Teardrops at the Lake: Chemistry of New Kingdom to Makuria Glass Beads and Pendants Between the First and Second Nile Cataracts

International expeditions extensively excavated Lower Nubia (between the First and Second Nile Cataracts) before it was submerged under the waters of Lake Nasser and Lake Nubia. The expeditions concentrated on monumental architecture and cemeteries, including sites at Qustul and Serra East, where the New Kingdom, and Napatan, Meroitic, Nobadian, and Makurian-period elites and common people were buried, ca. 1400 BC–AD 1400. Although the finds abound in adornments, including bead imports from Egypt and South India/Sri Lanka, only a few traces of local glass bead-making have been recorded in Nubia so far. Based on results of laser ablation–inductively coupled plasma–mass spectrometry (LA-ICP-MS) analysis of 76 glass beads, pendants, and chunks from Qustul and Serra East contexts, dated between the New Kingdom and the Makuria Kingdom periods, this paper discusses the composition and provenance of two types of plant-ash soda-lime (v-Na-Ca) glass, two types of mineral soda-lime glass (m-Na-Ca), and two types of mineral-soda-high alumina (m-Na-Al) glass. It also presents the remains of a probable local glass bead-making workshop dated to the period of intensive long-distance bead trade in Northeast Africa, AD 400–600.


Introduction
This article offers an overview of glass types in Lower Nubia, their provenance, and chronology based on the laser ablation-inductively coupled plasma-mass spectrometry (LA-ICP-MS) analysis of the New Kingdom through Medieval glass beads and pendants from Qustul and Serra East cemeteries, and the glass remains (beads, pendants, and chunks) from a Nobadian Serra East household. Historic Nubia encompasses the southern edge of modern Egypt and the northern part of modern Sudan (Dafalla, 1975;Osman, 1992;Taha, 2013Taha, , 2021. Due to the natural boundary of the rock-strewn rapids of the Second Cataract, Nubia is usually referred to as comprising two parts-Lower Nubia, between the First and the Second Nile Cataracts, and Upper Nubia, in the south (Fig. 1). As a part of the Nubian civilization, Lower Nubia witnessed one of the earliest events of social complexity in Africa. Thanks to its frontier location, it also benefited from contacts with its Egyptian neighbors. Archaeological finds suggest that the A-Group people (3700-2800 BC) and the C-Group people (2300-1550 BC) were wealthy individuals. Due to the strategic location at the junction of trade routes, these individuals controlled the trade routes and provided Egypt and the Mediterranean with raw materials and exotics. The Pan Grave people (2200( -1550, represented by small and dispersed populations from the Eastern Desert, also contributed to the rich history of the Lower Nubian region (e.g., Adams, 1977;Hafsaas, 2006;Török, 2008;Williams, 1983Williams, , 1986Williams, , 1989Williams, , 1993. The region was under Egyptian control in the New Kingdom (ca. 1570-1069. Between 747 and 656 BC, the 25th Dynasty, otherwise called Kushite Dynasty, which originated in Lower Nubia, controlled Ancient Egypt and other parts of northeast Africa, stretching from the confluence of the Niles to the Mediterranean. The Nubian kings promoted the revival of the arts, language, architecture, and religion of the New Kingdom, and Egyptian artisans and scribes were employed in Nubia (e.g., Fisher, 2012;Török, 2008;Welsby, 1996;Williams, 1990). The wealth of the Lower Nubian region during the Napatan period (ca. 750-350 BC) is assumed to have continued when the center of power moved to the kingdom of Meroë (ca. 350 BC-AD 350). Under Roman rule in the north and Meroë in the south, in the first through fourth centuries AD (Classic and Late Meroitic period), Lower Nubia was an intermediary between Upper Nubia and Egypt (e.g., Török, 2008;Williams, 1991). During the Kingdom of Nobadia (ca. AD 350-600), the region had unsettled relations with the Blemmyes, active in the Eastern Desert and the Red Sea ports, and the Egyptians in the north (Emery & Kirwan, 1938;Obłuski, 2013Obłuski, , 2014. By the early eighth century AD, Nobadia and Makuria (in Upper Nubia) were united under a Makurian king, which gave rise to the Makurian period (seventh-fourteenth centuries AD) (Fisher, 2012;Welsby, 2002).
In the sixth century AD, Lower Nubia adopted Christianity, and by the seventh century, there were changes in burial practices, especially in the number of burial goods. In contrast to the pre-Christian burials that abounded in grave goods, these were scarce or absent altogether in Christian burials. Before the Christian era, many beads were buried with deceased Nubians regardless of their social status, sex, and age. These included locally available ostrich eggshell and semi-precious stones and imported resources (Red Sea marine mollusk shells, Mediterranean Corallium rubrum sp., semiprecious stone). The beads were mostly made of faience and glass (Then-Obłuska, 2018b). Despite their apparent profusion in Nobadian graves, only one that assumed glass bead-making workshop has been recorded in a house at Serra East, in Lower Nubia (Williams, 1993, p. 229-230). Hence, the beads seem to testify to a rich history of wide-ranging contacts in the region. While Nubia's trade links with Egypt and the Mediterranean have long been well acknowledged, its eastern connections have only recently been recognized. Lower Nubia was close to Berenike, Marsa Nakari, and Quseir ports at the Red Sea, part of a busy commercial network connecting the Mediterranean world with the Indian Ocean during the Roman and Early Byzantine periods. The glass bead finds in Lower Nubia dating to these periods testify to the contacts with these ports and the Nubian Nile Valley (Then-Obłuska & Dussubieux, 2016;Then-Obłuska & Wagner, 2019a, b;Then-Obłuska, 2015a.

