Mycenaean Vitreous Artifacts: Overcoming Taxonomy Hurdles via Macro-XRF Analysis

Abstract

Mycenaean glass artifacts, such as beads and relief plaques, are highly susceptible to degradation, which can significantly modify their visual attributes and pose classification challenges. Corrosion on glass and faience artifacts has often led to misinterpretation, since the visual manifestations of degradation can be similar for both materials, impacting research conclusions. This paper presents a segment of a broader study conducted within the Myc-MVP project, utilizing advanced scientific methods to analyze the compositional changes in corroded vitreous artifacts. Through Macro-X-ray Fluorescence (MA-XRF) and LED microscopy, we aim to understand the correlation between compositional alterations and visual degradation manifestations. The use of MA- XRF was particularly crucial for non-destructively mapping the elemental distribution over large surfaces, allowing for a more comprehensive analysis of corrosion patterns. The results presented in this study are from a subset of artifacts examined using MA- XRF, highlighting critical insights into the spatial compositional shifts that contribute to visible deterioration. This paper discusses the first real-life contribution of Macro X-ray Fluorescence (MA-XRF) imaging to mapping the spatial compositional changes that occur when Mycenaean vitreous materials undergo degradation, yielding visible deterioration. MA-XRF scanning offers a fully non-invasive and non-destructive method for recording compositional data across the entire surface of an object. The results can be visualized as distribution images, which are more accessible and interpretable for a broader audience compared to the spectra generated by traditional spectrometric techniques. These findings aspire to inform strategies for the accurate classification, effective management, appropriate conservation treatment, and long-term preservation of vitreous artifacts.

1. Introduction

The Mycenaean civilization, flourishing from approximately 1600 to 1100 BCE, was characterized by advanced craftsmanship, including the production of glass and faience. These materials were not only used to create functional items but were also highly valued for their ornamental and symbolic roles. Artifacts such as beads, pendants, and inlays made of glass and faience were commonly found in elite burials, reflecting their association with status and cultural significance. The craftsmanship demonstrated a high level of technical skill and innovation, suggesting that Mycenaeans either developed or adapted complex manufacturing techniques from earlier Eastern Mediterranean traditions [1,2,3].

Glass in the Mycenaean context was often made by heating a combination of silica (quartz pebbles), alkali (plant ashes), and various colorants, yielding materials that ranged from translucent to opaque. Faience, on the other hand, was produced using a quartz core body coated with a vitreous glaze, creating a brilliant, glossy finish. Both materials were primarily employed to mimic precious stones like turquoise and lapis lazuli, showcasing the Mycenaeans’ ingenuity in achieving decorative splendor with available resources [3]. The typical values of both glass and faience are shown in

Table 1. Basic composition of Late Bronze Age vitreous materials of different types and origin.

	Vitreous Material Type	SiO2	Na2O	K2O	MgO	Al2O3	CaO
Data from [4,5,6,7]	Plant ash soda-lime glass (n = 142)	66	17	1.7	4	1.6	7
Data from [5] (only the plant-ash glasses).	Elateia and Thebes glass (n = 73)	67	17	1.3	3	2	6
Data from [6]	Mesopotamia glass (n = 25)	65	17	2.8	4	0.5	7
Data from [7]	Egypt glass (n = 30)	62	19	1.8	4.4	1	8
Data from [2]	Peloponnese Mycenaean glass (n = 11)	68	15	2.0	4	2	7
Data from [8]	Faience Mycenaean (body, n = 4)	84	2	1.2	1	4	3
Data from [8]	Faience Mycenaean (glaze, n = 6)	77	9	2.8	2	1	4

, where the differences in composition are evident.

The degradation of Mycenaean glass and faience is a significant concern for archaeologists and conservators, impacting both the physical appearance and the structural integrity of these ancient artifacts (Figure 1 and Figure 2). This corrosion process is exacerbated by several factors, including their small size, material composition, and environmental exposure over centuries.

