271 Arheološki vestnik 71, 2020, 271–282; DOI: https://doi.org/10.3986/AV.71.09 THE SITE The archaeological site at Korinjski hrib above Veliki Korinj, southeast Slovenia, was discovered as early as 1886 and systematically researched a hundred years later, in 1982 and 1983. The research was conducted by a team from the Institute of Archaeology ZRC SAZU, led by Slavko Ciglenečki (Ciglenečki 1985). The site is located at 728 m asl, in the karst region of Suha Krajina (Fig. 1). The earliest finds date to the late Eneolithic and the Bronze Age, the first habitation remains to the late La Tène period (Dular et al. 1995). Later, several small finds including coins point to the human presence in the 3 rd c. AD, while the main settlement phases date to the time after the 4 th c. AD. Around 400 AD, the site probably began to serve as a military post and was more permanently settled from the late 5 th or first half of the 6 th to the end of the 6 th c. Several characteristics of the site indicate a pronounced military function. Five towers (two of them also used as water cisterns) and a church of a simple ground plan represent the only architectural remains (Fig. 2); as such, Analyses of glass from late antique hilltop site Korinjski hrib above Veliki Korinj (Slovenia) Analize stekla s poznoantične višinske postojanke Korinjski hrib nad Velikim Korinjem Tina MILAVEC, Žiga ŠMIT Izvleček Analizirali smo vzorce steklenih posod in okenskega stekla s poznoantične utrdbe Korinjski hrib pri Velikem Korinju. Steklene najdbe s tega najdišča, datiranega v 6. st., kažejo značilen, vendar omejen spekter posod, predvsem kozarce brez noge, kozarce na nogi in okensko steklo. Rezultati arheometričnih analiz so pokazali nepričakovane sredozemske sestave steklenih mas, ki nakazujejo, da je bilo najdišče lahko vpeto v bizantinsko trgovsko mrežo ali sistem vojaške oskrbe. Ključne besede: Slovenija; Korinjski hrib; pozna antika; steklo; arheometrične analize Abstract Samples of glass vessels and window glass from late antique fortification Korinjski hrib near Veliki Korinj, Slovenia, were analysed. The glass assemblage from this 6 th c. site shows a typical, but limited set of vessels, which mainly com- prises beakers, goblets and window panes. The results of archaeometry analyses showed unexpected Mediterranean glass compositions, indicating that the site may have fitted into the Byzantine trade or army supply network. Keywords: Slovenia; Korinjski hrib; late antiquity; glass; archaeometric analyses 272 Tina MILAVEC, Žiga ŠMIT The analysed sherds from the church (Fig. 2) consist of one neck sherd of a flask, one cut rim sherd, one fire-rounded rim sherd, two base frag- ments of beakers, seven window pane fragments and one counter/gem (Fig. 4: 2–4,8–13,16,18–20). The pieces from T ower 1 comprise two fire-rounded rim sherds and one window pane fragment (Fig. 4: 21–23), those from Tower 2 consist of one rim sherd of a bottle and two body sherds of a vessel (Fig. 4: 24–26), from Tower 3 one window pane fragment (Fig. 4: 17), from Tower 5 one goblet foot (Fig. 4: 7), from Trench 2 two goblet feet and two fire-rounded rim sherds (Fig. 4: 1,5,14,15). The context for one of the sampled goblets is unknown (Fig. 4: 6). ARCHAEOMETRIC ANALYSES The analyses were performed using a combined method of proton induced X-rays and gamma rays (PIXE-PIGE). The measurements were made at the Tandetron accelerator of the Jožef Stefan Institute in Ljubljana, using the in-air measuring station. The proton nominal energy was 3 MeV, which decreased to about 2.77 MeV at the target, after passing an exit window of 2 μm thick tantalum foil and about 1 cm thick air gap between the window and target. The induced X-rays were detected by a Si(Li) detector with an 8 μm beryllium window placed at 135° with respect to the proton beam and at a distance of 6 cm from the target. Exact values of both air-gaps were determined from the measurements on the glass standard NIST 620. The X-ray detector was further equipped with a pin-hole filter made of 0.05 mm aluminium foil with a relative opening of about 10%. The exact values of the aluminium foil thickness and rela- tive opening were determined by the least-squares procedure from a series of targets. The detected X-rays enabled the analysis of elements from and including silicon onwards. For lighter elements, notably Na, Mg, Al, we exploited their character- istic gamma rays that were induced by inelastic nuclear