Lower Nubian sites and samples
Before disappearing under the waters of what is now Lake Nasser and Lake Nubia in the 1960s, Lower Nubia was the focus of extensive excavations that yielded, among others, a wide variety of beads, including glass ones. These beads were stored in museums and institutions around the world. One of the archaeological rescue missions was the Oriental Institute Nubian Expedition (OINE) of the University of Chicago (e.g., Williams, 2009;Williams & Heidorn, 2019). An assemblage of eighty-one beads and artifacts from Qustul and Serra East found during the OINE excavations in the early 1960s, and presently stored in the Oriental Institute Museum University of Chicago, was used in this study. Of this number, one bead (OINE67) appears to be a modern intrusion, three were made of stone (OINE19, 27, 51), and one was too corroded (OINE21). Hence, five beads were excluded from the LA-ICP-MS analysis.
The sites in Qustul and Serra East were excavated on the east bank of the Nile, just to the north of the Second Cataract (Williams, 1990(Williams, , 1991(Williams, , 1992(Williams, , 1993. The samples were found mostly in graves, at cemeteries Q, QC,R,VF,[20][21][22][23][24][25][26][52][53][54][55][56][57][58][59][60][61][62][63][64][65][66], as well as at cemetery B (OINE79) and its surface (OINE68-69) in Serra East. There are also glass beads from a household context in Serra East, site LB. This context is interpreted as a bead workshop associated with Nobadian culture (80)(81). The Nobadian cemeteries are generally tumuli superstructures with either single or multiple grave units. Various artifacts (metal tools, weapons, toilet utensils, and fittings, as well as some stone, clay, bone, and ivory objects, basketry, textiles, and glass vessels) have been found, although pottery and bead adornments dominate the grave assemblages. Bead adornments were discovered mostly loose, scattered, and in heaps. They accompanied the deceased regardless of their age or sex. Necklaces and bracelets are the main beadwork types; however, other forms have also been recorded (Williams, 1991(Williams, , 1993. Site LB was a house in a shallow wadi built against the south side of a rock outcrop on drift sand. The structure was built into the natural squarish inset or break in the rock face. Stones were loosely stacked with mud to form walls for one room with a rough stone floor paving. The walls stood only about 0.5 m high (Williams, 1993, pp. 229-230). The house or shelter, 2.7 × 3.7 m, contained a good collection of whole and smashed but complete pottery vessels. Disturbed skeletons of two goats or sheep in clean sand were found beneath the paving and deep debris. Outside, to the east, against the rock, several cooking or hearth constructions, with considerable ash deposits, were found. The doorway may have been in the east wall. Further to the west, 3.05 m northwest of the shelter, a fragmentary oven or kiln was also found built into a small corner against the rock. To the south, test pits in the wadi sand yielded occasional bead and pendant fragments 78,80,81), partially perforated stone beads, and some flint chipping debris and potsherds. Glass chunks (OINE76-77), ostrich eggshell fragments, unfinished ostrich eggshell beads, and a carnelian chunk were also collected from site LB. All of these suggest that there was a bead-making workshop at the site. The hearth was approximately 2.30 m deep and 2 m wide and consisted of upright stones with possible ash deposits. While the Qustul cemetery Q (220) was dated to ca. AD 370/380-410 (OINE01-42), the remaining cemetery sites were dated between the early fifth and sixth centuries AD (OINE43-66) (Williams, 1991(Williams, , 1993. The reuse of Meroitic beads in Nobadian graves was a common practice, and these are treated as Meroitic-dated bead types. Other beads (OINE58-61, 63-64) were of the 25th Dynasty date (Williams, 1990, and personal communication). A few samples belong to the New Kingdom (OINE62, 65) and Makurian (OINE57, 73) dates.
The glass assemblage for this study was selected to include all types of manufacturing techniques, color, and shape. Electronic Supplemental Material (ESM 1) presents the specimens arranged based on their Oriental Institute Museum (OIM) number, including information on the site, find context, find number, and all registered data about the techniques of manufacture, shape, color, and dimensions (mm scale). The assemblage includes monochrome, bichrome, and mosaic glass beads, as well as metal-in-glass beads with gold or silver foil between two glass layers. A variety of manufacturing techniques (e.g., drawing a glass tube; winding glass around a metal mandrel; folding a glass strip around a metal mandrel or joining glass strips around a mandrel; and rod-piercing glass), finishing processes (segmenting in molds and breaking apart the segments of a drawn tube; or cutting a drawn glass tube and heat-rounding the sections), and decoration (the application of trails, mosaic eyes) are recognized in the assemblage. The objects are illustrated in Fig. 2 and identified by their sample numbers.

Results and discussion
The compositions of the glass beads, including 56 major, minor, and trace elements, were obtained using LA-ICP-MS at the Elemental Analysis Facility at the Field Museum. More details on the instrumentation and protocol are in Then-Obłuska and , and ESM 2 presents major, minor, and trace element compositions of Corning Reference Glass B and D. Seventy-six glass beads, pendants, and chunks were analyzed. Different colors of the bichrome and polychrome beads were measured individually, and some measurements had to be repeated because, in some cases, the color meant to be targeted had been missed, and another one was measured instead. These resulted in 112 compositions (ESM 3). Soda is the main alkali in all glass samples, and four main glass types were identified based on the content of MgO and Al 2 O 3 (Fig. 3). Soda-lime glasses form the most numerous group with eight samples made of plant ash-soda-lime glass (v-Na-Ca) and 84 samples made of a mineral soda-lime glass (m-Na-Ca). Four samples feature mineral-or plant ash-soda composition (m/v-Na-Ca), and 16 samples are mineral soda-high alumina (m-Na-Al) ( Table 1).

