See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/306000319

Early Byzantine Glass Weights: Aspects of Function, Chronology and

Composition

Article · December 2015

CITATIONS READS

5 2,035

2 authors, including:

Andrew Meek
British Museum
51 PUBLICATIONS 222 CITATIONS

SEE PROFILE

All content following this page was uploaded by Andrew Meek on 15 December 2016.

The user has requested enhancement of the downloaded file.

Technical
Research
Bulletin
Volume 9
2015
Early Byzantine glass
SUMMARY This contribution presents a selection from the British
Museum’s collection of Byzantine glass weights, which comprises

weights: aspects of
180 pieces. Four distinct categories from within the overall corpus
have been chosen. In certain instances, some of these types can

function, typology and

be dated to the specific years of the primary issuing authority, the
eparch or prefect of Constantinople or to the reigns of individual

composition
emperors; others represent distinct regional types and secondary
production centres, perhaps under the control of different provincial
officials. They range in date from around ad 500 to the middle of

Chris Entwistle and Andrew Meek the seventh century. Some 44 of these glass weights were analysed
using scanning electron microscopy with energy dispersive X-ray
analysis and all were found to have been produced with soda-
lime-silica glasses made with a natron flux. The glasses could be
associated with previously established compositional types based
on minor variations in their chemical composition. ‘Strong’ HIMT,
Egypt I and Egypt II glasses were not found, adding evidence to
their suggested usage patterns during this period. Two common
glass types known to have been in circulation at this time, ‘weak’
HIMT and Levantine I, were both found in the assemblage. Some
other glass types that may result from variations in the recipes used
for these glass types, or the recycling of glass, were also found.
There are, however, few clear links between dating or provenance
and compositional type in this assemblage, suggesting that the
producers of these weights obtained their supplies of glass from
various sources.

Introduction
Towards the end of the fifth century and during the first
decades of the sixth century ad, the Early Byzantine
Empire in the eastern Mediterranean witnessed a series
of widespread fiscal and administrative reforms initiated
by the emperor Anastasius (ad 491–518) and continued
by Justinian I (ad 527–565). These included, inter alia, a
revamping of the copper coinage that saw the introduction
of major denominations such as the follis and half follis,
and an increase in the number of mints, which subsequently
led to a proliferation in the circulation of the gold coinage,
exemplified by increased releases of the nomisma and its
divisions, the semissis and tremissis [1; p. 4]. Further inno-
vations included the introduction of the five-stamp system
on silver vessels, which utilized both imperial and officials’
busts and monograms as a means of validation. It is against
this background of monetary reform that the initial intro-
duction of glass as a material for commodity and coinage
weights may have taken place. Unlike their Arabic counter-
parts, issued in vast numbers by successive dynasties from the
Umayyads to the Ayyūbids [2], early Byzantine glass weights
have not been the subject of systematic study. Indeed many
questions relating to their function, origin and date remain
unresolved. This contribution, which presents the results of
compositional analyses on 44 dated examples in the British
Museum’s collection, touches tangentially on some of these
issues. Its main focus, however, is an attempt to determine
the geographic origin of the four types discussed in detail
below: types A and B, which have long been considered to be
Constantinopolitan in origin, and types C and D, which are
thought either to represent provincial officials, and therefore
made outside the capital, or to have been made in Egypt
between ad c.642 and 692 [3, 4].

1
 Figure 1. Distribution map of Byzantine glass weights

centuries ad [8]. This conclusion is further supported by two

Function other statistical studies. A frequency table based on 222 exam-
Glass had obvious advantages as a material for weights: it was ples from various published sources, as well as unpublished
a relatively inexpensive medium, easy to manufacture, readily weights in the American Numismatic Society Museum in
detectable if tampered with and, unlike some metal counter- New York, found three distinct peaks: 37 examples weighing
parts, not prone to oxidation or corrosion. Byzantine glass between 4.41 and 4.60 g; 34 examples ranging from 2.01 to
weights are remarkably homogenous in appearance as the 2.20 g; and 19 specimens between 1.40 and 1.60 g [9; p. 78]. A
vast majority take the form of a disk with a central impressed second study, of the 158 examples in the De Menil Collection,
area enclosed by a rolled rim. A number of stages are appar- Houston, showed three similar peaks: 21 examples between
ent in their manufacture. A gob of molten glass was probably 4.31 and 4.50 g; 18 between 2.11 and 2.30 g; and 17 between
poured from a ladle onto a flat marble or metal surface. Then 1.31 and 1.50 g [10].
an intaglio die bearing the relevant design was pressed into the In addition to the statistical evidence, Byzantine glass
centre of the hot glass. Downward pressure from the die would weights frequently occur in archaeological contexts that leave
cause the gob to flatten outwards forming the rolled rim as an little doubt as to their function. Only a few can be mentioned
edge. The die employed was probably of metal as occasional here, but among the most important is the grave of an Avar
iron residues in the form of flakes and scale on the fronts and goldsmith discovered at Kunszenmárton in Hungary in 1932
backs of the weights indicate that the dies (and probably the [11; pp. 41–47, plates IV–VI], whose grave goods included
plates) were of this material [5, 6].1 The irregular shapes of equal-arm balances with pans together with nine Byzantine
the disks illustrate that they were not poured into depressions weights, five of copper alloy and four of glass. Similar jux-
in the plate prior to stamping. Any examples that proved to tapositions with weighing equipment can be observed at a
be severely over or under weight were probably remelted as number of other sites, including two further excavations in
cullet and reused. Hungary at the Avar cemeteries of Jutas [12; p. 32, plate
While it has generally been held that the majority of these VIII] and Pókaszepetk [13; pp. 423–426, plates 2 and 4].
glass disks were intended as coin weights, their precise function Comparable groups have been preserved in more obvious
has occasionally been disputed on the grounds that they do commercial contexts. One of a complex of 19 shops (E14) that
not always correspond exactly with known coin denominations lined the north side of the road known as the main avenue at
[7; pp. 84–85]. This counter-argument implies that they were Sardis yielded three glass weights all stamped with cruciform
originally intended to be highly accurate measures of coins monograms: these were found with two square, one nomisma
and not, as contemporary coin balances would suggest, simply bronze weights and a bronze half nomisma discoid weight.
rule-of-thumb weights for checking the tolerance above or Further finds inside the shop included a copper alloy steelyard
below which a coin would not be accepted. An unpublished with a lead weight and fragments of balance beams and pans
statistical analysis of over 500 glass weights suggests that the [14; p. 86, figs 464–468 and 477–482]. Similar glass weights
majority were intended to weigh the gold nomisma/solidus were also found in Room N of a ‘residential unit’ in Sector
(c.4.55 g), semissis (c.2.27 g) or tremissis (c.1.52 g), the three MMS/S in a western ‘suburb’ of Sardis [15; pp. 105–118].
principal gold coins in circulation during the sixth and seventh This discovery constituted the largest known discovery of

2 | Chris Entwistle and Andrew Meek

Byzantine weights from a discrete archaeological context.
a b
The hoard was composed of 21 greenish-blue glass weights
all stamped with the same cruciform monogram. Of great
interest is the metrology of this group: of the 21 examples,
12 fall in the range 4.15 to 4.36 g, four between 2.03 and 2.16
g and three between 1.40 and 1.42 g. The remaining two
weigh 3.91 and 3.85 g respectively, but both are damaged.
Statistically the three major groups again cluster quite con-
vincingly around the weights of the nomisma, semissis and
tremissis. Further south, the collapse of the second storey of
a workshop preserved a similar assemblage at Tel Naharon
in Israel. Finds at this site included four copper alloy scales – c d

three of which were balances and the other a steelyard – and

several sets of weights of glass, bronze and haematite. All the
coins found in the workshop dated to the sixth century ad [16].
Finally, at the unpublished site of Shiqmona, also in Israel,
eight discoid copper alloy weights were excavated, together
with three of glass, the latter all issued by the eparch Flavios
Gerontios, prefect of Constantinople in ad c.561: Nos 21–22
in the appendix.

