Professional Documents
Culture Documents
Proceedings of the Fourth Roman Military Equipment Conference, BAR International Series 394, Oxford
INTRODUCTION
The main aim of this paper is to identify and describe the elements
that went together to make up Roman military horse (or riding) harness
in the first century A.D.1 Closely linked with this, however, will be a
consideration of just how much the archaeological evidence can tell us
about the military use of mounts.2 It is to be hoped that the present
work will at least serve to generate new interest in cavalry equipment
and provoke discussion about the functions of its various components.
There are two main source areas for the study of Roman military
horse equipment: monuments depicting its use and artefacts recovered
from the archaeological record.
Pictorial evidence, most notably that represented by sculpture, can
be used to show how the various elements of horse harness functioned as
a whole. The value of such depictions is hotly debated, but it is
generally true that funerary monuments tend to give a more accurate
picture than official sculpture, although both categories vary widely in
quality.3 Two types of tombstone, the 'Reiter' and the 'Totenmahl' are
characteristic of the first century A.D. (although not exclusively so): 4
these show, in the first instance, the deceased cavalryman riding his
mount in combat, often with a cowering barbarian being trampled beneath
the horse;5 the other type depicts the dead man enjoying a funerary
banquet in the afterlife in an upper scene, whilst his horse is paraded
in all its equipment in a lower.6 Totenmahl depictions usually show the
horse being controlled from behind by means of long-reins and with its
saddle covered by an overblanket, whilst the trooper's calo carries
spare spears (javelins?). As has been noted, the quality could vary and
there are cases where the sculptor probably did not understand his
subject matter, but some of the best pieces, like the stone of T.
Flavius Bassus at Köln, appear to be detailed documents of the way in
which horse harness functioned.7
As well as private funerary monuments, mention must be made of
official sculpture, if only to sound the now customary note of caution
about interpreting it too literally. It is certainly important to
appreciate that Trajan's Column, which is traditionally regarded as a
prime source of information about the Roman army,8 includes a number of
clues to the fact that it is not a useful source for the study of
cavalry equipment, and these will be outlined later in some detail. Much
the same is true of most of the major monuments of the city of Rome, 9
but examination of some provincial sculpture of an official nature is
more productive. The triumphal arch at Orange,10 which is probably
Tiberian, is a good example of this, with important details about
harness being confirmed by the static weapons friezes, whilst the battle
scenes provide an interesting comparison with the depiction of Roman
cavalry on the Rhineland tombstones.
67
Archaeological evidence is usually manifested either as individual
items of equipment, recovered during the process of excavating a site,
or hoards of horse equipment which have been gathered for some
purpose.11 Site finds tend to include a wide range of types of any given
fitting, whilst hoards often have a narrower stylistic range. A large
proportion of published site finds come from excavations where
stratigraphy was inadequately understood and, frequently, where the
spatial distribution of finds was not considered to be of interest.12
Site finds can also usually be fitted within a broad date range for a
site, and nowadays, as excavation (and recording) techniques improve,
can be placed within a narrower chronological bracket.
The fact that a number of hoards of Roman horse equipment have been
recovered is extremely important for the interpretation of horse
harness. Such hoards include the collections from Doorwerth and Xanten,
Fremington Hagg, and Canterbury.13 The dating of these assemblages is
often problematic, but their value in interpreting the functional
relationships of the various elements is inestimable.
REPRESENTATIONAL EVIDENCE
A comparatively large number of figured first-century tombstones
which show mounted soldiers have survived. Some show the harness in
great detail, most are indifferent, whilst a few are positively crude in
the standard of their depiction of detail. The evidence presented is
equivocal, as the following selective survey of some of the more
important pieces shows.14 Unfortunately, some details are not very clear
on photographs and some tombstones, such as that of Primigenius, do not
reproduce at all well.15
68
Fig.1: Relief from tombstone of T. Flavius Bassus.
Fig.2: T. Flavius Bassus. Detail of haunch junction, showing haunch
straps passing beneath saddle pommel. Köln Museum.
69
Fig.3: T. Flavius Bassus. Detail of horse’s breast showing breast band
and pendants. Köln Museum.
Fig.4: T. Flavius Bassus. Detail of horse’s head with bridle, bit,
hackamore, and poll knot. Köln Museum.
70
3. Sextus Genialis (Reiter, ala Thracum)18 Cirencester
There is a phalera junction on the haunch, but no clear junction at
the shoulder. The harness straps are bordered, as are the reins and
haunch pendant strap (the latter is also fringed). The haunch strap
passes horizontally beneath the saddle blanket. The breast strap
broadens towards the centre of the animal's breast. No pendants are
shown, but the bridle is studded with phalerae. The saddle (with
pommels) is depicted, over a fringed saddle blanket. There appears to be
a suggestion of triplet straps beneath the sword, at the edge of the
blanket. Claudio-Neronian.
71
Fig.5: Relief from tombstone of Longinus Biarta. Köln Museum.
Fig.6: Relief from tombstone of Romanus Dardanus. Köln Museum.
72
Fig.7: Relief from tombstone of unknown cavalryman. Köln Museum.
Fig.8: Relief from tombstone of M. Sacrius Primigenius.
73
is tied in a poll-knot (Fig.4). Late Flavian.
74
Fig.9: M. Sacrius Primigenius. Detail of horse’s head, with bridle,
bit, poll knot, and phalera. Köln Museum.
Fig.10: Relief from tombstone of Lucius. Köln Museum.
75
Fig.11: Relief from tombstone of Oluper.
Fig.12: Relief from tombstone of Niger. Bonn Museum.
76
the saddle and the triplet straps, the short one being over the longer.
A neck strap, which may be decorated with pompons, is depicted, and the
bridle is decorated with a number of small phalerae. There is a poll-
knot, to which is fixed a large upstanding phalera, possibly decorated
with a bust (Fig.9). The mane is gathered into pairs of small tufts all
down the horse's neck and the tail is bound. The horse is being guided
on long-reins by a calo. Flavian.
77
Fig.13: Relief from tombstone of M. Aemilius Durises. Bonn Museum.
Fig.14: Relief from tombstone of Vonatorix. Bonn Museum.
78
14. Reburrus (Reiter, ala Frontoniana)29 Bonn
The shoulder and haunch junctions are formed from large phalerae,
but it is not clear how many straps pass from these to the saddle,
although they may be single in both cases. There is a large lunate
pendant on the horse's breast and the rear pendant strap survives. The
saddle pommels are evident, as is the curved lower edge of the saddle.
Triplet straps hang from the front and the rear, at the edge of the
short fringed shabracque. There is a small phalera on the bridle.
Claudio-Neronian.
79
80
Fig.16: Relief from tombstone of C. Romanius Capito.
Mainz Museum.
Fig.15: Relief from tombstone of Cantaber. Mainz
Museum.
saddle and a short shabracque is worn under it. A girth strap passes
over this cloth at its front. Tiberio-Claudian.
81
82
Fig.17: C. Romanius Capito Detail of haunch junction Fig.17: C. Romanius Capito Detail of breast
and tail binding. junction.
23. Abaius (Reiter, ala Picentiana)38 Mainz
Enough of this fragmentary stone survives to show that it had
phalera junctions, decorated with rosettes. The haunch junction has two
straps which appear to unite before passing to the saddle. There are no
pendants, but a pendant strap survives. There is a fringed saddle-cloth,
under which a bordered girth strap passes. The tail may be bound.
Flavian.