v-Na-Ca glass
Eight samples made from soda-rich glass have low alumina (Al 2 O 3 < 2%) and high magnesia (MgO > 3%) concentrations, indicating the use of plant-ash soda as a flux. Lime (CaO) concentrations range 3.5-8.0%. Plant-ash soda-lime-silica glass (v-Na-Ca) is the earliest known glass type. It was produced in Egypt (e.g., Rehren & Pusch, 2005;Smirniou & Rehren, 2011;Tite & Shortland, 2003)  and Mesopotamia (e.g., Shortland et al., 2018) as early as the middle of the second millennium BC. Later, v-Na-Ca glass was made by the Sasanians. This was followed by Islamic glass-makers in a region east of the Euphrates from the third century BC to about the seventeenth century AD, and by Islamic glass-makers in the East Mediterranean region, Egypt, and the Levant, starting from the mid-ninth century AD (Brill, 2005;Henderson et al., 2016;Mirti et al., 2008Mirti et al., , 2009Phelps, 2016). Based on different levels of MgO and trace constituents such as Ti, Zr, Cr, and La, two subtypes of plant ash-soda glass are distinguished in the OINE assemblage: New Kingdom (v-Na-Ca NK) and Medieval (v-Na-Ca OINE73) (Fig. 4A).

v-Na-Ca NK
A pendant of light blue wound glass decorated with a white spiral trail (OINE65Bl, W), found in Nobadian grave R 119 at Qustul and thought to be Meroitic in date (Williams, 1991, p. 146), is part of a v-Na-Ca sub-group. It has low levels of Rb (about 11 ppm) and Li (about 5 ppm). These chemical attributes exclude glass affiliation with contemporary v-Na-Ca Sasanian glass, featuring higher levels of both elements (e.g., Mirti et al., 2009;Then-Obłuska & Dussubieux, 2016). The two-colored glasses used to make this pendant have trace element levels indicating a Late Bronze Age date. Soda plant ash glass was produced since the middle of the second millennium in Egypt and Mesopotamia, and both production areas can be distinguished by plotting Zr/Ti and Cr/La ratios. A high ratio of 1000*Zr/Ti > 40 and a low ratio of Cr/ La < 4 suggest Egyptian provenance of the glass, while the ratios of 1000*Zr/Ti < 60 and Cr/La > 4 indicate Mesopotamian glass (Shortland et al., 2007). Comparison of the two ratios in OINE65 with those in Late Bronze Age glasses from Egypt and Mesopotamia (based on Shortland et al., 2007;Henderson, 2013, Fig. 6.10) reveals similarities with the Egyptian data (Amarna and Malkata) (Fig. 4A).
Undoubtedly, the blue glass (OINE65B) owes its color to the presence of CuO (2.5%), although a contribution of Fe 2 O 3 (0.5%) cannot be excluded. Traces of tin, SnO 2 (0.09%), Co (61 ppm), and Ni (62 ppm) might have been brought in accidentally by the copper. All the elements occur in proportions resembling the scores for the "Cu blue" glass from Amarna and Malkata (Shortland & Eremin, 2006, Table 1). The white glass (OINE 65 W) has CaO at a level of 6.5% and a very high level of Sb 2 O 5 (5.1%) and, like the New Kingdom v-Na-Ca glass from Amarna and Malkata, was decorated with   Table 1). Similar proportions of CaO in the blue and the white glass prompt the conclusion that the latter color was obtained solely by adding antimony to the colorless glass. Antimony precipitated with the calcium in the glass made with soda plant ashes, and calcium antimonate was obtained in this way (Shortland, 2002). A higher level of Al 2 O 3 and Sb 2 O 5 in the white glass (OINE65W) along with traces of cobalt in Fig. 4 A A biplot of chromium/lanthanum versus 1, 000 × zirconium/titanium in OINE v-Na-Ca glass in frames for late Bronze Age glasses from Egypt (Amarna and Malkata) and Mesopotamia (Tell Brak and Nuzi) (based on Shortland et al., 2007); B Biplot of Al 2 O 3 and MgO/CaO showing results for v-Na-Ca glass OINE73, and "Mesopotamian" glass from Samarra 1 & 2, mid-ninth century AD , Raqqa, ninth century AD (Henderson et al., 2016), Nishapur, ninth and tenth century AD (Henderson et al., 2016), and Veh Ardasir, AD 300-700 (Mirti et al., 2008(Mirti et al., , 2009), within range borders, according to Phelps (2016) copper blue glass (OINE65B) would exclude its affiliation with later New Kingdom glass from Lisht, dated to the end of the second millennium (Shortland & Eremin, 2006, p. 596-597, Table 2). Furthermore, the presence of traces of SnO 2 in OINE65B would also exclude its affiliation with early New Kingdom blue glass dated to the fifteenth century BC (Shortland & Eremin, 2006, p. 596-597, Table 2). The OINE65 affiliation with the Amarna and Malkata sites, assumed glass-making centers in the fourteenth century BC (Shortland & Eremin, 2006), makes this bead stand out from the Nobadian tomb collection. The bead may have been moved from one of the New Kingdom tombs at Qustul (Then-Obłuska, forthcoming; Williams, 1992). Bichrome pendants with a similar trail added spirally are dated to the New Kingdom period (Spaer, 2001, cats. 72-74). Another bead, OINE62, found in a 25th Dynasty grave, features higher K 2 O (1.3%) and high MgO (2.8%), and its translucent purple wound glass has 0.8% MnO and 0.7% Fe 2 O 3 . Although the New Kingdom glass usually features more elevated magnesium and potassium (Shortland & Eremin, 2006), OINE62 still seems to fit this glass group since it typologically resembles other New Kingdom beads (Then-Obłuska in press: cat. 324.1, from Qustul R 94 dated to New Kingdom Post-Amarna period; Metropolitan Museum of Arts, New York, MET 11.215.661, from Malqata, Palace of Amenhotep III, ca. 1390-1353 BC). Also, comparing its Cr/La and 1000*Zr/Ti ratios with records for Egyptian and Mesopotamian glass makes its attribution to New Kingdom Egyptian glass fairly apparent (Fig. 4A).

