Date post: | 13-Nov-2023 |
Category: |
Documents |
Upload: | independent |
View: | 1 times |
Download: | 0 times |
ORIGINAL PAPER
Glass ingots, raw glass chunks, glass wastes and vessels from fifthcentury AD Palatine Hill (Rome, Italy)
Elisabetta Gliozzo1 & Barbara Lepri2 & Lucia Saguì3 & Isabella Memmi1
Received: 4 May 2015 /Accepted: 17 September 2015# Springer-Verlag Berlin Heidelberg 2015
Abstract The research focused on a collection of 20 glassfragments, including raw glass chunks and ingots, wastesand vessels found at the Palatine Hill in Rome and dated tothe first half of the fifth century AD. The analyses performedby scanning electron microscopy (SEM)-energy-dispersivespectrometry (EDS), electron microprobe (EMPA), inductive-ly coupled plasma mass spectrometry (ICP-MS), laserablation-ICP-MS (LA-ICP-MS) and X-ray diffraction(XRD) showed that all samples are constituted by natron-based soda–lime–silica glass. De/colouring agents were leadstannates (yellow brownish), copper and lead antimonates(green), different Fe2+/Fe3+ ratios (green, light greenand yellow-green), Fe2+ (prevailing over Fe3+; aqua blue),cobalt (blue), metallic copper (reddish) and manganese(colourless). As for provenance, two samples were ofLevantine provenance, nine samples were likely ofEgyptian origin (HIMT glass) and, similarly, sevensamples (or maybe nine, adding ingots nos. 1–2) werelikely of North African provenance (HIMT/RNCBGY 1glass).
Keywords Glass . Late Antique . Palatine hill . Productionindicator . EMPA . ICP-MS . LA-ICP-MS
The archaeological background
The object of this research is a glass collection found in twobasement rooms of the Domus Aurea complex (Fig. 1), in thenorth-east Palatine slope in Rome (Italy).
The Domus Aurea project was planned by the EmperorNero just after the fire of 64 AD and led to a radical changeof much of the city of Rome. The glass fragments were foundin the Late Antique stratifications, corresponding to the aban-donment of this portion of the complex.
Brick masonry was used for the construction of these rect-angular basement rooms, arranged next to one another. Theopus spicatum was used instead for paving, judging at leastfrom the preserved portions. The barrel or cross vaultssustained a terrace overlooking the small artificial lake ofNero’s residence, which was replaced by the Flavian amphi-theater (better known as the Colosseum) at the end of the firstcentury AD.
The rooms were used for a long time, although sometimesmodified. One of them was also used to hide the signs attrib-uted to EmperorMaxentius (Panella 2011). During the secondhalf of the fifth century AD, they were definitely abandonedand filled with thick layers of earth and rubble, very rich inpottery and glass. In a later period, the vaults collapsed cov-ering this filling (Panella et al. 2014, in particular p. 190 ff.).
The story of these structures does not differ from that ofmany other public buildings in the city of Rome, which weresimilarly abandoned during the fifth century, due to the overalldegradation of the urban monuments. To give just a few ex-amples regarding the most famous buildings of the Palatinearea, similar phenomena have been observed in the ScholaPraeconum, the temple of the Magna Mater, the area ofVigna Barberini and the Domus Tiberiana. In these sites, aswell as in other areas of the city, a number of Late Antiquedeposits—miraculously survived the excavations of past
* Elisabetta [email protected]
1 Department of Earth, Environment and Physical Sciences, Universityof Siena, Siena, Italy
2 Archäologisches Institut, University of Cologne, Cologne, Germany3 Department of Antiquity Sciences, Sapienza University of Rome,
Rome, Italy
Archaeol Anthropol SciDOI 10.1007/s12520-015-0292-x
centuries—provided a decisive contribution to the knowledgeof the coeval urban economy.
As for the Late Antique contexts considered in this re-search, 46,000 pottery fragments and 2200 glass fragmentswere found in these two rooms of the Palatine substructures.
The studies of the archaeological findings, especially thatof pottery (Panella et al. 2010, in particular pp. 61–62) suggestthat the filling did not correspond to a dump which had grad-ually formed over time; conversely, it should represent theresult of a unitary event. The radical restructuring and cleaningof the neighbouring areas (actively frequented during thefourth and the first half of the fifth century AD) should haveled to the formation of this deposit in the decades immediatelyafter the mid-fifth century.
These last data are of particular interest in relation to theglass production indicators, as well as to finished objects,since it may suggest the presence of one or more glass work-shops, located not too far away from the place in which theindicators were found.
The glass found in the Late Antique deposit provides arange of useful information for the cultural and chronologicalframing of this archaeological context. The notable presenceof production indicators (glass ingots, raw glass chunks,moils, wastes and items with manufacturing defects) testifiesto the existence of glass workshops; moreover, the large con-centration of glass engraved with figurative scenes (one ofwhich was unfinished or broken during processing)strengthens the hypothesis that Rome was the main producingcentre of these vessels (Saguì 2009).
Glass is generally deemed less useful than pottery for anaccurate dating, but, in this case, it is possible to get an impor-tant result. In fact, among more than 2000 fragments found inthe examined layers, the late fifth century goblet-type Isings111 are completely lacking.
It is an argumentum ex silentio which strengthens the hy-pothesis that the dump was formed before the end of thecentury.
Materials and objectives
The research focused on a collection of 20 samplesincluding both production indicators and finished objects(Figs. 2 and 3).
All types of production indicators have been sampled andanalysed: three glass ingots (samples 1–2 and 4), six raw glasschunks (5–10), a lid-shaped moil (3) (Amrein 2001, p. 23,Fig. 12.2), five wastes (11–14 and 20) and a vessel withmanufacturing defects (15).
The first two ingots (nos. 1–2) are thick plates (2.5 cmmaximum) of a yellowish translucent glass. Based on thestraight and rounded form of the side preserved, thegreater ingot (no. 1) must have had a squared shape.Traces of the tool used for flattening the viscous glassmass are clearly visible on the surface. The shape is verysimilar to that of semifinished products used by mosaicists forthe making of tesserae, which were generally characterised bya rounded profile (Foy 2008).
The third ingot is of a green translucent glass and has ahemispherical shape (maximum thickness of 3 cm, diameterof 8.5 cm) (Uboldi 1999, p. 306, tav. CLIX, 5–6; Verità 1999).
Still within production indicators, all other items are trans-parent, except for waste nos. 11, 12 and 14, for which it is hardto say.
Together with production indicators, finished objects havebeen analysed as well. In this case, the main sampling criterionwas the relative abundance of a certain type both in the inves-tigated context and in the coeval Roman glass production.
Fig. 1 The archaeological area atthe north-east slope of thePalatine Hill at Rome (Italy)
Archaeol Anthropol Sci
The typical glass forms of this period are beakers with fire-rounded rims, bowls, plates and lamps with rims folded out-ward and downward, bowls decorated by a thick coil on therim of the same colour of the vessel, funnel-mouth flasks andconical hanging lamps (samples 17–18) (Saguì 2009).
In addition to the more common flat bases, tubular andbroad foot-ring (sample 16) and multiple-coil bases were alsofound (sample 19); this last type probably belonging to thebowls with the thick coil on the rim. As for the colours of thevessels, all the green tones are attested and in particular thedarker ones.
Overall, the research was aimed at characterising this glasscollection in order to define technological and provenanceissues, hence, the nature of the raw materials and their origin.
Experimental
Commonly used for imaging samples, scanning electron mi-croscopy (SEM) was used for textural observations, mainlyperformed in backscattered electrons. The instrument was aPhilips XL 30 SEM equipped with an EDAX-DX4 Energy-Dispersive Spectrometer (EDS) operating at 20 kV. For chem-ical microanalyses, a variety of natural phases (albite,almandine, biotite, chlorite, Cr-diopside, diopside, kadeite,
kaersutite, olivine, plagioclase, pyrope, rhodonite, sanidine)and synthetic materials and glasses (NIST 1831) was used asprimary and quality control standards. For SEM observations,a small fragment of glass was cut, mounted in resin, polishedand carbon coated.
For the quantitative determination of the major and minorglass components in all samples, an electron microprobe(EMPA) JEOL Superprobe JXA-8600, equipped with fourwavelength-dispersive spectrometers (WDS), was used underthe following operating conditions: 15 kV, beam current at15 mA, beam diameter 8–10 μm. PAP software was usedfor correction. Also for EMPA, a variety of natural phases(bustamite, stibnite, cuprite, galena, metallic cobalt and Sn)and synthetic glass materials (NIST 1831) was used as prima-ry and quality control standards. Precision was within 1 % formajor elements, about 3–4 % for minor elements and about8 % for trace elements. Accuracy was ≤1 % for SiO2, Na2O,CaO, MgO, K2O and MnO; equal to 2.5 % for Al2O3; 3.4 %for Fe2O3 and below 15 % for Cu, Pb and Sb. Generally, fiveto seven point microanalyses were performed per sample, andmean and standard deviations were calculated.