Typology e f
Well over 20 different types of Byzantine glass weight have
survived, but only four categories are considered here. These
have been selected because of the relative ease with which they
can be dated, sometimes to the reigns of individual emperors
or the terms of office of particular prefects of the capital. In
addition, all four types are thought to derive from specific
locations within the wide area in southeastern Europe and the
eastern Mediterranean throughout which glass weights were
distributed, Figure 1. Figure 2. Six glass weights stamped with imperial images,
Constantinople, ad c.500–650: (a) 1920,1104.1; (b) 1980,0611.2; (c)
1986,0406.18; (d) 1987,0703.21; (e) 1990,0601.18; and (f) 1984,0108.2.
Type A: ‘imperial’ weights Not to scale
This category is loosely defined by the presence of one or more
nimbed imperial busts, sometimes in conjunction with other
busts (the prefects of Rome and Constantinople or Christ) or
a b
accompanied by inscriptions or monograms, Figure 2. Given
the rather worn condition of many of these weights, details
of portraiture and costume rarely survive, although in some
instances items of jewellery – such as diadems and pendilia – can
be discerned. The imperial busts and monograms that appear
on type A weights show close parallels with those that appear
as stamps on the reverses of sixth and seventh century ad
silver vessels and serve a similar function, lending the weights
an official character by suggesting governmental validation
through the use of the imperial image [17; tables I–III].
Figure 3. Two glass weights stamped with a bust of the prefect of
Constantinople holding a mappa, Constantinople, ad c.500–650: (a)
Type B: weights with the bust of an eparch 1884,0509.13; and (b) 1990,0601.20. Not to scale
and inscription
This category is normally represented by weights stamped
with the frontal bust of an eparch enclosed by an identifying by both Flavios Gerontios and Flavios Zimarchos fall into this
Greek inscription, Figure 3. The eparch is generally depicted category: Nos 21–23 in the appendix. Sometimes the honorific
holding a mappa – the linen handkerchief thrown to indicate ενδοξοτατου (most blessed) is included, with a superscript ‘s’ as a
the start of races in the circus games – or a sceptre, or occa- contraction for the last five letters. A final variant, represented
sionally both. As with the busts employed on weights of type A, by a rare group of weights, employs the phrase PWMHC
details of portraiture and clothing are rarely evident. Various (Ρωμης; ‘of [New] Rome’, i.e. Constantinople) in columnar
inscriptions are employed on this type of weight: the standard form in the field: No. 20 in the appendix.
formula is prefaced with the word Επι (in the time of) followed The names of 22 different eparchs have been recorded
by the eparch’s name and title in the genitive case. Occasionally on this category of weight. Of these six can reasonably be
the title is omitted, but is clearly to be inferred: weights issued identified with sixth-century eparchs: Demosthenos in the

Early Byzantine glass weights: aspects of function, typology and composition | 3

 letters could be affixed internally or externally to the vertical,
a b
horizontal or diagonal bars. The cruciform monogram, as its
name implies, takes the form of a cross, with letters placed
at the ends of the arms. Sometimes the arms of the cross
emanate from a central ‘theta’, ‘pi’ or ‘phi’. Unlike in the
box type monogram, the ‘omicron’ and ‘upsilon’ are generally
combined in a υο-shaped formation placed at the top of the
vertical bar. The earliest Byzantine monograms to be found
on weights seemingly date to the third quarter of the fifth
century ad, when their emergence was possibly influenced
by the slightly earlier appearance of a rather basic form of
c d box type monogram on the bronze nummi of Theodosios II,
dating to the early 440s [20; p. 148, pl. 17]. It should be noted,
however, that these monograms, which first appear on copper
alloy exagia solidi of the eastern emperors Marcian (ad 450–
457) and Leo (ad 457–474), differ quite markedly from what
might be considered the typical sixth century ad monogram
described above in that they employ mainly Latin letters [21;
pp. 20–21]. There is a general consensus that the box type was
predominant in the period ad c.500–550, with the cruciform
Figure 4. Four glass weights stamped with box and cruciform
type gradually superseding it through the course of the second
monograms, eastern Mediterranean, ad c.500–650: (a) 1980,0611.8; half of the century.2
(b) 1984,0406.3; (c) 1987,0703.10; and (d) 1980,0611.34. Not to scale Hundreds of different monograms have survived on this type
of weight: in some instances they resolve as the same name, a
popular example being ‘of Timotheos’ here represented by two
a b
different forms (Figure 5), but the vast majority, insofar as it is
possible to decipher them with any reliability, seem to resolve
as individual names. The accepted chronology for these glass
weights is that their use lasted for around 150 years, from the
reign of Anastasius to the Persian and Arab invasions of Syria
and Egypt in the 630s and 640s ad. Given that it is known
from prosopographical evidence that the office of eparch of
Constantinople was held by the same person on more than one
occasion (e.g. Theodoros Teganistes is known to have held the
office at least four times in the late 510s/early 520s [22]), there
Figure 5. Two glass weights stamped with different cruciform
monograms resolving as the name ‘of Timotheos’, eastern
were probably only around 100 different individuals who were
Mediterranean, ad c.550–650: (a) 1987,0703.8; and (b) 1980,0611.33. prefects of the capital during these 150 years. As the weights bear
Not to scale hundreds of different individual box or cruciform monograms
that resolve as discrete names, these monograms must refer to
late 510s; Theodoros on four occasions in the late 510s or officials other than simply the prefect of the capital. Although
early 520s; Sergios during the latter half of the reign of there is a possibility that some weights were issued by officials
Justinian; Ioannes either with Ioannes Kokkorobios in ad 550, who were subservient to the prefect of Constantinople – the
or another Ioannes thought to have been prefect during the ninth-century Book of the Eparch refers to a boullotai [23; p.
reign of Maurice Tiberios (ad 582–602); Flavios Gerontios 317], literally ‘inspector of seals’ – it is more likely that they were
between ad 560 and 562; and Flavios Zimarchos on two occa- issued by the prefects of the major cities of the Late Antique
sions, the second being in ad 565. A further seven correspond world such as Antioch and Alexandria.
to known prefects of the first half of the seventh century:
Damianos under Heraklios before ad 630; Eulampios between Type D: ‘Arab-Byzantine’ weights
approximately ad 612 and 654; Kosmas in ad 609; Leontios in In 1948 Jungfleisch published an unusual glass weight with
ad 603; Rogatos early in the reign of Heraklios; Theodoros in no inscription or monogram but decorated with two imperial
ad 612; and Theopemptos in ad 605 or 607. The known date busts – tentatively identified as Constans and Constantine IV
range thus spans the reigns between Justin I (ad 521–527) and (ad 654–668) – holding a cross between them [3]. Pointing
Heraklios (ad 610–641) [18, 19]. out various iconographic anomalies with other Byzantine
‘imperial’ weights, Jungfleisch suggested that it was an Arabic
Type C: weights with a box or cruciform monogram copy and coined the phrase ‘arabo-byzantine’ to describe it.
By far the most common form of glass weight is that stamped He further argued that, after the fall of Egypt to the Arabs
with either a box/block or cruciform monogram, Figure 4. in ad 640–642, the Byzantine administrative system was so
The former is generally composed of a central ‘eta’, ‘mu’, ‘nu’ totally disrupted that the issuance of weights by the relevant
or ‘pi’, with the letters ‘omicron’ and ‘upsilon’, indicating the Byzantine authorities ceased. In the interval between then and
genitive case, placed at the apex of the vertical strokes. Other the introduction of a series of purely Arabic glass weights by

4 | Chris Entwistle and Andrew Meek

 a b a b

Figure 6. Two glass weights stamped with diademed busts with Figure 7. Two glass weights stamped with the bust of an eparch
c d
crosses in the field, Egypt, ad c.650–700: (a) 1980,0611.61; and (b) enclosed by a pseudo-Cufic inscription, Egypt, ad c.650–700: (a)
1872,1201.9. Not to scale 1980,0611.51; and (b) 1980,0611.52. Not to scale

Table 1. Selected previously published compositional types of natron-based glass from the first millennium ad
Type Site Na2O MgO Al2O3 SiO2 K2O CaO TiO2 MnO FeO
Levantine I Apollonia 15.2 0.63 3.05 70.6 0.71 8.07 <0.1 <0.1 0.35
Levantine II Bet Eli’ezer 12.1 0.63 3.32 74.9 0.46 7.16 <0.1 <0.1 0.52
Egypt I Wadi Natrun 21.4 0.90 2.70 66.5 0.34 3.70 0.27 <0.1 1.00
Egypt II Ashmunein 15.0 0.50 2.10 68.2 0.20 10.8 0.28 0.20 0.70
HIMT Carthage 18.7 1.29 3.18 64.8 0.44 5.24 0.68 2.66 2.07
Weak HIMT Bubastis 17.9 0.93 2.32 66.1 0.54 7.12 0.15 1.05 0.83
HIT Dichin 17.5 1.09 3.16 68.0 0.38 5.17 0.67 0.05 1.42

Note: Based on data from [25–27, 29, 30, 32].