The glass used by Mycenaean artisans was typically alkali-rich, making it highly susceptible to corrosion when exposed to moisture. Over time, this exposure resulted in a process called leaching, where network modifiers and stabilizers (soda, potash, lime) were depleted [9,10,11], leading to a weakened, porous surface often called silica gel. More often than not, the degradation manifested as a network of fine cracks that compromised the artifact’s structural integrity. Additionally, the surface could develop a white, powdery appearance due to the crystallization and precipitation of salts [12,13,14,15,16].

Faience, while similar in appearance to glass, has a different composition and faces unique challenges. Its core is composed of a non-vitreous, porous material that can absorb moisture. This inherent porosity allows water to infiltrate and disrupt the bond between the core and its glaze. The result is flaking or detachment of the glassy layer, leading to the significant loss of detail and coloration [12].

The small size of Mycenaean beads and other decorative items accelerates the degradation process. Smaller artifacts have a higher surface-area-to-volume ratio, meaning more exposure relative to their overall mass. This makes them more vulnerable to environmental stresses such as changes in humidity and temperature. Small beads, for instance, can experience faster surface corrosion, leading to a more rapid loss of their original shape and luster compared to larger glass objects. Beads and small decorative objects are often found in burial sites or as part of larger assemblages exposed to varying degrees of moisture, soil chemistry, and biological activity. The conditions of their excavation, first aid treatment, and subsequent conservation efforts are crucial in mitigating further deterioration. Handling such tiny, delicate items requires specialized techniques to avoid inadvertent damage during excavation, cleaning, and storage.

Archaeological glasses and glazes are characterized by intricate networks, primarily composed of silica with minor metallic oxides [17,18]. Ancient glassmakers used modifiers to lower the silica melting point by incorporating alkalis like sodium and potassium, as well as to enhance glass durability by the addition of alkaline earth metals such as calcium and magnesium [18]. The composition and proportion of these components influence the decay rate and overall corrosion of glass artifacts, particularly under varied environmental conditions. Mycenaean glass, which is plant ash-based [13], contains significant proportions of sodium, calcium, potassium, and magnesium oxides, which render it susceptible to corrosion [3,12]. The silica-to-alkali ratio is pivotal in determining glass durability, with higher alkali concentrations in wet environments increasing susceptibility to decay [18]. Alkali-deficient layers occasionally act as protective barriers, while alkali environments pose significant threats, accelerating glass dissolution, particularly at elevated temperatures [19,20]. Localized variations in burial environments can result in diverse patterns of glass corrosion within the same site [2,12,20].

The focus of this paper is on defining strategies for differentiating degraded vitreous materials, particularly Mycenaean glass, utilizing non-invasive spectroscopic techniques, notably MA-XRF. To our knowledge, this is the first ever application of MA-XRF on investigating the corrosion of Mycenaean vitreous materials; however, this technique has been applied before on much later glass artefacts, namely stained glass [21,22,23,24]. We address the challenge of distinguishing between faience and glass, which often presents difficulties for non-experts in the field and in a museum/storage setting. This paper builds on a selection of objects with clear preservation states, enabling secure material attribution, contrasting them with ambiguous counterparts from the same contexts. We present the preliminary results obtained from LED microscopy and macro-XRF applied to a subset of 5 objects within a collection of 120 Mycenaean artifacts from the Hellenic National Archaeological Museum (Collection of Prehistoric Antiquities).

2. Materials and Methods

2.1. Materials

This study is part of the Myc-MVP project, “Mycenaean Vitreous Production: A Novel Interdisciplinary Approach Towards Resolving Critical Taxonomy Issues”, funded by the Hellenic Foundation for Research and Innovation and hosted at the Institute of Nuclear and Particle Physics, NCSR Demokritos, Athens, in collaboration with the Department of Materials Science, the University of Ioannina, and the Hellenic National Archaeological Museum. The project focuses on vitreous objects from various Mycenaean contexts. Objects with varying degrees of corrosion were chosen, rendering their material attribution insecure, thus ensuring that each had a comparable counterpart in terms of typology. For this paper, five objects were selected and analyzed using an LED microscope (

Table 2. Investigation of the samples using LED microscopy.