83
28. Togitio (Reiter, unit unknown)43 Mannheim
Phalera junctions (decorated with rosettes) are depicted on this
stone at the haunch and shoulder positions. There are two shoulder
straps, but only one passing from the haunch junction to the pommeled
saddle. There are no pendants apart from the pendant straps and a lunula
on the breast. There is no sign of a saddle-cloth or girth. Claudio-
Neronian.
84
Fig.19: Depictions of harness on arch at Orange and Mausoleum at St.
Remy. 13: Orange, northeast face; 4: Orange, northwest face;
510: Orange, southeast face; 1113: St. Remy mausoleum; 12b:
breast and breeching straps; 12g: girth; 12p: pendant strap;
12t: triplet straps.
85
crescents. There are no obvious strap elements attached to the saddle.
The north-west face of the monument is quite severely damaged, but
it is possible to make out a saddle in the bottom centre of the
frieze(Fig.9,4)47. This has an interesting design along its lower edge,
consisting of cranellation interspersed by dots, above what may be a
fringe. A girth hangs down in the centre, and on either side of it are
sets of five straps. The right hand set is tangled up with a strap
coming from the right hand corner of the saddle, where there is the
suggestion of a ring.
The south-east face is the most informative of all and can boast
three bridles and three saddles48.The bridle in the top right-hand
corner (Fig.19,5) is draped across a shield, behind which the reins
disappear; the bit appears to be looped. A further bridle is to be found
in the centre left of the frieze (Fig.19,6), and cheekpieces, bit, and
reins can be distinguished. The third bridle (Fig.19,7), just to the
right and below the first, is partially hidden by a shield and a saddle.
Cheekpieces, headband, and bit are certainly depicted, but it is not
clear whether reins or a frontal are intended in the centre of the
bridle. The three saddles are symmetrically arranged, one on either side
in the centre, and one at the bottom centre of the frieze. The left-hand
example (Fig.19,8) appears to have some sort of panelling depicted on
the saddle itself; the pommels have suffered damage. There is a girth
strap hanging from the centre, terminating in a buckle, and decorated
with a voluted pattern. To the left side of this there are at least five
straps, with a further six to the right. A pair of straps hang from the
right hand corner, one vertically, the other caught up with the right-
hand group of straps. The left hand set of straps also appear to have
another strap entangled with them. The central saddle (Fig.19,9) has
very pronounced pommels and a lozenge pattern, on its surface (possibly
indicative of quilting?) and, unlike the other saddles on the friezes,
is taller than it is broad and has a straight lower edge. A central
girth strap and two groups of straps (five in each case) hang from it,
as do two straps from each of the two lower corners. The third saddle
(Fig.19,10), on the right-hand side, has similar panelling on it to the
first and is pommeled. The girth is hanging behind a shield above which
the saddle appears to be suspended. Two sets of five straps hang on
either side of it; the left hand set is caught up with a strap (possibly
the same one that is draped over the shield nearby), whilst two straps
descend from the right-hand corner of the saddle, where there is a ring.
b. The small battle scenes that concern us involve Roman cavalrymen
attacking barbarian foot warriors. On the north facade, there are two
such scenes, one in either corner. In the north-east example,49 a simple
ring junction at the shoulder of the horse can be discerned, from which
a pendant strap (with four horizontal divisions) hangs. The bit and
reins are visible, but the horse's head is too damaged to preserve any
other details of the bridle. At the north-west corner,50 a breast strap
that broadens to a point is all that can be seen.
The south-west corner scene51 again features a ring junction,
pendant strap, and broadening breast strap, but this time the details of
the bridle, bit and reins are quite well preserved. A small phalera
appears on the bridle at the junction of the throatlatch, cheekpiece,
86
headpiece, and browband. The horse on the scene at the south-east
corner52 has small phalerae on the bridle (one at the side and one in
the centre of the browband). A girth is also clearly visible beneath the
belly of the horse.
c). There are two large battle scenes, on the north and south faces of
the arch.
There are at least eight horses in the south frieze which are
Roman.53 They wear a simple harness consisting of a breast strap that
broadens to a point at the centre of the animal's chest, with ring
junctions at the shoulder and haunch positions,from which pendant straps
hang. The fallen horse in the bottom left-hand corner shows off the
harness, with the bordered saddle-cloth and girth, to advantage. The
other fallen horse, in the bottom centre right of the frieze, also has a
ring junction at the haunch with associated pendant strap, as well as
decorated breeching and a possible fringed saddle-cloth.
The north frieze54 has ten horses on it and they display the same
type of harness. One horse, at the extreme left-hand edge of the scene,
is particularly clear; apart from ring junctions and pendant straps, a
fringed saddle-cloth and girth are visible. As a rule, these two battle
scenes do not include representations of the saddle pommels, but a horse
in the top right centre is an exception to this. Tiberian.
31. Mausoleum (Fig.19)55 Saint-Rémy-de-Provence
The friezes on the base of the mausoleum at Glanum include a number
of depictions of horses.
On the north face of the monument,56 six horses are depicted -five
mounted and one fallen. The mounted figures show only a minimal amount
of detail, although the fact that some of them appear to have poll-knots
is noteworthy. Simple breast straps (or breast bands?) are attached to
bordered rectangular saddle-cloths. The fallen, riderless horse
(Fig.19,11), on the other hand, is extremely informative. 57 The pommeled
saddle, the surface of which is decorated with dots, sits on a small
rectangular saddle-cloth and is secured by a girth which passes over the
blanket. Straps hang down vertically (bearing in mind that the horse is
kneeling) from the two visible corners of the saddle, whilst breeching
can be seen passing round the rump of the horse and fastening to the
rear corners. Likewise, a probable breast strap is attached to the
visible front corner of the saddle. A pendant strap hangs on the haunch
of the animal.
Another mounted figure appears on the east face,58 but this has a
similar simplistic harness compared to the previous examples. The south
face, however, has another riderless horse (Fig.19,13)at the extreme
left-hand side which again presents a detailed picture of harness.59 A
decorated pommeled saddle is clearly shown, resting on a rectangular
saddle-cloth. The breeching passes back from the corner of the saddle
and under the tail of the horse. There appears to be a small phalera
junction at the haunch, from which a pendant strap hangs. The precise
arrangement of straps at the front is difficult to interpret, but there
would appear to be a set of triplet straps hanging down from the front
87
pommel, as well as the breast strap.
88
and may represent either the breast straps or a neck band.
89
Fig.20: Horseman from the Gundestrup cauldron.
Fig.21: Relief from tombstone of Rufus. Mannheim Museum.
90
of the fundamental problems with the representational evidence is
knowing what to believe. The quality of depiction ranges from the crude,
probably early tombstones (Fig.21), through the detailed later examples,
to the fine, classical official sculpture of the city of Rome itself.
Comparison of the equipment shown on these various images with material
from the archaeological record strongly suggests that the Tiberio-
Claudian and Flavian tombstones of the Rhineland were consistently the
most accurate representations of Roman cavalry.67 The Romanius stone
(and those like it) and that of Bassus seem to be the highpoint of
sculptural representation of horse equipment, and it is not surprising
that their harness makes the most sense when viewed from a purely
functional standpoint. However, this is very difficult territory - is
plausible equipment sufficient evidence for the accuracy of the
tombstone?68
91
Fig.22: Typical front saddle horns, showing A: profiles; B: variations
on lower opening; C: Weisenau ‘pockets’ at top of horn; D:
three types of edging.
92
Fig.23: Typical rear saddle horns, showing A: profiles of righthand
piece; B: possible variations in shape of rear of saddle; C:
squareended projection; D: roundended projection.
