news

     Several (mostly) blue-green glass beads from Iron Age sites in central Italy have been studied using a range of spectroscopic techniques: handheld X-ray fluorescence spectroscopy, fibre-optic reflectance spectroscopy, scanning electron microscopy coupled to energy-dispersive X-ray spectroscopy, micro-Raman spectroscopy and laser ablation-inductively coupled plasma mass spectrometry. Additional information is collected for each technique and discussed within the framework of the archaeological typology of the objects. The systematic evaluation of the results has allowed us to draw some conclusions about the raw materials used for primary production and to identify some indicators of the origin of the glass. Some beads found in Iron Age (IA) contexts have been tentatively assigned to Final Bronze Age (FBA) production based on their typology, and the compositional data obtained in this study confirm that they represent low-Mg-HH (LMHK) glass, typical of the Italian peninsula FBA. Other beads have been classified as low-magnesium (LMG) or high-magnesium (HMG) glass, providing additional information on the fluxes used in the Early Iron Age (EIA) and later. The differences in colour between beads reflect their chemical composition, with different types of beads being coloured differently. In some cases, different origins of the colouring raw materials can be identified. The provenance of the samples is difficult to determine, but the chemical data suggest that the original glass from which the beads were made was subdivided: one group of samples can be suggested by a local origin of the glass, while the majority of beads examined in this study point to several production sites in the Near East. For one typological group, some initial clues regarding local processing of imported glass are also highlighted.
       The aim of this paper is to provide new knowledge about glass in circulation during the Iron Age (IA) by using different analytical methods to obtain archaeological measurements of forty glass beads (whole beads or fragments) found in several burials in several historical regions.
       The location of the archaeological site where the glass samples were found. The map shows the modern administrative regions of Italy. The locations are: 1 – Vulci; 2 – Cerveteri; 3 – Veio; 4 – Capena; 5 – Terni; 6 – Osteria dell’Osa; 7 – Marino; 8 – Sermoneta; 9 – Verucchio
       The samples were chosen to represent different types of blue-green glass beads from burials of the Early Iron Age (EIA) I, EIA II, Early Orientalizing and Middle Orientalizing periods, i.e. from the 10th to the 9th century BC to the 7th century BC, with EIA beads being the most common. The beads are currently housed in two archaeological museums: the National Etruscan Museum of Villa Giulia and the Museum of Roman Civilization (both in Rome, Italy).
       All beads have been carefully examined and their types have been identified on the basis of macroscopic data to support the discussion of archaeological data, which is particularly relevant to the study of archaeological glass. Indeed, glass recipes changed markedly throughout the Mediterranean and Near East: by the early IA new shapes and decorations had appeared (reflected in bead types) as well as materials (probably reflected in the chemical composition of the glass).
       In Italy, the Late Bronze Age (FBA) was a golden age of glassmaking, and several archaeological studies have examined glass from this period: the chemical composition of beads from the Fratezina workshop and other sites such as the nearby Mariconda di Melara or Fondo Paviani (Veneto, Italy) has been determined and is associated with a transparent mixed alkali glass composition known as low-magnesium, high-potassium (LMHK) glass [1,2,3,4,5,6]. Chemical evidence suggests that this type of glass was probably produced locally, and isotopic analysis has provided further clues, pointing to the origin of the glass in a volcanic region, possibly near Rome.[7] Several beads included in this study have been typologically confirmed as FBA glass. Since LMHK glass production took place over a period of about 200 years, certainly in different workshops, its composition can be expected to vary.[8] Furthermore, evidence of low magnesium glass (LMG) from the Bronze Age has been found in the waste of the Sardinia [9] and Fondo Paviani [10] workshops.
       The composition of a sample from northern Germany dated to the 13th–12th centuries BC[11, 12] and other samples found in many places in Europe, Asia and Africa[2, 5, 13,14,15,16,17] suggests that HMG was used throughout the Bronze Age and beyond. In the Near East during the IA period, HMG was gradually replaced by LMG, which was produced from sodium-rich evaporites instead of plant ashes[16, 18, 19], leading to significant changes in glassmaking technology.