Banded and heterogeneous samples (nos. 1–2 and 12) werefurther submitted to bulk chemical analyses, using aPerkinElmer Elan 6100 inductively coupled plasma massspectrometer (ICP-MS). A homogenous sample (no. 4) was
Fig. 2 Images of the entire glasscollection investigated in thisresearch
Archaeol Anthropol Sci
further submitted to ICP-MS investigations, in order to com-pare the analytical results obtained by this technique withthose obtained by LA-ICP-MS. A 1 g sample was mechani-cally cleaned, ground and mixed with a flux of lithiummetaborate and lithium tetraborate and fused in an inductionfurnace. The molten melt was immediately poured into a so-lution of 5 % nitric acid containing an internal standard andmixed continuously until completely dissolved. Three blanksand five replicates of a single sample (three before the samplegroup and two after) were analysed per group of samples. Thedetection limits were 0.01 % for all the major and minor ele-ments. The accuracy for standards run with samples was betterthan 10 % relative to the certified values for the trace elementsand better than 1 % for all major elements except for Na2Owhich were better than 7 %. The duplicate precisionwas better than 1 and 5 % for major and trace elements,respectively. The blanks run with the samples were belowthe detection limit for all the elements reported. The results areprovided in Table 1.
The trace element content was determined for all sam-ples by laser ablation-inductively coupled plasma-massspectrometry (LA-ICP-MS at CNR, Pavia, Italy). The
instrument combines an ablation microbeam based on aNd:YAG laser source (Brilliant, Quantel) operating at266 nm (for details, see Tiepolo et al. 2003) and a quad-rupole ICP-MS (Drc-e, PerkinElmer). Thirty four massesfrom 7Li to 238U were acquired; the laser was operated at10 Hz of repetition rate, the power on the sample was1.5 mW, and spot size was set at 40 mm. Accuracy wasassessed on the USGS BCR-2 reference glass (analysed asan unknown in each analytical run) and was better than20 % at the sub-ppm level. Data reduction was carried outwith the software package GLITTER (van Achterberghet al. 2001) and using NIST SRM 610 and 29Si as exter-nal and internal standards, respectively. The obtainedvalues were normalised against the average compositionof the weathered upper continental crust (Kamber et al.2005). The results are provided in Table 2.
X-ray diffraction data on crystalline materials werecollected using the flat plate powder mounting inBragg-Brentano configuration. A Philips X’Pert PROPW 3040/60 was employed for data collection in the10–80 2θ range, equipped with a X’Celerator PW 3015 de-tector and Cu K radiation.
Fig. 3 Drawings of notable glassobjects investigated in thisresearch
Archaeol Anthropol Sci
Table1
The
results
obtained
byICP-MS(I)andby
EMPA
(E)on
theinvestigated
samples
#SiO2
TiO
2Al 2O3
CaO
MgO
Na 2O
K2O
Fe 2O3
MnO
Co
SbCu
Pb
Sn
TOT
as
as
as
as
as
as
as
as
as
as
as
as
as
as
1Y-Br
I55.54
–0.522
–3.12
–7.22
–0.98
–14.43
–0.73
–2.32
–2.62
––
–0.0077
–0.0113
–9.5621
–1.4996
–98.6
Ea
50.70
30.411
0.09
2.41
15.50
11.08
0.5
12.20
20.32
0.3
1.88
0.8
2.10
0.5
bdl
–bdl
–0.06
0.05
16.44
35.47
498.5
Eb
60.29
10.653
0.1
3.49
0.7
8.74
20.86
0.3
16.60
10.95
0.1
2.46
0.2
2.89
0.8
bdl
–bdl
–bdl
–2.11
0.6
0.08
–99.1
2Y-Br
I54.59
–0.659
–3.95
–7.38
–0.97
–13.67
–1.16
–2.65
–2.92
––
–0.0112
–0.0355
–9.1847
–1.4557
98.6
Ea
48.65
20.427
0.3
2.33
0.5
7.31
20.87
0.2
11.63
20.29
0.4
2.86
12.99
1.4
bdl
–bdl
–0.02
0.05
16.13
34.44
297.9
Eb
61.66
10.558
0.2
3.58
0.7
9.10
11.06
0.7
15.36
0.8
1.11
0.3
2.48
0.3
2.48
1.5
bdl
–bdl
–bdl
–1.97
10.32
0.1
98.9
3G(l.)
E67.34
0.6
0.401
0.07
2.91
0.1
6.31
0.2
1.37
0.2
17.89
20.32
0.06
1.76
0.4
1.14
0.7
bdl
–bdl
–bdl
–bdl
–bdl
–99.4
4G
I69.32
–0.108
–2.36
–6.54
–0.63
–17.58
–0.67
–0.88
–0.46
––
–0.658
–0.12
–0.214
–bdl
–99.5
E69.69
0.6
0.16
0.03
2.47
0.1
6.92
0.4
0.71
0.08
17.35
0.2
0.50
0.1
1.01
0.1
0.27
0.2
bdl
–0.15
0.3
0.13
0.01
0.29
0.1
bdl
–99.7
5YG
E65.37
0.4
0.509
0.09
3.00
0.2
6.50
0.3
1.01
0.3
18.51
10.34
0.2
2.24
0.4
2.21
1bdl
–bdl
–bdl
–bdl
–bdl
–99.7
6AB
E69.01
0.6
0.061
0.01
2.98
0.1
9.15
0.2
0.75
0.01
16.64
0.4
0.67
0.07
0.54
0.1
0.03
0.01
bdl
–bdl
–bdl
–bdl
–bdl
–99.8
7YG
E65.33
0.2
0.494
0.05
2.91
0.0
6.69
0.1
1.23
0.03
18.34
0.3
0.46
0.05
2.09
0.0
2.12
0.2
bdl
–bdl
–0.05
0.09
0.02
0.09
bdl
–99.7
8B
E66.94
0.5
0.22
0.06
2.56
0.3
7.87
0.1
0.79
0.3
18.63
20.27
0.2
1.97
10.20
0.07
0.08
0.01
bdl
–0.16
0.06
0.14
0.02
bdl
–99.8
9C
E66.48
0.4
0.053
0.07
3.28
0.3
9.24
0.1
0.43
0.03
17.83
0.3
0.82
0.3
0.38
0.1
0.89
0.06
bdl
–bdl
–bdl
–bdl
–bdl
–99.5
10YG
E64.93
0.4
0.488
0.1
2.82
0.4
6.69
0.0
0.97
0.3
18.88
20.27
0.2
1.94
0.6
2.32
0.2
bdl
–bdl
–0.03
0.01
0.03
0.01
bdl
–99.4
11G(l.)
E68.12
0.4
0.184
0.06
2.81
0.1
7.21
0.2
0.75
0.01
17.25
0.5
0.62
0.3
1.13
0.2
0.82
0.1
bdl
–0.16
0.01
0.25
0.06
0.26
0.01
bdl
–99.6
12Bk
I50.94
–0.418
–9.98
–16.67
–2.30
–1.42
–1.96
–8.80
–0.25
–0.036
0.01
0.0087
–0.0657
1.3495
–0.0532
–94.2
E51.26
30.480
0.03
9.29
0.4
16.73
22.52
0.1
1.51
0.1
1.80
0.09
8.42
0.7
0.19
0.1
bdl
–bdl
–bdl
–1.54
0.8
bdl
–93.7
13G
E66.58
20.15
0.2
2.74
0.3
7.09
0.6
0.79
0.2
18.01
0.8
0.42
0.02
1.49
0.6
1.41
1bdl
–0.18
0.08
0.11
0.08
0.52
0.2
bdl
–99.5
14R
E69.40
0.3
0.111
0.02
2.46
0.1
6.74
0.4
0.64
–17.98
0.7
0.52
0.1
0.67
0.2
0.43
0.1
bdl
–0.19
0.03
0.22
0.02
0.2
0.07
bdl
–99.6
15G
E66.82
0.3
0.605
0.09
3.02
0.3
5.88
0.1
1.09
0.3
17.83
20.18
0.2
2.73
0.4
1.87
1.0
bdl
–bdl
–bdl
–bdl
–bdl
–100
16G
E65.04
0.7
0.601
0.06
3.40
0.2
5.96
0.1
1.25
0.4
19.37
20.15
0.3
2.39
0.1
1.64
0.9
bdl
–bdl
–0.1
0.01
bdl
–bdl
–99.9
17YG
E66.26
0.7
0.567
0.2
2.82
0.2
6.43
0.2
1.11
0.2
18.60
10.27
0.08
1.96
0.0
1.59
0.8
bdl
–bdl
–bdl
–bdl
–bdl
–99.6
18YG
E66.11
0.2
0.765
0.03
3.11
0.1
5.37
0.4
1.26
0.1
18.14
0.2
0.41
0.03
2.30
0.1
2.18
0.2
bdl
–bdl
–bdl
–bdl
–bdl
–99.6
19G(l.)