93
Girth buckles (Table 2; Fig.36)
Apart from the fact that their size suggests they may have
fulfilled some such role, archaeological examples of these are virtually
impossible to prove. Most seem to have been of copper alloy with an iron
spindle. Some examples appear to have had the loop moulded complete
(Types 1-2 & 4), whereas others were clearly composite (Type 3), formed
from side pieces, cross-member, spindle, and tongue.74
94
decorative) or functional criteria.82 If we begin with the functional
characteristics (Figs.41-2), it is evident that a clear pattern can be
discerned.
a) One loop phalerae are fixed directly onto the strap, which passes
through the central loop and is fastened by a rivet passing through
the centre of the disc. These are usually the smallest phalerae
(suspension Type la).
b) Two loop phalerae (Types 2a-f) are found in a number of variants,
but are usually associated with a pendant. The first kind (Types 2a,
d-e) has two parallel loops through which the strap passes and is
fastened in the centre of the disc by a rivet. At the bottom of the
reverse face of the disc, a hinge for a pivoted pendant is situated
at the centre. The second variety (Type 2b) consists of two
converging loops, being closer at the bottom than the top, and
joined at the bottom by a short bar, which acts as the hinge for a
necked pendant. These normally form the intermediate size of
phalera.
c) Three loop phalerae (Type 3c) usually have one loop on either side,
one at the top, and a hinge at the bottom. Although intended for two
straps meeting at right-angles, the precise function of this form is
debatable and is considered in more detail below. Normally the
largest kind of phalera.
d) Four loop phalerae (Type 3d) are probably similar to the three-loop
type and are normally one of the largest varieties, although again
their exact manner of use is not immediately obvious.
e) Three ring phalerae (Types 4a-b) are from shoulder or haunch
junctions and are amongst the larger examples, the rings being used
for attaching junction loops.
f) Four ring phalerae (Type 5a) are also from shoulder or haunch
junctions and of a similar size range to three-ring examples.
g) External ring phalerae (Type 6a - usually with four rings equally
spaced around the circumference of the disc) are not found very
often and are known in both Augustan (Oberaden) and late-first to
mid-second century contexts (Newstead).
These functional criteria are totally separate from the decorative
regimes found (Figs.39-40) and the two do not appear to be connected in
any way. The most common decorative forms of phalera are Types 1 and 2,
the basic profile of which is a convex perimeter (narrow on Type 1,
broad on Type 2) around a concave centre (broad on Type 1, narrow on
Type 2). Amongst the less common decorative forms, are those with very
narrow convex perimeters and more-or-less complex concave centres (Type
3), the concave (Type 5), the convex (Type 9), the flat (Types 8 and
12), and the ornately decorated (Type 6). It may be that some of the
more bizarre examples, such as Type 6, were from officer's equipment,
given the unusual combination of silver and copper, as well as the
unique form, but (as always) this is a difficult argument to
substantiate.
95
Pendants (Table 6, Figs.24; 43-9)
As the name indicates, pendants are designed to hang freely from
harness. To this end, some means of suspension was necessary, and a
variety of forms were utilised by the Romans (Fig.24), but the most
common were the hinge and the neck. With a hinged suspension, a pierced
lug on the pendant was placed between two similarly pierced lugs on a
phalera and a hinge placed through all three (Fig.24,3). The main
alternative was to give the pendant a long neck which could be folded
forwards or backwards over a bar or ring on the phalera or stud from
which it was suspended (Fig.24,2). Although the method of suspension was
usually hidden behind the phalera, some forms of pendant which were not
suspended in this way developed decorated suspension necks (Fig.24,1).
Pendants were normally cast from copper alloy (brass); the silvered
and nielloed ones then having their exterior faces engraved as
necessary. The Xanten fittings appear to have had their silvering
applied before they were engraved.83
One of the commonest forms of pendant in the first century A.D. was
the so-called 'trifid' pendant (Type 1). These were always associated
with phalerae84 and were usually silvered and inlaid with niello. This
type is closely linked with the lunula (Type 9) with internal pendants
(Type 8), from which it was probably ultimately derived, and it is the
type found in the Doorwerth and Xanten hoards. There was a wide range of
variations upon the theme, but most versions had a central lobe
decorated in the form of a leaf. A common motif is the oakleaf central
lobe, flanked by acorn terminals. The Xanten and Doorwerth pieces also
exhibit 'eyebrows' and 'noses' (Type la) - in fact the similarities
between the two collections are quite remarkable and probably point to a
common workshop for their manufacture. The inlaid decoration on these
pieces is usually based on a theme of viticulture, with vine leaves,
tendrils, and bunches of grapes appearing in a variety of more or less
stylised forms.85 This type is found from the Claudian period onwards,
but cannot yet be shown to be earlier.86
Type 2 pendants have ovoid bodies and, like Type 1 pieces, are
normally suspended from phalerae (by hinge or neck), and decorated with
silvering and niello inlay (again with themes derived from viticulture).
The Canterbury examples are so far unique in being suspended from hinged
strap mounts, as well as being decorated with bosses. Like Type 1, these
probably date from the Claudian period.87
Examples of Type 3 are comparatively rare, but shown to belong to
the first century by the example from Doorwerth. They are interesting
insofar as most of them incorporate an openwork peltaform terminal.
Suspension is usually by a neck.
Type 4 pendants derive their shape from leaves, 4a and 4b directly
from vine leaves. Suspension methods vary, and include neck, loop, and
rivet.
Teardrop-shaped pendants (Type 5) are extremely common and
96
Fig.24: Methods of pendant suspension and martingale attachment. 1:
neck suspension; 2: loop suspension; 3: hinge suspension; 45:
possible methods of martingale attachment to breast junction.
97
fulfilled a variety of purposes. Some of the smaller ones formed
terminals to 'apron' straps or central pendants in lunulae (Type 9), but
the larger examples could be suspended from phalerae, as the finds from
Inota show.88 These normally employed neck suspension, but hinges are
also found.
'Bird-headed' or 'winged' pendants (Type 7) are one of the
commonest pre-Flavian types, the major forms being the simple 7b and the
more elaborate 7a. Tinning seems to be the most frequent form of
decoration, but pieces could incorporate punctim patterns. 7a pieces
usually have a stylised wolf's head, but some have more naturalistic
representations. 7b uses a stylised bird's head, either duck or goose,
but again some pieces are more realistic than others. The method of
suspension is always by neck, which was passed through a loop (often
suspended from a stud) riveted to the harness strap. This type of
pendant was popular in the pre-Flavian period (a fact illustrated by
their distribution in Britain), but has not so far been found in
Augustan or Tiberian contexts.89
Lunate pendants (Type 9) come in a wide range of forms and are
extremely common. Large, flat forms with a characteristically angular
crescent are rare (Types 9a-c & n), whilst the flat crescent with
knobbed terminals (Types d-e) is quite frequently found. Well-formed
crescents with an angular profile are very distinctive (Type f) and
there is little variation in this sub-type. A curved profile is also
quite common (Type g). There are a number of exotic variants which are
not very common, but amongst the earliest are the boar's tusk pendants
(Type s) which are found in both Roman and Celtic contexts. Suspension
is normally by means of a neck, but loops are found, as well as the
articulated double-hinge of Type 9s. Lunulae are found from the Augustan
period through to the second century A.D. in association with cavalry
harness.90 Type 8 pendants are normally found suspended (usually by a
neck) inside the arms of a lunula.