       Since glass from IA contexts in Etruria and Lazio has not been studied in detail from an archaeological point of view, the aim of this paper is to provide new compositional data to support archaeological interpretation of the complex historical reality of the region. In addition to examining the compositional characteristics of individual LBA glasses found in EIA settings, this study also examines the aforementioned technological transitions reflected in the chemical composition of several typological glass beads. Typology is the first piece of evidence that archaeologists take into account when discussing glass beads, so a typological framework is maintained throughout the paper. These beads were chosen to reflect the diversity of blue-green glass found in IA contexts in central Italy. Some of the bead types included in this paper were likely made in the Italian peninsula (see the Archaeological Glass Beads section), and determining their composition will therefore reveal whether the original glass was imported, thereby significantly deepening our understanding of the glass supply in central Italy at this time.
       Most of the beads included in this study were not permitted to leave the museum, so a two-stage approach was used to study this set of glass samples: non-invasive, on-site examinations using portable equipment, namely a digital optical microscope (DOM), a fibre optic reflectance spectrometer (FORS), and two portable X-ray fluorescence (p-XRF) spectrometers. A subset of eighteen samples were then analysed in the laboratory using scanning electron microscopy coupled to energy dispersive spectroscopy (SEM-EDS), micro-Raman spectroscopy (µ-Raman), and laser ablation inductively coupled plasma mass spectrometry (LA-ICP-MS). The sequence of scientific methods allowed for extensive screening of the composition of the beads included in this study, while achieving a depth of understanding that can only be achieved in a laboratory setting. Overview The analytical capabilities of the program allowed us to obtain informative data on silica sources and fluxes used in glassmaking, as well as to study colorants and opacifiers. The accuracy of the quantitative information was carefully checked in this study, as this is a prerequisite for discussing the data in terms of the elemental composition of LBA and IA glasses available in the literature.
       The selected set of glass beads (Figure 2) includes specimens from eight typological groups and nine archaeological contexts (Table 1). The reader should refer to these two elements (i.e. Figure 2 and Table 1) whenever a particular bead is mentioned in the text. In addition, the main characteristics of the eight typological groups are presented in Additional file 1: Table SI1. Each group raises specific questions that can be answered by supplementing the chemical characteristics presented in this paper with information obtained from the archaeological examination of the beads (i.e. typological classification) and the archaeological context.
       Glass beads (or bead fragments) included in this study. The specimens were classified into typological groups as described in the text. Photo courtesy of the National Museum of Etruscan Art and Villa Giulia.
       Below is a description of the beads selected for this paper according to their typological classification as well as the specific archaeological problems associated with each group, while the general structure and objectives of the entire study are presented in the introduction.
       In addition to being found in EIA burials, these beads can be attributed to an earlier production based on typological features. The group includes two forms characteristic of the Late Bronze Age (LBA): horn-shaped layered eyes (perla a occhi stratificati cornuti, Pfahlbaunoppenperle) and barrel-shaped beads decorated with a white spiral around the body (perla a botticella con decorazione spiraliforme, Pfahlbautönnchen). Both types of beads are common in Italy and other archaeological sites north of the Alps, especially in pile dwellings in Switzerland [20,21,22]. This is one of the types of beads produced during the FBA at the Frattesina di Fratta Polesine manufactory (Veneto, Italy). In particular, the barrel beads found at Frattesina show a wide range of colour variations, not just the simple combination of blue matrix glass and white decoration common elsewhere (compare 76 f. in [6] with fig. 1 in [11]).
       Corner eye beads are made by placing four sections of white glass on a spiral colored bead and then layering another layer of colored glass on top, forming the typical “eye” protruding from the bead (Figures 121-122, see [23]; No. 78, pp. 21-22 and 93-95, see [6]). With some variations, the most common and easiest to identify in FBA are the four-eye beads, such as VG25 and PG84. In some cases, one or even all of the eyes have fallen off the bead, leaving a flat (or concave) surface on which traces of the lost decoration can be found, such as PG87 and PG167. Both forms have a long tradition outside the FBA (pp. 85–91 in [6]; fig. 2 in [24]) and may have been produced as early as the 13th century BC in Italy using rusty-brown glass with a high magnesium content – ​​HMBG [25] (f. 177 pl. LVIa in [21]; summary pp. 81, 85 in [26]).