E66.49
0.03
0.437
0.04
3.07
0.1
6.91
0.3
0.97
0.1
18.11
0.4
0.53
0.06
1.71
0.1
1.48
0.2
bdl
–bdl
–0.08
0.05
bdl
–bdl
–99.7
20G
E67.90
0.8
0.133
0.03
2.63
0.1
7.64
0.4
0.84
0.01
17.82
0.1
0.48
0.1
1.01
0.0
0.62
0.04
bdl
–0.17
0.02
0.24
0.1
0.37
0.06
bdl
–99.9
Allvalues
areexpressedas
wt%
.EMPA
results
representthe
average(a)andstandard
deviation(s)values
offive
measurementspersample.[bdl=belowdetectionlim
it]
ABaqua
blue,B
blue,B
rbrow
nish,B
kblack,Ccolourless,G
green,(l.)lig
ht,R
reddish,Yyello
w,SDstandard
deviation
aLighter
band
bDarkerband
Archaeol Anthropol Sci
Table2
The
results
oftraceelem
entcom
positio
n(ppm
)obtained
byLA-ICP-MS
Sample
12
34
56
78
910
1112
1314
1516
1718
1920
n7
75
55
55
55
55
56
55
55
66
5
Li
a5.08
5.80
4.50
5.46
5.70
3.06
5.35
4.89
3.37
4.95
10.81
35.61
8.23
6.31
4.52
5.83
4.92
5.54
5.16
8.10
s2
0.6
0.8
0.5
0.3
0.6
0.2
0.3
0.1
0.5
23
0.5
0.7
0.3
0.6
10.3
0.5
0.3
Be
a1.34
1.43
0.46
1.39
1.92
1.14
1.13
1.92
1.08
0.86
2.92
2.83
1.55
0.65
–1.22
–0.73
0.88
3.29
s0.7
10.2
0.5
20.8
––
–0.5
20.2
0.6
––
0.4
–0.5
0.2
3
Ba
191.08
216.81
191.45
163.10
162.45
116.64
163.78
107.82
34.16
165.88
172.06
72.78
165.05
166.53
86.59
277.14
162.49
170.76
142.33
161.28
s16
206
03
210
22
710
163
85
711
55
7
Sca
6.32
7.01
6.06
4.80
7.58
3.31
6.93
4.53
2.30
5.43
2.76
11.21
4.40
3.24
6.99
6.99
6.96
8.45
6.01
3.97
s2
0.7
0.5
20.8
10.7
0.6
0.3
0.2
0.1
22
0.1
0.1
0.8
0.6
10.7
0.3
Va
48.05
48.78
46.25
18.22
60.09
7.39
58.67
24.85
18.32
57.75
25.57
86.13
23.02
18.75
67.17
63.64
58.33
69.35
49.92
25.80
s0.7
21
0.6
30.6
40.6
10.2
0.8
90.8
21
11
32
3
Cr
a49.73
61.36
52.65
15.87
63.36
14.86
63.24
45.70
9.77
60.51
25.26
80.67
18.71
16.67
73.61
70.69
72.41
104.45
61.14
20.46
s4
101
26
33
21
34
82
43
32
22
4
Co
a9.14
9.66
8.53
41.07
20.21
1.38
19.81
974.96
2.92
19.76
58.96
306.82
90.28
18.56
9.69
12.32
13.70
16.38
11.47
52.53
s2
0.8
0.3
190.1
0.1
0.6
107
0.4
0.6
267
164
0.4
0.5
0.5
0.03
0.2
6
Ni
a23.01
17.90
12.77
10.72
19.00
2.79
18.84
52.76
6.95
17.98
14.19
100.56
24.21
8.64
21.56
16.12
14.00
18.75
15.14
23.54
s9
31
21
0.9
0.8
20.03
20.4
303
20.9
21
0.6
0.7
11
Zn
a27.48
30.50
31.52
66.89
57.72
16.35
47.55
44.53
–35.03
70.92
1741.18
204.91
49.06
37.94
28.53
20.79
30.68
31.07
330.39
s8
1214
1819
213
4–
1018
719
4418
64
811
5286
Rb
a3.99
4.15
4.91
19.13
5.45
11.77
5.43
5.44
20.38
5.37
18.36
112.63
18.84
23.38
7.35
5.29
5.38
5.44
9.25
17.74
s0.3
0.7
0.4
1.0
0.3
0.6
0.2
0.3
0.3
0.1
112
22
0.3
0.06
0.2
0.3
0.3
4
Sra
742.98
742.64
516.73
501.80
593.57
473.59
583.77
606.45
513.59
572.07
503.29
592.03
512.57
465.16
447.73
453.93
506.36
457.26
526.68
516.84
s34
172
67
97
74
24
111
23
210
26
66
Ya
12.051
11.403
10.085
6.972
11.730
7.017
11.143
8.920
7.415
11.547
8.063
22.403
7.735
7.096
12.245
12.084
11.508
13.515
10.913
8.070
s0.7
0.4
0.4
0.3
0.3
0.4
0.3
0.5
0.07
0.4
0.2
50.1
0.4
0.1
0.4
0.3
0.4
0.3
0.5
Zr
a232.40
262.78
232.28
64.31
277.16
33.74
272.53
163.49
33.04
272.66
102.18
157.50
83.90
66.55
313.69
305.07
309.03
405.33
230.70
88.67
s44.7
3.5
6.9
1.3
0.8
1.5
3.4
14.9
12.9
520
48
99
28
74
Nb
a5.6725
5.2707
4.8630
2.0275
5.6858
1.1879
5.4790
3.5197
1.1628
5.4128
2.7978
9.9482
2.4847
2.0715
6.3249
6.6023
5.6603
7.6189
5.0242
2.7997
s0.4
0.3
0.2
0.1
0.2
0.1
0.09
0.02
0.1
0.5
0.3
0.7
0.05
0.1
0.05
0.3
0.4
0.6
0.2
0.4
Cs
a0.0618
0.0446
0.0594
0.3035
0.0835
0.1000
0.0962
0.0546
0.1482
0.0788
0.3782
4.5687
0.5327
0.5503
0.1140
0.0660
0.0666
0.0852
0.1276
0.6711
s0.05
0.02
0.02
0.08
0.03
0.01
0.006
0.02
0.002
0.03
0.09
0.5
0.08
0.09
0.03
0.01
0.03
0.01
0.02
0.5
Ba
a292.66
311.38
900.65
266.67
1188.20
236.24
1136.87
257.01
406.94
1126.67
434.35
558.69
356.96
286.13
917.98
649.34
1010.69
864.89
1122.80
360.81
s26
115
2755
64
378
616
120
515
1829
1814
126
4
La
a11.813
11.532
9.437
7.236
11.662
6.629
11.051
8.466
6.936
10.926
8.838
41.935
8.861
7.991
11.703
11.885
11.218
12.657
10.757
10.064
s2
0.2
0.2
0.3
0.1
0.2
0.3
0.3
0.2
0.3
0.3
110.8
0.1
0.2
0.2
0.5
0.6
0.3
2
Ce
a18.63
18.02
16.62
12.82
18.88
12.25
18.68
15.33
12.80
18.14
15.41
63.97
15.17
13.49
19.23
20.55
19.70
22.38
18.60
16.78
s1
0.5
0.2
0.8
0.9
0.4
0.2
0.2
0.2
0.2
0.7
61
0.6
0.4
0.4
0.2
0.6
0.2
4
Pra
2.650
2.411
2.134
1.636
2.491
1.490
2.405
1.921
1.635
2.429
1.811
7.951
1.813
1.718
2.680
2.624
2.371
2.777
2.279
2.161
s0.3
0.04
0.04
0.1
0.1
0.03
0.1
0.07
0.06
0.1
0.02
10.1
0.07
0.2
0.02
0.2
0.1
0.07
0.4
Archaeol Anthropol Sci
Table2
(contin
ued)
Sample
12
34
56
78
910
1112
1314
1516
1718
1920
n7
75
55
55
55
55
56
55
55
66
5
Nd
a10.596
9.884
9.026
6.410
10.554
5.855
10.582
8.252
6.927
10.334
7.838
30.712
7.980
7.211
11.304
11.470
10.060
12.074
10.058
8.617
s2
0.4
0.1
0.6
10.9
0.3
0.8
0.2
0.6
0.8
60.9
0.5
0.3
0.5
0.4
0.2
0.6
2
Sma
2.139
2.354
2.039
1.325
2.445
1.089
2.183
2.018
1.408
2.317
1.566
5.889
1.504
1.423
2.430
2.268
2.131
2.989
2.254
1.825
s0.7
0.4
0.04
0.3
0.2
0.2
0.5
0.2
0.08
0.6
0.1
10.07
0.2
0.1
0.3
0.08
0.2
0.3
0.4
Eu
a0.751
0.623
0.499
0.407
0.629
0.376
0.499
0.430
0.424
0.632
0.517
1.273
0.373
0.378
0.575
0.605
0.630
0.617
0.554
0.513
s0.2
0.06
0.04
0.01
0.1
0.09
0.04
0.07
0.06
0.07
0.03
0.2
0.1
0.1
0.1
0.1
0.2
0.02
0.09
0.02
Gd
a1.953
1.733
1.768
1.319
2.212
1.065
2.127
1.580
1.366
2.217
1.611
4.604
1.300
1.314
2.184
2.299
1.947
2.144
1.868
1.369
s0.2
0.6
0.3
0.09
0.2
0.2
0.2
0.3
0.07
0.2
0.1
10.1
0.1
0.1