scattering with protons. The lines used for analysis were 440 keV for Na, 585 keV for Mg, and 844 and 1014 keV for Al. Among those, the weakest line was that of Mg, which in practice limited the detection limit of MgO to about 0.3%. We further monitored the competing line at 583 keV from the natural background and subtracted its contribution. The induced gamma rays were measured by a 40% intrinsic germanium detector, the site shows similarities with other military fortifications in the southeastern Alps, along the east Adriatic coast and the Danube (Ciglenečki 2008, 500–501). Quantity of small finds in the settlement is generally low and does not indicate a large civilian population. The only exception is a relatively large number of fibulae and belt set parts, which could be linked to barbarian and Byzantine garrisons. As part of a comprehensive study of the site (Ciglenečki, Modrijan, Milavec 2020), several glass finds were analysed in detail. The main re- search questions of this analysis were how many and which primary glasses were used to produce the vessels and window panes, where they were coming from and what they can tell us about the site and its role in the wider Mediterranean trade network. W e were also interested in the comparison with other already analysed sites in Slovenia and Italy. The paper below presents the results of the archaeometric analyses in association with the archaeological context. GLASS FINDS Excavations at the site have unearthed 34 frag- ments of glass vessels and 56 fragments of window glass. Apart from a few fragments of early Roman vessels, which probably represent residual finds, the glass vessel assemblage is very simple, even for a late antique hilltop site. It consists of a few fragments of cut and mostly fire-rounded rims, a few beaker bases and feet of stemmed goblets (Figs. 3–4). There are no identifiable fragments of hanging glass lamps of either shape (with handles or stemmed). Pieces of window glass were mostly found in the church and outside Towers 1 and 3 (Fig. 2), suggesting that all buildings had glazed windows and that the inhabitants were using a limited amount of stemmed goblets, beakers and bottles or flasks. This assemblage – though modest – is very typical for late antique sites in the Mediterranean and its immediate hinterland (Milavec 2011, 101, 113). For the purposes of archaeometric analyses, we took 26 samples (Figs. 3–4) that were analysed at the Jožef Stefan Institute in Ljubljana. We strove to cover all periods (early Roman, late Roman, late antique), locations (trenches between tow- ers, inside towers and in the church) and types of glassware (beakers, goblets, flask, window glass, counter/gem). 273 Analyses of glass from late antique hilltop site Korinjski hrib above Veliki Korinj (Slovenia) Fig. 1: Late antique settlements in Slovenia. Sl. 1: Poznoantične naselbine v Sloveniji. Fig. 2: The ground plan of Korinjski hrib above Veliki Korinj. Locations of analysed glass finds. Sl. 2: Korinjski hrib nad Velikim Korinjem. Lokacije analiziranih steklenih predmetov. 274 Tina MILAVEC, Žiga ŠMIT Fig. 4: Korinjski hrib. Table of glass compositions (samples 1–26) in mass % (Na2O-Fe2O3) and μg/g (CuO-PbO). Sl. 4: Korinjski hrib, preglednica sestave stekla (vzorci 1–26) v masnih % (Na2O-Fe2O3) in μg/g (CuO-PbO). Fig. 3: Korinjski hrib. Analysed glass finds (samples 1–26; cf. Fig. 4). Scale 1:2. Sl. 3: Korinjski hrib, analizirani stekleni predmeti (vzorci 1–26, prim. sl. 4). M. = 1:2. 275 Analyses of glass from late antique hilltop site Korinjski hrib above Veliki Korinj (Slovenia) Sample Object Comp. Na 2 O MgO Al 2 O 3 SiO 2 SO 3 Cl K 2 O CaO TiO 2 Cr 2 O 3 MnO Fe 2 O 3 CuO ZnO Br Rb 2 O SrO Y 2 O 3 ZrO 2 SnO Sb 2 O 3 PbO 2 flask HIMT 17.2 1.38 2.47 64.9 0.39 0.92 0.62 6.87 0.37 0.008 1.89 2.91 75 31 6 - 443 8 206 - - 72 3 cut rim HIMT 15.8 1.43 3.00 66.0 0.28 0.80 0.90 5.90 0.56 0.010 1.40 3.73 163 56 15 - 572 24 410 - - 116 4 rim HIMT 17.8 1.46 2.51 67.5 0.41 1.04 0.45 5.98 0.30 - 1.41 1.04 69 25 5 - 488 7 150 - - 44 15 rim 3.2 19.4 0.65 2.09 67.5 0.68 0.76 0.61 6.62 0.11 - 0.96 0.60 80 14 8 6 525 - 90 - - 37 1 goblet 3.2 18.9 1.16 1.91 68.5 0.62 0.61 0.44 6.42 0.09 - 0.79 0.52 71 17 9 - 573 7 59 - - 64 13 window glass 3.2 18.9 1.63 2.08 65.7 0.45 0.86 0.55 7.70 0.11 - 1.25 0.68 70 18 7 5 699 - 59 - 154 51 17 window glass 3.2 18.7 1.35 2.18 64.5 0.59 0.71 0.66 9.06 0.14 - 1.17 0.86 57 19 8 - 798 - 89 - - 41 18 window glass 3.2 18.9 