v-Na-Ca OINE73
Other v-Na-Ca compositions belong to a polychrome bead fragment, OINE73A-E, found in a Nobadian house (fourth-sixth centuries AD), thus implying the v-Na-Ca OINE73 glass could have been produced during the Sasanian period. To date, the OINE73A-E data for the various colored glass fragments were placed in a MgO/CaO vs. Al 2 O 3 graph. The result demonstrated a distinction between the soda plant ash glass of Eastern Mediterranean provenance (Syria, Egypt, and Palestine/Levant) and the soda plant ash from the Mesopotamian region (northern Syria, Iran, and Iraq), dated between the eighth and tenth centuries AD (McIntosh et al., 2020;Phelps, 2016;Schibille et al., 2019). Further, two groups have been distinguished in the Mesopotamian glass: Mesopotamian Type 1 with samples from Veh Ardasir (third through seventh centuries AD) and Raqqa (type 4; ninth century AD), and Mesopotamian Type 2 with samples from Samarra (Schibille et al., 2018, Fig. 6). The results for OINE73, with Al 2 O 3 (about < 1.5%), fit the Mesopotamian Type 2 (Phelps, 2016;Schibille et al., 2018;Fig. 6), particularly the Samarra 2 subgroup (Fig. 4B).
Analyses of (de)colorants and opacifiers in OINE73 are compatible with an Islamic glassperiod affiliation. All glasses (yellow, black, red, and colorless) contain some concentrations of SnO 2 and PbO (> 2%). The white (PbO 2 and SnO 2 = 1.6%) and yellow (PbO = 8.7%, SnO 2 = 1.05%) colors were opacified with tin oxide and lead stannate, accordingly. The yellow color has some traces of antimonate (0.04%). Although, after the fourth century AD, tin, instead of antimonate, was used as an opacifier (Tite et al., 2008), occasional use of antimonate and arsenic was recorded for Merovingian-and Islamic-period yellow and greenish-yellow glass (Neri et al., 2019). With all the above in mind, the OINE73 fragment may be assumed to be a Medieval intrusion in the context of the Serra East Nobadian household rather than a Sasanian production. In fact, OINE73A-E is the fragment of a bead made of mosaic glass with a so-called checkerboard pattern already known in the Hellenistic period, and its production continued through Medieval times (e.g., Spaer, 2001, p. 120). The glass might have arrived through the Red Sea ports of Aidhab (used as a port at least from the time of the Fatimid conquest of Egypt in AD 969) and Suakin, some 230 miles south of ʿAidhab, founded in the ninth century.

m-Na-Ca glass
Most analyzed OINE samples (n = 84), featuring low alumina (< 3%), also have low magnesia (< 1.5%) concentrations, indicating the use of mineral soda as flux. Soda-lime glass, using mineral soda as a flux -usually in the form of natron from Wadi el Natrun in Egypt -was manufactured in Egypt and the Syro-Palestinian region for two millennia, between the tenth century BC and mid-ninth century AD (e.g., Phelps et al., 2016;. Dated to the beginning of the tenth century BC, glass vessels from Theban tomb 320 are characterized by low potassium and magnesium (< 1.2%), soda (18.2-23.4%), calcium (1.3-4.8%), and alumina (> 2.1%) and, most likely, were made from sand and mineral soda (Schlick-Nolte & Werthmann, 2003). Natron-based glass, usually containing low levels of magnesia and potash (< 1.5%) (LMLK glass) and moderate levels of Al 2 O 3 (0.5 to 3%) (e.g., Panighello et al., 2012), became widespread in the Southern, Eastern, and Western Mediterranean (e.g., . Some samples of the OINE m-Na-Ca type have very low levels of Al 2 O 3 (< 0.5%) and low levels of some trace elements (m-Na-Ca LT), suggesting the use of silica sources different from that of the m-Na-Ca glass produced between the Hellenistic and Islamic periods. In the sample discussed, this type of glass is found mainly in the first through sixth century-AD bead types. Hence, they are labeled Roman glass (m-Na-Ca R).
