From tiny to immense: Geological spotlight on the Alexander Mosaic

A key challenge in the art and archaeological field is the instrumental analysis of objects and materials while preserving their integrity. In this study, the world-renowned artwork Alexander Mosaic (The Issus Battle, collection of the National Archaeological Museum of Naples, IT), the most iconic representation of the face of the Macedonian king Alexander the Great coming from a Pompeii domus, was thoroughly analyzed with mobile and non-invasive methods, within a great project of restoration started in 2020. Representative areas of the Mosaic, overall consisting of ca. two million of tesserae, was studied by in situ videomicroscopy, infrared thermography (IRT), multispectral imaging, portable X-ray fluorescence (pXRF), Fourier transform infrared (FTIR) and Raman spectroscopy. Ten tesserae colors were discriminated, and hypotheses on their geological provenances are proposed. Plasters, mineral components and other substances of old protective materials were characterized. The information obtained with this approach paved the way to knowledgeable restoration.

Fig 1. The Alexander Mosaic (MANN, Naples, Italy) and in situ analyses.

The Alexander Mosaic and some pictures displaying the nondestructive noninvasive archaeometric campaign: A) The Alexander Mosaic (dimensions 583 x 325 cm) before the restoration intervention at MANN (September 2020), with indication of the examined areas (no. of point-analysis carried out by noninvasive techniques: area 1, 36 points; area 2, 23 points; area 3, 55 points; area 4, 106 points; area 5, 82 points). B) the endoscopic inspection of the back side. C) in-situ optical microscopy (OM). D) portable X-ray florescence (pXRF). E) Raman spectroscopy. F) External Reflectance Fourier transform infrared (ER-FTIR) spectroscopy. G) infrared thermography (IRT). H) tesserae counting. I) multispectral imaging.

https://doi.org/10.1371/journal.pone.0315188.g001

Conservation state and surface materials

A fundamental preliminary aspect related to the diagnostics of the mosaic was the evaluation of its state of conservation and surficial treatments. This was achieved by investigating the entire surface down to more selected areas, also interpolating different instrumental analyses (Fig 2A–2K). In particular, among the multispectral imaging techniques, reflected and photo-induced luminescence images captured on the entire surface of the mosaic provide, as a whole, information about the nature of the materials used for the manufacture of the Mosaic, as well as the occurrence of surface materials (both organic and inorganic) due to intentional treatments (e.g., protective coatings, consolidants, pigments) and/or alteration processes (revealed by the presence of oxalates, sulphates, nitrates). Reflected images (VIS, IRR, UVR) are obtained when the wavelength of the incident radiation and the emitted radiation are the same. Among them, VIS images capture objects as seen by our eyes suggesting, for this case study, that tesserae were composed of ten types of colors masterfully combined to enhance artistic effects characterizing the Alexander Mosaic. In this case of photo-induced luminescence imaging, the wavelength of the emitted radiation is shorter than the wavelength of excitation (incident) radiation. Capturing images in the visible region by the luminescence (fluorescence) effects of a body irradiated with UV radiation (400 nm), it is possible to obtain UVL images to appreciate significant differences among the tesserae beyond what is visible.

The UVL images show that Alexander’s face (Fig 2B), is composed of several shades of pink tesserae with appreciable changes in luminescence effects likely related to different chemical composition of the tesserae. A comparable effect on the luminescence can be observed on the chiaroscuro of the horse’s legs at the right bottom of the mosaic (Fig 2C and 2D), also in this case likely due to compositional differences of lithotypes. The UVL images then guided the density of analyses, where the artistic effects were accompanied by tesserae with different luminescence response, including restored areas.

Multispectral imaging also highlighted the distribution of additional materials onto the surface of the Mosaic (Fig 2E–2K). Mortars with different compositions, as well as superimposed decorative layers, are evident by UVL measurements (Fig 2E), whereas the velatino, quite widespread as consolidation and restoration intervention, appears very clearly in UVL and IRR images (Fig 2F and 2G). Dripping and heterogeneity of the adhesive surficial materials can be noted in Fig 2H and 2I (UVL and UVR images), whereas a likely organic coating with a strong luminescence effect is observed in Fig 2I–2K (UVL, UVR and VIS images).