1.07 2.02 65.2 0.62 0.69 0.66 8.65 0.12 - 1.15 0.89 51 25 5 9 740 - 80 - - 44 19 window glass 3.2 18.8 1.90 2.10 65.3 0.52 0.63 0.62 8.15 0.12 - 1.09 0.73 49 25 8 4 652 - 92 - - 33 20 window glass 3.2 18.8 1.37 2.06 64.6 0.65 0.66 0.64 8.96 0.14 - 1.29 0.85 64 24 7 6 760 - 94 - - 39 26 vessel 3.2 18.4 0.90 1.94 66.1 0.64 0.79 0.59 8.49 0.12 - 1.23 0.69 56 19 11 8 724 - 72 - - 40 21 rim 2.1 18.4 1.97 2.23 64.8 0.62 0.64 0.70 8.47 0.13 - 1.13 0.80 52 31 7 - 704 - 95 - 201 29 14 rim 2.1 18.5 1.73 2.28 63.8 1.01 0.72 0.74 8.66 0.14 - 1.33 0.94 240 32 6 8 874 - 112 - 233 85 22 rim 2.1 17.8 1.35 2.34 65.6 0.50 0.64 0.84 8.42 0.14 - 1.36 0.91 61 27 8 - 797 - 93 - - 41 8 beaker 2.1 18.0 1.64 2.75 66.2 0.46 0.65 0.84 6.91 0.16 - 1.28 0.98 67 39 9 - 768 8 127 - - 98 9 beaker 2.1 19.2 1.12 2.47 65.4 0.57 0.77 0.58 8.12 0.15 - 0.75 0.75 42 15 5 - 619 - 136 - 283 52 24 bottle 2.1 15.7 1.07 2.39 70.4 0.28 0.90 0.68 7.63 0.06 - 0.51 0.34 38 22 4 7 487 - 35 - - 23 5 goblet 2.1 17.4 1.40 3.00 62.9 0.83 0.69 0.90 10.29 0.19 - 1.02 1.10 120 54 7 - 1047 13 155 - 373 127 6 goblet 2.1 17.4 1.51 2.29 66.2 0.59 0.64 0.79 8.11 0.13 - 1.42 0.79 119 35 9 - 740 11 80 - - 122 7 goblet 2.1 19.6 1.25 2.89 62.0 1.26 0.84 0.89 8.52 0.19 - 1.38 1.12 112 38 10 - 799 9 91 - 93 117 25 vessel 2.1 18.8 1.54 2.38 63.7 0.80 0.79 0.68 8.38 0.17 - 1.66 0.95 111 36 0 6 836 - 67 - 283 16 23 window glass 2.1 18.3 1.42 2.21 64.9 0.53 0.89 0.73 8.35 0.13 - 1.55 0.85 92 18 9 - 749 - 76 - - 83 11 window glass 2.1 18.1 1.40 2.18 65.4 0.50 0.75 0.88 8.39 0.13 - 1.45 0.75 55 17 8 - 710 - 70 - - 37 12 window glass 2.1 17.6 1.58 2.33 65.0 0.46 0.66 0.90 8.78 0.15 - 1.57 0.94 84 25 10 10 894 - 157 - - 66 16 window glass 2.1 17.7 1.44 2.25 65.4 0.47 0.69 0.83 8.54 0.15 - 1.56 0.89 76 23 7 4 869 - 93 - - 44 10 counter/gem / 12.4 0.87 5.38 61.5 1.04 0.75 2.45 8.24 0.40 - 0.45 6.16 1975 130 8 97 464 - 72 232 408 501 276 Tina MILAVEC, Žiga ŠMIT positioned about 10 cm from the target. The front housing of the detector, made of aluminium, was additionally screened against the scattered protons by a plastic cover. For fitting the spectra, we used the AXIL code for X-rays and a program devel- oped in the lab for gamma rays. For evaluating the concentrations, we used a procedure developed in the lab that considers the matrix effects for induced X-rays and gamma rays simultaneously. The concentrations that follow from X-rays were calculated according to the method of independent physical parameters, while those from gamma rays were determined by the surface approximation ac- cording to the values in NIST 620 glass standard, measuring the number of incoming protons by a thin metal mesh intersecting the beam in vacuum. The concentrations were normalized according to the sum of all metal oxides giving unity; however, an additional test was performed according to the argon yield induced by protons in the calibrated air-gap. The sums of metal oxides determined ac- cording to the argon yield were allowed to differ from unity by less than 20%; larger differences signalled that the sample was misaligned or had a rough surface. Among the series of 27 measure- ments, one had the sum of metal oxides of about 1.7; the results of this measurement were removed from the list. Periodic analyses of NIST 620 and 621 standards measured as unknown samples revealed that our concentrations were accurate to within ±5%, while the uncertainties increased to about 10% for concentrations smaller than 0.1%. ANALYSES IN ARCHAEOLOGICAL CONTEXT The analysed glass from Korinjski hrib is roughly attributable to 5 th –6 th c. AD except the cut rim sherd from the church, which is late Roman. Individual finds are typologically and chronologi- cally determined in the monograph (Ciglenečki, Modrijan, Milavec 2020). The Al 2 O 3 -CaO and SiO 2 -Na 2 O diagrams (Figs. 5; 6) show two main groups of glass composition: Foy 3.2 (Fig. 3: 1,15; some not drawn) and Foy 2.1. (Fig. 3: 5–9,14,16,21–24; some not drawn). In addition, three fragments belong to the HIMT group (Fig. 3: 2–4): a cut rim sherd, a flask and a fire-rounded rim, all three discovered in the church. Foy série 3.2 from Korinjski hrib comprises half of the window glass sherds (Fig. 4: 13,18–20) Fig. 5: Korinjski hrib. Distribution of the measured glass samples according to the contents of aluminium and calcium oxides. The marked intervals of the 3.2 and 2.1 groups are 2σ ellipses according to the data of D. Foy (Foy et al. 2003). Sl. 5: Korinjski hrib. Porazdelitev izmerjenih vzorcev stekla glede na vsebnost aluminijevih in kalcijevih oksidov. Označena intervala za skupini 3.2 in 2.1 sta podana kot elipsi z dolžino glavnih polosi 2σ glede na podatke D. Foy (Foy et al. 2003). 