0.1
0.2
0.3
0.3
0.2
Tb
a0.3508
0.3365
0.3212
0.1841
0.3740
0.1794
0.3487
0.2385
0.1580
0.3669
0.1481
0.7418
0.2433
0.2360
0.3832
0.3140
0.3083
0.3255
0.2830
0.2862
s0.1
0.07
0.02
0.06
0.06
0.07
0.03
0.06
0.04
0.01
0.01
0.2
0.05
0.05
0.08
0.06
0.08
0.01
0.04
0.08
Dy
a1.9166
1.8550
1.6348
1.1727
2.0809
1.2255
1.7945
1.5640
1.1429
1.9559
1.3107
4.4493
1.4198
1.2292
2.2527
2.0381
1.9970
2.4193
1.9370
1.3591
s0.4
0.4
0.2
0.1
0.07
0.2
0.1
0.3
0.06
0.2
0.09
0.9
0.02
0.2
0.3
0.1
0.2
0.1
0.2
0.1
Ho
a0.4251
0.4268
0.3476
0.1997
0.4306
0.2530
0.4480
0.2940
0.2667
0.3818
0.3076
0.7600
0.2756
0.2560
0.4717
0.4454
0.4191
0.5304
0.3868
0.3357
s0.06
0.07
0.02
0.001
0.06
0.02
0.08
0.02
0.03
0.06
0.03
0.1
0.01
0.02
0.02
0.02
0.08
0.05
0.06
0.04
Er
a1.2331
1.2964
0.9398
0.7719
1.1974
0.5598
1.0508
0.9575
0.8232
1.1168
0.8358
2.1816
0.7899
0.6621
1.3475
1.3045
1.3511
1.4598
1.2067
0.7649
s0.09
0.2
0.07
0.1
0.09
0.1
0.03
0.2
0.05
0.2
0.05
0.3
0.2
0.1
0.2
0.06
0.05
0.08
0.1
0.03
Tm
a0.2637
0.2492
0.2514
–0.2818
0.2191
0.3062
0.1881
–0.1537
0.3192
0.3359
0.2401
0.1219
0.1570
0.1928
0.3225
0.2745
0.2909
–
s0.1
0.06
0.2
––
––
––
–0.02
0.07
––
0.03
0.07
0.06
0.1
0.1
–
Yb
a1.365
1.163
1.091
0.563
1.241
0.526
1.212
0.839
0.748
1.362
0.636
2.091
0.815
0.614
1.358
1.510
1.264
1.612
1.247
0.867
s0.2
0.04
0.04
0.1
0.1
0.2
0.05
0.1
0.3
0.1
0.01
0.5
0.2
0.1
0.09
0.1
0.2
0.2
0.2
0.2
Lu
a0.1676
0.1457
0.1810
0.0973
0.2139
0.0960
0.2023
0.1473
0.0836
0.1762
0.1085
0.3169
0.1288
0.0839
0.2193
0.2087
0.2088
0.2337
0.1661
0.1053
s0.05
0.05
0.05
0.05
0.03
0.02
0.04
0.04
0.03
0.04
0.02
0.06
0.01
0.009
0.02
0.03
0.02
0.02
0.02
0.02
Hf
a5.7906
6.2958
5.3401
1.5578
7.0690
0.8639
6.3979
4.0244
0.8283
6.4408
2.3174
4.1223
1.8361
1.7372
7.6927
7.2162
7.2111
9.5632
5.3783
2.2391
s1
0.4
0.1
0.1
0.2
0.2
0.1
0.3
0.1
0.2
0.1
0.8
0.1
0.05
0.3
0.4
0.3
0.6
0.1
0.5
Taa
0.3063
0.3110
0.2799
0.1361
0.3323
0.0857
0.3459
0.2711
0.1505
0.3038
0.1801
0.6742
0.1147
0.1311
0.4100
0.3579
0.3782
0.4864
0.3629
0.2254
s0.05
0.05
0.06
0.05
0.03
0.02
0.02
0.09
–0.05
0.06
0.1
0.04
0.02
0.1
0.04
0.06
0.1
0.09
0.1
Pba
119558
5854
102071
227
17216
1769
14203
3157
58872
6671
1275
1012
67
7048414
s262469
2653
0650
171
1190
05
447
19003
229
276
11
00
674344
Th
a1.964
2.087
2.071
1.355
2.350
0.879
2.380
1.635
0.761
2.253
1.824
11.382
1.691
1.571
2.412
2.538
2.518
3.154
2.178
2.092
s0.2
0.06
0.03
0.05
0.06
0.1
0.05
0.1
0.00
0.1
0.1
20.02
0.2
0.03
0.1
0.08
0.08
0.1
0.8
Ua
1.64
1.64
1.28
1.05
1.64
0.62
1.51
0.99
0.55
1.47
1.15
3.50
1.16
1.06
1.10
1.33
1.57
1.75
1.31
1.27
s0.1
0.1
0.07
0.04
0.07
0.07
0.1
0.09
0.03
0.03
0.06
0.3
0.07
0.03
0.04
0.06
0.06
0.08
0.06
0.2
Five
tosevenmeasurementswereperformed
each
sample.Significantfigures
aretothesecond
decimalplaceform
ajor
andminor
elem
ents,w
hereas
values
fortrace
andultra-traceelem
entsaresignificant
tothethirdor
fourth
decimalplace,follo
wingstandardisationandexperimentalconditio
ns
aaverage,sstandard
deviation,nnumberof
measurementspersample
Archaeol Anthropol Sci
Results and discussion
Technological issues are commented sample by sample, de-pending on the characterising features of each item.Provenance issues are discussed in relation to the greater orlesser compositional similarity of the investigated sampleswith the following reference groups:
– Levantine I (hereafter Levantine). Materials of Syrian–Palestinian origin, dated between the first and the eighthcentury AD (Brill 1988; Freestone et al. 2000; Foy et al.2003; Schibille et al. 2008; Foster and Jackson 2009);
– HIMT.Materials mainly dated between the fourth and theseventh century AD (Mirti et al. 1993; Foy et al. 2003;Foster and Jackson 2009; Freestone 1994), likely origi-nating from northern Sinai and Egypt (Freestone 2005;Nenna 2014). Further comparison has also been drawnwith respect to HIMT parental groups such as the HIMT 2group (Foster and Jackson 2009), the weak HIMT group(Rosenow and Rehren 2014), the CaO-rich HIMT group(Gliozzo et al. 2015b) and the HLIMT group (Ceglia et al.2015);
– RNCBGY 1. Naturally coloured blue-green and yellowglass dated between the first and the fourth century AD(Mirti et al. 1993; Jackson 2005; Silvestri et al. 2005;Silvestri et al. 2008; Foster and Jackson 2009), composi-tionally similar to the HIMT glass and, for this reason,associatedwith a North African (Egyptian?) area of origin(Gliozzo et al. 2013). It is worth clarifying that some glassgroups defined as ‘naturally coloured’ by their authorsand consequently included here can contain up to0.21 wt% Sb (see e.g. Foster and Jackson 2009); there-fore, their characterisation as naturally coloured is surelyquestionable;
– RNCBGY 2. Naturally coloured glass dated between thefirst and the fourth century AD (Mirti et al. 1993 andSilvestri 2008), compositionally similar to Levantineglass and, for this reason, associated with the Syrian–Palestinian area of origin (Gliozzo et al. 2013).Differently from the previous group, this glass does notcontain Sb, although data are not available for groupspublished by Mirti et al. (1993);
– RC-Mn and LAC-Mn. Mn-decoloured glass or Roman(first to third century AD) and Late Antique (fourth tosixth century AD) age of uncertain origin (Gliozzoet al. 2015a). Both groups are characterised by MnO>0.2 wt% and Sb <20 ppm.