An elevated cobalt (555-1629 ppm) content was the main colorant of the dark blue beads (OINE02BCo,09,13,15BCo,16BCo,25,30,31BCo,33BCo,34BCo,35,41,55BCo,72,74,81) and workshop chunks (OINE76-77). The cobalt used for the natron-type beads is not unambiguously associated with any particular impurities, a trademark of Roman cobalt sources (Gratuze et al., 1992). According to Gratuze et al. (2018), natron glass colored with cobalt features a relatively constant pattern with a high CoO/NiO ratio (CoO/NiO > 24). Sometime between the late fourth and the beginning of the sixth century, the CoO/NiO ratios experience a drastic decrease (2.2 < CoO/NiO < 5.1) (Gratuze et al., 2018, p. 18). In the OINE assemblage, only a few dark blue glass beads have a high Co/Ni ratio (> 24 for OINE16, 33, 72), while the glass of these Nobadian-dated beads may have been produced earlier, in the Meroitic period. Only one bead in the studied assemblage has a low Co/Ni ratio of 5.9 (OINE32). In contrast, most samples have ratios below 24 but higher than 5 (OINE02, 09,13,15,25,30,31,34,35,41,55,74,76,77,81), which may have resulted from recycling earlier glass mixed with the glass of a later date (Fig. 5C). Many of these samples were made of Egyptian glass (Fig. 5B). It seems probable that the period between the late fourth and the sixth century was the time when earlier, Early Roman cobalt glass of Co/Ni ratio above 24  would have been mixed with new resources characterized by the Co/Ni ratio between 2.2 and 5.1 . The m-Na-Ca beads were made with diverse techniques, well recognized in Roman and late antique northeast Africa (e.g., Then-Obłuska & Wagner, 2019b) and beyond. These are beads made of drawn and segmented glass and gold-in-glass as produced in Alexandria, Egypt (Kucharczyk, 2011;Rodziewicz, 1984). Other beads were made of the wound, folded, and rod-pierced glass, and many of them might have been manufactured locally.

m-Na-Ca LT/Classic
Nine m-Na-Ca samples (OINE59, 60Bl-Y, 61, 64Bl-B-Gr-W-Y) are characterized by very low levels of MgO, K 2 O, Al 2 O 3 ( ≤ 0.5%), and Fe 2 O 3 (average of 0.8%). They also have low levels of silica-related impurities and other earth trace elements (e.g., Ti, Sr, La, Rb, and Ba) compared with the m-Na-Ca Roman glass type. Although OINE58 and 63 feature MgO, K 2 O, or Al 2 O 3 levels higher than other examples in the m-Na-Na LT glass group, their trace elements (Sr) still fit the "low trace" natron group. For this reason, they have been assigned to the m-Na-Ca LT glass group.
The OINE m-Na-Ca LT composition suggests a very clean silica source, i.e., a better quality of sand or even quartz pebbles, for glass production (Shortland & Eremin, 2006). Based on the Zr level, two types of "classic natron" glass were distinguished: low-Zr and high-Zr natron glass (Conte et al., 2019). Low levels of Zr (< 9 ppm) and the lowest possible levels of alumina, magnesia, potash, iron, and REEs in a glass sample indicate, according to the authors, the use of quartz pebbles (Conte et al., 2019, table 4). However, the Zr level (> 21 < 272 ppm) in the OINE m-Na-Ca LT glass would point to a high-Zr affiliation, and this, in turn, would exclude the use of quartz pebbles. High lime concentrations in the m-Na-Ca glass came with sand collected from a beach and thus contained seashell fragments. A relatively low Sr content linked to a rather high CaO/SrO ratio (362) is thought to result from the addition of diagenetically altered shells, partly recrystallized once their initial strontium contents had been lost (Conte et al., 2019;Wedepohl et al., 2011).
As for the m-Na-Ca R glass, the authors compared yttrium to zirconium (Y/Zr) and cerium to zirconium (Ce/Zr) ratios of the m-Na-Ca LT glass samples with the ratios for glass produced in the Levantine and Egyptian regions, respectively. As shown in Fig. 5A, the m-Na-Ca LT glass samples follow a trend observed for glass samples of Egyptian provenance. Low Al 2 O 3 (< 0.5%) levels undermine any comparison of the m-Na-Ca LT glass with most of the low MgO and K 2 O (< 0.5%) natron glasses from the eighth through the fourth century BC Europe (e.g., Macedonia; Blomme, et al., 2017). Still, similar compositions have been reported from various French sites, dated to the beginning of the Iron Age, in the ninth through second centuries BC (Gratuze, 2009, Fig. 2). One of these groups features low potassium and low alumina, each at a level of about 0.5%. Some samples in this group come from the Champlay context dated to ca. 750-500 BC or 750-400 BC (Gratuze, 2009, Fig. 2). An antimony decolored glass sample from Sardis (Turkey), dated to ca. 700-500 BC (Ignatiadou, 2000), and a turquoise decoration of the Bologna eye bead from 500-300 BC-Etruscan context in northern Italy feature similar very low Al 2 O 3 results (Arletti et al., 2010, IG45). Interestingly, an opaque red chunk with MgO, K 2 O, and Al 2 O 3, < 0.5%, was found in Persepolis, dated to around the fifth century BC (Brill, 1999, IIH:198). Additionally, a glass of probable Egyptian provenance, featuring Mg, K, and Al, < 0.5%, low levels of some trace elements that resemble the m-Na-Ca LT glass, and a so-called "classic natron," was identified in the Iron Age Italy, ca. 800-500 BC (Conte et al., 2019).
The study by Conte et al. (2019), using the measurements of selected elements' levels, also presents ways to date the natron black glass more adequately. Some black samples with low lime, high iron, and high trace and REEs contents are dated to ca. 900-700 BC. Other samples (TG3bl, TG12bl, and TG13bl), characterized by lower alumina, titania, iron, and higher lime concentrations, are comparable with the OINE m-Na-Ca LT, and dated to ca. 700-500 BC.