277 Analyses of glass from late antique hilltop site Korinjski hrib above Veliki Korinj (Slovenia) Fig. 6: Korinjski hrib. Distribution of the measured glass samples according to the contents of silica and sodium oxide that served as flux. The indicated intervals of the 3.2 and 2.1 groups are 2σ ellipses according to the data of D. Foy (Foy et al. 2003). Sl. 6: Korinjski hrib. Porazdelitev izmerjenih vzorcev stekla glede na vsebnost kremena in natrijevega oksida. Označena intervala za skupini 3.2 in 2.1 sta podana kot elipsi z dolžino glavnih polosi 2σ glede na podatke D. Foy (Foy et al. 2003). Fig. 7: Korinjski hrib. Distribution of the measured glass samples according to the contents of strontium and magnesium oxides. The region of the HIMT glass and the time separation line were deliberately plotted according to the data of A. Cholakova (Cholakova et al. 2015). Sl. 7: Korinjski hrib. Porazdelitev izmerjenih vzorcev stekla glede na vsebnost stroncijevih in magnezijevih oksidov. Območje stekla HIMT in časovna meja sta povzeta po podatkih A. Cholakove (Cholakova et al. 2015) 278 Tina MILAVEC, Žiga ŠMIT from the church, a window glass sherd found in Tower 3 (Fig. 4: 17), a fire-rounded rim and a goblet foot (Fig. 4: 1,15) from Trench 2, as well as a body sherd of a vessel (Fig. 4: 26) from Tower 2. Foy série 2.1 consists of a fire-rounded rim and a goblet foot unearthed in Trench 2 (Fig. 4: 5,14), two fire-rounded rims and a piece of window glass from T ower 1 (Fig. 4: 21–23), fragments of a bottle and an unidentified vessel from Tower 2 (Fig. 4: 24,25), as well as a goblet foot from Tower 5 (Fig. 4: 7). Half of the window glass and two beaker bases from the church also belong to this group (Fig. 4: 8,9,11–12,16). The church further revealed a small round glass object (Fig. 3: 10; 4: 10; diam. 1.5 cm) that does not fit in any of the three groups. Too large for a ring bezel intaglio, it may be a game counter or a decorative gem. Its surface is dark and bubbly, probably damaged in the fire that destroyed the church at Korinjski hrib. The colour of the glass could not be established, probably also due to fire damage. Decorative glass gems can be found in basilicas and profane buildings in the East and West, they are red, blue and green and believed to imitate precious stones. Used for decorating various objects and parts of architecture, they substituted the more expensive stones (Antonaras 2018). Due to the generally modest decoration of the late an- tique churches in the southeastern Alps (Ciglenečki 2003), it is difficult to imagine what the gem from Korinjski hrib could have decorated, perhaps a book cover or a small artefact. Another possibility is that the object represents a gaming token, though these are usually somewhat larger (Cool 2016). The glass from Korinjski hrib shows no correlation between glass composition and vessel form. There is also no perceptible difference between the glass assemblages from the different buildings within the settlement. The great majority of fragments is of a very similar colour, varying between yellow and green; no specific colour could be assigned to individual glasses. It seems the glasses were used completely interchangeably. It can reasonably be doubted whether the secondary producers, blow- ers of glass, noticed the difference between the chunks at all. It is possible that the only difference is chronological. In the 4 th and 5 th c., HIMT is the most strongly represented glass in western Europe. Recent research indicates that it represents the continuation of the 1 st –4 th c. antimony-decolourised glass production, probably located in Egypt, which ceased to be made in the 4 th c. (Freestone et al. 2018). A similar situation, with typologically early fragments of the presumably early HIMT glass, was observed at Tonovcov grad (Milavec, Šmit 2018, 356). The predominance of HIMT in the western Mediterranean and its hinterland in the late Roman period as opposed to the prevailing Levantine I in the 6 th c. can be linked to the general characteris- tics of pottery circulation and foodstuff supply in the region. Evidence shows that the predominant directions of supply changed from north Africa (mainly Tunisia and