Average values of each reference group have been provid-ed in Table 3. The assignment of a single sample to a referencegroup was considered reliable only when the sample plotted inthe same group field in all proposed diagrams, unless other-wise specified.
Glass ingots: the yellow sample nos. 1 and 2
Samples 1 (Fig. 4a) and 2 (Fig. 4b) are texturally very similar.Both are compositionally banded (Fig. 4c) and show relicts ofrounded quartz (Fig. 4d) and newly formed euhedral crystalsof wollastonite (Fig. 4e). Devitrification phases such as (non-stoichiometric) Ca–Na silicates are further present, especiallyin association with lead stannate (Fig. 4f).
The great compositional heterogeneity of darker and lighterbands does not make their analyses entirely reliable (see thehigh standard deviations in Table 1); however, it is evident thatlighter bands are greatly enriched in Pb and Sn with respect tothe darker ones.
In Fig. 4g–l, it is possible to get an insight into the twodifferent types of glass. Dark bands are typically characterisedby (a) low amounts of lead, (b) absence of tin and antimony and(c) presence of large aggregates of Mn oxides (Fig. 4g–h).Conversely, the light bands are characterised by (a) very highlead and tin contents, (b) absence of MnO oxides, (c) rare pres-ence of (lead?) antimonates (Fig. 4i) and (d) presence of leadstannate (Fig. 4j). The composition of the lead stannates mea-sured by EMPA throughout the sample was variable due to thesmall size and heterogeneity of these aggregates, but XRD in-vestigation was able to identify the PbSnO3 cubic polymorph.The SEM-EDS and EMPA investigation of the antimonates wascontroversial; in fact, they were too small to impede any reliablequantitative measurement by these techniques; however, also inthis case, XRD identified these phases as lead antimonates.
The bulk analyses indicate both samples as soda–lime–silicaglass, with Na2O contents ranging between 10.7 and 13.4 wt%,CaO between 7.2 and 7.4 wt% and SiO2 between 54.3 and55 wt%. In addition, the MgO/K2O ratio (Fig. 5) indicates thatthe flux consists of natron, while the Al2O3 values (>3 wt%)indicates a siliceous sand as the network former of the baseglass. As previously observed, lead and tin are responsible forthe yellow colour of these samples, where traces of copper andantimony were further found. Interestingly, MnO contents arealways higher than those of Fe2O3, suggesting that a yellowglass containing the yellow lead stannate (i.e. ‘the lead–tin yel-low anime’, see Tite et al. 2008)was added to a colourless glass.The PbO/SnO2 ratio of these samples was very high, rangingfrom 5.35 to 5.41. Based on Tite et al. (2008), this feature mayindicate a low firing temperature (<700 °C) of a very fluid melt(high Pb), in order to prevent the formation of cassiterite. Apartfrom this, a low-temperature fluid melt must have createdfavourable working condition, prolonging the workability ofglass without favouring crystallisation processes.
The present data were compared with the available referencegroups (Figs. 5 and 6). It is possible to notice that samples 1 and2 always plot close to the HIMT reference group rather than theLevantine glass. This similarity would be even closer if normal-ising data after excluding both Pb and Sn amounts, but it is alsoworth considering that the best comparison is found with
Archaeol Anthropol Sci
tesserae rather than with vessels and that many of thosetesserae were found in Italy or in Israel. In Italy, several sixthcentury tesserae are closely comparable to the samples investi-gated here: some tesserae from the palaeo-Christian glass mo-saic of St. Prosdocimus at Padua (e.g. the brown tessera no.M5in Silvestri et al. 2014) and some others from the Basilica of SanVitale at Ravenna (e.g. yellow-green tesserae C20 and C24 inFiori et al. 2004). An earlier example is further provided by theslab no. OS-R2 found at Porta Marina at Ostia which, however,shows 12 wt% of PbO, 0.5 wt% of SnO2 and a low MnOamount (Verità et al. 2008). Outside Italy, several sixth to sev-enth century AD tesserae from the exedra of Beit She’an inIsrael (Shugar 2000) are particularly similar to the investigatedsamples, especially the Israeli sample no. 9 which howevershows a lower PbO content (5.3 wt%). In terms of PbO andSnO2 contents, also several tesserae from Shikmona are similarto these samples although showing higher SiO2, CaO andMnOcontents (see e.g. samples 29355x and 29352s in Freestoneet al. 1990). The REE pattern (Fig. 7) is Sr richer and Ba poorerthan that of the HIMT glass, anyway much more similar to theHIMTrather than to the Levantine glass. Also, Zr andHf valuesare relatively high and much more similar to other HIMT sam-ples than to the Levantine ones. In this instance, it is also worthnoting that the REE pattern of these wastes could be influencedby several contaminants introduced with the colouring agents.
Glass ingot: the green sample no. 4
Sample 4 is completely different from samples 1 and 2. It istexturally very homogeneous; in fact, compositional bandingis absent. Relics are extremely rare and always constituted byrounded quartz (<100 μm), while newly formed phases arerepresented by very rare crystals of wollastonite (<20 μm).Extremely rare Cu and lead antimonate droplets of appreciabledimensions were also found; the former never exceeded 2 μm,while the latter varied from 1 to 15 μm. Unfortunately, thesmall dimensions of these droplets impeded a detailed analysisboth in terms of qualitative description and quantitative com-position. The bulk chemical composition is that of soda–lime–silica glass, where a siliceous sand (network former) wasmixed with natron (flux). As for colouring agents, MnO andFe2O3 contents are below 1 wt%while Cu, Pb and Sb were allabove 1000 ppm and, therefore, likely to have been intention-ally added. The green colour must have been obtained bymixing yellow lead antimonates to blue Cu2+. It would notbe surprising if the colouring agents had been added to anaturally coloured glass, in order to produce a new raw glass.Comparing the present data with the available referencegroups (Figs. 5 and 6), sample 4 is always included in theRNCBGY 1 group although it is not a naturally coloured glass(like some of the RNCBGY 1 samples). Similarly, the REE
Table 3 Average (a) andstandard deviation (s) values ofthe reference groups cited in text
Group Levantine HIMT RNCBGY 1 RNCBGY 2 RC-Mn LAC-Mn
n 152 396 290 77 50 58
SiO2 a 69.9 67.8 69.8 70.6 69.6 67.1
s 1.8 2.1 0.9 1.4 2.0 1.7
Al2O3 a 2.9 2.4 2.4 2.5 2.5 2.4
s 0.2 0.3 0.2 0.2 0,2 0.3
TiO2 a 0.08 0.24 0.1 0.1 0,1 0.2
s 0.02 0.16 0.0 0.0 0 0.1
CaO a 8.7 6.0 6.7 7.7 7.8 7.6
s 0.9 0.6 0.5 0.5 0.8 1.3
MgO a 0.6 0.9 0.6 0.5 0.6 1.0
s 0.1 0.2 0.1 0.1 0.2 0.6
Na2O a 15.6 19.4 17.5 16.1 16.3 17.1
s 1.2 1.1 0.7 0.9 1.7 1.9
K2O a 0.8 0.5 0.7 0.8 0.6 0.7
s 0.2 0.1 0.2 0.3 0.2 0.4
Fe2O3 a 0.4 1.1 0.5 0.4 0.3 1.0
s 0.1 0.6 0.1 0.2 0.2 0.5
MnO a 0.6 1.3 0.4 0.5 1.1 1.2
s 0.6 0.5 0.2 0.4 0.4 0.5
The compositional values of single collections included in the Levantine and HIMT glass are listed in text.Average values of RNCBGY 1 and 2 glass groups are from Gliozzo et al. (2013). Preliminary values for RC/LAC-Mn glass groups are proposed in Gliozzo et al. (2015a )
n number of samples included in each group
Archaeol Anthropol Sci
pattern (Fig. 7) is ‘intermediate’ between the HIMT and theLevantine ones, and, again, the introduction of colouringagents may have altered the REE pattern.