       The beads of the first group were found in funerary tombs in two Lattieri cemeteries, Osteria dell’Osa and Sermoneta-Caracupa, and in a famous cremation tomb in the Etruscan cemetery of Vulci-Cavalupo, which has been called the “Sardian Bronze Tomb” because it contained three imported Sardinian bronzes.[27] The specimens considered here (i.e. VG25, VG26, and VG29) were found in this tomb along with other beads, most of which showed signs of secondary firing, as reflected in changes in colour and texture. This group includes two beads from Osteria dell’Osa: one is a conical, irregularly twisted green bead from tomb 82 (PG169), and the other is a fragment of a green bead with a white speckled pattern (PG112). The latter corresponds to Frattesina type 18 [6] and can be considered a typical FBA composition. Both beads have a slightly translucent turquoise-green matrix glass, which is unusual for EIA. The main purpose of the analysis of these samples was to check their compliance with the LBA glass composition.
       (Perlina ad anello; pearl of the ring). This set of beads includes ring-shaped (and in some cases flattened) beads in dark, blue and red colours. Small ring beads, made by winding hot glass on a mandrel, played an important role in the funerary customs of the Italian FBA and EIA [28], since more than 95% of the beads excavated at Frattesina are undecorated ring beads (Fig. LVII in [21]; 71f. figs. 1 and 12 in [6]). Furthermore, translucent turquoise beads have been found in large quantities outside the Frattesina area and throughout the Italian peninsula ([29], p. 19, with fig. [30]; f. 81 in [6]), as well as in some IA tombs (figs. 11, 74–76 in [31]). Characterizing the composition of the granules of this group will provide some elementary data for an interdisciplinary framework aimed at tracing the circulation of these granules during the FBA and IA.
       This group includes five beads from the cemeteries of Osteria dell’Osa, Sermoneta Caracupa, and San Agnese Acciarie (Table 1). Three of them are ring beads with two or three spiral eyes (green pearls with black spiral eyes; green pearls with yellow spiral eyes). These ornaments are unique to the EIA and are obtained by heat-assembling two pieces of glass of different colors, which are then drawn into a twisted net pattern (Figure 12 on page 30 [32]). Two or three pieces of this wire are then placed over the bead, either protruding plastically (e.g. PG122) or smoothly recessed into the bead (PG166). The second type is a smaller ring or cylinder bead with a black (or possibly yellow) line soldered to it but not covering the entire circumference of the bead (piccola perla verde con filo nero (-giallo?); kleine grüne Perle mit schwarzer (-gelber?) faded decoration). The broken beads examined here show only the shallow grooves of the now lost decoration (PG121), while another bead (PG65) retains some remnants of the decorative lines. The general appearance of the beads distinguishes them from other known EIA bead types. Their chemical properties provide additional information about their origin.
       The eyeball (blue pearl with white eyes and simple rings; blau-weiße Ringaugenperle) is one of the earliest, most durable and most widespread ornaments in the EIA [33]. This ornament typically consists of three “eyes” in the form of white glass rings embedded in a matrix of thermally blue glass, which are sometimes lost in the eyeballs (but not in the beads examined in this study). In addition to the typological attribution to eye beads, this small group of beads was also distinguished in a previous analysis of the larger group of cobalt blue eye beads [34], since both copper and cobalt contributed to the blue colour of these specimens. Differences in size, colour, opacity and production technology indicated that different components were found within this group [34]. However, this particular group of blue beads attracted attention as they are among the first LMG samples from central Italy, and they were studied in this paper to further understand their composition.