Mauretania) in the 4 th and first half of 5 th c. to the eastern Mediterranean (Levant and the Aegean) in the second half of the 5 th and 6 th c. (Modrijan 2011, 155–158). The série 3.2 glass was initially dated to the late 5 th and early 6 th c. by Danièle Foy and colleagues, who posited that it originated in the vicinity of the Belus river (Foy et al. 2003). Recently, is has been convincingly argued that its origin is in Egypt (Schibille et al. 2016; Balvanović et al. 2018). It is a heterogeneous group so far recorded in North Africa, southern France, Britain, Italy (Aquileia, Classe, Padua), the Balkans (Jelica and Caričin grad in Serbia, Butrint in Albania, Bulgaria) and datable between the late 4 th and the early 6 th c. (Foy et al. 2003; Gallo et al. 2014; Maltoni et al. 2015; Maltoni et al. 2016; Balvanović et al. 2018; Freestone et al. 2018, 162; Cholakova, Rehren 2018). The Foy 2.1 glass is slightly later; it spans from the second half of the 6 th to the second half of the 7 th c. and also originates from Egypt. It has been recorded in western Europe, North Africa, Cyprus and the Balkans (Jelica, Caričin grad, Butrint, Bulgaria) (Foy et al. 2003; Balvanović et al. 2018; Freestone et al. 2018, 162; Cosyns, Ceglia 2018). There is an interesting blurring of compositional differences between the two glasses, which led Cholakova and Rehren to assume that the primary production of the later 2.1 glass contained recycled 3.2 cullet (Cholakova, Rehren 2018, 64–67). In the case of Korinjski hrib, this assumption is confirmed by the lack of a marked compositional differen- tiation between the plotted samples (Figs. 5; 6). The chronological distribution according to the SrO-MnO diagram (Fig. 7) indicates that most of the 3.2 and 2.1 glass sherds correspond with the glass from the Balkans datable to the 6 th century (Cholakova et al. 2015). Only five sherds from Korinjski hrib match the 5 th century glass: two 3.2 sherds from Trench 2 (Fig. 4: 1,15), two 2.1 sherds from Tower 2 and the church, respectively (Fig. 4: 9,24), and the game counter/decorative gem (Fig. 4: 10). 279 Analyses of glass from late antique hilltop site Korinjski hrib above Veliki Korinj (Slovenia) The samples that do show a marked difference between types of glass belong to window panes. Considering the chronological difference for the two compositions, we may posit that the church windows at Korinjski hrib were glazed in the first building phase, at the end of the 5 th c. Later, in the presumed Justinianic renovation phase, some window panes may have been replaced with those of the later 2.1 glass. On Cyprus, the use of the 2.1 glass in basilicas was also attributed to the Justinianic phase (Cosyns, Ceglia 2018, 86). DISCUSSION The glassware from Korinjski hrib is of three different glasses, all originating in Egypt. The main difference between them at the site appears to be in chronology, starting with HIMT and continuing with série 3.2 in the 5 th and série 2.1 in the 6 th c. These chronological differences are not reflected in the typology of the Korinjski hrib glassware, except perhaps the typologically earliest HIMT fragments. The greatest surprise of the results is that the glasses represented at Korinjski hrib differ from those present at Tonovcov grad near Kobarid, also a late antique hilltop site, but in western Slovenia (Fig. 1) (Šmit et al. 2013; Milavec, Šmit 2018). The glass from Tonovcev grad shows similari- ties with Italian sites in that it is predominantly made of Levantine I glass. Despite numerous similarities between the two Slovenian sites, such as contemporary existence and a strategic location on a hilltop, Korinjski hrib shows a completely different repertoire of glass composition. It also differs greatly from the large distribution centres in Aquileia and Ravenna, with mainly HIMT and Levantine I compositions (Gallo et al. 2014; Maltoni et al. 2015; Maltoni et al. 2016; Maltoni et al. 2018; Silvestri et al. 2018), that are identified as the most probable intermediate points of raw glass trade from Egypt and Syria-Palestine to the Adriatic hinterland. Fig. 8: Sites with the Foy 3.2 and 2.1 compositions in the Balkans (adapted from Balvanović 2018, Fig. 9). Sl. 8: Najdišča s steklom sestave Foy 3.2 in 2.1 na Balkanu (prirejeno po Balvanović 2018, sl. 9). 