discussion of the epigraphic script of the late Roman period presents
considerable problems. A model of clarity is represented by the table of letter
forms given by C. Bradford Welles in his publication of the inscriptions of
Gerasa, which are largely late Roman; but as M. Sartre pointed out in his
publication of the inscriptions of Bostra, such a presentation can be
misleading, in suggesting a clearer pattern of development than that which
actually exists. 24 The complexities of the situation have
also been illuminated by the publication of the early third-century Jewish
inscription from Aphrodisias; as the editors have pointed out, although the
script of that text has many apparently 'late' features, all can be attested in
use by the third century, in inscriptions, and, far more extensively, in
papyri. 25 The first conclusion to be drawn from
this is that in looking at the changes in epigraphic styles in the late Roman
period, we are not confronting the development of completely new scripts, but rather
a change in the range and type of letter forms considered appropriate for
Jewish inscription, although large and imposing, was private — that is, it was
put up not by the civic authorities, but by the Jewish community; and it is
clear that in assessing and comparing scripts it is essential to look first at
the nature and purpose of the documents. During the Roman period, when all
public inscriptions were put up in a characteristic and consistent script, a
far wider range of scripts were acceptable in a private context. This
distinction can still be seen in the earliest inscriptions presented here. Texts 1, 2, 3, 4, 5, 6, 234 and 7 are public inscriptions of the second half of the third century,
and they all use the traditional Aphrodisian public script of the preceding
centuries, based on essentially square letters. But the funerary inscriptions
which appear to be of the same period, 147
— 152, use a far wider range of
scripts: only in 152 is there an
attempt to use the standard forms, but even there cursive forms intrude.
Thereafter, as the script of public inscriptions alters, it is less easy for us
to detect any distinction between the letterforms of public and private texts;
but it seems likely that such a distinction continued. I propose here,
therefore, to consider only the script of public inscriptions, in order to
compare like with like. I have not included consideration of the three
definitely Byzantine inscriptions — 99,
110, 173 — which are so dated on the basis of their script.
principal interest of the study of scripts is in order to determine a system of
dating inscriptions. In presenting the inscriptions, I have referred to script
several times for this purpose; but there is, of course, considerable danger of
producing circular arguments, by assigning all texts of a certain appearance to
a certain date, and then describing their appearance as characteristic of that
date. I shall therefore limit myself here as far as possible to the discussion
of texts for which there is some evidence of date apart from the script.
major and underlying difficulty, however, is a different one. As I have pointed
out, Introduction.4 following, in the later
third century there were major changes in the nature and presentation of
inscriptions; the major factor is the enormous reduction in quantity, and that
seems to be accompanied by an important change in attitude to the script. Consistency
in appearance no longer appears to be a primary aim; instead, we have several
groups of inscriptions honouring the same individuals, each of which uses fundamentally different epigraphic styles
(see Introduction.9). It will be seen from what
follows that it is difficult to analyse the inscriptions of this period in
terms of 'development' of scripts. One indication of date may be provided by
the overall look of an inscription — whether the lines are regularly laid out
or not — although this too can be shown to vary between inscriptions of the
same period. Another sign of 'lateness' is an increasing tendency to use
different forms of the same letter in one and the same text; but the first
example of this is 9, which I have
dated to the early fourth century.
The Α with a dropped bar was already
standard in the Roman period, and continues to the sixth century (e.g. 85 —8). We also find a straight bar, from the early fourth century (e.
g. 11, 12, 13 14,) and well into the fifth (e. g. 39, 40), and sometimes
both in the same text (e.g. 23);
sometimes this is combined with slightly curved sides (e.g. 17, 18, both sides curved; 39,
one side curved). Another development is a slanted bar (e.g., in the fourth
century, 24, in combination with
dropped bar; in ? the sixth, 89). A
perhaps late development, found also in L
is to flatten the top of Α (89).
This may be a development from the accentuated upper serif; this feature is
also developed to slant sideways, perhaps a late feature (62, 65, both late fifth
or early sixth century). This is the closest that Α
comes in public texts to the cursive form found in 'private' texts (e.g. 118. ii), and in 65, where the standard form is also used (cf. also 90); significantly, it is not found in 83, which includes several cursive
The Β of the second and third centuries was
a complex letter (see 1b, 5
a) which was quickly abandoned for a
simpler form. The two chief variants are either two bowls touching the main
stem (from the fourth century, 21, 24) or an outline of the two bowls,
which does not touch the stem (also from the fourth century, 14, 15, 23 and later ? 72). A further, and significant variant
is the development of a form with two separate bowls, touching the main stem,
but not one another; this appears not to develop before the mid or late fifth
century (31 and 41, early examples, are not
as accentuated as later examples, 62,
65, 73, 76, 82, 83), and it is largely on this basis that I have dated 81 to the sixth century.
offers no remarkable features.
There are no very major variants on the standard form: there is a tendency to
accentuate the upper serif upwards from the fourth century (e. g. 14, 23) which continues into the fifth (39); an apparently later development is a side-ways slanted serif,
as in A and L
(62, 66, 74). A completely
cursive form of δ is found in the Albinus texts ( e.g. 83. vi, x, xvi,
xix) as well as the standard form (83. vii); it also appears in the private text of 551, 164.
The two chief variants here are between a squared or a lunate form; but the
determining factor here seems to be the overall script of a particular text,
rather than any matter of date. It is perhaps characteristic of later scripts
when squared and lunate letterforms are used in the same text (e.g. 81, 83.i)
does not appear often enough to offer any significant variants.