An analysis of opacifiers in the m-Na-Ca LT group confirms its early date. Yellow glass in OINE60Y and 64Y features significant Sb 2 O 5 (1.2%, 1.1%) and PbO (15%, 9.7%) levels and a complete lack of tin. Antimony-based opacifiers (i.e., lead antimonate yellow) were used, in the Near East and Egypt, from the onset of glass production, ca. 1500 BC, through the Roman period (Turner & Rooksby, 1959). Towards the end of the Roman period (especially fourth century AD onwards), the production of opaque yellow glass would fall back on the use of stannate instead of lead antimonate (Tite et al., 2008). It was not until the late fifteenth century AD that the latter was reintroduced into glass production (Molina et al., 2014). OINE58 is translucent blue with CuO (1.2%), MnO, and Fe 2 O 3 (0.4%). Like the blue glass in v-Na Ca NK (OINE65A), it does not contain tin.
An m-Na-Ca LT type bead (OINE64), decorated with colorful spots, belongs to a group of so-called crumb beads reported from contexts dated between the Late Bronze and the Medieval times (Spaer, 2001, p. 127). It was found along with other m-Na-Ca LT glass beads in a 25th Dynasty grave. The same context also yielded a quadruple wedjat eye, typical of the Third Intermediate Period (Williams, 1990). Comparing OINE64 black glass compositions with Italian samples (see above) suggests ca. 700-500 BC as a probable date. PXRF analysis of red "natron sodium-lead-calcium-magnesium-silica" glass beads from the Nubian site of Gala Abu Ahmed in Wadi Howar, dated ca. 1100-400 BC, provided no trace elements feasibly comparable with OINE glass (Daszkiewicz & Lahitte, 2013). Some compositional similarities in OINE58 can be recognized in an orange bead of Egyptian glass, featuring low levels of MgO, K 2 O, and Al 2 O 3 (< 0.5%) (Then-Obłuska & Wagner, 2019b, SNM07). Its elemental levels (e.g., Al 2 O 3 0.25%, Sr 93 ppm, Zr 13.8 ppm, and Ti 319 ppm) resemble those in m-Na-Ca LT glass; however, the NaO (2.98%) and CaO (1.8%) levels are much lower. The bead was found in a Sedeinga grave, accompanied by several other beads of the same type and Napatan amulets (Then-Obłuska, 2015b). A Napatan date for this glass type can be supported by evidence from Nag Shayeg, where beads of this type have been found in a probable Napatan tomb, T131 (Then-Obłuska & Wagner, 2019b, Pl. 28.1-28.2).

Fig. 6
Principal components 1 and 2 calculated using the concentrations of MgO, CaO, Sr, Zr, Cs, Ba, and U for samples belonging to glass groups m-Na-Al 1, 2, 3, 4, and 6 and for samples from Nubia. The m-Na-Al 1 glass samples are unpublished data from Sri Lanka and South India, the m-Na-Al 2 glass samples are beads from Chaul (Dussubieux et al., 2008), the m-Na-Al 3 glass samples are beads from Kopia (Dussubieux & Kanungo, 2013), the m-Na-Al 4 are glass vessel fragments from Sumatra (Dussubieux, 2009), and the m-Na-Al 6 glass are from the site of Juani Primary School (Dussubieux & Wood, 2021) m/v-Na-Ca glass Four glass samples (OINE03,45,46,53) with low alumina levels, moderate K 2 O (< 1.5%) level, and elevated concentrations of MgO (> 1.5%) suggest the use of mineral-soda and plant ash or the specific soda plant ashes. The K 2 O or MgO > 1.5% values are commonly believed to indicate the use of organic material in the form of plant or wood ash in the glass batch. Glass with higher concentrations of MgO or/and K 2 O was identified and discussed for early Roman glasses in Egypt (Nenna & Gratuze, 2009;Then-Obłuska & Dussubieux, 2016), and the Egyptian m/v-Na-Ca glass was found mainly in the first through mid-fourth centuries AD Nubia (Then-Obłuska & Dussubieux, 2021;Then-Obłuska & Wagner, 2019a). The present study confirms their Egyptian provenance (Fig. 5A), and the OINE m/v-Na-Ca glass beads in this assemblage were most probably Meroitic items reused in Nobadian graves.

m-Na-Al glass
Sixteen samples have high alumina (> 7%) and low magnesia (< 2%) concentrations, indicating the use of mineral-soda flux (m-Na-Al). The mineral-soda high alumina glass beads, with relatively high (> 5%) concentrations of alumina and trace elements, are particularly common in India where, undoubtedly, they were manufactured (Brill, 2003;Dussubieux et al., 2010. Low magnesia (< 2%) concentrations indicate the use of mineral-soda flux. Based on different trace element levels, two sub-types have been identified within the OINE assemblage: 15 samples were made of m-Na-Al 1 and one sample of m-Na-Al 2 (Fig. 6). m-Na-Al 1 Fourteen samples in the OINE collection have high Al 2 O 3 contents ranging from 7.1 to 13.4% (OINE01,04,07,08,12,14,18,20,23,39,49,50,52,79). The MgO concentrations in this glass are usually low (< 1%), while other trace elements, such as uranium with 4-24 ppm, have the highest concentration in this study. Dussubieux and co-authors distinguished a few subtypes of high-alumina mineral soda glass (m-Na-Al 1-4 and 6) based on the contents of five elements: Sr, Zr, Ba, U, and Cs (Dussubieux et al., 2010, tab. 3;Dussubieux & Wood, 2021). Using principal component analysis (PCA) and glass constituents, MgO, CaO, Zr, Sr, Ba, Cs, and U, the m-Na-Al glass beads found in Nubia were compared with already defined m-Na-Al subtypes (m-Na-Al 1-4 and 6), and they showed similarities with the m-Na-Al 1 glass (Fig. 6). The compositions of the 14 high alumina samples of this m-Na-Al 1 glass group (formerly known as "low uranium-high barium glass," Dussubieux et al., 2010) have average contents of Ba and U that match the m-Na-Al 1 type (Table 2).