280 Tina MILAVEC, Žiga ŠMIT In glass composition, the glass from Korinjski hrib is rather comparable with that from Serbia and Bulgaria with predominant 3.2 and 2.1 glass compositions (Fig. 8; Balvanović 2018; Cholakova, Rehren 2018). Alongside other factors explained below, this seems a compelling argument in favour of the overall military character of the site. A military character of Korinjski hrib can be inferred from its strategic location, specific ar- chitecture (five towers and only a small church) and simple ground plan of the church. As such, it closely resembles the 6 th century. fortresses along the Danube limes and the eastern Adri- atic (Ciglenečki 2008, 500–501). In addition, the small finds from these sites all share a relatively large percentage of items ascribable to barbarian populations, as well as weapons, belts and fibulae with Byzantine analogies (Ciglenečki, Modrijan, Milavec 2020). As mentioned above, our results of the glass composition analyses underscore the architectural and artefactual parallels with the Danube region; if Korinjski hrib hosted garrisons of the Byzan- tine army, perhaps the supply was organized in a similar way to that for the Danubian forts. The results also show a lack of similarities with the sites in western Slovenia and its marked North Italian links. The architectural and artefactual evidence suggests that the fort on Korinjski hrib was con- structed in the (late) 5 th c. and renovated in the time of Justinian I in the mid-6 th c.; the results of glass composition analyses support this. At other sites, such dating and phasing is corroborated by numismatic evidence (Ciglenečki 2008, 519). Korinjski hrib is one of a group of sites in southeastern Slovenia that line the route leading eastwards (Zidani gaber above Mihovo, Sv. Lambert near Pristava above Stična, Limberk near Mala Račna; see Fig. 1) and show a similarly marked military character (Ciglenečki 2008, 500–501). The glass finds from these sites have not been analysed yet, but it would certainly be very interesting to see the results and consider their implications. ANTONARAS, A. Ch. 2018, A special group of early Christian glass ‘gems’ from Greece. –In: Rosenow et al. (eds.) 2018, 1–21. BALV ANOVIĆ et al. 2018 = BALV ANOVIĆ, R., M. MARIĆ STOJANOVIĆ, Ž. ŠMIT 2018, Exploring the unknown Balkans: Early Byzantine glass from Jelica Mt. in Serbia and its contemporary neighbors. – Journal of Radio- analytical and Nuclear Chemistry 317/2, 1175–1189. CHOLAKOVA, A., T. REHREN 2018, A Late Antique manganese decolourised glass composition: Interpre- ting patterns and mechanisms of distribution. – In: D. Rosenow, M. Phelps, A. Meek, I. Freestone (eds.), Things that Travelled. Mediterranean Glass in the First Millennium CE, 46–71, London. CHOLAKOVA, A., T. REHREN, I. C. FREESTONE 2016, Compositional identification of 6 th c. glass from the Lower Danube. – Journal of Archaeological Science. Reports 7, 625–632. CIGLENEČKI, S. 1985, Potek alternativne ceste Siscija– Akvileja na prostoru zahodne Dolenjske in Notranjske v času 4. do 6. stoletja (Der Verlauf der Alternativ- strasse Siscia-Aquileia im Raum von Westdolenjsko und Notranjsko in der Zeitspanne vom 4. bis zum 6. Jh.). – Arheološki vestnik 36, 255–284. CIGLENEČKI, S. 2003, Frühchristliche Kirchen in Slowenien und die Elemente ihrer Innenausstattung (Ranokršćanske crkve u Sloveniji i elementi njihova unutarnjeg uređe- nja). – Hortus Artium Medievalium 9, 11–20. CIGLENEČKI, S. 2008, Castra und Höhensiedlungen vom 3. bis 6. Jahrhundert in Slowenien. – In: H. Steuer, V. Bierbrauer, M. Hoeper (eds.), Höhensiedlungen zwischen Antike und Mittelalter von den Ardennen bis zur Adria. Ergänzungsbände zum Reallexikon der Germanischen Altertumskunde 58, 481–532, Berlin, New York. CIGLENEČKI, S., Z. MODRIJAN, T. MILAVEC 2020, Korinjski hrib in poznoantične vojaške utrdbe v Iliriku / Korinjski hrib and late antique military forts in Illyricum. – Opera Instituti Archaeologici Sloveniae 39. COSYNS, P ., A. CEGLIA 2018, Glass production and con- sumption in Cyprus in Late Antiquity (fourth-seventh century CE). – In: Rosenow et al. (eds.) 2018, 72–91. COOL, H. E. M. 2016, Recreation or decoration: what were the glass counters from Pompeii used for? – Papers of the British School at Rome 84, 157–177. DULAR et al. 1995 = DULAR, J., B. KRIŽ, D. SVOLJŠAK, S. TECCO HV ALA 1995, Prazgodovinska višinska nase- lja v Suhi krajini / Vorgeschichtliche Höhensiedlungen in der Suha krajina. – Arheološki vestnik 46, 89–168. FOY, D. 1995, La verre de la fin du IVe au VIIIe siècle en France Méditerranéenne. Premier essai de