According to the combined use of the vibrational spectroscopic techniques (i.e., Raman and ER-FTIR), the surface materials mainly consist of gypsum, natural waxes and traces of oxalates (S1 Table). The presence of gypsum is also confirmed by the very frequent occurrence of sulfur in pXRF measurements (S2 Table), along with the ubiquitous calcium signals. Gypsum was detected in OM as sporadic whitish deposits covering jointing mortars tesserae. Oxalates also occurs as the most stable calcium oxalate phase, namely whewellite [Ca(C₂O₄)·H₂O]. As far as waxes are concerned, the typical spectral signal of natural ester-based waxes was identified characterized by the presence of a small derivative-like effect around 1740 cm^-1 in their ER-FTIR spectra (S2 Fig) attributable to the C = O bond vibration.

This feature allowed us to exclude the use of mineral wax (ceresin) or similar synthetic products, such as paraffin, and microcrystalline wax. However, it is hard to identify the exact composition of the natural wax applied [17] since this category of products (such as beeswax, carnauba, montana, candelilla wax, etc.) show signals that are quite comparable to each other, or often compromised by spectral deformation and/or by the overlap with the other detected substances’ absorption bands [17]. Actually, natural wax could not be ascribable to an original coating, as its low thermal stability [18] could not have withstood the high temperatures derived from the unexpected event of 79 CE.

IRT has revealed some thermal anomalies in different areas (Fig 3) for instance, around the black horse’s head or near the dry tree trunk. It is worth noting that the IRT information pointed out that the areas affected by deterioration do not correspond to the ones that had been treated with the velinatura, which seems to be fulfilling its consolidation purpose. Rather, these anomalies could be attributed to volumetric changes in the mortars in response to temperature and humidity variations and/or deformations triggered by kinematic stresses (natural or artificial), inducing a depression in the right side of the mosaic, as also reported in [10]. This depression led to a non-perfect planar orientation of the tesserae, thus affecting the emissivity and consequently the recorded temperature.

The endoscopic observations on the back sides (S3 Fig), among the wooden beams used for supporting the Mosaic in the current vertical position, as also detected by the georadar surveys [10] and inferred from the historical sources, detected many spaces filled with mostly paper, including newspaper; moreover, evidences of binder depositions are found, probably composed of vinyl or gypsum-based substances, likely in order to attempt to secure the structure of the Mosaic with a very fluid adhesive substance.

Composition and textural features of tesserae

The results of the combined spectroscopic analyses are reported in this section with the aim of inferring the compositional features of the mosaic tesserae. As aforementioned, the Alexander Mosaic is composed of ca. 1.9 ·10⁶ tesserae displaying ten kinds of colors and a wide range of micro-textures masterfully combined to enhance artistic effects of the artworks (Fig 4). These tesserae are joint by a whitish mortar well observed by means of OM images (Fig 4) showing faded chromatic variations resembling those of the tesserae (S4 Fig), likely made to give continuity to the depicted scene as in a painting, as also mentioned in literature [5]. Raman spectroscopy also detected traces of barium sulphates (i.e., baryte BaSO₄) and lead carbonates (i.e., cerussite PbCO₃) probably attributable, according to previous investigation [13], to contamination by restoration mortars.

Portable XRF measurements detected a very variable chemical composition of all tesserae in terms of type of elements and their NET intensities (S2 Table). Ca, Fe and S are ubiquitous (Table 1).

The occurrence of Ca is mostly ascribed to chemical composition of both carbonate-based components (i.e., tesserae and/or possible contamination of jointing mortars) and, in the case of S, of sulphates as surface materials (together with oxalates). Fe is a common element both in carbonate and silicate rocks. According to the PCA statistical treatment of pXRF dataset (Fig 5), Ca and Fe are the elements that most affect the dataset influencing the contribution of other elements on the total variance. First and second components (PC1 and PC2) have a very good score, responding to the 96% of the total variance, 80.8% and 15.1% respectively (Fig 5). The statistical treatment of pXRF data permitted discrimination in four groups of tesserae with the only exception of an outlier represented by a single green tessera.

The nature of the coating materials

Analyses of the materials found onto the surface of the Mosaic display an almost ubiquitous presence of calcium sulphate (gypsum), natural wax as well as traces of calcium oxalate.