Phallic pendants (Types 6 and 10) are very common in Roma-military
contexts, but they exhibit a great deal of variety, so much so, that no
one subtype predominates amongst Type 10. The basic symbolic elements
are a stylised phallus and fist (making the mano fica sign against the
Evil Eye) in an upward-curving crescent, from which testes and a further
phallus normally hung. It is interesting that these pendants were almost
invariably suspended by means of a ring, usually large enough to form a
junction. In fact, some pieces are known with junction loops still
attached, but none of these are forms of strap mount which we can
associate with cavalry, which leads to the question of whether phallic
pendants belong with vehicle harness or beasts of burden.91
The same is true of Type 6 pendants, most of which have similar
loops at the top, and an example from Strasbourg still has its junction-
loops attached.92
Although there is no evidence that the piece is from cavalry
harness, the final type of pendant (Type 11), the lozenge, is included
here because of its resemblance to one of the forms of pendant worn by
Bassus (above, No.6, and Fig.3).
98
99
Fig.25: Reconstructed typical CeltoRoman harness with terminology.
Strap fittings (Figs.26; 50-6)
Roman harness was adorned with a number of different types of
fitting, most of which were functional, rather than purely decorative.
It is necessary to consider the five main categories as a whole, as well
as individually, since each set of harness usually contained examples
from each category.
An examination of the published material shows us that there are
ten basic stylistic groups of strap fitting, but at least two of these
(Types 7 and 10) probably belong to either vehicle or non-military
riding harness. In many cases, we do not posses examples of a category
for each sub-type, so some interpolation is necessary to reconstruct
sets.
Type 1 fittings are those of the 'spectacle' type, normally having
a flat figure-of-eight shaped body with two dome-headed rivets securing
them to the strap. These are common throughout the two Germanies and
Britain and appear to date from Julio-Claudian up to Hadrianic times. 93
Type 2 fittings are similar, but have a slightly moulded profile to the
body and a more-or-less stylised volute at one end: their distribution
is more limited, being rare in Britain.94 Type 3, or the 'double-
spectacle' fittings, enjoy a very similar distribution to Type 1. They
exhibit pairs of rivets securing them to the strap; whilst there are a
number of variants, the basic form is very distinctive. Type 4 fittings
are the ones with 'moulded' decoration, using themes that include the
bordered ovoid and the acorn, and they might also be tinned or silvered
and inlaid with niello; these seem to be largely pre-Flavian in their
use.95 Type 5 fittings are rectangular with a broad decorated band at
one end and these appear to be in use throughout the first century A.D.,
but - interestingly - do not include strap mounts. Type 6 are amongst
the most familiar, since these are found in the Doorwerth, Xanten, and
Fremington Hagg assemblages. This type is flat, although shaped,
silvered, and inlaid with niello, using tendril motifs related to those
found on Type 1 and 2 pendants and their respective phalerae. Type 6 was
probably introduced in the Tiberio-Claudian period.96 Type 7 is based on
the triangle and only junction loops are known for this type, hinting
that they may not belong with cavalry harness. Type 8 is the simple
rectangle, some sub-types being inlaid, others not. Type 9 are quite
rare, so the absence of some categories of fitting may not be
significant, but it suggests that they had a very limited distribution.
Type 10 includes various rectangular forms and probably belongs to
vehicle or pack animal harness, given their crudity and the absence of
other categories apart from junction loops.97
100
101
i. Junction loops (Table 7, Figs.50-1)
The first of our four main types of strap fitting, junction loops,
were normally cast from copper alloy (brass). In form, they were simply
a decorated upper plate which was bent over on itself and the back
element riveted to the front, the bend forming the loop which was
attached to the ring junction or phalera. They were not ordinarily
detachable from their parent junction.
A clay mould for a Type 1 junction loop has been found in Nijmegen,
demonstrating that some at least were cast in the folded position. Some
objects from Arlaines, however, which appear to be Type 2 junction
loops, have straight loops, whilst retaining their casting flash.98
A distinction existed between junction loops intended for use with
rings and those that were to be attached to phalerae: ring junction
loops had moulded or decorated loops, frequently arched up from the body
of the fitting, whereas those for phalerae were unadorned and simply
folded back on themselves (as was so with Type 6 and the Newstead Type
5e junction loops). The commonest junction loops in Britain and the
Germanies are Types lc and Id, with their characteristic (and presumably
symbolic) incised triangle on the loop.99
Junction loops suffered a number of different forms of damage,
almost certainly the result of their manner of use, and equally likely
the reason for their high discard rate compared to some other categories
of fitting (they are arguably the most common of the harness fittings
recovered from the archaeological record).100 We may classify these
faults as follows:
a) Broken loop. Commonly the loop fractures just below the apex, so that
the whole of the rear of the fitting is missing. Probably caused by
metal fatigue (a programme of metallographic examination would be
useful here) brought about by the fact that this part takes the
strain and friction of contact with the junction ring - complete
examples show clear signs of wear, due to the fact that the loop was
hollow and not shaped to fit the curvature of the ring.
b) Distorted loop. Here the rear strip and loop are twisted or bent out
of place, sometimes straightening out the fitting. This suggests
forcible removal from the strap, either accidentally (such as a
rivet failing) or deliberately (a wish to change the fitting for
some reason).
c) Body fracture. The body fractures below the loop and first rivet, but
above the second - very common on Type 1 and 3 fittings. Possibly
due to the strap stretching and causing the inflexible fitting to
fail.
d) Rivet failure. A majority of fittings are found without their rivets,
suggesting that this was another important reason for their being
discarded. Like c, possibly caused by movement of the strap; this
can lead to the whole item being removed from harness.
Alternatively, pieces that failed for one of the above reasons would
102
have had their surviving rivets removed in order to take them off the
strap.
More recently, this type of fitting has suffered somewhat at the
hands of scholars, who have coined a variety of names for it, but
'junction loop' is the most descriptive of its form and function. They
have even been mistaken for baldric fastenings (e.g. 'baldric clip'), as
have categories iii and iv below. 101
103
fact that two-rivet 'spectacle' fittings have three rivets in their
strap mount form. Type 6 mounts often feature decorated rivet heads,
either domed or dished.105
HARNESS RECONSTRUCTION
The Saddle
The most important piece of harness for a cavalryman, in order that
he should be truly effective in close combat, was the saddle. The
outstanding qualities of this were that it should distribute the weight
of the rider upon the horse's back without causing it any discomfort,
whilst also providing a comfortable and safe seat for the rider.106 To
achieve the first, the saddle has to have a wooden 'tree' (or frame)
with padding, whilst the second is accomplished by shaping the seat to
fit the rider.107
Peter Connolly has reconstructed a Celto-Roman saddle of the first
century A.D., based upon the available evidence. This was mainly
archaeological (leather saddle covers and copper alloy 'saddle horns'),
but was supported by sculptural evidence (taken from tombstones and the
St.Rémy mausoleum).108 The efficacy of this reconstructed saddle was
demonstrated in its actual experimental use.
Connolly suggested that the saddle pommels were wooden and that the
holes around the edge of some of the saddle horns were for nailing these
objects to the pommels. These horns, which were covered with leather,
cannot have been attached to the outside of the saddle, since nail holes
are not found on surviving leather fragments. Connolly felt that the
balance of evidence lay in favour of their being used to shape the
pommel on the inside,109 but the peripheral holes, if indeed used for
nailing, would imply that the wooden pommels themselves would have been
shaped, thus rendering the horns unnecessary. It might be argued that
they were a form of protection for the pommels, protecting thee against
damage in battle or mishandling when the saddle was removed from the
horse.
