iron oxide (red ochre—Fe₂O₃ (iron(III) oxide)), realgar (arsenic(II) sulfide: As₄S₄), red lead, and possibly Egyptian blue and ultramarine to create a purple. The latter could potentially explain the elevated aluminium (see also the XRD results, below). The elevated calcium here may indicate gypsum as a substrate for pigment application. The mixing of Egyptian blue and haematite (red ochre) is confirmed for Vesuvian wall paintings [41], and Egyptian blue and red lead (a mixed lead oxide—Pb₃O₄) dominate pre-mixed pigments for Pompeian wall paintings [40]. Purple Emperors’ cloaks on spectacular marble reliefs from Nicomedia were created by overlapping cinnabar red and a mixture of vestigial organic pink lake and Egyptian blue [47,48].

The microphotographs of purple areas on our inscription panel, both in situ and under the microscope, confirm the presence of blue elsewhere on the altar, most likely Egyptian blue and/or ultramarine (see below).

Fillet Features

• Red

Red is the dominant colour on the features delicately carved in the fillet to the left of the deity, as indicated by elevated iron and manganese on several features, including

○

A red ochre and realgar mix for the area immediately above the carved features, the top horizontal line, the ring of the circular feature, and the vertical line below the circle;

○

Red ochre for the central vertical line beneath the crescent moon.

• Yellow

The combined peaks of calcium and strontium, as well as iron and traces of arsenic, on what appears to be the original surface of the lower right sunbeam radiating above the deity’s head, are intriguing. The absence of manganese here might suggest the iron derives from yellow ochre (Fe₂O₃·H₂O) as opposed to red or brown, while the trace of arsenic could derive from realgar. However, on balance and given the feature depicted, this is most likely a mixture of yellow ochre and orpiment to create a bright, vibrant yellow to emulate the sun’s rays when illuminated by the recessed torch.

• Purple

Purple is also indicated in the left triangle below the crescent moon on the right fillet through the elevation of copper and mercury, suggesting a mix of Egyptian blue and cinnabar/vermilion, with calcium phosphate indicated by elevated phosphorus, which may have darkened the hue.

• Black

Calcium phosphate is indicated by the elevated peak of phosphorous on the circle in the centre of the circular rim on the fillet to the right of the deity.

• Green

Elevated manganese, iron, and arsenic on the crescent moon here suggest a mix of green earth and orpiment to create a muted yellowish–green consistent with this feature.

Elevated aluminium, potassium, and iron, combined with traces of arsenic on the central line of the linear feature separating the two bottom triangles, indicate green earth mixed with a small amount of realgar or orpiment. Given that realgar mixed with green earth would produce a murky brown, it is here proposed that the arsenic most likely derives from orpiment to produce a bright and vibrant green.

The combination of elevated phosphorus, sulphur, calcium, and potassium, along with copper on the deity’s wreath, might be suggestive of a green earth and ultramarine mix with calcium phosphate, perhaps depicting a deep green for the leafy wreath. However, given its proximity to the fragments reconstructed by conservators here, this is uncertain.

• Brown

The combined peaks of magnesium, manganese, iron, copper, and traces of arsenic on the frame for the carved fillet on the left suggest a mixture of green earth, red ochre, Egyptian blue, and either realgar or orpiment (probably the former, given the XRD results for this feature). The latter would desaturate the other individually vibrant colours to create a complex and muted greyish–brown hue for the carved frame of the fillet on the left (and presumably also the right, but no sample was collected from this feature).

4.3. The Inscription Panel

There are numerous areas of visibly extant pigment on the inscription panel, although the remaining traces are challenging to detect with the naked eye. White is visible in the first letter of the inscription, and red in some letters, as well as purplish–blue and orange spots on the top frame and gold on the bottom frame. Yellow and blue are visibly present on the left panel frame grooves, along with purple on the outer frame, flanked on the exterior by yellow.