Sample Number Location	Material	Photo	LED Microscopy	Preservation State	Corrosion Effects
MVP-7 Tiryns HΝAΜ Π 1616	Glass			Medium corroded	Pitting with characteristic concentric layers; forming of local crust and flaking; stained by metal oxides
MVP-10 Spata HΝAΜ Π 2078	Unclear (glass in the archaeological record)			Heavily corroded	Pitting with characteristic concentric layers; forming local crust; stained by metal oxides
MVP-15a MVP-15b Spata HΝAΜ Π 2189a HΝAΜ Π 2189b	Glass			Medium to heavily corroded	Pitting with characteristic concentric layers; forming of local crust; discolouration occurred by the semi opaque milky like surface from the concentric layers.
MVP-37 Mycenae HΝAΜ Π 3117	Unclear			Good to medium preservation state	Local crusts; discolouration by metal oxides
MVP-83 Mycenae HΝAΜ Π 2893	Glass			Medium to heavily corroded	Pitting with characteristic concentric layers; forming local crust

) and MA-XRF (Figure 3, Figure 4, Figure 5, Figure 6 and Figure 7). These samples date broadly to the Late Bronze Age (LBAIIA-LBAIIIC, 1500-1100 BCE) and exclusively consist of relief plaques.

MVP-7 is a glass square plaque from a sum of fourteen similar ones retrieved from the palace of Tiryns, dating to LHIIIA (1420–1315 BC). On the front side, it bears an incised circle, inscribed in the square. There is a small cavity in the center of the circle, and it is obviously worn and chipped in places. The turquoise-blue glass has flaked off along the periphery, revealing the off-white glass matrix. Interestingly, the circle retains its color completely, which is also likely due to differences in the taphonomic conditions [25,26].

MVP-10 comes from a set of chamber tombs in Spata [27], dating to LHIIIA–LHIIIB (1420–1315 and 1330–1200 BC, respectively). It is referred to in the archaeological record as a large, almond-shaped, blue glass bead, decorated with long, well-spaced incisions, covered with light brown accretions [28].

MVP-15 (a–b) were retrieved from the same set of chamber tombs in Spata [28] and date to LHIIIA (1420–1315 BC). In fact, two samples of the same artistic choices were found. The lower halves of a pair of mold-made glass relief rectangular plaque-beads were decorated with two ivy leaves, preserving the horizontal, ribbed, raised band that separates the plaque from the tassel-like endings. One was made with off-white glass and the other one with dark-colored glass. Both objects show signs of wear and chipping. The dark-colored one had flaked in places, revealing a light brown hue indicative of corrosion (light brown glass paste, as previously stated) [29,30].

MVP-37 was retrieved in Mycenae from Chamber tomb 81, which was excavated by Tsountas (1895) and dates to LHIIIA. It is a mold-made, flat-convex plaque-bead in the shape of figure-of-eight shield, with a raised ridge all around the object’s periphery. It bears a horizontal perforation through the middle and exhibits signs of wear and limited fractures in places. Some bits are missing, especially on one half. It must have been a blue-green color, which has now oxidized, with brown accretions on the surface [31].

MVP-83 is derived from Mycenae. It is a part of a dark blue glass bead in the form of an ivy leaf and four curved lines of granulation above its volutes. Due to corrosion, the original color has been altered, resulting in a yellowish surface/crust. There is a string-hole at the bottom of the ivy leaf.