Raw glass chunks: the yellow (greenish) sample nos. 5, 7
and 10
The texture is very homogeneous in all samples, rarely inter-spersed with newly formed wollastonite crystals and extreme-ly rare quartz relics. All samples showed a soda–lime–silicacomposition, with Na2O contents ranging between 18.3 and
18.9 wt%, CaO between 6.5 and 6.7 wt% and SiO2 between64.9 and 65.4 wt%. SiO2, Al2O3 (comprised between 2.8 and3.0 wt%) and CaO contents further indicate that a siliceoussand was used to produce all these vessels. As for fluxes, thehigh Na2O contents and the K2O/MgO ratio indicate the mainuse of natron for these samples. As for colouring agents, sam-ples 5, 7 and 10 are similarly coloured by Fe2+/Fe3+, but theyare further characterised by high levels of MnO (>2 wt%).Given that these samples retain some colour, a mistake madeduring melting may have interfered with the discolourationprocess. However, it is also likely that these values are due
Fig. 4 SEM-BSE images of theglass items, showing theirtextures or notable phases. Pleasenote that scale bar changes
Archaeol Anthropol Sci
to the composition of the base glass and are therefore notindicative of the colouring process. The trace amounts of Cu
and Pb in samples 7 and 10 are too high to be ‘natural’ and toolow to be intentionally added; therefore, the recycling practice
Fig. 5 MgO-K2O, SiO2-Na2Oand Al2O3-CaO binary diagramsshowing both the samplesinvestigated in this research andthe compositional fields of thereference groups. Samples 1 and 2have been omitted in SiO2-Na2Odiagram
Archaeol Anthropol Sci
Fig. 6 Al2O3-Fe2O3, Fe2O3-Fe2O3/MnO, MgO-Al2O3 andMnO-TiO2 binary diagramsshowing both the compositionalfields of the reference groups(same references used for Fig. 5)and the samples investigated inthis research
Archaeol Anthropol Sci
which is a common feature of Late Antique glass productions(Freestone and Hughes 2006; Cagno et al. 2012) must betaken into account. Comparing the present data with the avail-able reference groups (Figs. 5 and 6), these samples alwaysplot within the HIMT compositional field and their REE pat-tern (Fig. 7) is perfectly compatible with that of HIMT glass.
Raw glass chunk: the blue sample no. 8
Sample no. 8 is characterised by a very homogeneous texture,free of compositional banding and rarely showing relics (onlyquartz) or newly formed phases (wollastonite). The bulk chem-ical composition is that of soda–lime–silica glass, made bymixing siliceous sand with natron. As for the colouring agents,cobalt must play the major role. The concomitant presence ofrelatively high amounts of Cu and Pb does not seem to have aninfluence on colour while they could testify to the recyclingpractice or they may have been part of the suite of metals intro-duced with cobalt. In the latter case, however, it should be pos-sible to observe an analogous increase of Ni and Zn or of Al,Mg, Fe andMn, while in this sample, only a slight increase of Niis measured. In any case, the common practice of recyclingcullet may be identified as the most likely source for cobalt.
Comparing the results obtained for this sample with theavailable reference groups (Figs. 5 and 6), it is possible to
notice that it mainly resembles the composition of HIMT glass(esp. in relation to SiO2, Na2O, Al2O3, MgO and Fe2O3 con-tents), with respect to which, however, it shows higher CaOand lower MnO amounts. The REE pattern is much moresimilar to that of other HIMT and RNCBGY 1 samples thanto the Levantine ones, but also in this case, the problem of theinclusion of contaminants must be taken into account.
Raw glass chunks: the aqua blue sample no. 6
and the colourless sample no. 9
The aqua blue sample is the most homogeneous one of theentire collection; neither relics or newly formed phases werefound in the analysed fragment. Similarly, the colourless sam-ple showed very few quartz relicts only. The bulk chemicalcomposition of both samples is that of soda–lime–silica glass,where siliceous sand was mixed with natron.
The aqua blue sample must be coloured by Fe2+/Fe3+, whilethe colourless sample showed very low Fe2O3 amounts andMnO >0.2 wt%; hence, the latter must have been deliberatelyadded. Both samples did not show traces of other colouring/decolouring agents and, therefore, are considered ‘fresh’ glass.
The comparison with the available reference groups ad-dresses the Levantine origin as highly likely for both samples,although Na2O and Al2O3 contents of sample 9 are slightly
Fig. 7 REE pattern. Valuesnormalised against the averagecomposition of the weatheredupper continental crust (Kamberet al. 2005). The diagramsrepresent the REE patterns of theinvestigated samples, divided intothe different compositionalgroups. The blue and the red linesrespectively indicate the REEpatterns of Levantine and HIMTglass as reported by Schibille2011
Archaeol Anthropol Sci
higher than those of the Levantine glass while SiO2 amounts arelower. The colourless sample has been further compared withthe reference groups available for Roman and Late AntiqueMn-decoloured glass, RC-Mn and LAC-Mn. Both groups are simi-lar to this sample but show higher SiO2, MgO and MnO con-tents counterbalanced by lower Al2O3, CaO, Na2O and K2Ocontents. Fe2O3 contents are compatible with the RC-Mn groupwhile much lower than the LAC-Mn group. The REE pattern(Fig. 7) is perfectly comparable to that of Levantine glass.
Lid-shaped moil: the light green sample no. 3
This waste is both texturally and compositionally similar tothe yellow (greenish) raw glass sample nos. 5, 7 and 10; there-fore, it is possible to draw the same considerations regardingthe use of siliceous sand as the vitrifying agent and the use ofnatron as the fluxing agent. As for the light green hue of thisglass, it is evident that it is coloured by Fe2+/Fe3+, as iron is theonly colouring agent present in sufficient amounts. As forprovenance, the HIMT glass represents the closest compari-son. To be precise, sample no. 3 shows slightly higher MgOcontents and a differentMnO-TiO2 ratio, but it always remainsin a range of acceptable variation. Also, the REE pattern(Fig. 7) is perfectly comparable to that of the HIMT glass.
Wastes: the green samples 11, 13 and 20, the blackish
sample no. 12 and the reddish sample no. 14.
The dark red sample no. 12 is not a proper glass but a collateralby-product; however, the correlation to the glass production isguaranteed by stratigraphic issues. The fragment does notshow layering, and its composition resembles that of a Ca-rich clay. The melting of a crucible seems less likely, basedon the Ca-rich composition of this fragment. High amounts ofPb and Cu, determined by bulk chemistry, can be correlated tothe presence of CuS (resembling covellite composition), CuSn(i.e. tin bronze) and Pb droplets observed by SEM-EDS.
The green samples 11, 13 and 20 are proper soda–lime–silica glass, with SiO2 contents ranging from 66.6 to68.1 wt%, Na2O contents from 17.5 to 18.0 wt%, Al2O3 from2.6 to 2.8, CaO from 7.1 and 7.6 wt% and bothMgO and K2Ocontents below 1 wt%. A siliceous sand was mixed with na-tron and likely coloured by the joint effect of copper and leadantimonates. In fact, these samples are all characterised by Sbcontents ranging between 1575 and 1841 ppm, Cu contentsranging between 1103 and 2475 ppm and Pb contents rangingbetween 2640 and 5150 ppm. In detail, Cu2+ ions develop ablue colour, while the yellow Pb antimonates promote boththe opacification and the colour shift to a light green hue. Thelow amount of iron ions could have affected the developmentof the green colour, while the MnO contents (>1 wt% in sam-ple 13 only) must not have had any effect. Both Cu dropletsand lead antimonates have been identified by SEM-EDS.
Texturally, a barely perceptible compositional banding isobserved in samples 11 and 14. Both showed increasingFe2O3 and CaO amounts counterbalanced by decreasingNa2O contents in lighter bands with respect to the darker ones.However, the drop no. 14 is very similar to the previous greenones in terms of Sb and Cu contents, while it shows slightlylower amounts of Pb, Fe2O3 and MnO. In sample 14, thedimension of the metallic prills decreases to a few micronsand the glass retains a reddish colour which must be due tothe presence of the metallic Cu identified by XRD. Theamounts of Pb must have been too low to favour the growthof cuprite crystals (Freestone 1987) but enough to act as areducing agent as well as Sb, although a strong reducing at-mosphere would be able to reduce Cu also in the absence ofother components (Angelini et al. 2004). Interestingly, theabsence of cuprite may be also responsible for the translucenteffect of this glass. With respect to red glass tesserae, theFe2O3 contents are much lower (see e.g. Santagostino et al.2008; Gliozzo et al. 2012); this feature may open very differ-ent hypotheses. Among the latter, (a) this further reducingagent was not added because the glassmakers were lookingfor a green colour as in samples 11, 13 and 20; (b) high levelsof iron were not needed in the making of this type of redtranslucent glass. Indeed, from a to b, the reconstructionchanges radically from a mistake to a technologically ad-vanced procedure.
Comparing the results obtained for these sample with theavailable reference groups (Figs. 5 and 6), it is possible tonotice that they mainly plot in the HIMT and the RNCBGY1 compositional fields. Indeed, they are not naturally colouredglass and they do not show a REE pattern perfectly compara-ble to that of the HIMT glass; however, the Levantine prove-nance would represent a weaker hypothesis. It is worth re-membering the possibility that the REE pattern of these wastescould be influenced by several contaminants introduced withthe colouring agents.
Vessels: the green samples 15, 16 and 19
and the yellow-green samples 17 and 18
The texture is very homogeneous in all samples, rarely inter-spersed with newly formed wollastonite crystals and extreme-ly rare quartz relics. All samples showed a soda–lime–silicacomposition, with Na2O contents ranging between 17.8 and19.4 wt%, CaO between 5.4 and 6.9 wt% and SiO2 between65.0 and 66.8 wt%.