Additionally, one sample, a green drawn and rounded bead OINE11, has a high concentration of MgO (3.3%), K 2 O (2.2%), and Al 2 O 3 (5.6%), and low CaO (3.1%). The trace element levels match the m-Na-Al 1 group (Table 2). Also, the PbO (4%), CuO (0.8%), and SnO 2 (0.5%) levels fit within the range for the green glass in the m-Na-Al 1 group. OINE11 was Table 2 Average concentrations and standard deviations of important elements crucial for separating m-Na-Al subtypes (data from Dussubieux et al., 2010), followed by data for the high-alumina glasses (m-Na-Al 1, m-Na-Al 2) from the Lower Nubian OINE collection found in a Nobadian grave together with glass beads made of m-Na-Al 1 glass, which would confirm its affiliation with high alumina glass of South Indian/ Sri Lankan provenance. The presence of lead and tin in OINE14 (PbO = 4.6%; SnO 2 = 0.54%) suggests the yellow bead was probably colored and opacified by lead stannate. Six semi-translucent pale green samples, OINE08,18,23,39,50,79, contain significant quantities of CuO (0.3-0.8%), PbO (2-5%), and SnO 2 (0.3-0.6%), suggesting lead stannate may have contributed to the opacification of the glass. Seven beads, OINE01,04,07,12,20,49,52, are orange and contain high concentrations of copper (CuO 7.2-9.2%), but also a higher concentration of iron (2.3-3.1%) when compared to the blue, red, and black m-Na-Al 1 glass from both South Asia (Lankton & Dussubieux, 2006, p. 129, Table 2) and Nubian sites (Then-Obłuska & Wagner, 2019b). The orange samples are also characterized by high levels of MgO, K 2 O, and P 2 O 5 . Phosphorus and lime-rich inclusions were found in an orange m-Na-Al 1 glass sampIe from South Asia. These suggest a possible addition of an apatite-rich ingredient for internal reduction to convert the Cu 2+ into Cu 2 O (Dussubieux et al., 2010) that usually colors glass orange. The m-Na-Al 1 glass was most probably manufactured in Sri Lanka or South India. Beads made of this glass are found in Sri Lanka and South India, between the second/first century BC and fifth century AD; and in Southeast Asia between the fifth century BC and tenth century AD (Carter, 2016;Dussubieux et al., 2010, tab. 4;Dussubieux & Gratuze, 2013). Aside from the Southeast Asian finds, the presence of South Indian/Sri Lankan glass beads has also been confirmed at the Early Roman Red Sea port of Quseir, Egypt (Then-Obłuska & Dussubieux, 2016), in Merovingian-period Europe (Pion & Gratuze, 2016;Poulain et al., 2013), and Zanzibar, AD 700-1100 (Sarathi et al., forthcoming;Wood et al., 2017).
The South Indian/Sri Lankan glass (m-Na-Al 1, -green, orange, black, yellow, and orange-on-red) has been found in the Nubian Nile Valley between the First Cataract and the confluence of the Niles (Then-Obłuska & Wagner, 2019a, b), including the mid-fourth century AD samples from the cemeteries of nomadic peoples (Blemmyes) around Kalabsha, representing the northernmost presence of these glass beads in the Nile valley (Then-Obłuska & Dussubieux, 2021). Beads made of m-Na-Al 1 glass were produced using a technique diagnostic of Indian origin-drawing a glass tube and heat-rounding its sections (Francis, 2002). South Indian or Sri Lankan glass beads have also been macroscopically identified at other sites associated with the Blemmyes: the Early and Late Roman Red Sea port sites of Berenike and Marsa Nakari (Francis, 2002(Francis, , 2007Then-Obłuska, 2016, 2017b, 2018a and the Eastern Desert sites of Shenshef and Sikait (Then-Obłuska, 2017a, thus pointing to the east-west direction of South Asian bead distribution in northeast Africa. m-Na-Al 2 One sample, OINE57, has low levels of MgO (1.1%) and K 2 O (1.8%) and a high level of Al 2 O 3 (8.5%), pointing to its mineral soda high alumina affiliation. When compared with m-Na-Al 1 glass (Table 2, Fig. 6), it displays higher concentrations of U and Cs and lower concentrations of Ba, Sr, and Zr and fits the m-Na-Al 2 group as defined by Dussubieux et al. (2010). The CuO (0.7%), PbO (2.4%), and SnO 2 (0.4%) levels in the green glass suggest the use of lead stannate. The m-Na-Al 2 glass was previously identified at sites dating from the ninth to the nineteenth century AD, located on the west coast of India and the east coast of Africa (Dussubieux et al., 2010). A recent analysis of more beads from the East African coast has helped revise the chronology for this glass and suggested its presence from around the fourteenth century AD onwards. The Indo-Pacific Khami beads from Southern Africa and m-Na-Al 2 beads on the East African coast have been identified as sharing the same composition. Therefore, both can be assigned to around the fourteenth century AD (Dussubieux & Wood, 2021). Although the m-Na-Al 2 glass beads might have been manufactured in Maharashtra, a recent study using Sr, Nd, and Pb isotope analysis suggests that the raw glass was likely procured from a different region, possibly western Uttar Pradesh . The beads would have been traded across the Indian Ocean through Chaul, south of Mumbai (Wood, 2019). OINE57, found in grave VF68, originally was strung together with other drawn green and black beads and a Mediterranean Sea coral (Corallium rubrum sp.). Since the latest evidence for m-Na-Al 2 glass suggests a new fourteenth century AD dating for the Qustul VF68 grave, it appears fairly probable that the beads and grave may belong to the Islamic period in Lower Nubian history.