2.2. Macro-XRF

The μ-XRF scanning was realized with the M6 Jetstream (Bruker) scanner (Available online: https://www.bruker.com/en/products-and-solutions/elemental-analyzers/micro-xrf-spectrometers/m6-jetstream.html (accessed on 9 April 2024)), which allows scan areas of 80 × 60 cm². The M6 Jetstream is equipped with a 30 W rhodium X-ray tube. In the present measurement, the X-ray tube was operated at a high voltage of 50 kV and a current of 600 μA, while no absorption filter was applied on the beam path of the exciting radiation. The measurement was conducted by applying helium flux to enhance the detection of low-energy transitions from light elements such as Mg, Al, and Si. The incoming X-ray beam from the source was focused using a polycapillary glass optic and impinged perpendicularly on the target surface. A silicon drift detector with a 30 mm² active area was used for photon detection at an angle of about 60° relative to the target surface, with an energy resolution of 138 eV at the Mn Kα-energy. Each scan was analyzed with a scanning speed of 10 mm s⁻¹, a dwell time/pixel of 20 ms, a step size of 200 μm, and an adjustment of the lens’ focal distance to offer a spatial resolution of 100 µm at Mo Ka.

Changes in the intensities of the characteristic transitions in the elemental maps enable the extraction of information concerning the elemental distribution on the object. It is important to note that variations in intensity can arise not only from changes in atomic concentration but also from geometric factors, particularly concerning nonplanar and rough surfaces.

3. Results and Discussion

The selection of the samples was realized based on the nature of the material and drew on the expertise of the research team. Heavily corroded samples, whose nature might appear ambiguous to non-experts, were deliberately selected. The criterion for selection was that a ‘healthy’ counterpart of clear and equivalent nature was also available within the same context. Nevertheless, a sample of unclear nature (Sample MVP-37), despite the lack of an uncorroded counterpart, was also incorporated to exploit the advantages of the chosen technique towards deciphering its nature.

Figure 3. Sample MVP-7 showing a preserved glassy phase with localized calcium near the edges and higher potassium concentrations in the central area. Copper is notably concentrated in corroded sections, while calcium migration indicates environmental leaching. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Figure 3. Sample MVP-7 showing a preserved glassy phase with localized calcium near the edges and higher potassium concentrations in the central area. Copper is notably concentrated in corroded sections, while calcium migration indicates environmental leaching. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Sample MVP-7 (Figure 3) preserved a glassy phase and was assigned as glass via visual examination. In this sample, silicon is distributed almost evenly in the whole scanned area, while calcium is localized towards the edges of the sample (e.g., upper left corner). At the same time, in the area where the glassy phase is detected, even macroscopically, there exists a higher alkali concentration as expected, with the potassium content being higher in the center left part of the sample. As expected, potassium is depleted in the corroded area. Interestingly, the colorant—copper on this occasion—is more concentrated in areas which seem rather corroded and not in the turquoise glassy phase of the sample. Finally, the whitish corroded areas (mostly in the corners of the sample) have higher amounts of calcium. In the period under study, lime (CaO), incorporated into the glass network through plant ashes, acts as a stabilizer. Over time, exposure to moisture and environmental factors can cause leaching, and the alkali components and calcium migrate to the surface [32]. This can result in calcium-rich layers forming on the exterior as part of the corrosion process. Thus, in this case, the localization of calcium is indicative of this process of the elements leaching out of the glass.

Figure 4. Sample MVP-10, a highly corroded glass bead, displaying uniform silicon and potassium distribution. Copper is concentrated in colored areas, while manganese browning is observed. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Figure 4. Sample MVP-10, a highly corroded glass bead, displaying uniform silicon and potassium distribution. Copper is concentrated in colored areas, while manganese browning is observed. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Sample MVP-10 (Figure 4), while highly corroded and with the absence of any glassy phase or glaze, was assigned to the glass group in the archaeological record, although it could be considered as having been made of faience. This almond-shaped bead exhibits a uniform distribution of silicon and potassium and, as expected, both elements display similar distribution patterns. Particularly, areas with elevated silica also show elevated potassium, and vice versa. Interestingly, a higher concentration of lime is found in the upper right corner, which, in terms of visual characteristics, appears similar to the rest of the sample. The rest of the sample appears rather depleted in calcium. Areas which visually exhibit a color have elevated amounts of copper, and some very restricted areas with elevated amounts of manganese can be characterized as suffering from the manganese browning effect which is rather typical in vitreous materials [17]. In general, the amount of copper is at the level of 1% wt. in the bulk analysis. However, there are areas that have higher copper concentrations due to the presence of undissolved copper particles; in such cases, the copper content can rise to a few tens of % wt. The weight concentration of Mn is an order of magnitude smaller and varies from 0.1 to 0.5% wt. Since the quantitative analysis of the samples under investigation is not the focus of this paper, it will not be discussed any further. In glass, Si is typically the most stable component, even in corroded areas. Alkalis (K, Na) and alkaline earth (Ca) are leached out first, but the complete depletion of Si in Cu-rich regions is unusual for glass, unless the structure is heavily altered. In faience, the glaze is often unstable in burial environments, and if the glaze contained Cu, it could have degraded completely, leaving behind a Cu-rich residue while losing Si, K, and Ca, which is evident in the high-copper areas. Therefore, there is an indication that the material is made of faience.