Mixing and layering of pigments is now increasingly studied and understood as a primary method of Roman artists creating desired hues for realism in their articulation of sculpted stone [23,30,49,50], but polychromy on relief-sculpted inscriptions remains an unexplored area of study. It should raise no surprise then that, as noted above, the panel background is dominated by a rich palette of colours, including purple, blue, yellow, black, red, and gold, with no visible distinction between some features or areas. This results in a complex sequence of colours appearing together, for example, on the top left, close to the first letters, and other spots. These traces appear in the same matrix in most instances, confirming that a highly complex mixture of pigments created a purple background to the panel (Figure 11, top), which was then covered with a ‘gilding’ material. It is not possible to determine whether this gilding derives from a later episode of colouration or is an intended feature of the original polychromatic scheme, but it has clearly been applied across the entire panel, since it visibly frames inscribed letters and between lines (see Figure 11, middle and bottom; Figure 12, top), confirming its application respected the existing red-painted letters. Some pigments have evidently degraded at a molecular level so that they are invisible to the naked eye, but colour leaching is observable under the light microscope (Figure 12, bottom), a process that has also been identified on Antonine Wall distance sculptures [23].

Initial visible inspection of samples under the light microscope, then, appears to identify the vibrant yellow of orpiment, the orange of realgar, the distinctive red of cinnabar/vermilion, and a bold blue matrix turning to purple in some areas that evidently derives from a compound of different pigments and a highly reflective golden topcoat.

The results of pXRF analysis on the framed inscription are contained in Table 3, followed by a full discussion on the contents.

4.3.1. Panel Background

Elemental peaks associated with the panel are, unsurprisingly, as complex as the visibly surviving polychromy and must be carefully disentangled. For instance, a bright and vibrant blue survives on several areas across the frame and panel background, on some features, such as the background to the left of DEO (first line). On the external left frame, the blue visibly alters to purple, for example, in the centre of microsamples M8 and M3 (see Figure 12).

Sandstone is characteristically high in silica, which may mask peaks of this element on some pigments, for example, the blue; see below. Iron is also characteristically elevated in sandstone [51], but the ground samples provide a base for the latter, confirming elevations where encountered.

• Blue

Turning firstly to the blue observed in several spots, elevated potassium in some samples could indicate ultramarine; here, the elevated sulphur also predominantly corresponds with high calcium, indicating the additional presence of a gypsum substrate. Higher levels of iron, corresponding with copper, combine with relatively low calcium and silica, which might suggest the presence of azurite, a pigment known to lose colour when finely ground [52], which could explain the blue colour leaching. However, on balance and taking account of the XRD results below, Egyptian blue is confirmed as the source of elevated copper, with the silica naturally present in the sandstone interfering with silica levels detected by the pXRF instrument.

• Red

The red colourant is evidently derived from a compound of different pigments, possibly mixed with the blues in advance of painting onto the altar to create purple. This is a practice confirmed through analysis of pigment receptacles from Vesuvian and Pompeian contexts containing similar recipes [40,41].

The compound comprises:

• Elevated manganese and iron indicate red ochre;
• Traces of mercury confirm cinnabar/vermilion;
• Traces of arsenic suggest realgar and could explain bright orange spots visible in some areas (see Figure 11, top, and Figure 13);
• Traces of lead suggest red lead.
• Green

Elevated aluminium combined with some elevated potassium and iron in several spots confirms green earth in this mixture, perhaps used to darken the hue; see also XRD results.

• Gold

The presence of elevated titanium is enigmatic since this element is not associated with any known Roman pigment, yet it appears consistently across all painted features of the panel. Unlike the carved relief above, this may not derive from conservation materials since the inscription shows no visible signs of intervention (though see the XRD results, below). Intriguingly, high titanium corresponds with elevated copper, aluminium, and zinc, suggesting a symbiotic relationship with these elements that may be associated with the highly reflective gold-coloured material applied as a final layer across the panel.