the reforming caliph Abd’ al Malik in ad 692, someone must whether these weights were imitating imperial or eparchal
have assumed administrative responsibilities for the produc- busts, but it is clear from a number of examples in the British
tion of weights. While Byzantine glass weights would have Museum’s collection that they were probably copying both.
remained in circulation for some time, eventually most would Examples of the former include two weights whose busts bear
have become unusable through a combination of breakage crude radiate crowns and pendilia, Figure 6. There are no iden-
and wear. Jungfleisch argued that this lacuna was filled by tifying inscriptions, but the busts are flanked by two crosses.
‘Coptic’ merchants issuing weights on their own authority, The latter type is represented by two very unusual weights,
but using as their prototypes the old Byzantine series. Balog both with a frontal bust holding up an object in either hand,
subsequently developed Jungfleisch’s thesis and proposed presumably a mappa and a sceptre (Figure 7: Nos 41–42 in the
three different categories of ‘arabe-byzantine’ weights with: appendix). Enclosing the bust is an inscription that is, unfor-
a debased monogram; a denominational mark; or with a bust tunately, very blurred but which might be in Greek, although
but no inscription [4]. there are hints that the letters are Cufic. If correct, this would
Of these three categories only the third merits serious con- be the clearest evidence for ‘arabe-byzantine’ glass weights.
sideration. The problem with Balog’s first type is that all the
examples of debased monograms cited by him are of the box Composition
rather than the cruciform type. It seems unlikely that Coptic
merchants would be copying a type of monogram which, if Introduction
the general consensus on its dating is correct, had been out of In recent years a large number of first millennium ad glass
widespread use for nearly 50 years. His argument would have assemblages has been analysed compositionally and published,
been more convincing had the debased monograms been of and compositional types have been defined based on various
the cruciform type.3 At the time of publication Balog knew of characteristics associated with the raw materials used to make
only one example of his suggested second category – weights the glass [24]. The major component of these glasses, and that
with a denominational mark – although many more examples most commonly used to differentiate between types, is sand,
have been published subsequently that include both commod- with impurities in the silica-rich sand sources giving rise to
ity and coinage weights. There seems no reason why weights the majority of the compositional differences seen in Table 1.
with a mark alone should be classified as ‘arabe-byzantine’ Some of the types have chronological or geographical limits
as they simply imitate the most common type of Byzantine that can be used to draw conclusions about the glass objects
metal weight. analysed. The glass weights analysed in this study date to the
A more compelling case can perhaps be made for Balog’s sixth and seventh centuries ad, with production attributed
third category – those weights stamped with a bust only. to Constantinople, the eastern Mediterranean and Egypt.
His examples are almost totally lacking in facial features or The most significant natron-based glass types in circula-
details of costume and insignia, very often representing a mere tion during this period, and at these locations, are termed
approximation of the human form. Balog does not discuss Levantine I, Levantine II, HIMT (with high manganese, iron

Early Byzantine glass weights: aspects of function, typology and composition | 5

 Table 2. Accuracy and standard deviation measurements (wt%) for the Corning glass standard A
Corning A Na2O MgO Al2O3 SiO2 P2O5 SO3 Cl K2O CaO TiO2 MnO FeO CoO NiO CuO ZnO Sb2O3 PbO
Known 14.52 2.81 1.01 66.56 0.14 0.16 0.10 2.93 5.30 0.80 1.18 1.09 0.15 0.03 1.22 0.04 1.72 0.08
Measured (n = 14) 14.78 2.63 0.83 66.85 0.08 0.24 0.15 2.97 5.05 0.98 1.04 0.98 0.17 0.03 1.24 0.04 1.55 0.08
Accuracy 1.8 6.8 21.7 0.4 75.0 33.3 33.3 1.4 5.0 18.4 13.5 11.2 11.8 0 1.6 0 11.0 0
Standard deviation 0.08 0.03 0.01 0.15 0.03 0.05 0.01 0.03 0.04 0.02 0.02 0.03 0.02 0.02 0.03 0.02 0.13 0.05