However, an alternative explanation for the copper alloy horns may
be that the pommels were not wholly wooden. If we accept that the horns
were intended to shape the pommels, then it is conceivable that some
form of padding was used to form them. A fabric cover, stuffed with
horsehair, would be fairly rigid (comparison with recent upholstery
techniques confirms this), yet easier to produce than carved wooden
pommels; if the horns were then stitched to the outside, these would not
only give the pommels the desired shape, but also add an extra degree of
rigidity.110 It is also possible that such padded pommels had a simple
wooden core, which would then be attached to the tree, but this would
not seem to be essential.
The set of horns from Pit XXII at Newstead include two names on one
of the objects, which also has a different numeral punched in it to the
others. It is conceivable that one pommel on Senecio's saddle (number
XII) had to be replaced with one from Cresces' (number XV); Senecio
scratched his name inside, but in the nominative case, not the genitive,
as he had done in the other three.111 The similarity between
104
these two different rear horns and their implied interchangeability
suggests that saddlers used the same basic shape of horn for each saddle
they made.112
The Harness
It should by now be apparent that the evidence for the
reconstruction of Celto-Roman harness is both archaeological and
iconographic, the former providing various artefacts, the latter
suggesting ways in which they may have been used. The chief means of
fastening the saddle to the horse was provided by the girth strap; this
passed under the animal's belly and attached to the saddle on either
side. Modern saddles require leather 'flaps' to prevent the buckles
which attach the saddle to the girth from chafing either horse or rider,
so it is reasonable to expect the Roman system to have included
something similar.113 As we have seen, buckles that are large enough to
have served as girth fastenings are known, but it is usually difficult
to confirm their identification as such; however, the reliefs on the
triumphal arch at Orange depict what are presumably captured Gallic
saddles, and these have girth straps (some of which appear to be
decorated), each with a single large buckle.114 Girth straps are shown
on some tombstones, although usually without a buckle.115 Surviving
buckles suggest a width of between 50mm and 60mm for the girth strap
itself.116
The girth, however, serves only to hold the saddle onto the horse,
but it cannot prevent it from moving backwards or forwards on the
animal's back, and it is for this reason that modern horse equipment may
include the 'breastplate' and 'crupper'.117 The breastplate fastens to
the saddle and passes around the front of the animal, thus preventing
the saddle from slipping backwards. The crupper, on the other hand,
passes under the horse's tail and prevents the saddle moving forwards.
Thus the three elements, girth, breastplate, and crupper, serve to keep
the saddle firmly in place and provide a firm seat for the rider. In
modern harness, the girth may be connected to the breastplate by means
of a 'martingale', which may also be connected to the bridle in some
cases.118
Roman harness (Fig.25) appears to have used a similar system,
although differently constructed. Four important 'junctions' can be
discerned from the tombstone evidence, two on the haunches, and two on
the shoulders, whilst a fifth is found on the breast on some of the
horses.
a) The breast junction is normally composed of three straps, one
passing down to the girth, forming the 'martingale', and one on
either side connecting with the shoulder junctions.
b) The shoulder junction comprised three elements: one strap from the
breast junction, one passing towards the saddle, and one forming a
pendant strap, the function of which appears to have been purely
decorative. In some cases, an additional strap passed between the
junction and the saddle.
c) The haunch junction also comprised either three or four elements
105
and was a mirror image of the shoulder assembly: one strap passing
around the horse's rear to attach to the other haunch junction, one
attaching to the saddle D-rings, and one forming another pendant
strap. A fourth strap might pass to the saddle.
Together, a) and b) are equivalent to the breastplate, whilst c) is
the Romano-Celtic prototype of the crupper. This arrangement, suggested
by detailed examination of tombstones, is apparently confirmed by the
archaeological evidence.
With the exception of a few mineralised pieces on the rear of some
of the Xanten phalerae,119 there are no recognised examples of harness
straps surviving, possibly because the leather was oiled rather than
tanned.120 These show the strap to have been 20mm broad with a double
row of stitching c. llmm apart. Jenkins thought that the stitching
implied double thickness leather,121 but since cow hide - with a
thickness of about 5mm - was probably used, doubling would have proved
unnecessary; the stitching would simply help prevent stretching of the
straps.122 These dimensions are confirmed by examining the loops on the
rear of phalerae: widths of around 20mm and thickness of about 5mm are
found there too. It is possible that straps had rounded ends, which
would avoid the problem of the redundant corners of the straps becoming
tatty; leather straps with rounded ends (although almost certainly not
from horse harness) are known from Vindonissa.123
106
Fig.27: Types of Celto-Roman harness. A: half harness; B: full harness;
C: three-quarter harness (front); D: three-quarter harness
(rear).
Pendants
Why did Roman military riding harness (and, presumably, its Celtic
predecessor) include pendants? It has been suggested that lunulae from
Dangstetten may have been a direct introduction by Celtic cavalry. They
were certainly the earliest form of first-century pendant. More
importantly, they were (and still are) very powerful magical symbols. It
is, of course, possible that harness decoration was purely aesthetic in
nature, intended to make the cavalryman and his mount look as splendid
as possible, but it must be remembered that symbolism played an
important part in both Celtic and Roman superstition and religion.126
The lunula was symbolic of the moon (and femininity), and the
phalera may have been used to represent the sun (and masculinity).127
This imagery is enhanced in the early pendants by the use of boar's
tusks,128 but it may be that the trifid pendants of later years,
although containing a memory of the shape of these early examples, had
lost something of the meaning originally attached to them. 129 whatever
the case, there seems to be some reason at least for attributing an
apotropaic role to the decorative elements of horse harness. In the
nineteenth century, heavy-horse harness included horse brasses,
107
beautifully crafted pieces with significant symbols such as the cross,
the sun, a harvest scene, or the crescent moon (or lunula).130 It has
been suggested that these were protecting the horses from malign
influences.131
Pendants may have been something of a nuisance to a horse, in which
case some attempt may have been made to shield the animal from any
harmful abrasion. In heavy-horse harness, pendants were usually backed
by a shaped piece of leather and it is not inconceivable that similar
measures were adopted in Roman harness. Some pendants have rivet holes
in their body which may have been used to fasten them to just such a
backing, whilst breast phalerae may have used the martingale strap to
protect the horse from the pendant (Fig.24).132 Certainly, the pendants
found at Inota do not appear to have been free for much movement.133
The Bridle
The bridle is important, because it is one of the prime means the
rider has of controlling his mount, although control by means of his
legs will have been essential in combat - protecting himself
with a shield on his left side and using a weapon on the right meant
that the reins lay on the horse's neck at this crucial stage
(and this is usually indicated on the sculptural evidence - see above).
In modern equestrian practice, a variety of bits are available to suit
the temperament of the animal and the same appears to have been true in
the first century A.D., but it is important to remember that in combat,
the relationship between animal and rider would have been of paramount
importance and, at that point, the type of bit would have been largely
irrelevant.134
The basic elements of the bridle - cheekpieces, noseband, headband,
throatlatch - are all visible in the representational evidence (see
above). Some bridles, like that of Primigenius (Fig.9), were decorated
with small phalerae, which we may presumably identify with suspension
Type la. These pieces are usually quite diminutive, but the loop on
the rear face nevertheless suggests that, if our identification is
correct, the straps of the bridle were of similar (perhaps marginally
narrower) proportions to those of the rest of the harness.135
It was never the intention of this paper to go into the bridle or
bits in any detail, but it seems clear that the range of bit and
hackamore types speaks eloquently of the sophistication of the Celtic
horsemen.136
The Saddle-Cloth
Roman saddle-cloths or shabracques were apparently fringed in a
similar manner to vexilla. The shabracque was worn under the saddle and
covered the horse's flanks in most cases; the horses of Sacrius
Primigenius and Lucius (above, Nos.10-11) have extremely long examples
with a second, shorter cover on top (possibly over the saddle - see
above). Some other horses have their saddles covered by long cloths,
108
but the only sculptural representations of horses being ridden with long
saddle-cloths are on Trajan's Column (No.33).