Visibly, this golden pigment does not cover the inscribed letters since it halts at the letter margins (see Figure 11, middle). Physically, the material is tar-like in character, to the extent that microsamples stuck to the scalpel on extraction. Elementally, it is similar to orichalcum (aurichalcum), a prestigious brass alloy of copper and zinc, sometimes lead, though susceptible to degradation through zinc depletion [53]. This material has been used since at least the 6th century BCE [54] to manufacture various artefacts [55,56] and for casting high-denomination coins [57]. Orichalcum could also potentially explain traces of cadmium, tin, antimony, and chromium in several pXRF samples, as well as arsenic and lead traces. This uncertainty prompted a deeper exploration of samples using FTIR-ATR and XRD, which produced surprising and deeply significant results (see below).

4.3.2. Panel Frame

• Yellow and orange

Vibrant yellow spots clearly visible in the inner groove on the left inscription panel frame, as well as the external edge, are characteristic of orpiment (see Figure 8 and Figure 12, sample M2, and Figure 13). Although arsenic is not elevated in this sample (2i), it is entirely possible that the instrument was not well placed to detect all elements, given the depth of the groove. Intriguingly, unlike the newly confirmed practice of using carbon black to emphasise garment folds on frontier statuary [30], the brightness of the yellow suggests its use here to highlight the panel groove and to contrast the outer frame’s blue- and purple-coloured features (see below for detail).

Orange spots visible on the underside of the top frame have the appearance of realgar and serve the opposite effect, i.e., playing with darker shades in the shadows of the frame. Certainly, arsenic is present at trace levels across the entire inscription panel, confirming orpiment and/or realgar are components of the artist’s palette here [40]; this too is confirmed with XRD below.

• Blue

The different hue of bright blue traces visible on the left inner frame could suggest an alternative pigment or at least a different mixture of the ultramarine and Egyptian blue already identified above on the panel background, since there is no visible hint of purple tone, in contrast to the outer frame (see Figure 13). This cannot be confirmed in the pXRF results since the uneven surface may have, again, interfered with the reading, but the visibly different character of this blue compared with those on the panel suggests a bright blue internal frame and purple outer, both flanked with yellow.

4.3.3. Inscribed Letters

• Gypsum

Elevated levels of calcium and sulphur confirm a sulphate-rich gypsum in letter D of DEO in the first line through pXRF samples 4i and 5i, M of SIMPLEX in the third row (sample 17i), and possibly P of PREF in the fourth row. This corresponds with the visible white colouration in the D of DEO and the microsample (see Figure 9, M5).

• Red

Red microsamples observed under a light microscope are characteristic of cinnabar/vermilion, and the pXRF results with elevated mercury corroborate this pigment as a trace element across the entire panel but at very significant elevations in the letters, confirming they were depicted in bright, vibrant cinnabar/vermilion red.

4.4. FTIR-ATR and XRD Results

Given the unusual complexity of some samples, particularly the golden topcoat on the inscription panel background, it was decided to expand the analytical repertoire to include FTIR-ATR and XRD.

4.4.1. FTIR-ATR

Firstly, FTIR-ATR was deployed on samples M4 and M9 to identify the tacky, likely organic material present (Figure 14, top). The spectrum shows the characteristic absorption bands of beeswax at 2915, 2849, 1742, 147, 1463, 1175, 730, and 720 cm⁻¹ [58]. Thus, while recognising these samples are rather ‘noisy’ due to their small and heterogeneous character, we can confirm beeswax as the organic component mixed in, or over, the golden top layer to create a shiny and highly reflective surface. Although the blue in sample M1 could not be identified, FTIR does confirm traces of wax here also, as well as calcium sulphate (Figure 14, bottom), corresponding with the pXRF results.