Figure 5. Sample MVP-15, a heavily corroded glass pendant, showing even silicon distribution with localized copper concentrations. Blue glass areas exhibit higher calcium and potassium, with lower iron levels, indicating better preservation. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Figure 5. Sample MVP-15, a heavily corroded glass pendant, showing even silicon distribution with localized copper concentrations. Blue glass areas exhibit higher calcium and potassium, with lower iron levels, indicating better preservation. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Sample MVP-15 (Figure 5) is part of a heavily corroded glass pendant. The analysis was performed on the back flat surface. Silicon is distributed evenly across most of the surface, except for an area in the middle of the sample, which has suffered the most from milky-like corrosion. Copper shows a similar distribution pattern, with some localized areas of excess concentration. Regions that preserve their glassy phase and blue coloration exhibit higher amounts of calcium and potassium and lower concentrations of iron.

Figure 6. Sample MVP-37 displays distinct corrosion patterns, with excess silicon in the glassy phase and elevated calcium in the corroded areas, suggesting crust formation. A direct correlation between manganese and cobalt indicates the initiation of manganese browning, typical of cobalt-based colorants. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Figure 6. Sample MVP-37 displays distinct corrosion patterns, with excess silicon in the glassy phase and elevated calcium in the corroded areas, suggesting crust formation. A direct correlation between manganese and cobalt indicates the initiation of manganese browning, typical of cobalt-based colorants. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

In Sample MVP-37 (Figure 6), silicon is present in areas where, visually, a glassy phase is noted. The corroded surfaces show elevated amounts of calcium, suggesting the formation of crusts. Regarding the colorants, a direct association between manganese and cobalt is observed, which can be explained by the nature of the colorant, as manganese is considered an accessory to the cobalt colorant [33,34]. In particular, cobalt in glass can be derived from either specific cobalt minerals or/and cobaltiferrous alums [3,35,36]. In the first instance, apart from cobalt, these minerals are related to other elements, one of which is manganese. Therefore, it is clear that in areas where cobalt is present, the initiation of the manganese browning effect takes place more easily, given the presence of manganese.

Based on the examination of the surface, both visually and with the aid of MA-XRF, we concluded that the sample is made of glass. The presence of elevated calcium in the corroded areas, which can form crusts, aligns with the common corrosion patterns seen in glass over time. Calcium oxide (CaO) is often used as a stabilizer in glass, and its leaching out over time due to environmental exposure is a typical process for glass degradation, whereas in faience, calcium is not typically a dominant component, and leaching behavior like this is less common. The relationship between manganese and cobalt in the colorant also suggests a glass composition. Cobalt, commonly used in glass production for blue coloring, and its interaction with manganese, which can cause browning, supports the idea of a glass object rather than another material, such as faience or ceramic, which would exhibit different compositional characteristics. Colorants are used in faience making, but cobalt, which is more commonly used in glassmaking, would not typically be present in faience to the same extent.

Figure 7. Sample MVP-83 with central copper accumulation in the better-preserved area, with even silica and cobalt distribution throughout. White calcium crusts on the surface hint at chemical alteration. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

Figure 7. Sample MVP-83 with central copper accumulation in the better-preserved area, with even silica and cobalt distribution throughout. White calcium crusts on the surface hint at chemical alteration. The y-axis shows normalized intensities for the maximum (100%) and minimum (0%) detection of each specific element.