Many aspects of the depiction of cavalry horses on the Column,
particularly when compared to Totenmahl sculpture, suggest that they are
modelled on the partially-understood equipment of horses in parade gear;
the extra-long saddle-cloths are one example, whilst the triplet straps
have turned into folds at the front and rear edge of the shabracque. 137
Similarly, the harness proper does not appear to have been understood by
the sculptors and is, in many examples, completely impractical.138
In the case of tombstones which do not have a saddle-cloth, it is
likely that it was added in paint. It should be noted that a number of
inconsistencies occur on tombstones, when girth and harness straps can
be shown passing under or over saddle cloths (see above).
The girth is shown passing beneath the saddle-cloth in so many
cases that it is necessary to ask whether it was not in fact fitted over
the saddle and girth, but under the harness straps.139 If this was
indeed the case, then there may have been a small under-blanket of some
kind.140
109
Fig.28: Techniques of saddle plate attachment to triplet straps. A:
Type 2 plates; B: Type 3 plates; C: detail of method of
fastening; D: cross-section.
Saddle Plates
Any discussion of the triplet straps must be related to a
consideration of so-called saddle plates. A series of rectangular
decorative plates found in the Doorwerth hoard were identified by
Holwerda with the objects depicted on the tombstone of M. Sacrius
Primigenius (above, No.10 and Fig.8). Brouwer cast doubt upon his
interpretation, but was unable to offer an alternative.143
Any reconsideration of the role of these objects must begin with
the sculptural evidence. Apart from the well-known stone of Primigenius,
that of Lucius (No.11 and Fig.10) may have had similar adornments added
in paint and Oluper (No.12 and Fig.11) has two rectangular plates half-
way down his triplet straps (see above). More importantly, the Celto-
Roman saddles on the arch at Orange (No.30 and Fig.19) have rectangular
plates affixed to their triplet straps.144
The fact that the sets of triplet straps seem to have come in two
widths, if we believe the evidence of the saddles themselves, and that
two standard widths of saddle plate are known145 must be more than
coincidence. The matter is surely put beyond reasonable doubt by the
examination of the rear faces of saddle plates, where provision for
attachment to three straps is the norm (Fig.28).146
These objects only appear on Totenmahl tombstones and the arch at
Orange, so it may be that their use was purely decorative: a full set
would have been extremely heavy and would have presented an unacceptable
nuisance to both horse and rider when in motion. Many Totenmahl stones
show the whole saddle being covered with a blanket (Figs.6; 13), so it
is possible that more pieces have been concealed in this way.
110
Other Details
Examination of the tombstones reveals a number of other aspects of
military horse equipment that are worth commenting upon. Many of these
features could not be reconstructed from the archaeological evidence, so
representational evidence comes into its own.
Tail binding occurs quite frequently on Roman funerary sculpture
from the provinces, consisting of a binding of material around the base
of the tail (Fig.17). Quite apart from presenting a pleasing 'streamer'
effect, it would help prevent the animal's tail becoming fouled. This
may be inherited from Celtic practice, if the Gundestrup cauldron is to
be believed (No.36 and Fig.20).
Neck straps are worn in many cases, again presumably for decorative
effect. Phalerae with suspension type 3a were almost certainly intended
to accommodate a neck strap. In a few instances, collars or neck bands
appear to have been worn under the front harness elements. In the case
of Bassus, the band sits under the breast straps, junctions, and
pendants, suggesting that it may have been intended to afford the horse
some protection from the decorations (Fig.3).
A few of the Totenmahl tombstones show the horses with their manes
gathered into knots along the neck, with one large poll-knot between the
animals ears. The latter is accompanied by a large phalera apparently
seated vertically (Fig.9). The poll-knot also seems to be shown on the
Saint-Rémy mausoleum.147
The Horse
Although not directly relevant to the subject matter of this paper,
the horses used by the Roman cavalry nevertheless deserve brief
consideration, if only because their size would have affected their
harness. The discovery of one or more cavalry cemeteries at Krefeld-
Gellep has provided data on the size of horses, which can be compared
with evidence from other sites, particularly Newstead.148 It is worth
stressing that representational evidence cannot be used reliably to
assess the size of Roman horses.149 The excavated data, on the other
hand, seems to suggest an average height of around 14 hands, most
specimens being between four and seven years old, but analysis of the
Krefeld-Gellep material also revealed a high proportion of male animals,
possibly stallions.150 This last point appears to be reflected on
cavalry representations, where examples of ungelded males are known. 151
Cavalrymen appear to have bought their mounts from the army for a
nominal fee and it is possible that the same was also true of their
harness and other equipment.152
111
of study.153 Most reconstructions chose to follow sculptural depictions
more or less closely, and a significant number of those used Trajan's
Column.
In 1985, Ian Jenkins published fully for the first time the hoard
of horse trappings (reputedly from Xanten) in the possession of the
British Museum. Believing that this constituted most of the fittings
from a single set of harness, Jenkins provided a reconstruction drawing
of the harness from which they came. The Xanten material contains only
four diagnostic phalerae, Al-4: Al has two rectangular loops to secure
it to the harness, two loops above the strap, and a hinge at the bottom
for its pendant (suspension type 3a); A2 and A3 are three-loop phalerae
(type 4a); A4 is a four-loop junction, with one loop on one side, two on
the other, and one at the bottom (type 5a). Jenkins' reconstruction uses
A2 and A3 as the shoulder junctions, with one strap passing horizontally
under the saddle blanket; on the other hand, he uses A4 (and its missing
partner) for the haunch junctions, with one strap passing horizontally
under the blanket, one directed towards the saddle (but again under the
blanket), and one forming the breeching.154 Although some
representations show straps apparently joining under the saddle blanket,
these are seldom the most reliable. As such, the harness makes no sense,
unlike those of Bassus and Romanius, which are perfectly functional.
If the Xanten fittings all belong to one set of harness, then the
evidence of the phalerae suggests that it was a three-quarter harness.
However, there are problems with interpreting the pieces in the
collection as one homogeneous set. The fact that different names are
punched on some of the pieces may be of no consequence, but it is
interesting that pendant B6 bears the vertical inscription 'VERCUNDI',
whilst Al reads 'T. CAPITONI MARIAN' horizontally. Moreover, the scheme
of decoration on the two objects is markedly different. 155 At least
three decorative types of small phalera are present in the group,156 so
it is not impossible that the Xanten material, like that from Doorwerth,
is a collection of bits of harness, rather than one complete set.157
112
Fig.29: Reconstruction of typical Tiberio-Claudian harness.