4.4.2. XRD Results

Six samples were selected for XRD analysis from both the inscription and the relief sculpture (M1, M2, M3, M4, M9, and M16), and the results are enlightening (Figure 15 and Figure 16). They support the pXRF results, including the confirmed presence of Egyptian blue, ultramarine, orpiment, green earth, cinnabar/vermilion, calcium phosphate, red lead, realgar, and yellow ochre. Carbon black is typically amorphous, but crystalline phases can be found in the form of graphite, which produces a characteristic peak at around 20° in 2θ, making it also identifiable in sample M4. The results further confirm that pigments do survive on some of the relief-sculpted features, such as the carved fillet to the left of the deity, which contains a compound of realgar, yellow ochre, and metacinnabar (a different polymorph of cinnabar that is still HgS but black in colour, whereas cinnabar can transform into more stable metacinnabar at high temperatures). In addition, XRD identifies kosmochlor (NaCr³⁺Si₂O₆), a sodium chromium pyroxene mineral from the same family as jadeite which could have been used as a green pigment, as well as FeS, an iron(II) sulfide metallic black pigment. Other results are likely the products of conservation, including strontium dichromate trihydrate (Sr (Cr₂O₇)(H₂O)₃)—a yellowish compound not available until the 18th century; bismuth vanadate and clinobisvanite (a product of bismuth vanadium oxide); sodium chlorate and dickite. Sr (Cr₂O₇)(H₂O)₃ and bismuth vanadate are likely to have been used by conservators to colour-match repairs to the original material. Sodium chlorate can be derived from many bleaching and cleaning products that may have been used to remove dirt and debris from the surface of the altar but will have a leaching effect. Dickite is a clay mineral possibly used in repairs as part of a cement or plaster.

Critically, XRD identifies the ‘gold’ colourant as pyrite (FeS₂)—commonly known as fool’s gold. The presence of Felsobanyaite, which occurs as a weathering product under acidic conditions associated with pyrite decomposition, as well as tochilinite (a reaction of fool’s gold) and pyrrhotite (fool’s gold which has lost its iron content, Fe_1−x), corroborates this. Pyrite originates from the Greek word for ‘fire’ and is well known to Greeks and Romans as a stone that sparks on percussion [59]. Pliny (Book 36, Chapter 30) [60] states pyrites from mines near Acamus in Cyprus:

‘Resembles copper one variety of it being silver, another of golden colour. There are various methods of melting these stones, some persons fusing them twice, or three times even, in honey, till all the liquid has evaporated; while others, again, calcine them upon hot coals, and, after treating them with honey, wash them like copper. It is roasted in various ways. Some roast it two or three times in honey until it is no longer fluid. Others firstly roast it on hot coals and then in honey. Afterwards it is drenched like copper. The medicinal properties which these minerals possess are of a calorific, desiccative, dispersive, and resolvent nature, and, applied topically, they cause indurations to suppurate. They are employed also, in a crude state and pulverized, for the cure of scrofulous sores and boils.’

Pyrite (fool’s gold) has never previously been recorded as a pigment; this result breaks new interpretive ground and transforms our understanding of the materials Roman artists were exploiting for their palette. If our pyrite was mixed with honey as Pliny [60] suggests, we can confirm another product of the bee, along with the beeswax, identified with FTIR-ATR.

○

Chemically, Green Earth is a potassium aluminosilicate/aluminosulphate material with doping of magnesium and iron (K[(Mg, Fe²⁺)(Al, Fe³⁺)](AlSi₃, AlS₃)(OH)₂). Biotite is a variant in this mineral class with the chemical composition K₂[Mg₄Fe₂](AlSi₃O₁₀)₂(OH)₄.

○

Lazurite is the mineral identifier for the pigment ultramarine, having the chemical formula (Na_3.5Ca_0.5)₈[S₃(SO₄)₆(Al₆Si₆O₂₄)₄].

○

Kosmochlor comes from the jadeite mineral class and is green in colour.

○

There are different iron sulfide variants: pyrite (FeS₂), otherwise known as fool’s gold; tochilinite and the oxidised form of pyrite formed over time when exposed to the environment; and pyrrhotite. This form has a reduced iron content (Fe_(1−x)S) where 0 < x > 0.125. This is a mineral that often has a red colouration, hence the name coming from the Greek for ‘flame-coloured’.