6 | Chris Entwistle and Andrew Meek

Table 3. Normalized VP-SEM-EDX results for the glass weights
No. in appendix Production place Colour Date Type n Na2O MgO Al2O3 SiO2 P2O5 SO3 Cl K2O CaO TiO2 MnO FeO CoO CuO ZnO PbO Glass type
1. 1980,0611.2 Constantinople (?) Green 582–602 A 2 19.6 1.2 2.9 64.3 0.2 0.5 0.8 1.0 7.5 0.2 0.4 1.3 <0.1 <0.1 <0.1 <0.2 wHIMT (low Mn)
2. 1990,0601.16 Constantinople (?) Olive green 582–602 A 2 19.6 1.3 3.2 63.8 0.2 0.5 1.0 1.0 6.5 0.2 0.8 1.9 <0.1 <0.1 <0.1 <0.2 wHIMT (high Fe)
3. 1920,1104.1 Constantinople (?) Green 540–565 A 4 19.3 1.2 2.4 65.7 0.1 0.5 0.9 0.6 6.8 0.1 1.3 0.9 <0.1 <0.1 <0.1 <0.2 wHIMT
4. 1997,0218.1 Constantinople (?) Pale brown 540–565 A 2 20.4 1.1 2.3 64.7 0.1 0.7 1.0 0.8 6.7 0.1 1.3 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT
(green)
5. 1986,0406.18 Constantinople (?) Green 540–565 A 2 21.7 1.0 2.2 63.1 0.1 0.4 1.0 0.5 7.0 0.2 1.5 1.1 <0.1 <0.1 <0.1 <0.2 wHIMT
6. 1980,0611.60 Constantinople (?) Brown (602–610?) A 2 13.9 3.3 2.2 69.8 <0.1 0.3 0.1 1.1 6.4 0.1 <0.1 0.2 <0.1 <0.1 1.3 <0.2 Modern
7. 1997,0218.4 Constantinople (?) Blue 550–600 A 2 18.2 1.0 2.4 64.9 0.1 0.4 0.9 0.6 6.6 0.1 1.2 1.0 0.1 0.3 <0.1 2.2 wHIMT
8. 1980,0611.30 Constantinople (?) Blue 550–600 A 2 19.2 0.9 2.4 64.1 0.1 0.4 0.8 0.6 6.9 0.1 1.2 1.2 0.1 0.3 <0.1 1.4 wHIMT
9. 1893,0409.2 Constantinople (?) Blue 550–600 A 2 19.8 1.0 2.5 64.3 0.1 0.4 0.8 0.6 6.6 0.1 1.1 1.1 0.1 0.3 <0.1 1.3 wHIMT
10. 1990,0601.17 Constantinople (?) Red-brown/ 550–600 A 2 19.2 1.2 2.9 63.7 0.1 0.4 0.8 0.7 7.3 0.2 1.3 1.9 0.1 0.2 <0.1 <0.2 wHIMT (high Fe)
black
11. 1884,0509.13 Constantinople (?) Blue 510s B 2 18.2 1.1 2.3 67.4 0.1 0.5 0.9 0.6 7.3 0.2 0.2 1.0 0.1 0.1 <0.1 <0.2 wHIMT (low Mn)
12. 1987,0703.20 Constantinople (?) Blue 510s B 2 18.6 1.2 2.2 66.9 0.1 0.4 0.9 0.7 7.0 0.2 0.2 1.2 0.2 0.2 <0.1 0.2 wHIMT (low Mn)
13. 1980,0611.48 Constantinople (?) Green 522–523 B 2 16.9 0.9 2.7 68.4 0.1 0.5 0.9 0.7 5.2 0.2 1.1 2.3 <0.1 <0.1 <0.1 <0.2 wHIMT (high Fe)
14. 1986,0406.12 Constantinople (?) Green 522–523 B 2 17.6 1.5 2.8 63.0 0.2 0.5 0.9 1.2 7.9 0.2 1.3 2.8 <0.1 <0.1 <0.1 <0.2 wHIMT (high Fe)
15. 1891,0512.11 Constantinople (?) Green 522–523 B 2 18.2 1.1 2.8 65.3 0.2 0.5 0.9 0.8 6.2 0.2 1.1 2.6 <0.1 <0.1 <0.1 <0.2 wHIMT (high Fe)
16. 1986,0406.15 Constantinople (?) Aqua 518–526 or 612 B 2 17.5 1.1 3.5 66.5 0.1 0.6 0.4 0.9 8.0 0.1 0.5 0.7 <0.1 <0.1 <0.1 <0.2 Levantine I / wHIMT
17. 1997,0218.2 Constantinople (?) Blue 518–526 or 612 B 2 19.3 1.1 2.3 63.4 0.3 0.4 0.9 0.7 7.9 0.2 1.3 1.1 0.1 0.2 <0.1 0.7 wHIMT
18. 1892,0613.61 Constantinople (?) Pale green 582–602 B 2 18.9 1.1 2.4 65.3 0.1 0.5 1.0 0.8 7.4 0.2 1.5 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT
19. 1987,0703.19 Constantinople (?) Pale green 582–602 B 2 20.9 1.1 2.3 63.5 0.1 0.5 1.0 0.8 7.5 0.2 1.2 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT
20. 1884,0509.14 Constantinople (?) Pale green 582–602 B 4 14.9 0.7 3.5 70.4 <0.1 0.2 1.0 0.5 8.1 0.1 <0.1 0.5 <0.1 <0.1 <0.1 <0.2 Levantine I
21. 1892,0613.59 Constantinople (?) Green 560–562 B 2 18.2 1.1 2.7 66.9 0.2 0.5 0.8 0.9 6.4 0.2 1.4 0.8 <0.1 <0.1 <0.1 <0.2 wHIMT
22. 1892,0613.60 Constantinople (?) Green 560–562 B 2 16.7 1.2 2.8 67.2 0.1 0.4 0.8 0.9 7.5 0.2 1.1 0.9 <0.1 <0.1 <0.1 <0.2 wHIMT
23. 1987,0703.18 Constantinople (?) Light green 565 B 2 16.6 1.3 3.8 66.4 0.2 0.4 0.3 1.8 6.3 0.2 1.1 1.3 <0.1 <0.1 <0.1 <0.2 wHIMT
24. S.322 Constantinople (?) Blue-green 609 B 2 16.2 0.9 3.2 67.4 0.1 0.2 0.9 0.5 10.0 0.1 <0.1 0.5 <0.1 <0.1 <0.1 <0.2 Levantine I
25. 1987,0703.17 Constantinople (?) Green 609 B 2 15.4 0.8 3.2 69.8 <0.1 0.1 1.0 0.6 8.5 0.1 <0.1 0.4 <0.1 <0.1 <0.1 <0.2 Levantine I
26. 1879,0522.49 Constantinople (?) Aqua 610–612 B 4 16.3 0.7 3.2 66.3 0.2 0.2 0.7 0.9 10.8 0.1 <0.1 0.4 <0.1 <0.1 <0.1 <0.2 Levantine I
27. 1980,0611.90 Constantinople (?) Green 600–650 B 2 18.9 1.0 2.9 66.6 0.1 0.4 0.7 0.8 7.1 0.1 0.6 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT (low Mn)
 28. 1983,1108.1 Constantinople (?) Blue-green 610–630 B 2 19.4 1.0 2.3 65.9 <0.1 0.5 1.0 0.7 7.6 0.2 0.2 0.8 <0.1 0.1 <0.1 <0.2 wHIMT (low Mn)
29. 1882,0510.20 Constantinople (?) Blue 610–630 B 2 21.8 1.1 2.2 64.2 <0.1 0.4 1.0 0.5 7.4 0.1 0.3 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT (low Mn)
30. 1987,0703.4 Eastern Mediterranean Purple 500–599 C 2 18.2 0.6 2.0 69.4 <0.1 0.4 0.9 1.0 4.7 0.1 2.0 0.5 <0.1 <0.1 <0.1 <0.2 ‘Decoloured’
31. 1980,0611.4 Eastern Mediterranean Brown-green 500–599 C 2 18.0 1.1 2.4 66.4 0.1 0.5 0.9 0.9 7.5 0.2 1.3 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT
32. 1990,0601.12 Eastern Mediterranean Blue 500–599 C 2 18.3 1.1 2.5 63.9 0.1 0.5 0.9 0.7 8.9 0.2 0.6 1.9 0.1 0.1 <0.1 0.3 wHIMT (high Fe)
33. 1987,0703.3 Eastern Mediterranean Aqua 500–599 C 2 16.9 0.9 3.2 66.6 0.2 0.1 0.7 1.0 8.6 0.1 0.7 0.7 <0.1 0.1 <0.1 0.2 Levantine I / wHIMT
34. 1980,0611.13 Eastern Mediterranean Dark blue 500–599 C 2 17.2 0.7 3.2 64.0 0.1 0.3 0.8 0.8 10.3 0.1 0.2 1.0 0.1 0.3 <0.1 1.1 Levantine I / wHIMT
35. 1981,0601.3 Eastern Mediterranean Clear 550–650 C 2 18.1 1.1 2.6 66.6 0.1 0.5 0.8 0.7 7.1 0.1 1.4 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT
36. 1984,0109.1 Eastern Mediterranean Light green 550–650 C 2 17.8 1.1 2.7 66.5 0.1 0.3 0.6 1.4 7.8 0.1 0.7 0.7 <0.1 <0.1 <0.1 <0.2 wHIMT (low Mn)
37. 1891,0512.13 Eastern Mediterranean Green 550–650 C 3 15.2 0.5 2.8 72.4 <0.1 0.1 1.0 0.5 6.8 0.1 <0.1 0.4 <0.1 <0.1 <0.1 <0.2 Levantine I
38. 1987,0703.16 Eastern Mediterranean Brown-green 550–650 C 2 17.6 1.3 2.8 66.3 0.2 0.4 0.8 1.0 6.8 0.2 1.6 0.9 <0.1 <0.1 <0.1 <0.2 wHIMT
39. 1980,0611.21 Eastern Mediterranean Aqua 550–650 C 3 14.4 0.5 3.1 71.2 0.1 0.1 0.7 0.8 8.3 0.1 <0.1 0.4 <0.1 <0.1 <0.1 <0.2 Levantine I
40. 1923,1107.1 Eastern Mediterranean Green 550–650 C 2 19.6 1.3 2.7 64.4 0.2 0.5 1.0 0.9 6.2 0.2 0.9 2.0 <0.1 <0.1 <0.1 <0.2 wHIMT (high Fe)
41. 1980,0611.52 Egypt Blue 642–692 D 2 20.0 1.0 2.3 64.2 <0.1 0.4 0.8 0.6 7.2 0.1 1.2 1.2 0.1 0.3 <0.1 0.6 wHIMT
42. 1986,0406.14 Egypt Blue 642–692 D 2 20.1 1.0 2.4 64.4 0.1 0.3 0.8 0.6 6.9 0.1 1.2 0.8 0.1 0.2 <0.1 0.9 wHIMT
43. 1980,0611.51 Egypt Blue 642–692 D 2 20.5 1.2 2.6 62.5 0.1 0.3 0.8 0.6 7.2 0.1 1.2 2.1 0.1 0.2 <0.1 0.5 wHIMT (high Fe)
44. 1980,0611.71 Egypt Dark blue 642–692 D 2 17.1 0.6 3.1 66.8 0.1 0.2 0.9 0.8 8.1 0.1 0.2 0.8 0.1 0.2 <0.1 0.9 Levantine I / wHIMT

Note. The values are reported as percentages of oxides (except Cl) and are averages from analyses on more than one area: n = the number of areas analysed. Sb2O3 and NiO were not found above the detection limits for this
methodology in any of the objects analysed.

Table 4. Average VP-SEM-EDX data for each of the compositional types found
Type n Na2O MgO Al2O3 SiO2 K2O CaO TiO2 MnO FeO
wHIMT (high Fe) 8 18.74 1.20 2.78 64.36 0.83 6.93 0.16 1.04 2.20
wHIMT 17 19.01 1.09 2.53 65.15 0.78 7.01 0.15 1.28 0.92
wHIMT (low Mn) 7 19.18 1.12 2.50 65.97 0.81 7.38 0.15 0.36 0.91
Levantine I / wHIMT 4 17.17 0.83 3.25 65.99 0.87 8.74 0.10 0.38 0.78
Levantine I 6 15.39 0.70 3.17 69.58 0.64 8.75 0.09 0.03 0.44
‘Decoloured’ 1 18.19 0.64 2.00 69.44 1.04 4.67 0.11 2.00 0.46

Note. The values are reported as percentages of oxides and are averages from analyses on more than one area:
n = the number of areas analysed.