In the second and third centuries commonly had the cross bar detached from the
sides; this feature still appears in 9,
but thereafter a simpler form, with a straight attached cross bar, prevails and
does not offer many significant variants, although the cross bar is sometimes
used decoratively (e. g. 49).
Most commonly appears in a simple form, with a straight and attached cross-bar,
from the fourth century (e.g. 25) to
the sixth (62); but the cross bar is
sometimes used decoratively, either extending outside the letter (89) or detached and decorated inside
the letter (57, honouring Pytheas).
An alternative development is to form the letter with straight sides; this is
found in 55, also honouring Pytheas,
as well as 85).
The principal variation is between forms where the arms end below and above the
extent of the upright (e.g. 24, 73, 74), and those where they extend to the full depth of the upright
(e.g. 9, 36, 83); another variant
is that the arms are sometimes detached from the upright (e. g. 28). But these variants do not appear
to be related to date.
The developments are very similar to
those of Α and Δ.
It sometimes appears with slightly curved sides (e.g. 17, 18, both sides
curved; 39, one side curved), and
once with a flattened top (89). The
upper serif is sometimes accentuated vertically (e.g. 24) and in later texts slanted sideways (62, 65, 74; less markedly, 83.vii, x).
The forms vary chiefly between those which start the slanting elements from the
top of the uprights (e. g. 28, 74), and those which start some way
down the uprights (e. g. 13, 38, 60, and cf. 82, with an
accentuated serif where the sloping bars meet). But both forms seem to be in
use throughout the period, and can be found in the same text (e. g. 62). There is a fully cursive Μ in the carefully cut fourth century text 19, and in some of the acclamations for
Albinus the cross bar is a single curve rather than two slanting lines (e. g. 83.xvi; but contrast vii
and xix) as also in 89.
Again, the chief variation is between forms
where the oblique goes from the top of one upright to the bottom of the second
( e. g. 19, 73, 83), and those where
it starts — and sometimes finishes — some way down the uprights (e. g. 11, 33, 36, 82, 88); both these usages seem to extend throughout the period.
This is one of the most varied letter forms; the shapes vary from a simple form
like a double Z (37,
39, 64), to a form where the upper and the lower elements meet in a
sort of scroll (33, 73, 83.xvi). Again, these
forms seem to be in use over the same period, although the second may be
slightly later. Occasionally the central element does not touch the upper and
lower cross bars (19).
Shows no significant developments.
Shows no significant developments.
Varies considerably in form. In the simplest form, the bowl is simply attached
to the upright at the top (e. g. 17,
39, 90); occasionally the bottom of the bowl is left open, with a
flourish (45). The bowl can be attached to an upright which extends above it
(9, 37, 74, 82) sometimes with an accentuated serif
(33, 73), or the bowl left open and decorated (57). There is one example of the top of the bowl left open, with a
flourish (19, line 2).
The standard form prevailed throughout the Roman period, and is found in 1 — 7, and in 9 and 11, both early fourth century. But
thereafter it is replaced by either a lunate or a square form — the lunate form
is already in use in 12 and 13, which are closely contemporary with
11, and there is a square Σ in the last line of 9. As with E, the choice of form tends to depend on
the overall style of the text.
Shows no significant variations.
The chief variation in forms is between the simple form (e. g. 12,
23, 65) and that with a cross-bar (e. g. 24, 42, 61, 73); both continue in use throughout the period.
Offers extensive possibilities for variation. It often appears with a very
exaggerated upright, especially when it is the initial letter of Φλάβιος (so in 25
— 27, 35, 69, 81). The shape of the bowl is sometimes
altered from a pure circle: it can be heartshaped (50) or angular (62 l. 8,
but a circular form in l. 10), or formed from two semi-circles (24, 54, 82) which sometimes
do not touch the upright (69; but
contrast the contemporary 68).
Is normally square, or elongated upwards when the script is elongated; but there
is sometimes a tendency to exaggerate one cross bar (e. g. 42, 58, 68, 74, 86); this may be a
later development, but the simpler form also continues (e. g. 62).
Does not appear very often; it is usually angular rather than curved.
The standard form, which prevailed throughout the Roman period, is found in 1, 2, 3, 4, 5, 6, 7, 9 and 11; thereafter the
cursive form prevails, and is found in 12,
closely contemporary with 11. The
old form is revived once, in 33,
about whose date I am very unsure. Otherwise the chief variation is between
three forms of the double-u shaped Ω: the angular
form of two Ωs (37,
62); the flat-bottomed form (36, 54, 62); and the rounded
form of two υs. Within this latter form there is
considerable further variation: the central upright may be the same height as
the other two (13, 22, 42, 83), or lower (24, 53, 61, 73); a further variant is when the two
sides form a single curve, with an upright inserted in the centre (14, l.7, but contrast ll. 3-4, 20, 25). Of all these shapes, I think that it is possible that a
relatively late one may be that with well rounded sides and a central upright
which does not rise very high — e. g. 45,
49, 74; but this clearly co-existed with several other forms.
Abbreviations undoubtedly increased in use during
this period, and their extensive use is characteristic of later fifth or sixth
century texts. The best example here is provided by the funerary text of 551, 164. Ο
and Θ in ligature occur several times (e. g.
66, 69) and abbreviated καί is quite
common. The most common mark of abbreviation is a scroll, either above (so
usually for Φλάβιος) or below the line (so usually for καί); a relatively late development appears to be
that of placing the last letter of the abbreviated syllable above the line (e.
g. 65, 69, 163, 164).