○

The other two iron sulfide phases are FeS, a metallic black pigment, and Felsobanyaite, a weathering product under acidic conditions associated with pyrite decomposition.

○

HgS was identified in the form of metacinnabar, a black pigment rather than the red pigmentation seen with cinnabar.

○

The NaMg(PO₃)₃ identified is likely from a fossilised creature in the stone prior to sculpting, as is generally found in fossilisation of bones; another phase likely from the stone (or from later repairs) is dickite, a natural clay mineral.

○

The final two phases identified are bismuth vanadate (bismuth vanadium oxide), a known modern pigment likely from trying to match the colour of repair works to the original stone material, which is found naturally as a rare rock mineral, and sodium chlorate, a bleach cleaning byproduct.

To further explore the potential materials used for conservation, XRD was undetaken on an additional sample (M13), which was extracted from the mixture used to plaster/cement over the crack joining the top sections of the altar (Figure 17). This too detected bismuth vanadate along with calcite, quartz, and metanatrolite (Na₁₂Al₈Si₁₂O₄₀), a sodium aluminosilicate stone. This provides irrefutable evidence that bismuth vanadate, calcite, quartz, sand, and other crushed stone were likely mixed to blend the restored areas with the original sandstone during conservation treatment.

5. Polychromy Palette

The results have revealed a rich and unprecedented palette of pigments that permit the re-imagining of this monumental altar. Some colours were evidently mixed into previously unrecorded recipes to create desired hues depicting different sculpted features and inscribed letters (

Table 4. Pigments on the Carrawburgh altar.

Pigment	Chemical Formula	Feature
Gypsum	Calcium Sulphate Dihydrate: CaSO4·2H2O	- Relief sculpture of deity - Features in fillet left of deity - Brooch on deity’s cloak - Inscribed letters
Bone Black	Calcium Phosphate: Ca3O8P2	- Deity’s cheek, forehead, hair, eye socket - Features in left and right fillets - Purple compound featured in left fillet - Dark green compound on deity’s wreath
Carbon Black	C	- Inscription background with gold sample (M4)
Green Earth	Phyllosilicate Mineral: K[(Al,FeIII),(FeII,Mg](AlSi3,Si4)O10(OH)2	- Background of deity - Deity’s cheek, forehead, hair, eye socket - Features in left and right fillets - Greyish–brown compound on left fillet - Vibrant compound feature on left fillet - Dark green compound on deity’s wreath
Orpiment	Arsenic(III) Sulfide: As2S3	- Crescent moon on left fillet - Linear feature on right fillet - Carved sunbeam beside deity’s head - Vibrant green compound feature on left fillet - Vibrant yellow compound on sunbeam - Carved inner groove of side panel frame
Egyptian blue	An artificial copper calcium silicate - CaCuSi4O10	- Purple compound on deity’s brooch - Purple compound on inscription panel - Purple compound on feature in left fillet - Bright blue inner frame of panel - Greyish–brown compound on left fillet - Dark green compound on deity’s wreath
Ultramarine	Na7Al6Si6O24S3	- Purple compound on deity’s brooch - Purple compound on inscription panel - Purple compound on feature in left fillet
Cinnabar/Vermilion	Mercury(II) Sulfide: HgS	- Inscribed letters - Purple compound on inscription panel - Purple compound on feature in left fillet
Realgar	Arsenic(II) Sulfide: As4S4	- Purple compound on inscription panel - Purple compound on deity’s brooch - Greyish–brown compound on left fillet - Underside of top carved panel frame - Identified with XRD on carved fillet left of deity
Red Lead	Dilead(II) Lead(IV) Oxide: Pb3O4	- Purple compound on deity’s brooch - Purple compound on inscription panel
Red Ochre	Fe2O3 (iron(III) oxide)	- Purple compound on inscription panel - Features on left fillet
Madder Lake?	Rubia Tinctorum: C14H8O4	- Possibly part of the compound on deity’s skin, hair and eye socket
Yellow Ochre	Hydrated Iron(III) Oxide-Hydroxide Limonite: Fe2O3·nH2O	- Vibrant yellow compound on sunbeam
Pyrite (Fool’s Gold)	FeS2	- Top layer over inscription panel
Kosmochlor	NaCr3+Si2O6	- Identified by XRD on sample M1 on the left inscription frame
Metacinnabar	HgS	- Identified by XRD on the carved fillet left of deity
FeS	Iron(II) sulfide—metallic black pigment	- Identified by XRD on the left panel frame