       (Vago or pendente a tubetto; Röhren pearls). These rare and unusual beads from Verucchio are wound around a metal rod like simple bead rings. The beads are then decorated with hot white glass thread, which is wound around the bead more than 20 times and “combed” four or five times with a metal point or knife to create an arched or chevron pattern.[35] The decorative threads in the specimens examined in this paper have been lost, changing their original appearance. Similar beads have only been found in two other graves in the Verruciolipi cemetery, dating to the mid-8th century BC (19–20 BC, type 13A, fig. 31, pp. 228–230, in [31]). Determining their composition will shed light on their interpretation, as their rarity has led to a general lack of archaeological conclusions.
       (Perla costolata or melon beads; Melonenperle). The set consists of three spherical beads in the shape of a melon. The technique for producing these beads was already known from the LBA and earliest IA (type 9 in [6]; pp. 37–39 in [36]; figure 1 in [37]). They became more common in the sixth and fifth centuries BC, and archaeological evidence documents the existence of two workshops where they were produced.[38, 39] The spherical beads are formed on rods and then their hot bodies are worked with tools such as the back of a knife to create relief shapes. The thin pendant PG97 was produced using a similar technique, but the thin bronze rod inside suggests that it was intended for hanging. This group also raises the question of their origin; chemical characterization will therefore allow us to incorporate these examples into a wider framework of information on blue glass.
       (Vago per il revestimo dell’arco di fibula; Fibelbügelperle). The blue leech-shaped beads PG59 and PG60 are specially made for the decoration of brooches. The base of the bead has indentations of a round pattern, widening the ‘belly’, but the bead must have been wrapped around a curved rod (photographs on pp. 58-61 on the CD in [40]; fig. 7 in [41]). They are decorated with yellow glass, wrapped around the bead and ‘combed’ into a herringbone pattern while the matrix glass is still hot, since the decorative glass is completely immersed in the matrix. This shaping procedure distinguishes the fibular beads of this group from the more common, larger fibular beads found in the Bologna and Verucchio areas [40, 41]. The moulding technique unique to the beads discussed here is found only in a few other examples from Veio (from the same tomb, but using brown glass) and in a unique green-red bead from Narseh, Falerii, Capena and Picenum (catalogue numbers 154–161 in [40]; figure 8 in [41]). Archaeological and technological evidence suggests that they were probably made somewhere in southern Etruria, possibly at Ager Faliscus; their elemental composition may thus shed light on whether they were made from imported or local glass. Of particular interest for the interpretation of these bulbous beads is a comparison with the composition of possibly imported beads found in the same burial (PG63, group 8).
       This set includes an assortment of blue beads. Three thick ring beads from Osteria dell’Osa. Two of the beads (PG168 and PG171) are dark turquoise and translucent, and similar beads have been found at archaeological sites throughout Italy.[33] The third bead (PG170), from the same tomb, is made of opaque turquoise or light blue glass. Opaque turquoise glass tinted with copper was used in Bronze Age Egypt,[42] but is very rare in the EIA context of Italy. Incidentally, the opaque turquoise glass decorating the bird-shaped beads found in the same tomb (not included in the current sample set) is also unusual.[43] The remaining beads in the set (PG33, 63, and 116) are made of translucent blue glass. Finally, PG136 is a small fragment that has been assigned to this group (tentatively) based on closeness of color/transparency.
       Determining the chemical composition of these beads will help contextualize them within the overall structure of glass bead production and exchange.
       The list of samples and the analytical methods used for each sample can be found in Additional file 1: Table SI2.
       The beads were examined using a Dino-Lite AM4815ZT–Edge digital microscope to properly record their characteristics and select analysis points. To improve the overall quality of photo documentation of the examined beads, digital images were acquired at 20x and 100x magnification.