Early Byzantine glass weights: aspects of function, typology and composition | 7

 Figure 8. Plot of wt% calcium oxide against wt% aluminium oxide

Figure 9. Plot of wt% iron oxide against wt% aluminium oxide

Figure 10. Plot of wt% iron oxide against wt% manganese oxide

8 | Chris Entwistle and Andrew Meek

and titanium), Egypt I and Egypt II, Table 1 [25–32]. The published elsewhere (compare Table 1 and Table 4). As the
first two of these groups are associated with production in the weights are believed to have been formed by remelting previ-
Levant and the final two in Egypt, while HIMT is thought to ously produced glass in an open flame, this may be the result of
have been produced in northern Egypt, although the precise contamination by fuel ash during this process: see the section
production location has not been ascertained. on ‘function’ above, and [37] for illustrations of objects pro-
While the Egyptian and Levantine types have been well duced in a similar manner. An alternative explanation for these
established for some time, the group of HIMT types has differences in the potassium levels found in Levantine glasses,
remained a subject of debate since their definition in the 1990s suggested by Tal et al. [38; p. 92], is that the potassium released
[33; p. 290, 34]. There are a number of different subtypes from wood ash during primary glass production may contami-
under the umbrella term HIMT. A ‘weak’ HIMT type (here- nate the glass melt, resulting in variations in the potassium
after termed wHIMT) and the similar HIMT 2 type [31], levels in the glass, even within a single batch.
contain significant but lower levels of iron, manganese and The glasses analysed in this study can be divided into six
titanium [29, 31, 35] as well as other elements. Another type, compositional types, Table 4. The data were assigned to these
without elevated manganese levels, is known as HIT [30]. As types by comparison with previously published data and by
more studies are published, a clearer picture of this family of using scatter plots to map the data, Figures 8–10. The types
glass types will hopefully be revealed. are based on the pre-established categories discussed above
Previous analysis of 38 Byzantine glass pendants from the and are related to the raw materials used to produce the
British Museum’s collection by Röhrs et al. concluded that glasses, Table 1. However, some new sub-types were created
they were produced from a variety of glass compositions, to explain differences between the data in this study and the
including Levantine I, HIMT and coloured glass of Roman pre-established groups.
composition [36]. It was not possible to link these composi- Six of the glasses in the assemblage were found to have
tions to discrete production sites and instead it seemed likely been produced from Levantine I glass, Table 4. This glass type
that those producing these objects obtained their glass from was also found in the Byzantine pendants analysed previously
various primary production locations, either directly or via and its chronological and geographical range, discussed above,
the recycling of glass objects. This study aims to continue made it very likely to be found in the glass weights [36, p. 180].
that work and to improve understanding of the production The largest compositional type represented in this assem-
of Byzantine glass objects. blage is wHIMT. This group coincides with Rosenow and
Rehren’s ‘weak’ HIMT type [29], and shares some similarities
Analytical method with Foster and Jackson’s HIMT 2 type [31]. The glass in one
Variable pressure scanning electron microscopy with energy of the type B weights (1987,0703.18) is exceptional, with very
dispersive X-ray spectrometry (VP-SEM-EDX) analysis was high aluminium and fairly high potassium levels, although the
carried out on the reverse side of the weights. A Hitachi compositional data for other elements in this glass still suggest
S3700N equipped with an Oxford instruments INCA x-act that it is of the wHIMT type, Figures 8–10 and Table 3. The
detector was used. A chamber pressure of 40 Pa was employed most likely explanation for this object’s compositional differ-
so that charge was conducted away from the sample. All the ences is contamination from the ceramic crucible in which it
measurements were carried out at a working distance of was remelted.
10.0 mm, a voltage of 20 kV, a count time of 180 seconds There are three further types that share characteristics with
and count rates of c.12000 counts per second. This analytical wHIMT type glass:
method resulted in detection limits for most metal oxides of
around 0.1 weight percent (wt%). 1. wHIMT (high iron) glasses are similar in composition to
The weights were not sampled as they fitted easily into the wHIMT glasses, but have higher iron levels, similar to
sample chamber of the SEM and no coating was needed for those of HIMT glass (Figures 9 and 10). This may result
analysis at a pressure of 40 Pa. A small area on each object was from the addition of iron to a wHIMT glass or its produc-
polished with 6 μm diamond paste to remove the weathered tion from a sand source different to that used in either the
surface of the glass as this crust would influence the analytical wHIMT or HIMT type glasses.
results; for full details see [36; p. 180]. 2. wHIMT (low manganese) glasses also share almost all
Corning glass standard A was analysed to check for accu- of the chemical characteristics of wHIMT glasses, but
racy and precision, Table 2. The percentage accuracies of have lower manganese levels, Figure 10. The manganese
the results for phosphorus pentoxide (P2O5), sulphur trioxide levels are not as low as those found in HIT glasses, so
(SO3) and chlorine (Cl) are sufficiently high that data for these they represent an intermediate composition between
components must be considered semi-quantitative. wHIMT and HIT [30]. It has been suggested by Rehren
and Cholakova that manganese was an optional addition
Analytical results and discussion to the family of HIMT glasses [30]. It may be the case,
therefore, that these low manganese wHIMT glasses are
Compositional types the result of the addition of less manganese to the glass
All the weights were produced from soda-lime-silica glass and batch than was used for the wHIMT or HIMT glasses.
exhibit low magnesium and potassium contents typical of glass 3. Levantine I/wHIMT glasses have similar calcium and
produced using a sodium-rich mineral known as natron, Table aluminium levels to Levantine I glass, with iron and man-
3. However, the potassium levels in many of the weights in ganese levels comparable to the wHIMT low-manganese
this assemblage are higher than the majority of similar glasses glasses, Figure 8. This may be the result of mixing glasses

Early Byzantine glass weights: aspects of function, typology and composition | 9

 addition of further manganese to an already manganese-
a b
decoloured glass.
The final glass weight to be discussed is 1980,0611.60, the
busts on which suggest that it dates to the early seventh century
ad. However, its composition has characteristics that are not
associated with Byzantine glasses and indicate a modern
production: low chlorine (0.1% Cl), high barium (c.1 wt%
BaO from semi-quantitative analysis not reported in Table 3)
and zinc (1.4% ZnO: Table 3). It is therefore assumed that the
glass used to produce this weight is not ancient.

c d Colours
Green, blue-green, aqua and brown colours in glasses of the
types discussed in this study are normally associated with the
presence of iron and copper, Figure 11. Various shades can be
produced using these two metals by altering the furnace tem-
perature and adjusting the environment to make it more or less
reducing or oxidizing. In the samples analysed here, copper
is only present at levels above 0.1 wt% CuO in glasses that
are coloured a strong blue by the addition of cobalt. The four
e f
colours seen in this assemblage are, therefore, principally the
result of the presence of iron in the glasses. This contrasts with
the pendants analysed previously, where copper was present at
a level above 1 wt% CuO (and not associated with significant
cobalt levels) in six of the 38 pendants analysed [36; p. 180].
The majority of green or green-brown weights were pro-
duced from wHIMT glasses (Table 5); previous studies have
established that shades of green and brown are associated
with glasses of the HIMT compositional family [31; p. 189,
36; p. 183]. As some green weights were also found in the
g h Levantine I group, these colours are not exclusively associ-
ated with wHIMT glasses. Five of the six aqua or blue-green
weights were produced from Levantine I or Levantine I/
wHIMT glasses; the majority of Levantine I glasses analysed
previously were of a blue-green colour [31; p. 190]. The
single blue-green weight produced from wHIMT (low Mn)
glass – 1983,1108.1 – is significantly darker in colour than
the other five.
The blue weights were all produced from wHIMT glasses
Figure 11. Examples of glass colours found in the weights analysed:
(a) lightly coloured – 1981,0601.3; (b) aqua –1879,0522.49; (c) green
(Table 5); 12 of the 13 were coloured with cobalt and are a
– 1990,0601.16; (d) green-brown – 1986,0406.18; (e) dark blue – very dark blue, so dark as to appear almost opaque black. All
1884,0509.13; (f) translucent blue – 1882,0510.20; (g) purple – but one (1884,0509.13) also contain significant levels of lead
1987,0703.4; and (h) red-brown/black – 1990,0601.17. Not to scale (0.2–2.2 wt% PbO). The presence of lead and antimony in
translucent cobalt blue glasses of the post-Roman period is
of these two compositional types or the addition of a raw normally attributed to the use of lead-antimonite-opacified
material rich in manganese and iron to a Levantine I glass. cobalt-containing blue glasses as a raw material [32, 40, 41].
The calcium and aluminium contents and the colours In the case of the blue weights, antimony was not, however,
of these glasses correspond more closely to Levantine I detected at a significant level and previous studies have sug-
glasses than wHIMT, suggesting that the second of these gested that the presence of lead may be associated with the
two options is perhaps more likely. particular cobalt-rich raw material used to colour the glass
[42; pp. 240–242], which seems likely also to be the case in
The single object in the ‘decoloured’ compositional group these glass weights.
is a purple glass weight from the eastern Mediterranean One glass weight (1882,0510.20) is a more translucent
(1987,0703.4). Even though this glass is purple in appearance, blue and does not contain cobalt or copper at a detectable
it is of a compositional type associated with glasses decoloured level. The blue colouration may be due to the presence of
with manganese (see [31] and [39], type 2a). It has low levels very low levels of cobalt, below the detection levels of the
of iron, which discriminates it from all types of HIMT glass, analytical technique employed, but a contribution is likely to
and low calcium and aluminium, which differentiates it from be made by iron, which can produce a blue colour if the glass
Levantine I glass. This glass may have been produced by the is remelted in a particular environment. The difference in
colouration between this object and others with fairly similar

10 | Chris Entwistle and Andrew Meek

Table 5. Distribution of glass colours in the compositional types
Colour wHIMT wHIMT (low Mn) wHIMT (high Fe) Levantine I / wHIMT Levantine I ‘Decoloured’

Green / green-brown 10 3 5 0 3 0
Lightly coloured 1 0 0 0 0 0
Red-brown / black 0 0 1 0 0 0
Purple 0 0 0 0 0 1
Aqua / blue-green 0 1 0 2 3 0
Blue 6 3 2 2 0 0