; Figure 18 and Figure 19).

6. Discussion and Digital Reconstruction

Interventions by conservators prioritising the reconstruction of fragmented top sections have undoubtedly negatively impacted the survival of pigments on the altar’s relief-sculpted features, at least on the visible level. The repeated elevations of titanium on several features in this area, particularly around the deity’s head, wreath, and most of the radiate sunbeams, likely derive from that work and have interfered with the pXRF results, making it challenging to identify pigments that may once have adorned the carved features. Intervention is further confirmed through elevated levels of sodium chlorate, probably deriving from a cleaning agent, combined with bismuth vanadate/bismuth vanadium oxide and dickite for colour-matching identified through XRD. The additional XRD results from sample M13, which is certainly a conservation product, verify that these materials were used in the altar restoration treatment. This is regrettable, but a salutary lesson to conservators, excavators, and curators to prioritise the engagement of a heritage materials scientist to undertake a detailed survey and analysis of any stone sculptures, sculpted reliefs, or inscriptions from antiquity before undertaking any conservation treatments.

Despite this intervention, pigments do survive on some features along with gypsum, including a vibrant yellow sunbeam emulating the sun’s rays radiating above the deity’s head, created from a mixture of orpiment and yellow ochre. Additional mixed recipes on the relief include possible blue–green eyes of the deity who is wearing a vibrant cinnabar/vermilion red cloak held in place with a purple-rimmed annular brooch. Their head is crowned by a deep green wreath, and the surviving features delicately carved into the fillets flanking him were painted variously in purple, red, yellow, and green.

Like the exceptionally surviving pigments recently reported on from Pompeian contexts, where a compound of Egyptian blue and red lead dominates colours mixed for shading [40], the artist painted our altar with a recipe comprised of Egyptian blue, ultramarine, cinnabar/vermilion, red ochre, realgar, and red lead mixed to create desired hues, as well as yellow ochre and black metacinnabar. This is enlightening and confirms that a wide range of materials was readily available to artisans on the northwestern boundary of the Empire.

Extrapolating these results confirms a highly complex practice of pigment mixtures employed to depict sculpted and inscribed features. The consistency of elevated elemental peaks across the panel, including letters, confirms the entire panel was painted purple in advance of letters being depicted in cinnabar/vermilion red. Then the background was painted gold, not necessarily contemporaneously. This stratigraphic sequence is confirmed by in situ visual inspection as well as on microsamples.

pXRF and XRD results corroborate the visual inspection to identify a complex practice of mixing various colours to depict each carved feature. Critically, the identification of a purple inscription background is deeply enigmatic and a groundbreaking result that completely transforms our understanding of epigraphic practice and performance.

This finding is unparalleled in Roman epigraphy and speaks to the status of individual Mithraic initiates who appear to be elevating their social, political, religious, and material positions by aligning themselves with a colour traditionally associated with imperial reserve [47]. Certainly, purple is a well-documented status symbol in Antiquity [61,62].