The use of wax as a protection—for modern and ancient mosaics and paintings—is well-documented and recognized in archaeometry and conservation science. Wax is a natural product, of animal or plant origin, widely used as a protective coating for its compatibility with all kinds of materials (natural stone, wood, cellulosic materials, etc.). Spectroscopic studies [19] suggest that waxes can be easily distinguished from other organic substances (e.g., proteins and resins), whenever used as protective or strengthening agents in cultural heritage [20, 21]. The wax coating detected from this study, could have been applied on the Mosaic as a part of the restoration works occurred over time, both before and after the detachment from the archaeological site and its transfer to its current position at the MANN. To support this hypothesis, Melillo in 2013 [5] reports some steps of the restoration work carried out until 1843 by the Neapolitan mosaicist Piedimonte. He used to conclude the cleaning process by applying a wax diluted in aqua regia onto the surface of the Mosaic after filling the lacunae with a slurry made of gypsum. According to a note of the architect Pietro Bianchi in 1827, this treatment, along with the high humidity of the site, led to the progressive deterioration of the Alexander Mosaic [5]. Then, the famous Roman mosaicist Vincenzo Raffaelli, director of the Vatican mosaics in Rome, was chosen by the Minister of Home Affairs to examine the conservation state of the Alexander Mosaic. Raffaelli found a coating that was similar to a “resinous varnish”, responsible for the decay of the colors of the tesserae and of the general bad state of the artwork; he also pointed out the use of three different types of bedding mortars for the repositioning of the mosaic tesserae, i.e., a lime-based mortar, that was applied originally (probably compatible with the supranucleus that was typically produced for the bedding of Roman mosaic tesserae [22], a greyish lime mortar, used for more particular and detailed repairs, and a third type of mortar, made of a gypsum-based grout. According to Raffaelli, the use of gypsum was not appropriate in humid places because it could cause expansions and detachments around the gripping surface [5].

The widespread occurrence of gypsum detected on most of the areas analyzed using pXRF, Raman and ER-FTIR can be attributed to remnants of the gypsum applied, along with layers of paper and fabric, to protect the surface of the Alexander Mosaic during the phases of detachment and displacement of the artwork in 1843. In fact, when the mosaic was placed in a wooden box and moved to the Real Museo Borbonico in November 1843, during transportation it stayed in direct contact with the gypsum layer inside the box for several days and was also affected by heavy rain. Such prolonged contact between the Mosaic and the gypsum could have led to the additional precipitation of gypsum on the Mosaic and its wax coating, permeating the surface through the outermost part of the supranucleus. This may have caused the formation of a thin layer of gypsum, which is often detected by ER-FTIR spectroscopy as a reflective optical surface, in which the specular components distinctly prevail over the diffusive ones, unlike what one might expect from a gypsum normally scattered in a rough material such as mortar, when it occurs as a binder or a product of alteration [23, 24].

Oxalates could be the result of both the alteration processes of the organic protective coatings [18].

Geological materials and sourcing of tesserae

According to Marcaida et al. [25], Pompeian mosaics—the most important decorative patterns in Ancient Roman cities, then in the Bay of Naples and surroundings of Mount Vesuvius—have not been deeply analyzed from a chemical point of view and only few works report the chemical and mineralogical composition of some tesserae from Pompeian mosaics [18, 26, 27]. The materials used for the realization of the Roman tesserae, made up of individual pieces of cubic form, are mainly composed by carbonate and volcanic rocks, colored glasses or ceramic materials, whereas the base materials used to join tesserae were lime and/or other binders [23, 25, 28, 29]. The tesserae used in Ancient Rome were mainly made of calcareous rocks obtained from local sources of natural stone, with additions of crushed brick, tile and pottery creating colored shades of dominant black, red, white, blue and yellow, commonly forming polychrome patterns [25].