113
Fig.30: Reconstruction of typical Flavian harness.
laid upon missile handling in the manoeuvers described in Arrian's
Techne Taktika.166
In the early stages, cavalry units were apparently taken into the
army from friendly tribes and commanded by tribal nobles. This practice
led to units being known by their first commander's name even when they
were thoroughly integrated into the Roman army.167 Even ordinary
cavalrymen must have been wealthy by the standard of the day, since they
would have to have provided their own equipment.168 Moreover, the
relative affluence of cavalrymen seems to have continued under Roman
control, because the pay of auxiliary cavalry was probably superior to
that of their fellow (auxiliary) infantry and it is noteworthy that over
50% of figured military tombstones of the first century A.D. from
Britain and the Rhineland show mounted troops of some sort.169
Apart from legionary horsemen and the troopers of the alae and
cohortes equitatae, officers also used horses. The tombstone of the
centurion T. Calidius Severus illustrates this,170 but the equestrian
and senatorial officers of auxiliary and legionary units were likewise
mounted,171 so it is obviously possible that at least some of the
surviving 'cavalry' equipment belonged to them.172 However, the
proportions of these compared to troopers in the cavalry units of the
army makes it more likely that the material found did belong to the
114
Fig.31: Reconstruction of typical Flavian parade harness, showing horse
on long-reins.
cavalry (unless there is some unseen bias affecting the way that
material is deposited in the archaeological record, and the evidence
would seem to suggest that this was not so).173
It is fairly clear that two distinct traditions of niello
decoration of military equipment were to be found in the first century
A.D. In the first place, belt plates display one particular grammar of
ornament (largely botanical in origin),174 the combination of which
appears to have been characteristic of infantry (and this was also
picked out, albeit somewhat stylised, by sculptors175). In addition, it
would seem that these plates were usually tinned.176 Elements of horse
harness, on the other hand, reveal a different decorative tradition,
closely linked with viticulture; moreover, these objects are usually
silvered, rather than tinned.177
It is therefore extremely interesting when items of harness are
found that appear to fall into the first of these two traditions. Almost
invariably, these pieces do not conform to the most commonly found types
of harness fitting.178 This might be seen as hinting at differentiation
between horses in infantry units and the true cavalry of the auxilia,
but it is hardly proven.
115
In the final analysis, it must be the numerical argument that
convinces us that most elements of horse harness recovered from the
archaeological record belonged to the troopers of the cohortes equitatae
and alae. However, this assumption begs the question of whether such
material could not also belong to civilian harness.
It seems fairly certain that in both the Celtic and Roman worlds,
horse riding was largely the province of the nobility and was frequently
associated with war. We have already seen that Celtic nobles formed the
backbone of early Imperial cavalry, and we need only note that, in Roman
society, the nouveau-riche class were known as equites, or that it was
young aristocrats who participated in the 'Trojan Game'.179 our evidence
for civilian horse-riding in the provinces during the first century A.D.
is negligible, so again the numerical argument, plus the military
context, suggests that it is unlikely to be 'civilian' horse harness
that is normally found.180
CONCLUSION
The pictorial and archaeological evidence all seems to point to the
fact that Celto-Roman cavalry (or riding) harness was completely
practical and not merely somewhere to hang decorations. Moreover, it
seems to confirm that the Romans adopted the equipment of the 'Celtic'
cavalry wholesale, since the functional parts of the harness do not
change significantly between the time of the Gundestrup cauldron and the
tombstone of Bassus, probably well over a century later. However, this
is not to deny that change did take place: if decorations were only
lunulae and phalerae when Celtic cavalry first saw service in the
Imperial army, they soon took on a developmental life of their own,
establishing a tradition that was independent of their original Celtic
heritage and their new-found allegiances.
It has hopefully been demonstrated that there is little reason for
confusion over the attribution of horse equipment recovered from the
archaeological record; the majority of it must, logically, have
originated with the Celtic cavalry of the alae and cohortes equitatae,
and thus it is not misleading to talk of it as 'cavalry equipment.'
Moreover, some pointers have been offered towards the identification of
trappings from the horses of officers and those belonging to vehicle
harness.
ACKNOWLEDGEMENTS
Since its original composition in 1982, many people have helped me
with this study of Roman cavalry equipment. David Kennedy and Phillip
Freeman were kind enough to read drafts of the first version, whilst the
revised text was read and commented upon by Jon Coulston and Martha
Andrews, for which I thank them. Peter Connolly has been the source of
much fruitful discussion on the subject of Celto-Roman harness and has
shared with me the results of his experiments, as well as redrawing my
rather lame Roman horse. I am grateful to the former and present
curators of the Hadrian's Wall Museums, Bill Hubbard and Georgina
Plowright respectively, for permission to publish the Corbridge finds.
Mr Hubbard also produced the illustrations of these objects. John
116
Hedges and West Yorkshire Archaeological Services permitted me to
examine the Castleford material, whilst I have received assistance from
Canterbury Archaeological Trust, the Grosvenor Museum Chester, London
Museum, Peterborough City Museum, and Wiesbaden Museum, when examining
collections of cavalry equipment. Others who have aided me at some point
include Lindsay Allason-Jones, Marijke Brouwer, Mike Dobby, Liviu
Petculescu, and Pat Southern. This seems the appropriate place to
acknowledge the various museums who have permitted me to examine and
photograph their cavalry tombstones. The photographs are reproduced here
by kind permission of the following museums:
Römisch-Germanisches Museum, Köln (Köln Museum)
Rheinisches Landesmuseum, Bonn (Bonn Museum)
Mittelrheinisches Landesmuseum, Mainz (Mainz Museum)
Städtisches Reiss-Museum, Mannheim (Mannheim Museum).
I should like to thank all of those mentioned above for their help.
If there is any merit in this work, then it is surely due to their
unstinting efforts, whilst all remaining shortcomings must remain my own
responsibility.
117
APPENDIX 1: THE PUBLICATION OF FINDS
During the preparation of this article, it has proved necessary to
rely very heavily upon published archaeological evidence. Consequently,
some of the shortcomings of this corpus have become all too apparent, so
it is the intention of this brief appendix to outline some of these and
to suggest ways in which the publication of cavalry equipment (and,
indirectly, military equipment in general) could be improved.
Many archaeological reports, both old and more recent, suffer from
deficiencies in their descriptions of items, particularly in the
provision of accurate (or even any) dimensions. In most cases where
measurements have been given in this paper, they have had to be provided
by measuring illustrations with Vernier callipers and then scaling up
the results - clearly not a very satisfactory technique, and yet a
majority of small-find reports still ignore vital measurements.181
Illustrating cavalry equipment by means of photographs, as was
common in earlier reports, introduces complicated distortions due to the
imperfection of even the best camera lenses.182 Although it is possible
to correct such deficiencies by means of a computer, such a task seems
unnecessarily complex purely in order to obtain the true dimensions of
an artefact. Measured archaeological illustrations are better, but even
then a proportion of these, where it was possible to check published
dimensions with those provided by Vernier callipers (even allowing for
rounding errors), proved to be inaccurate. However, these are the
examples that are reproduced with the correct reduction: some are
wrongly reproduced or labelled and this sort of mistake can only be
detected by studying large amounts of material and having some idea of
what to expect.183
It is common to answer such criticisms by saying that the expert
must study the original collections, but this surely negates the whole
point of publishing archaeological finds in the first place? It is,
moreover, ignoring the fact that it is perfectly possible to produce
useful and informative reports on artefacts. Rigorous standards need to
be enforced for archaeological small-find illustration and the path has
already been marked out by a round-table organised in France,184 where
the principles of publishing a line drawing at 1:1 (with a scale) along
with a photograph were established. It also should be added that views
from all angles are essential, particularly for the study of phalerae
(many were rendered useless for the purposes of the present study
because neither the text nor the illustration gave any idea of what lay
on the reverse face of the object).185
To this end, the following guidelines for the publication of Roma-
cavalry equipment may be suggested:
1. Line drawings of the front and reverse face of an object, preferably
at 1:1, but in every case to include a drawn scale, rather than a
captioned one.186
2. Cross-section of the object, particularly useful in the case of
phalerae (but also very rare at the moment - most of the examples
included in this paper are reconstructed from highlights in the
118
illustration).187
3. A concise textual description of the item, any ancient or modern
damage, corrosion, and any decoration remaining.188
4. Detailed measurements, which should always include maximum width (or
diameter), maximum height, and maximum thickness, but should also
include any other important dimensions.