That the panel is thereafter covered in a golden material (see Figure 12, top) further reinforces the indisputable significance of these results since this is derived, not from gold gilding, but from golden reflective material identified as pyrite. This ‘fool’s gold’ is equally unreported in polychromy research; the only potential parallel for this compound derives from pyrite, more commonly used as a firelight or for medicinal purposes according to Pliny [60]. On our altar, it is mixed with beeswax, as confirmed by FTIR results, to create a shiny, spreadable substance. The low FTIR bands on samples at 1583 and 1541 cm−1 could derive from saponification of the beeswax known from Punic wax production recorded in purple pigment samples from Vesuvian wall paintings [41]. This is further supported in the historical sources by Pliny’s [60] (Vol 9, 33–35) recipe of boiling beeswax in seawater and straining to remove impurities, as well as Vitruvius’ recording of beeswax varnish to prevent the blackening of cinnabar, a pivotal component of our purple recipe. The same FTIR peaks are present on all microsamples analysed (M1, M2, M3, M4, M9, and M16), confirming beeswax was used across all painted features of the altar, a technique that has been recorded as far back as the 26th Egyptian Dynasty (664–525 BCE) on painted sarcophagi [58]. Pliny [60] (book 45) also states that cinnabar can turn black if ‘adultered’, but if genuine, it retains its original colour, which lends further credence to the presence and purpose of beeswax to ensure this red component of the compounds remains steadfast.

The specific combination of golden and purple colouration marks this monument out as possessing an elevated social status of Roman luxuria, associated with, and controlled by, the State [63,64]. Pyrite in malleable form is recorded as being used for medicinal purposes but never previously identified as a pigment. Is it then possible that a powdered form of pyrite was mixed with beeswax to provide the illusion of shiny gold gilding covering the panel? Certainly, the uniquely angular character of our pigment (see Figure 15, bottom) combined with the FTIR results lends weight to this proposition.

While the scientific analysis of this iconic monument has proven invaluable, they do not help us to fully engage with the powerful performative qualities of the altar that ancient audiences would have experienced. For that, we must turn to digital reconstruction. Therefore, based upon these results, it is now possible to present an authentic reconstruction of how the altar may have originally been decorated to reaffirm its status as a powerful and hugely impactful monument (Figure 20).

7. Conclusions

This multi-technique analysis cautions that the symbolic significance of monumental inscriptions can only be fully appreciated by delving deeper than their physical attributes, epigraphic articulation, and material properties. The combined approach to depositional contexts, textual content, artistic articulation, and performative components supported by scientific analysis provides an innovative platform for reinvestigating curated collections of carved stones. This unparalleled research breaks new interpretive ground by untangling these interconnected elements of an altar to Mithras from the military zone on the boundary of the Empire, including its iconography, inscription, and polychromatic decoration. Through this, we identify, for the first time, high-ranking Mithraic initiates on the frontier. The ramifications are unprecedented, providing fresh insights into one of the most mysterious cults in the Roman pantheon. Critically, we can now establish the symbolic significance of inscribed altars as the physical embodiment of central people and their symbiotic relationships with deities that are integral to Mithraic practice.

We can now cast light on the pivotal role of polychromy, including complex compounds that created previously unrecorded hues, in the performative activities of Mithraic rituals. What we cannot determine with certainty is whether the application of the golden topcoat on the inscription panel is contemporaneous with the purple it overlies. If it is, then it raises interesting questions on why the gold covers a colour historically restricted to the imperial realm—was it deliberately hiding the dedicator’s illicit appropriation of purple? If not, then it speaks to more than one episode in the trajectory of this altar to Mithras and changing practices over time. On balance, and given the artist’s respect for the inscribed red letters when applying the gold, it is likely contemporaneous with the purple and an intriguing, secretive act of subterfuge.

In any event, the gold finishing layer would have elevated this stone monument to an entirely unprecedented level. Embedding the illusion of a gold inscription created an exceptional reflective surface to bounce the open flames from lanterns, casting light into the Mithraeum. In the context of this deliberately darkened space, surrounded by open lamp flames casting shadows and the unique scent of burning pinecones, cult participants would have been completely immersed in sensory memories and heightened emotions [65,66] woven through their olfactory experience and entangled in the powerful performance of this majestic monument.