typo-chrono- logie. – In: D. Foy (ed.), Le verre de l’ Antiquité tardive et du Haut Moyen Âge. Typologie–Chronologie–Diffusion, 187–380, Guiry en Vexin. FOY et al. 2003 = FOY, D., M. PICON, M. VICHY, V. THIRION-MERLE 2003, Caractérisation des verres de la fin de l‘Antiquité en Méditerranée occidentale: l‘emergence de nouveaux courants commerciaux. – In: D. Foy (ed.), Échanges et commerce du verre dans le monde antique, Monographies Instrumentum 24, 41–86. FREESTONE et al. 2018 = FREESTONE, I. C., P . DEGRYSE, J. LANKTON, B. GRATUZE, J. SCHNEIDER 2018, HIMT, glass composition and commodity branding 281 Analize stekla s poznoantične višinske postojanke Korinjski hrib nad Velikim Korinjem in the primary glass industry. – In: D. Rosenow, M. Phelps, A. Meek, I. Freestone (eds.), Things that Tra- velled. Mediterranean Glass in the First Millennium CE, 159–190, London. GALLO et al. 2014 = GALLO, F., A. MARCANTE, A. SILVESTRI, G. MOLIN 2014, The glass of the ‘Casa delle bestie ferite’: a first systematic archaeometric study on Late Roman vessels from Aquileia. – Journal of Archaeological Science 41, 7–20. MALTONI et al. 2015 = MALTONI, S., T. CHINNI, M. V ANDINI, E. CIRELLI, A. SILVESTRI, G. MOLIN 2015, Archaeological and archaeometric study of the glass finds from the ancient harbour of Classe (Ravenna- -Italy): new evidence. – Heritage Science 3:13. [https:// heritagesciencejournal.springeropen.com/articles/10.1186/ s40494-015-0034-5]. MALTONI et al. 2016 = MALTONI, S., A. SILVESTRI, A. MARCANTE, G. MOLIN 2016, The transition from Roman to Late Antique glass: new insights from the Domus of Tito Macro in Aquileia (Italy). – Journal of Archaeological Science 73, 1–16. MALTONI et al. 2018 = MALTONI, S., F. GALLO, A. SIL- VESTRI, M. VANDINI, T. CHINNI, A. MARCANTE, G. MOLIN, E. CIRELLI 2018, Consumption, working and trade of Late Antique glass from north Adriatic Italy. An archaeometric perspective. – In: Rosenow et al. (eds.) 2018, 191–214. MILAVEC, T. 2011, Steklene najdbe / Glass finds. – In: Z. Modrijan, T. Milavec, Poznoantična utrjena naselbina Tonovcov grad pri Kobaridu. Najdbe / Late Antique fortified settlement Tonovcov grad near Kobarid. Finds, Opera Instituti Archaeologici Sloveniae 24, 83–119. MILA VEC, T ., Ž. ŠMIT 2018, Analyses of late antique glass from Tonovcov grad near Kobarid in archaeological context / Analize poznoantičnega stekla s Tonovcovega gradu pri Kobaridu v arheološkem kontekstu. – Arhe- ološki vestnik 69, 351–368. MODRIJAN, Z. 2011, Keramika / Pottery. – In: Z. Modrijan, T . Milavec, Poznoantična utrjena naselbina T onovcov grad pri Kobaridu. Najdbe / Late Antique fortified settlement Tonovcov grad near Kobarid. Finds, Opera Instituti Archaeologici Sloveniae 24, 121–219. ROSENOW, D., M. PHELPS, A. MEEK, I. FREESTONE (eds.) 2018, Things that Travelled. Mediterranean Glass in the First Millennium CE – London. SCHIBILLE et al. 2016 = SCHIBILLE, N., A. STERRETT- KRAUSE, I. C. FREESTONE 2016, Glass groups, glass supply and recycling in late Roman Carthage. – Ar- chaeological and Anthropological Sciences 8, 1–19. [DOI: 10.1007/s12520-016-0316-1]. SILVESTRI et al. 2018 = SILVESTRI, A., F. GALLO, S. MALTONI, P . DEGRYSE, M. GANIO, A. LONGINELLI, G. MOLIN 2018, Things that travelled: A review of the Roman glass from northern Adriatic Italy. – V / In: Rosenow et al. (eds.) 2018, 346–367. ŠMIT et al. 2013 = ŠMIT, Ž., T. MILAVEC, H. FAJFAR, Th. REHREN, J. W. LANKTON, B. GRATUZE 2013, Analysis of glass from the post-Roman settlement To- novcov grad (Slovenia) by PIXE-PIGE and LA-ICPMS. – Nuclear Instruments and Methods in Physics Research B 311, 53–59. Arheološko najdišče Korinjski hrib nad V elikim Korinjem v Suhi krajini (sl. 1) je bilo odkrito že leta 1886, ekipa z Inštituta za arheologijo ZRC SAZU ga je sistematsko raziskala med letoma1982 in 1983 pod vodstvom Slavka Ciglenečkega. Najdišče je bilo verjetno v uporabi kot vojaška postojanka okrog leta 400 in bolj trajno naseljeno od poznega 5. st. ali prve polovice 6. st. do konca 6. stoletja. Ima kar nekaj značilnosti, ki nakazujejo poudarjeno vojaški značaj. Pet stolpov (dva od teh sta bila uporabljena tudi kot vodna zbiralnika) in cerkev preprostega tlorisa so edine arhitekturne ostaline (sl. 2) in kažejo podobnosti z vojaškimi utrdbami v jugovzhodnih Alpah, na vzhodni jadranski obali in v Podonavju. Glavno raziskovalno vprašanje tega prispevka je, koliko in katere primarne sestave steklenih mas so uporabili za izdelavo steklenega posodja