Table 6. Comparison of compositional types and production locations

Production Production Entwistle weight wHIMT wHIMT wHIMT Levantine I / Levantine I ‘Decoloured’
location date type (low Mn) (high wHIMT
Fe)
Constantinople (?) ad c.500–650 A: ‘Imperial’ 6 1 2 0 0 0
Constantinople (?) ad c.500–650 B: Bust of an eparch 6 5 3 1 4 0
and inscription
Eastern ad c.500–650 C: Box or cruciform 3 1 2 2 2 1
Mediterranean monogram
Egypt ad c.642–692 D: ‘Arab-Byzantine’ 2 0 1 1 0 0

compositions is most likely to be the result of remelting in a B) are produced from wHIMT glass (24 of 28). The same can
reducing environment [43; p. 23]. also be said of weights attributed to Egypt (type D: Table 6).
The single red-brown/black weight (1990,0601.17) appears The two most likely explanations for this are that the produc-
to be a mixture of two different glasses, although its composi- ers in these areas were preferentially using wHIMT glass, or
tion is not significantly different from many of the other glasses that at the time of production this was the main glass type
analysed in this study. Previously, Freestone et al. analysed a available to them.
black glass mosaic tessera that contained streaks of red and The weights from the eastern Mediterranean (type C) show
found that the colouration was caused by the precipitation examples of each compositional type. The small number of
of copper sulphide in the black areas and metallic copper in eastern Mediterranean weights analysed in this assemblage
the red streaks [44; p. 274]. They suggested that variations does not allow any definite conclusions to be drawn about
in melting conditions were able to cause this colouration and the relative popularity of the different glass types, although it
it seems likely that this is also the case for the streaked glass seems that the producers used glasses that originated from a
analysed here [44]. variety of primary sources.
There are two weights that exhibit characteristics associ- Four of the type A weights from Constantinople were
ated with ‘decoloured’ glasses. First, the purple weight produced using the same die (1997,0218.4; 1980,0611.30;
(1987,0703.4), which is discussed in detail above, and second, 1893,0409.2; and 1990,0601.17), suggesting that they were
a pale coloured glass weight (1981,0601.3) produced from made in a single workshop. Weight 1997,0218.4 is produced
wHIMT glass. The most likely explanation for the lack of colour from red-brown/black glass and the other three from blue
is decolouration of the glass by the addition of manganese. As wHIMT glass. The compositions of these three blue glasses
discussed above in relation to the purple weight, manganese are so similar that they may have been produced from a single
can act as a decolourant by counteracting the green/brown glass source, Table 3. The only significant differences are in
colouration caused by iron oxide in the glass. As this weight is their lead and cobalt levels, suggesting that these components
not compositionally distinct from the other wHIMT glasses it may have been added separately to each of the glasses. The
seems likely that, as with weight 1882,0510.20, the environ- red-brown/black weight is produced from wHIMT (high iron)
ment in the furnace during remelting was responsible for the glass. The use of two different glass types in these four objects
difference in colouration [43; p. 35]. shows that the single workshop producing them used more
than one glass type glass to make weights.
Regional and workshop groups
The weights analysed in this study are from three produc- Chronological patterns
tion locations: Constantinople (‘imperial’ weights and weights The weights in this study date to the sixth and seventh centuries
with the bust of an eparch and inscription: types A and B), ad, Tables 3 and 7. There was no correlation between date and
the eastern Mediterranean (weights with a box or cruciform glass type, with every compositional type except ‘decoloured’
monogram: type C) and Egypt (‘Arab-Byzantine’: type D), found in each century; the only ‘decoloured’ glass dated to the
Table 6. The glass types used in these locations are not distinct sixth century ad. This lack of correlation limits the extent to
from each other, but some patterns can be observed. First, the which chronological patterns can be established in the glasses
majority of weights attributed to Constantinople (types A and that were analysed. However, because the weights are securely

Early Byzantine glass weights: aspects of function, typology and composition | 11

Table 7. Chronology of glass types found in this study century ad. That analyses of type C weights – considered to
have been made throughout the eastern Mediterranean in
Glass type Date range
the major provincial centres – show a greater compositional
wHIMT (high Fe) Sixth to late seventh century ad range is perhaps also to be expected, given a need to exploit
wHIMT Sixth to late seventh century ad a wider range of sources for the glass. Finally, the inconclu-
wHIMT (low Mn) First quarter of sixth to mid-seventh century ad sive evidence for weights of type D, which appear to belong
Levantine I / Sixth to late seventh century ad entirely to wHIMT, seems to suggest their manufacture in
wHIMT
northern Egypt, which would fit well with the theory that this
Levantine I Mid-sixth to mid-seventh century ad
type represents a transitional series produced in Egypt in the
‘Decoloured’ Sixth century ad period ad c.642–692.
Further analyses of the remaining 136 glass weights in
dated, determining their compositional types provides useful the British Museum’s collection would be needed to take this
information for future studies. Of particular importance is the research further, and it would be particularly interesting to
discovery that the wHIMT glass type was still in use, in Egypt see if any of the remaining 18 weights of type D show com-
at least, in the mid-to-late seventh century ad, for example in positional characteristics associated with Egypt I type glasses.
the type D weights 1980,0611.52 and 1986,0406.14.
Some more interesting patterns present themselves when Appendix: dated examples of Byzantine
what is absent from this assemblage is considered. First, glass weights
there are no glasses of the ‘strong’ HIMT type (HIMT 1), a
compositional type that was in circulation in the late fourth Type A: weights with imperial busts
to the sixth century ad [29, 31]. Although the majority of 1. 1920,1104.1. Nimbed and diademed bust of Justinian
the weights in the current study date to the sixth century ad, I (ad 527–565) enclosed by a Latin inscription: DN
none is of this glass type, which contrasts with the pattern IVSTINIANVS PP AVC. Diameter (Ø) 25 mm; mass
found for the Byzantine pendants in the British Museum 4.17 g (nomisma).
collection, where 10 of the 37 objects analysed were found 2. 1997,0218.1. Nimbed and diademed bust of Justinian
to have been produced from HIMT glass [36, p. 186]. Those I above a retrograde cruciform monogram resolving
pendants dated from the fourth and early fifth century ad, as ‘of Sergios’ enclosed by a Latin inscription: DN
so it is interesting to note that the compositional differences IVSTINIANVS PP AVC. Ø 27 mm; mass 4.46 g
can potentially be linked to a chronological change in the (nomisma).
production of two similar object types. 3.  1986,0406.18. As No. 2. Ø 25 mm; mass 4.24 g
Second, there are no glasses of the Egypt I or Egypt II (nomisma).
compositional types. This is also an important finding as Egypt 4.  1980,0611.2. Nimbed and diademed imperial bust
I type glass has a likely start date in the late seventh century above a box monogram resolving as ‘of Martinos’;
ad and an end date in the eighth century ad, while Egypt II is dated by bust type to the reign of Maurice Tiberios
associated with the eighth to the ninth century ad. The well- (ad 582–602). Ø 27 mm; mass 4.21 g (nomisma).
dated glass weights analysed here (sixth to late seventh century 5. 1990,0601.16. As No. 4. Ø 18.5 mm; mass 1.35 g
ad) were, therefore, very unlikely to have been produced from (tremissis).
Egypt II type glasses, but it is interesting to note than none is 6. 1980,0611.60. Standing figures of the emperor Phocas
produced from Egypt I glass, reinforcing the existing evidence (ad 602–610) and his wife Leontia, both crowned and
of its suggested start date in the late seventh century ad, but wearing imperial costume. Ø 30.5 mm; mass 3.48 g.
not earlier. The unparalleled composition of the glass of this
weight strongly suggests that this is a modern concoc-
Conclusions tion, a glass disk having been impressed with a genuine
This study set out to examine various groups of Byzantine follis of Phocas.
glass weights in an attempt to establish their individual com- 7–10. 1997,0218.4; 1980,0611.30; 1893,0409.2; and 1990,
positional signatures. A number of tentative conclusions may 0601.17. Stamped with a nimbed and diademed impe-
be drawn from these analyses. Weights of types A and B, by rial bust above a cruciform monogram resolving as ‘of
virtue of their iconography, have long been assumed to have Euthalios’ flanked by two smaller busts; all four weights
been made in the capital, Constantinople. There was a slight from the same die. Ø 24.5, 24, 20 and 21 mm; masses
possibility, therefore, that these weights might have exhibited 4.59 g (nomisma), 3.81 g (nomisma?), 2.01 g (semissis)
different compositional characteristics to the main groups and 2.21 g (semissis).
already established for the eastern Mediterranean in Late
Antiquity, namely Levantine I, Levantine II and HIMT type Type B: weights with a bust of the prefect
glasses. That this is not the case is hardly surprising. Clearly of Constantinople
the capital either imported glass from these southern produc- 11. 1884,0509.13. Bust of an eparch enclosed by a Greek
tion centres or exploited recycled glass from these sources. It inscription resolving as ‘of Demosthenos’; dated to his
is surely no coincidence that the prosopographical evidence prefectship in the late 510s. Ø 23.5 mm; mass 4.03 g
for the dating of these weights ceases at much the same time (nomisma).
that the glass-producing provinces of Syria/Palestine and 12. 1987,0703.20. As No. 11. Ø 25 mm; mass 3.95 g
Egypt were lost to the Arabs, by the middle of the seventh (nomisma?).