Glass Bead-making in Lower Nubia
During the New Kingdom and the 25th Dynasty, glass beads in Lower Nubia were made of Egyptian glass (v-Na-Ca and m-Na-Ca LT) and most probably imported. Although glass and metal-in-glass beads of Egyptian and Levantine m-Na-Ca R glass became very common in Nubian tombs (e.g., Then-Obłuska & Wagner, 2019b), during the Meroitic period, no evidence of local bead-making has yet been found. Whereas many beads found at the Nobadian sites were imported (Egypt and South India/Sri Lanka), there is also some scarce evidence for possible beadmaking (re)using glass imported from Egypt and the Levant. Samples OINE70-78 and 80-81 were found at a Serra East Nobadian household bearing traces of a fireplace and the remains of an oven or a kiln interpreted as a bead workshop. A question then arises whether cobalt blue chunks, OINE76 and 77, could have been used there for bead and pendant manufacture. As recent experiments prove, glass can be processed in rudimentary household fire pits where, with the help of blowpipes, temperatures high enough to produce glass beads are achievable (Hodgkinson & Bertram, 2020).
Chunks and dark blue beads found in this Serra East workshop (OINE74, 76, 77, 81) have comparable Co/Ni ratios of between 13.2 and 15.6 that also resemble other dark blue specimens in the Nobadian collection (OINE02BCo,09,13,15BCo,25,30BCo,31BCo,34BCo,41,55BCo) having a Co/Ni ratio of between 11.25 and 18.8 (Fig. 5C). The chunks are of Egyptian (OINE76) and Levantine (OINE77) origin, also attested for cobalt blue glass beads in Lower Nubia (Fig. 5B). This observation seems to verify the hypothesis that the Serra East chunks were used in the local bead-making process.
Translucent and semi-translucent copper blue beads with yellow spots (OINE66B, 71B) and teardrop pendants (OINE37,44,75,78,80) were found in the workshop and graves. A lack of antimonate characterizes the blue glass, and the biplot of Y/Zr to Ce/Zr ratios indicates its Levantine origin (Fig. 5B). While the Levantine glass beads themselves are uncommon finds in Nubia in the period under discussion (Then-Obłuska & Wagner, 2019b), a glass of that type was undeniably used in the production of diagnostic Nobadian ornaments, most probably locally made from reused Levantine glass. Indeed, teardrops of Levantine glass and those of high-alumina glass of uncertain provenance have already been recorded from Lower Nubia (Then-Obłuska & Wagner, 2019a, b: SJE02 and SJE25 accordingly). It thus appears these pendants may have been produced locally using glass from different sources.
Only two Makuria period specimens were analyzed: an imported glass (v-Na-Ca OINE73) and an imported item (m-Na-Al 2). We cannot provide chemical compositional results supporting the idea of glass bead-making during the Makuria period, but we must mention a workshop at the Early Christian site of Debeira in Lower Nubia that yielded much ash and large pieces of unworked glass remains and a bead of similar glass (Shinnie & Shinnie, 1978, p. 44). This, in turn, implies a need for further typological and archaeometric evidence to test the hypothesis of local glass bead production in Lower Nubia.

Conclusions
The analysis of glass beads from Lower Nubia, a now-submerged region, reveals developments in bead glass chemistry over three millennia in Northeast Africa, encompassing the New Kingdom (v-Na-Ca NK), 25th Dynasty (m-Na-Ca LT), Early Roman/ Meroitic (m-Na-Ca R, m/v-Na-Ca), Late Roman/ Nobadian (m-Na-Ca R, m-Na-Al 1), Makurian (v-Na-Ca OINE73), and possibly Islamic (m-Na-Al 2) periods. This study of glass provenance presents the first-ever dataset attesting to Egyptian glass from the New Kingdom (v-Na-Ca NK) and 25th Dynasty periods (m-Na-Ca LT/Classic) in Lower Nubia. It also presents new evidence for Egyptian and Levantine m-Na-Ca glass in the Early Roman/Meroitic and Late Roman/Nobadian periods. Moreover, the study offers new data for South Indian/Sri Lankan glass bead imports in Late Antique northeast Africa (m-Na-Al 1). Furthermore, it provides the first evidence for the presence of "Mesopotamian" Islamic glass (v-Na-Ca OINE73) and Indian glass (m-Na-Al 2) in Medieval northeast Africa.
Lower Nubia mostly imported glass beads between the fourteenth century BC and fourteenth century AD. However, the presence of cobalt blue chunks and beads, both having similar Co/Ni ratios and found in a Late Antique bead workshop, seems to corroborate the hypothesis of the local bead and pendant manufacture. This assumption seems to be further confirmed by the presence, in the workshop, of a diagnostically Nobadian pendant type, i.e., copper blue teardrops, locally produced using different but mainly Levantine glass sources. Beadmaking in medieval Nubia requires further investigation.