In MVP-83 (Figure 7), the even distribution of silica and cobalt reflects a standard glass composition, while the concentration of copper in the center might be explained by the fact that as the glass degrades or ages, the copper that is initially part of the glass composition might be more prone to leaching out from the surface; in this case, it is practically localized in the more corroded/whitish areas. The coloration of glass with a combination of two oxides (CoO and CuO) was very common during the Mycenaean period [2]. There is also a degree of co-localization of Mn and Co in the area where a corroded (crust-covered) part of the Mycenaean glass exists, suggesting that these elements were preferentially concentrated in the altered layer. Their spotty distributions could suggest that they may have migrated and re-precipitated from solution. A similar behavior is also noted with Fe, which is mostly concentrated on the left side of the object where the corroded surface is more evident. Calcium crusts point to chemical alteration, with calcium leaching or migrating to the surface to form white crusts, a sign of interaction with external factors.

4. Conclusions

The MA-XRF analysis has revealed distinctive elemental distributions and corrosion patterns across the five vitreous materials selected, allowing for a better understanding of the degradation processes that are related to their composition.

Based on the analytical results, it was possible to identify the nature of the vitreous materials through the identification of the elemental distribution in the materials’ surfaces.

Elevated calcium on corroded surfaces points to crust formation, which is typical of glass corrosion. Calcium oxide (CaO), used as a stabilizer in glass, leaches out over time with environmental exposure, unlike in faience, where such behavior is rare. High levels of calcium indicate the formation of lime crusts, as expected due to leaching resulting from corrosion, and it is evident in areas showing white crusts.

Regarding alkali depletion, which is one of the most common corrosion effects, potassium distributions show a variable behavior, particularly in relation to the areas that are visually corroded. The patterns noticed in the samples suggest that alkali metals like potassium play a significant role in glass stability and can be easily distinguished using macro-XRF.

The distribution of colorants on the surface of glass may give insights into the migration of colorant metals, such as copper and cobalt, and hence reveal corrosion effects and degradation. Cobalt, used for blue coloring, often appears with manganese, supporting the identification, as the manganese browning effect is characteristic of glass but not seen in faience or ceramics. In addition, cobalt and manganese follow the same pattern, and this association suggests that cobalt minerals or cobaltiferous alums (often containing traces of manganese) were used to create blue glass, explaining the consistent occurrence of manganese alongside cobalt in these areas.

Copper shows a complex distribution across the samples. Its concentration varies locally, with some samples showing isolated copper-rich zones, possibly due to differential corrosion or different manufacturing technologies (addition of mineral or/and scrap metal). For example, we noticed its concentration in corroded regions of a sample rather than the colored turquoise glassy areas, suggesting that copper migration may occur as the glass matrix breaks down, which has also been noticed in previous research [12].

Finally, the variations in silicon levels are mostly minor and, when present, are often accompanied by shifts in the distribution of other elements such as potassium, calcium, or colorants, suggesting that the base glassy matrix remains largely consistent, despite corrosion in specific areas.

The patterns that have been found, such as calcium enrichment in crusts and potassium depletion in corroded zones, which are significantly affected by the exposure to soil and moisture, serve as indicators of corrosion that can help conservators classify vitreous materials and determine the degree of degradation of related glass artifacts.

The use of Macro-XRF in studying the corrosion of vitreous materials marks a significant innovation in vitreous materials studies. Its non-invasive, transportable nature allows for the in situ analysis of delicate artifacts, preserving their integrity while delivering highly detailed compositional insights. The technique’s ability to produce intuitive distribution maps provides a clear understanding of elemental behaviors, enabling the identification of corrosion patterns and material composition with precision. This approach bridges the gap between scientific investigation and conservation practices, offering conservators a powerful tool for assessing and mitigating deterioration. With its versatility and potential for broader applications, Macro-XRF can revolutionize the study and preservation of cultural heritage in general, and there is optimism for future advancements in non-destructive material analysis.