       This method is used for non-invasive detection of dyes. Not all samples could be used to obtain informative spectra, and samples with surface weathering did not show any characteristic spectra. Light from an Ocean Insight HL-2000-HP-FHSA 20 W tungsten-halogen source is fed through a 2 m long reflectance/backscatter fiber bundle terminating in a probe with a 400 µm core diameter and a 6.35 mm tip diameter. The sample is fixed to the sample holder and the probe angle is adjusted to approximately 45° to capture the scattered light and exclude the reflected component. With this arrangement, the area of ​​the analyzed spot is a few square millimeters, which is usually suitable for testing the dominant color of beads, but not for testing the smallest jewelry. The diffuse reflected light was transmitted to an Ocean Insight QEPro CCD detector with an HC1 grating, operating from 248 to 1038 nm and having an optical resolution of 6.78 nm (full width at half maximum). The system was calibrated using a high reflectivity Spectralon standard. Integration times were set between 0.019 and 0.029 seconds, and 40 or more scans were averaged to obtain a single spectrum. Multiple spectra were collected from different parts of each bead. The diffuse reflectance spectra were then normalized to 100% to allow direct comparison of the absorption characteristics of glasses of different color saturations.
       Two p-XRF spectrometers were used. An ELIO instrument (XGLab SRL, Milan, Italy) was used to analyze the PG series samples. It is equipped with a Rh anode source with a beam focusing diameter of about 1.2 mm and a silicon drift detector (SDD) with an area of ​​25 mm2. The settings were as follows: acquisition time: 90 s; current: 40 μA; voltage: 40 kV.
       The “VG” group was analyzed using a home-made portable X-ray fluorescence spectrometer developed by the Italian National Institute of Nuclear Physics (INFN Frascati, Italy). The device is equipped with a W-anode source, and the beam is focused into a spot of 300 μm in diameter using polycapillary optics. The effective area of ​​the SDD detector is 20 mm2, and the Mn Kα resolution is 173 eV. The instrument parameters were set as follows: time: 200 s; current: 80 μA; voltage: 40 kV.
       All p-XRF spectra were fitted using PyMCA software [54] to obtain elemental concentrations. Previous publications [55] discussed the optimization of the archaeological glass analysis procedures and the tests carried out to verify the quality of the equipment data. The p-XRF results were processed using principal component analysis (PCA) using the open source data analysis software Instant Clue [56], which was used to examine the heterogeneity of the average composition of the samples (excluding decorative parts) in the data set. Values ​​below the assumed limit of quantification (LOQ) were set to 0 and row and column Z-score normalization was used for data pre-processing.
       Eight samples were analyzed using this technique: five samples were prepared as polished cross sections and analyzed under high vacuum, and three samples (PG39, VG22, and VG106) were placed in the sample chamber without any preparation and analyzed under low vacuum (50 Pa). The equipment was a JSM-IT300LV from JEOL (Akishima, Japan) connected to an energy dispersive spectrometer (Oxford Instruments, Abingdon, UK) with an SDD detector. The spectra were recorded under the following conditions: 1) voltage: 15 kV; 2) current: about 2 mA; 3) acquisition time: 40 s; 4) working distance: 10 mm. The compositional data of the cross-sectional samples were collected from a square area of ​​approximately 10 μm on a side at a magnification of 5000×. The data from the five regions were averaged to calculate the mean bulk composition (normalized to 100%) and standard deviation. In addition to providing information on the composition of the glass matrix, these data were used to further validate the accuracy of the Ca concentrations determined by p-XRF, as these concentrations were used as internal standards in the LA-ICP-MS analysis. Inclusions, if present, can be analyzed by focusing the electron beam on the target particle and collecting qualitative elemental information from the point of study.
       Raman analysis was performed using a LabRAMHR Evolution spectrometer (Horiba Ltd., Kyoto, Japan) equipped with a Peltier-cooled charge-coupled device (CCD) detector, helium-neon (633 nm) and argon (488–514 nm) lasers, a BH2 microscope (Olympus Corp., Tokyo, Japan), and an ultra-low wavenumber module that can obtain Raman spectral information in the region below 100 cm-1 with a spatial resolution of 1 μm and a spectral resolution of approximately 1 cm-1. The laser beam was focused onto the sample using four objectives, namely 20X, 50X and 100X (Leica DMLM microscope) and a 50X long focal length objective (Olympus (Japan). The instrument was calibrated at the 520.5 cm-1 peak of the Si (111) standard. The analysis parameters were adjusted to optimize the signal-to-noise ratio by limiting the laser power below 25% (514 nm excitation) and 75% (633 nm excitation) to avoid sample damage. Spectra were acquired in the spectral range of 60–1560 cm-1 using an 1800 g/mm grating. The linear baseline was subtracted from the raw spectra using LabSpec6 software (Horiba, Kyoto, Japan), with the reference points set at approximately 60, 300, 700, 800 and 1300 cm-1. Band deconvolution was then performed using the curve fitting application Fityk [57].