The identification of base materials of the tesserae and their supply sources are a challenge for the archaeometric research, especially when it is not possible to take advantage of laboratory (more or less) destructive quantitative analyses (i.e., thin section, X-ray powder diffraction, isotopic analyses) on (micro)samples. From a mineralogical-petrographic perspective, our data show that the investigated tesserae of the Alexander Mosaic belong to four compositional types: carbonate-bearing tesserae (Group 1), silicate-bearing tesserae (Group 2), natural and vitreous tesserae with intermediate composition (Group 3) and iron-rich tesserae with relatively high Fe concentrations (Group 4). Vitreous (i.e., synthetic) tesserae are characterized by a peculiar composition as shown in Fig 5, and mostly include grey ad green colors, with the green ones richer in Cu (S2 Table). For comparison, portable XRF measurements were performed on blue and green vitreous tesserae from other mosaics exposed at MANN (Mosaic Room 59). The analyses, reported in Fig 5 with black circles mostly included in the dashed ellipse, confirm this assumption.

Concerning the geological provenance of geomaterials, we can only speculate about possible source areas for some of the previously mentioned groups, considering qualitative remarks and similarity with some lithotypes likely employed at Roman times [30–35] (Fig 6).

Group 1 includes tesserae of various colors. Among these, the following can be ascribed to famous lithotypes.

Some of the white tesserae (Figs 4A and 6) could be related to the Marmor Lunensis from the Apuan Alps quarries (Italy), which started to be mined for marble in the 1st century BCE and was no longer used starting from the 3rd century CE; currently the analogous that is extracted in the Carrara basin is Bianco Carrara, in the C and CD varieties which can be distinguished by the presence of a whiter (C) or slightly grayer (CD) background color.

Pale pink tesserae (Figs 4H and 6) can reasonably be associated with the Breccia Nuvolata lithotype is quite reasonable, despite historical sources reported it was extracted starting from the 2^nd century CE from various sites of the Mediterranean area (Italy, Tunisia, Algeria, and Libya); this material was used in several different ways in Leptis Magna (Libya) and even some walls of the Temple of Serapis in Pozzuoli (near Naples) are covered with this rock.

Intense pink tesserae (Figs 4I and 6) could be ascribed to the Marmo Rosa of Portugal, but a further origin could be the Breccia Appenninica (Apuan Alps); the absence of white veins, typical of the so-called Portasanta lithotype (Marmor Chium), does not support this hypothesis, although this lack could depend on the small size of the tesserae.

Some of the yellow tesserae (Figs 4G and 6) show blackish lineation as in the Giallo di Siena stone [31–33]. Other possible raw material could be the so-called Giallo Antico (Marmor Numidicum), extracted starting from the 2nd century BCE until the 3rd century CE in the roman city of Simitthus (modern Chemtou, Tunisia).

Light Red tesserae (Figs 4J and 6) could belong to the aforementioned Breccia Appenninica from Apuan Alps, or to a variety of Diaspri Rossi (Sicily).

Light blue tesserae (Figs 4L and 6) are composed of a carbonate lithotype with minor inclusions containing Si and Fe; this feature could be related to a marble like Marmo Bardiglio or Cipollino.

Group 2 mostly includes black and grey tesserae and some green ones. Black silicate tesserae (Figs 4N and 6) can be reasonably ascribed to a melanocratic microcrystalline rock such as the so called “Granito ad erbetta”, a very fine-grained magmatic rock of gabbroid composition [36].

Group 3 gathers tesserae of different color including black, brown, yellow, red, grey, and mostly green.

Dark red tesserae (Figs 4R and 6) could possibly be attributed to the Rosso Antico (Marmor Taenarium, originating from Cape Matapan, Greece), known from the Romans since the end of the 2nd century BCE. This marble derives from low-grade metamorphism on impure carbonates: in addition to calcium carbonate, it also contains silica. The predominant mineralogical constituent of the lithotype is calcite with subordinate constituents as quartz, plagioclase, mica, chlorite [37].

For the green types, Fig 6 reports the green silicate tessera of Fig 4T (see also Table 1) and indicated with #1, as well as another green proxy indicated with #2. These lithotypes could have as a possible origin from the so-called Verde Issorie (a serpentinite rock). Also, the Verde Antico (Marmor Thessalicum or Marmor Atracium, originating from different areas of the Thessalia region) cannot be ruled out, being this rock mentioned in Diocletian’s edict; it is a compact and fine-grained marble, with a bright green matrix and showing different hues, with the darker shade variety being quite rare.

Yellow tesserae from this group (Figs 4Q and 6) could be related to the Diaspro Giallo/Giallo Venato, supplied from Spain or Sicily (Italy), which also shows the previously mentioned black spots. It should be remarked that variations in the hues of the yellow tesserae, but in general for all the color here discussed, could be obtained depending on different surface polishing degree.