12 | Chris Entwistle and Andrew Meek

13. 1980,0611.48. Bust of an eparch enclosed by a Greek 32. 1990,0601.12. Box monogram resolving as ‘of
inscription resolving as ‘in the time of the eparch Konstantinos’. Ø 18 mm; mass 1.75 g (semissis).
Theodotos’; dated to the prefectship of Theodotos 33. 1987,0703.3. Box monogram resolving as ‘of Patrikios’.
Kolokynthios in ad 522–523. Ø 22 mm; mass 2.22 g Ø 20.5 mm; mass 2.11 g (semissis).
(semissis). 34. 1980,0611.13. Box monogram resolving as ‘of Methodios’.
14. 1986,0406.12. As No. 13. Ø 20.5 mm; mass 1.98 g Ø 20 mm; mass 2.08 g (semissis).
(semissis). 35. 1981,0601.3. Box monogram resolving as ‘of Kyrillos’. Ø
15. 1891,0512.11. As No. 13. Ø 19.5 mm; mass 1.17 g 16 mm; mass 0.93 g (tremissis).
(tremissis). 36. 1984,0109.1. Cruciform monogram resolving as ‘of the
16. 1986,0406.15. Bust of an eparch enclosed by a Greek eparch Aristomachos’. Ø 27 mm; mass 4.48 g (nomisma).
inscription resolving as ‘in the time of the eparch 37. 1891,0512.13. Cruciform monogram resolving as ‘of
Theodoros’; dated to the prefectship of Theodoros in Akakios’. Ø 25 mm; mass 4.46 g (nomisma).
either ad 518–526 or 612. Ø 17.5 mm; mass 1.19 g 38. 1987,0703.16. Cruciform monogram resolving as ‘of
(tremissis). Prokopios’. Ø 20 mm; mass 2.00 g (semissis).
17. 1997,0218.2. As No. 16, but the inscription reading ‘of 39. 1980,0611.27. Cruciform monogram resolving as ‘of
Theodoros’. Ø 19 mm; mass 4.48 g (nomisma). Theodoros’. Ø 25 mm; mass 4.49 g (nomisma).
18. 1892,0613.61. Bust of an eparch enclosed by a Greek 40. 1923,1107.1. Cruciform monogram resolving as ‘of
inscription resolving as ‘in the time of the eparch Ioannes’; Timotheos’. Ø 21 mm; mass 2.02 g (semissis).
dated to the prefectship of Ioannes in either ad 550 or
582–602. Ø 25 mm; mass 4.43 g (nomisma). Type D: Arab-Byzantine’ weights
19. 1987,0703.19. As No. 18. Ø 25 mm; mass 3.63 g. 41. 1980,0611.52. Bust of an eparch enclosed by a debased
20. 1884,0509.14. As No. 18, but with the inscription reading Greek or Cufic inscription. Ø 23 mm; mass 4.27 g
‘in the time of Symeonos, eparch of [New] Rome’. Ø (nomisma).
17 mm; mass 1.45 g (tremissis). 42. 1986,0406.14. Bust of an eparch enclosed by a debased
21. 1892,0613.59. Bust of an eparch enclosed by a Greek Greek or Cufic inscription. Ø 23 mm; mass 3.27 g
inscription resolving as ‘in the time of Flavios Gerontios’; (nomisma).
dated to his prefectship between ad 560–562. Ø 25 mm; 43. 1980,0601.51. Bust of an eparch enclosed by a debased
mass 4.27 g (nomisma). Greek or Cufic inscription. Ø 23 mm; mass 4.18 g
22. 1892,0613.60. As No. 21. Ø 21.5 mm; mass 2.21 g (nomisma).
(semissis). 44. 1980,0611.71. Indistinct frontal bust. Ø 16 mm; mass 1.37
23. 1987,0703.18. Bust of an eparch enclosed by a Greek g (tremissis).
inscription resolving as ‘in the time of Flavios Zimarchos’;
dated to his prefectship in ad 565. Ø 19 mm; mass 1.64 g Acknowledgements
(tremissis). The authors would like to thank the many colleagues at the British
Museum who assisted with this work. They also thank two anonymous
24. S.322. Bust of an eparch enclosed by a Greek inscription
reviewers and Daniela Rosenow (University College London) for helpful
resolving as ‘in the time of the eparch Kosmas’; dated comments and discussion.
to his prefectship in ad 609. Ø 25 mm; mass 4.41 g
(nomisma). Authors
25. 1987,0703.17. As No. 24. Ø 17.5 mm; mass 1.35 g Chris Entwistle (centwistle@thebritishmuseum.ac.uk) is a curator in
(tremissis). the Department of Britain, Europe and Prehistory, and Andrew Meek
(ameek@thebritishmuseum.ac.uk) is a scientist in the Department of
26. 1879,0522.49. Bust of an eparch enclosed by a Greek
Conservation and Scientific Research, both at the British Museum.
inscription resolving as ‘in the time of the eparch Rogatos’;
dated to his prefectship in ad c.610–612. Ø 21 mm; mass References
1.82 g (semissis). 1. Grierson, P., Byzantine coins, Methuen and Co., London (1982).
27. 1980,0611.90. Bust of an eparch enclosed by a Greek 2. Balog, P., Umayyad, Abbasid and Tulunid glass weights and vessel stamps,
inscription resolving as ‘of the eparch Mousilios;’ dated American Numismatic Society, New York (1976).
3. Jungfleisch, M., ‘Un poids monétaire en verre arabo-byzantin’, Bulletin
to his prefectship in ad c.600–650. Ø 26 mm; mass 3.56 g
de l’Institut d’Egypte 29 (1946–1947) 13–17.
(nomisma). 4. Balog, P., ‘Poids monétaires en verre byzantine-arabes’, Revue Belge de
28. 1983,1108.1. Bust of an eparch enclosed by a Greek Numismatique 104 (1958) 127–137.
inscription resolving as ‘of the most blessed Domianus’; 5. Miles, G.C., Early Arabic glass weights and stamps; with a study of the
dated to his prefectship in ad 610–630. Ø 25 mm; mass manufacture of eighth century Egyptian glass weights and stamps by Frederick
R. Matson, American Numismatic Society, New York (1948).
4.41 g (nomisma). 6. Matson, F.R., ‘Glass standards of weight and volume in eighth
29. 1882,0510.20. As No. 28. Ø 25 mm; mass 4.41 g century Egypt’, The Glass Industry 30 (1949) 487–491.
(nomisma). 7. Ross, M.C., Catalogue of the Byzantine and Early Mediaeval antiquities
in the Dumbarton Oaks Collection. Volume I: Metalwork, ceramics, glass,
glyptics, painting, Dumbarton Oaks Research Library and Collection,
Type C: weights with a box or cruciform monogram
Washington DC (1962).
30. 1987,0703.4. Box monogram resolving as ‘of Thomas’. 8. Freestone, I.C. and Leese, M.N., ‘Byzantine glass weights: composition,
Ø 23 mm; mass 3.78 g (nomisma). manufacture and weight standards’, in C.J.S. Entwistle, A catalogue of
31. 1980,0611.4. Box monogram resolving as ‘of Paulos’. Ø the late Roman and Byzantine weights and weighing equipment in the British
26 mm; mass 4.35 g (nomisma). Museum, British Museum Press, London (forthcoming 2016).