       Seventeen samples were analysed using a NexION 300 × Perkin-Elmer ICP single quadrupole mass spectrometer (Waltham, USA) coupled to an ESI NWR 213 laser ablation system (ESI New Wave Research Co., Cambridge, UK) (Additional file 1: Table SI2). Concentrations of 38 elements were determined. The measured masses are listed in Table 2, where the accuracy of the quantitative results verified against certified reference materials is also shown. In addition, the settings applied during data collection are listed in Table 3. The laser pre-irradiation allowed micro-sampling of the clean glass just below the surface, and the experimental setup ensured the highest sensitivity of the equipment, suppressing interatomic interference and the formation of doubly charged ions. Between 3 and 5 acquisitions were obtained for each colour on a single glass bead. Quantification was then performed using 44Ca as an internal standard, based on the p-XRF values ​​for this element. The data for each sample were averaged and normalized to 100% for major and minor elements (expressed as mass percent oxide).
       At the beginning and end of each analytical run and after every 5 samples, blank values ​​(gas flow without ablation) were taken and photographs of the reference material were taken to check the accuracy and identify possible deviations. The primary standard used for calibration was NIST612, and the quality control standards were CMOG A (major and minor elements) and NIST614 (trace elements). Mg, K and Ti were found to be systematically overestimated or underestimated. It was decided to compensate for the analytical uncertainty by averaging these elements using certified values ​​obtained for CMOG B, C and D reference materials.
       Detailed results of the FORS study are presented in Additional file 1: Table SI3, and examples of patterns are shown in Figure 3. The interpretation of the spectra is supported by the studies of Micheletti et al. [58] and references therein.
       Representative FORS spectra with bands for spectral interpretation and images for colour/texture comparison
       A large number of beads in our sample set yielded spectra with a reflectance peak in the blue-green region with typical absorption bands for octahedrally coordinated Cu2+ ions (Figure 3, top). All beads from groups 3, 5, and 7 and most of the beads from groups 6 and 8 have Cu2+ bands, although this feature is not uniformly distributed among all type groups. Some beads have other bands in addition to the Cu2+ band: PG136 has a weak Fe3+ band at 380 nm, and PG63 has a Fe3+ band at 450 nm; In addition, the spectral features indicate the possible presence of Mn3+ in the group 7 samples, which needs to be confirmed by other elemental methods.
       Another set of spectra (Figure 3, middle) consists mainly of representatives of groups 2 and 4, but also includes spectra of other group types, namely VG23 and 106, PG109, 111, 158, 159 and 171. These spectra contain Fe3+ bands at 380 nm and 420 nm, which may also indicate the presence of Mn2+, as well as characteristic Co2+ triplet states found at 540, 595 and 650 nm.
       In this general picture, different trends are visible in the near-IR region: the samples in group 4 mainly have clear Cu2+ bands, while the other samples mainly show Fe2+ bands. In addition, around 400 nm, the reflectivity diverges: PG158, 159 and VG106 show lower reflectivity in this region, with a maximum occurring after 700 nm; samples PG138 and VG23 show medium reflectivity, while the other samples reflect strongly around 400 nm (Figure 3, middle). This difference may be caused by the different relative contents of Fe3+ and Mn2+ in the glass matrix.[59]
       The red glasses (VG24, 25, 26, 29) show a Cu0 band around 560 nm, and some samples show broad Fe2+ bands in the near infrared region (Fig. 3, bottom), confirming the data on heating under reducing conditions.
       With this technique it is impossible to obtain any information about small decorative details made of white or yellow glass, since the spectrum does not have any characteristic bands other than that characteristic of the main part of the bead.