Similarly to the light green sample no. 3 and the yellow(greenish) sample nos. 5, 7 and 10, the contents of SiO2,Al2O3 (comprised between 2.8 and 3.4 wt%) and CaO indi-cate that siliceous sand was used as the network former, whilehigh Na2O contents and low K2O and MgO contents indicatethe main use of natron as a flux.
Archaeol Anthropol Sci
Different ratios of Fe2+/Fe3+ must be responsible for theyellow to green colours of these samples, where the amountof other colourants is negligible. Fe2O3 values range between1.7 and 2.7 wt% and exceed 2 wt% in samples 15–16 and 18.Interestingly, also MnO contents are always above 1 wt%, butthe glass still preserves some colour. In fact, MnO valuesexceed 2 wt% in sample 18, while it is comprised between 1and 2 wt% in samples 15–17 and 19. The Fe2O3/MnO ratio isbetween 0.9 and 1.1 in sample 18 or above 1.1 in the remain-ing samples 15–17 and 19. Sb and Pb are absent in all thesesamples. Cu is below the detection limit in samples 15 and17–18, while comprised between 100 and 1000 ppm in sam-ples 16 and 19. Therefore, the samples 15 and 17–18 may beconsidered as ‘fresh glass’, while the green samples 16 and 19are likely to maintain traces of recycling.
As for provenance, both the comparison with availablereference groups (Figs. 5 and 6) and the REE pattern(Fig. 7) indicate Egypt (HIMT glass) as the most likely areaof origin of these items. Furthermore, comparing these sam-ples with the others investigated here, the best comparison isfound with the light green and the yellow (greenish) samplesof raw glass (nos. 3, 5, 7 and 10).
Conclusions
The results have been summarised in Table 4.All samples are soda–lime–silica glass made by mixing
siliceous sand (vitrifying agent) and natron (fluxing agent).For the preparation of ingot nos. 1 and 2,Mn-decoloured glasswas used as the ‘base glass’, to which the yellow leadstannates were added as a colourant.
Apart from lead stannates, the colouring agents were vari-ous. Green colours were obtained both by the combined actionof Cu2+ (blue) and lead antimonates (yellow) and by differentratios of Fe2+/Fe3+. Similarly, different Fe2+/Fe3+ ratios deter-mined the light green, yellow (greenish) and yellow-greenhues while Fe2+ prevailing over Fe3+ must be responsible forthe aqua blue colour. Cobalt was used for deep blue glass,while metallic copper imparted the translucent reddish colourof sample 14.
Looking for similarities within the investigated collection,it was possible to correlate (a) the raw glass sample no. 4 withthe waste nos. 11, 13 and 20 and (b) the raw glass nos. 5, 7 and10 with the lid-shaped moil no. 3 and the vessel nos. 15–19.The sample nos. 1 and 2 are not comparable with any of theother samples and show an overall composition which is clos-er to that of tesserae than that of vessels. The shape of theseingots, flat and squared, may be a further indication of theiruse for this specific glass production.
As for provenance, it is worth considering that Table 4represents an oversimplification which does not take into ac-count the numerous stages (from the procurement of raw T
able4
Summaryof
theresults
Colour
Vitrifying
agent
Fluxingagent
Colouring
agent/s
Group
REE
Provenance
Raw
glass
1–2
Yellow(brownish)
Colourlessglass(siliceoussand)
Natron
Leadstannates
HIM
T/none
≈HIM
TNorth
Africa(Egypt)?
4Green
Siliceous
sand
Natron
Cu2
++lead
antim
onate
RNCBGY1
RNCBGY1/H
IMT
North
Africa(Egypt)?
5,7,10
Yellow(greenish)
Siliceous
sand
Natron
Fe2+/Fe3
+prevailin
gover
MnO
HIM
THIM
TEgypt
(north
Sinai?)
8Blue
Siliceous
sand
Natron
Cobalt
HIM
T(RNCBGY½
ornone)
HIM
T/RNCBGY1
North
Africa(Egypt)?
6Aquablue
Siliceous
sand
Natron
Fe2+/Fe3
+Levantin
eLevantin
eLevantin
ecoast
9Colourless
Siliceous
sand
Natron
Mn(purewhitesand)
Levantin
e(H
IMT)
Levantine
Levantin
ecoast
The
collarandthewastes
3Light
green
Siliceous
sand
Natron
Fe2+/Fe3
+HIM
T(none)
HIM
TEgypt
(north
Sinai?)
11Green
(light)
Siliceous
sand
Natron
Cu2
++lead
antim
onate
HIM
T/RNCBGY1/none
HIM
T/RNCBGY1
North
Africa(Egypt)?
13,20
Green
Siliceous
sand
Natron
Cu2
++lead
antim
onate
HIM
T/RNCBGY1
HIM
T/RNCBGY1
North
Africa(Egypt)?
14Reddish
Siliceous
sand
Natron
Metallic
Cu(Pb,Sb
ineffective)
RNCBGY1
HIM
T/RNCBGY1
North
Africa(Egypt)?
Vessels
15–16,19
Green
Siliceous
sand
Natron
Fe2+/Fe3
+(M
nineffective)
HIM
T(none)
HIM
TEgypt
(north
Sinai?)
17–18
Yellow-green
Siliceous
sand
Natron
Fe2+/Fe3
+(M
nineffective)
HIM
T(none)
HIM
TEgypt
(north
Sinai?)
Archaeol Anthropol Sci
materials to the abandonment) that the objects may havepassed in different areas.
Most materials showed close compositional similaritieswith North African products, while the Levantine productionswere attested by two raw glass samples only: the aqua blue no.6 and the colourless no. 9. Among the North African products,it was possible to distinguish proper HIMT glass, hence likelyof Egyptian origin, from RNCBGY 1 parent group of un-known provenance and, more frequently, to identify ‘interme-diate’ glass whose composition was in between the HIMTandthe RNCBGY 1 groups.
As for provenance of the glass ingot nos. 1 and 2, on theone hand, the comparison traced among tesserae composi-tions would better indicate a Levantine origin than a NorthAfrican one; on the other hand, the comparison with the avail-able reference groups would indicate a closer similarity to theHIMT rather than the Levantine glass. Indeed, the composi-tional fields in Figs. 5 and 6 are highly reliable for vesselinvestigation, while the tesserae should be treated separately.Also, the REE pattern (Fig. 7) is similar to that of HIMT glass,although Sr richer and Ba poorer and likely contaminated bythe introduction of the colouring agents. Given the contradic-tory results, the provenance assignment of these two samplesremains open. The possibility that the decoloured base glasshad a different provenance with respect to the yellow glass isalso likely, and it would also represent a further explanation ofwhy the REE pattern matches that of the HIMT glass onlypartially. Furthermore, the presence of apparently useless con-tents of Cu (and maybe also Sb) seems to testify that even rawglass manufacture used recycled materials.
Glass recycling must have been widely applied although itis still difficult to trace sure boundaries in separating concen-trations that are likely to be natural, from those reflecting theincorporation of glass cullets and those clearly indicating thedeliberate addition of a certain component. Assuming, forinstance, that Cu, Pb and Sb contents comprised between100 and 1000 ppm are indicative of the recycling practice,the Levantine sample nos. 6 and 9 and the vessel nos. 15, 17and 18 would be the only ‘freshly made’ glass of the collec-tion. Moreover, one must not forget, however, that self-recycling may add even these few samples to the category ofrecycled.
The results obtained on recycling appear of great interestbecause they regard raw glass, not only shaped products.There is primary glass produced both along the Levantineand the North African coasts, and there are raw chunks ofraw glass or even glass ingots in Italy which are likely madeusing recycled materials. Whether this practice (a) wasadopted in the primary ateliers or in secondary ones and (b)with a sole technical purpose (e.g. changing the glass colourlike in the cobalt blue sample) or economic grounds (e.g. lowcost of cullets, decline in imports of freshly made glass), it isnot possible to tell, but these are questions to keep well in
mind until the state of the art of the studies will provide suf-ficient tools for a convincing explanation.