in okenskih stekel z najdišča, od kod so prišle in kaj nam lahko povejo o najdišču in nje- govi vpetosti v širše sredozemske trgovske tokove. Med izkopavanji je bilo odkritih 34 odlomkov steklenih posod in 56 odlomkov okenskega stekla. Celo za poznoantično višinsko najdišče je sestava steklenih najdb zelo preprosta. Nekaj odrezanih in predvsem zataljena ustja dopolnjujejo dna in visoke noge kozarcev (sl. 3). Prepoznavni deli visečih svetilk, ročaji ali votli tulci, niso bili naj- deni. Okensko steklo je bilo odkrito predvsem v cerkvi in zunaj dveh stolpov (stolpa 1 in 3; sl. 2). Predvidevamo lahko, da so vse stavbe imele oken- Analize stekla s poznoantične višinske postojanke Korinjski hrib nad Velikim Korinjem Povzetek 282 Tina MILAVEC, Žiga ŠMIT ska stekla in da so prebivalci uporabljali omejeno količino kozarcev (na nogi in brez nje) ter večjih in manjših steklenic. Šestindvajset vzorcev (sl. 3–4) je bilo odvzetih in analiziranih na Institutu “Jožef Stefan” v Ljubljani. Glede na diagrama CaO-Al 2 O 2 in SiO 2 -Na 2 O (sl. 5; 6) so analize pokazale dve glavni vrsti steklenih mas, Foy 3.2 (sl. 3: 1,15; nekateri kosi niso risani) in Foy 2.1 (sl. 3: 5–9,14,16,21–24; nekateri kosi niso risani). Trije odlomki pripadajo vrsti HIMT (sl. 3: 2–4). Iz mase HIMT so bili izdelani odrezano ustje, odlomek stekleničke in zataljeno ustje, vsi trije najdeni v cerkvi (sl. 3: 2–4). Iz mase Foy série 3.2 je polovica odlomkov okenskega stekla iz cerkve (sl. 4: 13,18–20), odlomek okenskega stekla iz stolpa 3 (sl. 4: 17), zataljeno ustje in noga kozarca (sl. 4: 1,15) iz sonde 2 in ostenje posode iz stolpa 2 (sl. 4: 26). Iz mase Foy série 2.1 so zataljeno ustje in noga kozarca iz sonde 2 (sl. 4: 5,14), dve zatalje- ni ustji in kos okenskega stekla iz stolpa 1 (sl. 4: 21–23), odlomka steklenice in ostenje posode iz stolpa 2 (sl. 4: 24,25) ter noga kozarca iz stolpa 5 (sl. 4: 7). Druga polovica odlomkov okenskega stekla iz cerkve in dva odlomka dna kozarcev prav tako spadajo v to skupino (sl. 4: 8,9,11,12,16). V cerkvi je bil odkrit tudi majhen okrogel predmet (sl. 3: 10; 4: 10; premer 1,5 cm), ki ne spada v nobeno od omenjenih skupin. Prevelik je za vložek za prstan, morda gre za igralni žeton ali okrasni kamen. Med tipom steklene mase in oblikami posod ni bilo mogoče opaziti povezave. Prav tako ni opazne razlike med vrstami steklenih posod iz različnih stavb na naselbini. V ečina odlomkov je zelo podobne barve, ki variira med rumeno in zeleno. Nobene barve ni mogoče pripisati posamezni stekleni masi. Vse tri vrste steklene mase s Korinjskega hriba izvirajo iz Egipta, glavna razlika med njimi na najdišču je očitno v kronologiji. Prva je HIMT, nato série 3.2 (v 5. st.) in série 2.1 (v 6. st.). Te kronološke razlike se ne zrcalijo v tipologiji posod, razen morda v tudi tipološko najzgodnejših kosih iz mase HIMT. Največje presenečenje opisane analize je bila raz- lika med rezultati s tega najdišča in s T onovcovega gradu nad Kobaridom, poznoantičnim višinskim najdiščem v zahodni Sloveniji. Rezultati so poka- zali veliko podobnost z italijanskimi najdišči in večinsko zastopano maso Levantine I. Čeprav sta si najdišči podobni v tem, da sta sočasni višinski točki s strateško lego, ima Korinjski hrib povsem drugačno sestavo steklenih mas. Razlikuje se tudi od velikih distribucijskih centrov v Akvileji in Raveni, ki sta veljali za najverjetnejši vmesni postojanki v prodaji steklene mase med Egiptom, Sirijo in Palestino ter sredozemskim zaledjem. T am sta zastopani večinoma masi HIMT in Levantine I. Korinjskemu hribu podobne sestave steklenih mas najdemo v Podonavju, Srbiji in Bolgariji (sl. 8). T e povezave potrjujejo tudi druge, arhitekturne in artefaktne analogije s Podonavjem. Če je bila na najdišču bizantinska vojska, je bil morda hrib oskrbovan podobno kot podonavske utrdbe. To bi pojasnilo razlike z zahodno Slovenijo in njenimi severnoitalijanskimi povezavami. Analize stekla morda lahko pritrdijo tudi predvideni kronologiji gradnje trdnjave na Korinjskem hribu v poznem 5. st. in prenove v času Justinijana v sredini 6. stoletja, ki je na drugih najdiščih potrjena z novci. Tina Milavec Oddelek za arheologijo Filozofska fakulteta Univerza v Ljubljani Aškerčeva 2 SI-1000 Ljubljana tina.milavec@ff.uni-lj.si Žiga Šmit Fakulteta za matematiko in fiziko Univerza v Ljubljani Jadranska 19 SI-1000 Ljubljana and/in Institut Jožef Stefan Jamova 39 SI-1001 Ljubljana ziga.smit@fmf.uni-lj.si https://orcid.org/0000-0001-8178-7930