Early Byzantine glass weights: aspects of function, typology and composition | 13

9. Miles, G.C., ‘On the varieties and accuracy of eighth century Arab other than pottery. British Academy Monographs in Archaeology No. 4, vol. II,
coin weights’, Eretz-Israel 7 (1964) 78–87. ed. H.R. Hurst, Oxford University Press, Oxford (1994).
10. Vikan, G., ‘Glass weights’, in G. Vikan, Byzantine objects of everyday life 34. Mirti, P., Casoli, A. and Appolonia, L., ‘Scientific analysis of Roman
from the Menil Collection (unpublished). glass from Augusta Praetoria’, Archaeometry 35 (1993) 225–240.
11. Csallány, D., A Kunszenmártoni avarkori ötvössír. Goldschmiedegrab aus der 35. Foy, D., Picon, M., Vichy, M. and Thirion-Merle, V., ‘Caractérisation
Avarenzeit von Kunszentmárton (Ungarn), Szentes (1933). des verres de la fin de l’antiquité en Méditerranée occidentale:
12. Rhé, G. and Fettich, N., Jutas und Öskü: zwei Gräberfelder aus der l’émergence de nouveaux courants commerciaux’, in Échanges et
Völkerwanderungszeit in Ungarn (Skythika 4), Seminarium Kondakovianum, commerce du verre dans le monde antique: actes du colloque de l’Association
Prague (1931). française pour l’archéologie du verre, ed. D. Foy and M.-D. Nenna, Éditions
13. Sós, Á., ‘Frühmittelalterliche Brandbestattung mit Feinwaage in Monique Mergoil, Montagnac (2003) 41–86.
Pókaszepetk’, Slovenská Archeológia 26(2) (1978) 423–429. 36. Röhrs, S., Meek, A. and Entwistle, C., ‘A scientific study of Late
14. Stephens Crawford, J., The Byzantine shops at Sardis: archaeological Antique glass pendants in the British Museum’, in New light on old
exploration of Sardis, Monograph 9, Harvard University Press, Cambridge glass: recent research on Byzantine mosaics and glass, ed. C. Entwistle and
MA (1990). L. James, British Museum Press, London (2013) 178–188.
15. Fulghum, M.M. and Heintz, F., ‘A hoard of Early Byzantine glass 37. Entwistle, C. and Finney, P.C., ‘Late Antique glass pendants in the
weights from Sardis’, American Journal of Numismatics 10 (1998) 105–120. British Museum’, in New light on old glass: recent research on Byzantine
16. Vitto, F., ‘Notes and news: Tel Naharon’, Israel Exploration Journal 30 mosaics and glass, ed. C. Entwistle and L. James, British Museum Press,
(1980) 214. London (2013) 131–177.
17. Dodd, E.C., Byzantine silver stamps: Dumbarton Oaks Studies 7, Dumbarton 38. Tal, O., Jackson-Tal, R.E. and Freestone, I.C., ‘Glass from a Late
Oaks Research Library and Collection, Washington DC (1961). Byzantine secondary workshop at Ramla (south), Israel’, Journal of
18. Martindale, J.R., The prosopography of the later Roman Empire. Volume 2: Glass Studies 50 (2008) 81–95.
A.D. 395–527, Cambridge University Press, Cambridge (1980) 39. Meek, A., ‘Gold glass in Late Antiquity: scientific analysis of the
19. Martindale, J.R., The prosopography of the later Roman Empire. Volume 3: British Museum collection’, in New light on old glass: recent research on
A.D. 527–641, Cambridge University Press, Cambridge (1992). Byzantine mosaics and glass, ed. C. Entwistle and L. James, British
20. Grierson, P. and Mays, M., Catalogue of Late Roman coins in the Dumbarton Museum Press, London (2013) 121–130.
Oaks Collection and in the Whittemore Collection: from Arcadius and Honorius 40. Meek, A., ‘The scientific analysis of the British Museum’s gold glass
to the accession of Anastasius, Dumbarton Oaks Research Collection and collection’, in D. Howells, A catalogue of the Late Antique gold glass in the
Library, Washington DC (1992). British Museum, ed. C. Entwistle and L. James, British Museum Press,
21. Bendall, S., Byzantine weights: an introduction, The Lennox Gallery, London (2015) 30–40.
London (1996). 41. Henderson, J., ‘Chemical characterization of Roman glass vessels,
22. Cameron, A., ‘Theodorus τριcέπαρχοc’, Greek, Roman and Byzantine enamels and tesserae’, in Materials Issues in Art and Archaeology II, ed.
Studies 17(3) (1976) 269–286. P.B. Vandiver, J. Druzik and G.S. Wheeler, Materials Research Society,
23. Kazhdan, A.P. (ed.), Oxford Dictionary of Byzantium, vol. 1, Oxford Pittsburgh (1991) 601–607.
University Press, Oxford (1991). 42. Henderson, J., ‘Localized production or trade? Advances in the
24. Freestone, I.C., ‘Glass production in Late Antiquity and the Early study of cobalt blue and Islamic glasses in the Levant and Europe’,
Islamic period: a geochemical perspective’, in Geomaterials in Cultural in Patterns and process: a festschrift in honour of Dr Edward V. Sayre, ed. L.
Heritage, ed. M. Maggetti and B. Messiga, Geological Society, London van Zelst, Smithsonian Center for Materials Research and Education,
(2006) 201–216. Maryland (2003) 227–245.
25. Freestone, I.C., Gorin-Rosen, Y. and Hughes, M.J., ‘Primary glass 43. Möncke, D., Papageorgiou, M., Winterstein-Beckmann, A. and
from Israel and the production of glass in Late Antiquity and the Zacharias, N., ‘Roman glasses coloured by dissolved transition metal
Early Islamic period’, in La route du verre: ateliers primaires et secondaires ions: redox reactions, optical spectroscopy and ligand field theory’,
du second millénaire av. J.-C. au Moyen Âge, ed. M.-D. Nenna, Maison de Journal of Archaeological Science 46 (2014) 23–36.
l’Orient méditerranéen - Jean Pouilloux, Lyon (2000) 65–83. 44. Freestone, I.C., Bimson, M. and Buckton, D., ‘Compositional
26. Nenna, M.-D., Picon, M. and Vichy, M., ‘Ateliers primaires et categories of Byzantine glass tesserae’, in Annales du 11e Congrès
secondaires en Egypte à l’époque gréco-romaine’, in La route du verre: de l’Association Internationale pour l’Histoire du Verre, Association
ateliers primaires et secondaires du second millénaire av. J.-C. au Moyen Âge, ed. Internationale pour l’Histoire du Verre, Basel (1990) 271–280.
M.-D. Nenna, Maison de l’Orient méditerranéen - Jean Pouilloux,
Lyon (2000) 97–112.
27. Freestone, I.C., Wolf, S. and Thirlwall, M., ‘The production of Notes
HIMT glass: elemental and isotopic evidence’, Annales du 16e 1. The only published technical study of ancient glass weights is that of
Congrès de l’Association Internationale pour l’Histoire du Verre, Association eighth-century Arabic glass weights by Matson, in which he described
Internationale pour l’Histoire du Verre, Nottingham (2003) 153–157. flakes of iron scale on the reverse or folded into the glass [5, 6]. An
28. Bimson, M. and Freestone, I.C., ‘The discovery of an Islamic glass- as-yet unpublished report by Freestone of some of the Byzantine glass
making site in Middle Egypt’, Annales du 10e Congrès de l’Association weights in the British Museum’s collection remarks that “although
Internationale pour l’Histoire du Verre, Association Internationale pour present in some cases, iron scale is considerably less frequent on
l’Histoire du Verre, Madrid (1987) 237–244. the Byzantine weights … and thus there is likely to have been some
29. Rosenow, D. and Rehren, T., ‘Herding cats: Roman and Late Antique difference in the tools or operating procedures used by the Byzantine
glass groups from Bubastis, northern Egypt’, Journal of Archaeological and Islamic artisans, although the significance and precise nature of
Science 49 (2014) 170–184. this difference is not yet fully understood” [8].
30. Rehren, T. and Cholakova, A., ‘The early Byzantine HIMT glass 2. The transition from box to cruciform monogram can be observed
from Dichin, Northern Bulgaria’, Interdisciplinary Studies 22/23 (2010) at an early stage on the brass collars and marble capitals of three
81–96. churches in Constantinople. At Hagia Sophia, it is the imperial
31. Foster, H.E. and Jackson, C.M., ‘The composition of ‘naturally monogram of Theodora that changes from the box to the cruciform
coloured’ late Roman vessel glass from Britain and the implications type, while Justinian’s remains of the box form. Hagia Sophia was
for models of glass production and supply’, Journal of Archaeological completed by ad 537, so the change presumably occurred in the later
Science 36 (2009) 189–204. years of its construction, probably ad c.535–536. Similar cruciform
32. Bimson, M. and Freestone, I.C., ‘Analyses of some glass from Anglo- monograms of Theodora can be seen at the Church of Saints Sergios
Saxon jewellery’, in Glass in Britain and Ireland 350–1100 AD: British and Bacchos (completed by ad 536) and at Hagia Eirene (rebuilt by
Museum Occasional Paper 127, ed. J. Price, British Museum Press, ad 548).
London (2000) 131–136. 3. It is noteworthy that some of the earliest Byzantine box monograms
33. Freestone, I.C., ‘Appendix: chemical analysis of ‘raw’ glass fragments’, are extremely crude and these weights may represent the prototypes
in Excavations at Carthage, 1. The circular harbour, north side: the site and finds of the more developed box monogram.

14 | Chris Entwistle and Andrew Meek