       On-site p-XRF analysis is necessary to track compositional relationships between different samples, which are then used to determine whether the subset of samples accepted for analysis in the laboratory correctly represents the entire sample set.
       The composition of the samples determined by p-XRF in this study is presented in Additional file 1: Table SI4. For comparison, published data for group 4 cobalt-copper beads [34] are also included. Figure 4 shows the results of a preliminary analysis of the p-XRF data using PCA (the first three PCAs explain 59.94% of the explained variance).
       Dual PCA plot of p-XRF data. PC1 vs. PC2 (top) and PC1 vs. PC3 (bottom). Data for decorative parts is not shown in the figure.
       The type groups were divided by PC1, where K2O and CaO played the main roles. Groups 1 and 2 samples had higher K2O contents than the other groups, with group 1 having the highest contents (4–5 wt.% K2O) and groups 3 and 7 having the lowest contents (approximately 1.5–3% K2O). For all other samples, the K signals were mostly below the LOQ (i.e. 1.2% K2O), with some exceptions in group 8 (PG33, 63 and 170). CaO values ​​showed a different picture, with the highest concentrations found in groups 2 (Terni samples only), 3, 4, 5 and 8, while group 1 and the Vulci samples of group 2 had the lowest concentrations among the entire sample set.
       For PC2, the key elements are Ti, Zr, Cu, Mn and Sn, although no specific separation occurs for this PC. PC3 is separated due to the different contents of Co and Ni.
       The PCA plot in Figure 4 shows that composition largely reflects typology, with a few notable exceptions. PG112 (group 1) was added to groups 5, 6, and 8, confirming the results of [34]. Likewise, PG169 (group 1) was more closely related to group 3. Although cluster 8 is within the PC2 trend, it appears to be quite heterogeneous in terms of composition. Groups 1 and 2, although widely scattered across the plot, are the only groups with positive PC1 values. Tubular beads, although easily distinguishable typologically, are closely related compositionally to groups 6 and 8.
       In addition to describing the distribution of typological groups in terms of composition, the purpose of the p-XRF analysis was to monitor the representativeness of the smaller sets of samples analyzed by the laboratory, since all groups except groups 5 and 7 had some representativeness in the subset studied by minimally invasive methods. The samples in the lower right quadrant of the PCA plot in Figure 4 (e.g., the red glass from the Bronzetti Sardi tomb) are not representative enough to be obtained for analysis in the laboratory. This issue should be further considered in the discussion of the data presented in the Discussion section.
       A deeper understanding of the general trends revealed by the PCA analysis of the p-XRF data can be gained from some 2D plots related to the glass properties (Figure 5). The plot of Fe2O3 versus TiO2 shows that samples PG88, 89 (Group 6) and 136 (Group 8) have higher Ti contents than other samples in their groups, without a proportional increase in Fe content. The plot of CaO versus SrO shows the same trend line for almost all samples, while different sample types show slightly different contents of these elements. As for the red samples in Group 1, they are all within the concentration range of Groups 1 and 2, allowing us to assume that the compositions of Group 1 and Group 2 samples represent the same glass type. Group 5 samples were related to Groups 4, 6 and 8. Despite the obvious typological differences, sample PG63 is compositionally closer to Group 7 (fibular beads). These assumptions about compositional similarities allow us to extend the discussion of data obtained from laboratory analysis of samples to samples analyzed in museums using different analytical methods.
       Due to the small size of the decorations (small thickness and width), the instrument was unable to detect a unique response from these glasses, but some rudimentary evidence was obtained from the opacifiers. High Ca and Sb peaks in the white decoration of Group 4 indicate the presence of calcium antimonate in the white glass, a substance that has been documented in Egypt since the LBA [60]. This was not observed in the Group 1 samples, where no Sb was detected in the white part, but high Ca concentrations were clearly detected in VG25 and 29. In the yellow part (present on the beads from Groups 3 and 7), Pb and Sb signals indicate the presence of lead antimonate, which gives the beads their color and opacity. The dark decorations did not show significant changes in composition compared to the bulk of the same bead.