References
Amrein H (2001) L’atelier de verriers d’Avenches. L’artisanat du verre aumilieu du Ier siècle après J-C, AventicumXI (Cahiers d’Archéologieromande 87), Lausanne
Angelini I, Artioli G, Bellintani P, Diella V, Gemmi M, Polla A, Rossi A(2004) Chemical analyses of Bronze Age glasses from Frattesina diRovigo, Northern Italy. J Archaeol Sci 31:1175–1184
Brill RH (1988) Scientific investigations of the Jalame glass and relatedfinds. In: G. D. Weinberg (ed.), Excavations at Jalame: site of aglass factory in late Roman Palestine, Columbia, pp. 257–294
Cagno S, Favaretto L, Mendera M, Izmer A, Vanhaecke F, Janssens K(2012) Evidence of early medieval soda ash glass in the archaeolog-ical site of San Genesio (Tuscany). J Archaeol Sci 39:1540–1552
Ceglia A, Cosyns P, Nys K, Terryn H, Thienpont H, Meulebroeck W(2015) Late antique glass distribution and consumption in Cyprus:a chemical study. J Archaeol Sci 61:213–222
Fiori C, Vandini M, Mazzotti V (2004) I colori del vetro antico, Il PratoFoster HE, Jackson CM (2009) The composition of ‘naturally coloured’
late Roman vessel glass fromBritain and the implications formodelsof glass production and supply. J Archaeol Sci 36:198–204
Foy D (2008) Les Revêtements muraux en verre à la fin de l’Antiquité:Quelques témoignages en Gaule méridionale. Journal of GlassStudies 50:51–65
Foy D, Picon M, Vichy M, Thirion-Merle V (2003) Caractérisation desverres de la fin de l'Antiquité en Méditerranée occidentale:l'émergence de nouveaux courants commerciaux. In: D. Foy, M.-D. Nenna (eds.), Échanges et commerce du verre dans le mondeantique, Actes du Colloque de l'AFAV, Aix-en-Provence etMarseille 2001, Monographies Instrumentum 24, ÉditionsMonique Mergoil, pp. 41–85
Freestone IC (1987) Composition and microstructure of early opaque redglass. In: M. Bimson, I.C. Freestone (eds.), Early vitreous materialsBritish Museum Occasional Paper, 56, British Museum, pp. 173-191
Freestone IC (1994) Appendix: chemical analysis of ‘raw’ glass frag-ments. In: H.R. Hurst (ed.), Excavations at Carthage. The BritishMission, vol. II,1, Oxford University Press, p. 290
Freestone IC (2005) The provenance of ancient glass through composi-tional analysis. Mater Res Soc Symp Proc 852
Freestone IC, Hughes, MJ (2006) The origins of Jarrow glass. In: R.Cramp (ed.), Wearmouth and Jarrow monastic sites vol. 2. EnglishHeritage, pp. 147–155
Freestone IC, Bimson M, Buckton D (1990) Compositional categories ofByzantine glass tesserae. In: Annales du 11e Congrès del’Association Internationale pour l’Histoire du Verre, Bâle 1988.Amsterdam, pp. 271–281
Freestone IC, Gorin-Rosen Y, Hughes MJ (2000) Primary glass fromIsrael and the production of glass in late Antiquity and the earlyIslamic period. In: M.-D. Nenna (ed.), La route du verre. Ateliersprimaires et secondaires du second millénaire av. J.-C. au MoyenÂge, Travaux de la Maison de l’Orient Méditerranéen, 33, 65–83
Gliozzo E, Santagostino Barbone A, Turchiano M, Memmi Turbanti I,Volpe G (2012) The coloured tesserae decorating the vaults of theFaragola’s balneum (Ascoli Satriano, Foggia, southern Italy).Archaeometry 54:311–331
Gliozzo E, Santagostino Barbone A, D’Acapito F (2013) Waste glass,vessels and window-panes from Thamusida (Morocco): grouping
Archaeol Anthropol Sci
natron-based blue-green and colourless Roman glasses.Archaeometry 55:609–639
Gliozzo E, Lepri B, Saguì L, Turbanti Memmi I (2015a) Colourless glassfrom the Palatine and Esquiline hills in Rome (Italy). New data onantimony- and manganese-decoloured glass in the Roman period. JArchaeol Anthropol Sci. doi:10.1007/s12520-015-0264-1
Gliozzo E, Turchiano M, Giannetti F, Santagostino Barbone A (2015b)Late Antique glass vessels and production indicators from the townof Herdonia (Foggia, Italy). New data on CaO-rich / weak HIMTglass, Archaeometry (in press). doi:10.1111/arcm.12219
Jackson CM (2005) Making colourless glass in the Roman period.Archaeometry 47:763–80
Kamber BS, Greig A, Collerson KD (2005) A new estimate for the com-position of weathered young upper continental crust from alluvialsediments, Queensland, Australia. Geochim Cosmochim Acta 69:1041–1058
Mirti P, Casoli A, Appolonia L (1993) Scientific analysis of Roman glassfrom Augusta Praetoria. Archaeometry 35:225–240
Nenna MD (2014) Egyptian glass abroad: HIMT glass and its market. In:D. Keller, J. Price, C. Jackson (eds.), Neighbours and successors ofRome, Oxbow Books, pp. 177–193
Panella C (ed) (2011) I segni del potere. Realtà e immaginario dellasovranità nella Roma imperiale, Edipuglia
Panella C, Saguì L, Casalini M, Coletti F (2010) Contesti tardoantichi diRoma: una rilettura alla luce di nuovi dati. In: Menchelli S, SantoroS, Pasquinucci M, Guiducci G (eds) LRCW3. Late Roman CoarseWares, Cooking Wares and Amphorae in the Mediterranean.Archaeology and archaeometry. Comparison between western andeastern Mediterranean, BAR Int Ser 2185, I-II. Archaeopress,Oxford, pp. 57–78
Panella C, Zeggio S, Ferrandes A (2014) Lo scavo delle pendici nord-orientali del Palatino tra dati acquisiti e nuove evidenze. Scienzedell’Antichità 20(1):159–204
Rosenow D, Rehren T (2014) Herding cats – Roman to Late Antiqueglass groups from Bubastis, northern Egypt. J Archaeol Sci 49:170–184
Saguì L (2009) Ateliers de verre gravé à Rome au IVe siècle ap. J.-C.:nouvelles données sur le verre gravé Bà relief negatif^. In: Annalesdu 17e Congrès de l’Association Internationale pour l’Histoire duVerre, Anvers 2006, University Press Antwerp, pp. 206–216
Santagostino Barbone A, Gliozzo E, Turchiano M, D’Acapito F, MemmiTurbanti I, Volpe G (2008) The sectilia panels of Faragola (Ascoli
Satriano, southern Italy): a multi-analytical study of the red, orangeand yellow glass slabs. Archaeometry 50:451–473
Schibille N (2011) Late Byzantine mineral soda high alumina glassesfrom Asia Minor: a new primary glass production group, doi: 10.1371/journal.pone.0018970
Schibille N,Marii F, Rehren T (2008) Characterization and provenance oflate antique window glass from the Petra church in Jordan.Archaeometry 50:627–642
Shugar AN (2000) Byzantine opaque red glass tesserae from Beit Shean,Israel. Archaeometry 42:375–384
Silvestri A (2008) The coloured glass of Iulia Felix. J Archaeol Sci 35:1489–1501
Silvestri A, Molin G, Salviulo G (2005) Roman and medieval glass fromthe Italian area: bulk characterization and relationships with produc-tion technologies. Archaeometry 47:797–816
Silvestri A, Molin G, Salviulo G (2008) The colourless glass of IuliaFelix. J Archaeol Sci 35:331–341
Silvestri A, Tonietto S,Molin G, Guerriero P (2014) The palaeo-Christianglass mosaic of St. Prosdocimus (Padova, Italy): archaeometric char-acterisation of tesserae with copper- or tin-based opacifiers. JArchaeol Sci 42:51–67
Tiepolo M, Bottazzi P, Palenzona M, Vannucci R (2003) A laser probecoupled with ICP-double focusing sector-field mass spectrometerfor in situ analysis of geological samples and U-Pb dating of zircon.Can Mineral 41:259–272
Tite M, Pradell T, Shortland A (2008) Discovery, production and use oftin-based opacifiers in glasses, enamels and glazes from the late IronAge onwards: a reassessment. Archaeometry 50:67–84
Uboldi M (1999) I vetri. In: Brogiolo GP (ed.) S. Giulia di Brescia gliscavi dal 1980 al 1992. Reperti preromani, romani e alto medievali.All’Insegna del Giglio, pp. 271–307
van Achterbergh E, Ryan CG, Jackson SE, Griffin WL (2001) Datareduction software for LA-ICP-MS. In: Sylvester, P. (ed.), Laser-ablation-ICPMS in the Earth Sciences: Principles and Applications,vol. 29. Mineralogical Association of Canada, pp. 239–243
Verità M (1999) Analisi di reperti vitrei e scarti di lavorazione di tarda etàromana provenienti dagli scavi del monastero di Santa Giulia aBrescia. In: Brogiolo GP (ed), S. Giulia di Brescia gli scavi dal1980 al 1992. Reperti preromani, romani e alto medievali.All’Insegna del Giglio, pp. 309–314
Verità M, Arena MS, Carruba AM, Santopadre P (2008) Materiali vitreinell’opus sectile di Porta Marina (Ostia antica). Bollettino ICR 16–17:78–94
Archaeol Anthropol Sci