Group 4 includes tesserae with a relatively high amounts of iron and different colors consisting in light blue, grey, black, green, and red.

Black carbonate tesserae of this group (Figs 4W and 6) are likely represented by fine-grained and compact carbonate types, rich in Fe and carbon-bearing inclusions, which could be assigned to the Nero Antico (Djebel Aziz, Tunisia). The use of an analogous black lithology from the Tunisian Chemtou site, where Giallo Antico was also extracted, cannot be excluded. This particular typology of tessera, with size less than 2mm, denotes, even more than the others, a strong compositional interaction with the (red-coloured) mortar surrounding it. This evidence can be seen on the Fig 4W and is reflected in the particularly high amount of Fe in the XRF analysis.

The Mosaic’s surface

Combined in-situ analyses (OM, ER-FTIR, Raman, pXRF, multispectral imaging, IRT) revealed many details of selected surficial areas of the mosaic. The investigations performed through multispectral imaging have shown high luminescence both in limited areas, due to the compositional varieties of the mosaic tesserae, and in irregularly large areas, due to protective coatings. The velatino was a quite common consolidant for the restoration intervention. Infrared thermography revealed some areas with temperature variations, but these did not correspond to areas treated with velinatura consolidation, which appears to have fulfilled its purpose. However, in assessing the conservation state, thermal imaging revealed some critical areas in different parts of the mosaic, which deserve special attention.

Gypsum, natural wax, and traces of calcium oxalate are the common phases detected on the Mosaic’s surface. It is likely that wax, a common protective coating for artworks, was applied during the restoration process, as the restorers used to do in the past. Calcium oxalate, found in traces, may be a by-product of the degradation of organic protective coatings. Gypsum, which is widespread throughout the Mosaic, probably originated from protective layers applied during detachment and transport at the Museum in the 19^th century, resulting in a thin layer of this component in many areas of the surface.

The Mosaic’s tesserae and mortars

A careful micro and macro inspection of the tesserae along with spectroscopic analyses allowed for drawing some hypothesis on their geological nature and tentatively assign potential provenance areas. Further than surficial investigation endoscopic observations from the back side of the artwork detected many empty portions, probably representing areas that were not reached by gypsum-based substances added to secure the structure of the mosaic during the former dismantling from its original site and transport operations. These potential areas of weakness should be taken into due consideration during restoration.

Based on OM, pXRF with PCA data processing, ER-FTIR and Raman analyses, the investigated tesserae can be sorted into four main groups: calcium carbonate-based (the most heterogeneous group and with many colors, i.e., black, white, grey, red, brown, pink, yellow, green, light blue), silicate-based (grey, black, green), a mix of both and vitreous types (black, grey, brown, yellow, red, green), and Fe(Ca+Mn)-bearing tesserae (black and red, but also grey, green, light blue). Microcrystalline texture is more widespread among the considered lithologies. As far as the possible origin of the rock raw materials that made up the mosaic tesserae is concerned—a challenging topic in archaeological provenance studies—only hypotheses can be made, given the completely non-destructive mode of the study. Based on lithological similarities with rocks from known mining districts of Roman times, several mining areas in the Mediterranean region can be considered as geological sources, such as Italy, Greece, Iberian Peninsula and Tunisia. The jointing mortars can have different compositions (even though commonly Ca-bearing) and colors, which vary from whitish to different shades sometimes akin to those of the tesserae.

All the data produced in this study have been gathered in a QGIS project and available also for the restorers, for a constant monitoring of the areas of intervention and of the related analytical information.

The mosaic is currently undergoing restoration, and new data on the underlying mortars have recently been acquired by the present research group on the backside mortars and are currently being processed. The combination of these new data, along with information obtained from a new instrumental investigation campaign planned for the mosaic surface in the final phases of the restoration operations, will further enrich our knowledge of this superlative work of ancient art.

S3 Fig. Representative endoscopic images on the back sides of the mosaic.

Evident spaces filled with binder depositions (A), probably composed of vinyl or gypsum-based substances, along with paper, including newspaper (B, C, D).

https://doi.org/10.1371/journal.pone.0315188.s003

(PDF)