5. In addition to these requirements, a photograph of the object is
sometimes useful, if only to gauge how accurately the surface detail
has been rendered by the graphic artist.189
An example report of this nature is included (below, Appendix 2,
although the descriptions are deliberately more elaborate than would be
normal); it is to be hoped that these basic standards will at least be
considered by archaeologists, for the present lack of any conformity
only serves to hinder the study of Roman military horse equipment.
119
APPENDIX 2: SOME UNPUBLISHED HORSE EQUIPMENT FROM CORBRIDGE,
NORTHUMBERLAND
Amongst the unpublished and unstratified material from the Roman
site at Corbridge now held in the site museum, are a number of pieces of
Roman harness equipment. Eight junction loops and one junction phalera
of common first or early second century A.D. forms are published here
for the first time. No information about the provenance of the objects
survives, but they are almost certainly from Corbridge and presumably
come from the 1906-14 campaigns of excavation, although no explicit
reference is made to any of them in the published reports.
1. 'Spectacle' junction loop (type 1d; inv.no.75.1338). Fig.32. The
loop is U-sectioned, with a shallow scarp at the junction with the
body of the fitting; this part is decorated with a moulding in the
form of a broad relief band bordered by two narrower bands. There
are traces of an incised triangle on the loop, with its broadest
end against the moulding. The loop shows signs of wear at the apex,
but it has fractured where the loop has been bent under to form the
return. To judge by the nature of the fracture, it is possible that
the damage occurred subsequent to its discovery. The body is the
classic figure-of-eight shape with a slightly convex upper face
(and flat rear face), although there is slight damage to the end of
the object. At the broadest points of the fitting, two flat
circular recesses surround the rivet holes. A dome-headed rivet
survives in situ in the area nearest the loop, whilst traces of the
other rivet head are to be seen around the edges of its recess. The
body has been bent up slightly at the loop end, causing the rivet
head not to sit quite flush with the surface. Patinated with no
areas of bright metal.
Maximum length: 51mm; maximum width of body: 14mm; maximum width of
loop: 7mm; internal diameter of loop: 8mm; height of scarp between
loop and body: 2mm; thickness of body: 1.75mm; diameter of rivet
head: 11mm; height of rivet head: 5mm; diameter of rivet hole: 3mm
2. 'Spectacle' junction loop (type lc; inv.no.75.1344). Fig.32. This
fitting has a U-sectioned loop with a high scarp between the loop
and body, the scarp being surmounted by a moulding consisting of a
narrow (rather crude) beaded band with a narrow plain band on
either side of it. Corrosion makes it impossible to be certain that
an incised triangle is present, or whether there are any traces of
wear near the apex, which is the point at which this fitting has
fractured. The body is flat and bears no relief decoration; it is
in the form of a combination of the figure-of-eight with stylised
volutes. The centres of the two circular areas have been pierced
(almost certainly punched) from the upper surface, whilst the lower
face shows signs of having had the edges of the holes burred. The
end of the body nearest the loop has again been bent slightly
upwards. Heavily patinated, with severe pitting on the loop itself.
Maximum length: 63.5mm; maximum width of body: 15mm; maximum width
of loop: 10mm; height of scarp between loop and body: 6mm;
thickness of body: 1mm; diameter of rivet holes: 2mm
3. 'Spectacle' junction loop (type lc; inv.no.75.1339). Fig.32. Fitting
with a massive U-sectioned loop with a high scarp at the junction
with the body, this being surmounted by a similar moulding to that
120
121
Fig.32: Junction loops from Corbridge (Nos.13). Scale 1:1.
of No.2, although no beading can be discerned. The incised triangle is
clearly visible. Signs of wear are only clear on one side of the loop,
which may suggest idiosyncratic use. The loop is broken on the
underside, almost certainly before loss. The lower end of the body, with
its rivet area, is missing, but it is broadly similar to that of No.2,
although the volutes are not present, the upper surface is slightly
convex, and the circle around the surviving rivet hole has been slightly
recessed. No signs of burring on the rear face around the rivet hole.
Heavily patinated and slightly pitted on the loop.
Maximum length: 47mm; maximum width of body: 13mm; maximum width of
loop: 10mm; internal diameter of loop: 9.5mm; height of scarp between
loop and body: 8mm; thickness of body: 2mm; diameter of rivet hole: 3mm
4. 'Spectacle' junction loop (type lc; inv.no.75.1340). Fig.33. This
fragment is recognisable as being closely akin to No.3. The U-
shaped loop is decorated with an incised triangle and there is a
high scarp between the loop and the body of the fitting, consisting
of a medium band with narrow bands on either side of it. The loop
has been bent up beneath the body in such a way as to suggest that
the fitting was wrenched out of shape; the 'return1 of the loop, to
which the rivets fastened, has been torn and distorted where it is
broken off. There are signs of wear around the apex of the loop.
Only a small fragment of the body around one rivet hole survives
and that is bent at an acute angle in relation to the loop. The
body is made of extremely thin metal, possibly the reason for its
failing. Hardly any patination; 'bright' metal objects from
Corbridge normally originate in anaerobic conditions, since the
soil conditions are so hostile to copper alloy artefacts, and one
of the various ditches discovered by Forster and Knowles seems the
most likely provenance for this item.190
Maximum length: 29mm; maximum width of body: 10mm; maximum width of
loop: 9mm; internal diameter of loop: 9mm; height of scarp between
loop and body: 6mm; thickness of body: 0.25mm
5. 'Spectacle' junction loop (type 1d?; inv.no.75.1342). Fig.33. The
loop does not appear to be hollow on this example and there is no
trace of an incised triangle on its surface. The moulding at the
junction between the loop and body is a broad band with a narrow
one above and below it, but there is no significant scarp. There
are no clear signs of wear at the apex of the loop, but this area
is quite heavily corroded. The body is slightly convex and broken
off below the first rivet base, the domed rivet head still being in
place. The object is heavily patinated, as well as corroded in some
areas (notably the rear face of the loop and body).
Maximum length: 33mm; maximum width of body: 12mm; maximum width of
loop: 6mm; internal diameter of loop: 5.5mm; height of scarp
between loop and body: 0.5mm; thickness of body: 1.5mm; diameter of
rivet head: 10mm; height of rivet head: 2mm
6. 'Spectacle' junction loop (type 1d?; inv.no.75.1341). Fig.33. A
fragmentary example which appears to have been (mechanically?)
cleaned fairly vigorously at some stage since recovery. The loop
bears no trace of the normal incised triangle. The moulding is
unlike that of the previous junction loops, consisting of a broad
U-shaped groove with double narrow bands above and below it. These
122
Fig.33: Junction loops from Corbridge (Nos.48). Scale 1:1.
123
are not executed in very high profile and the scarp between the loop
and body is not very pronounced. There are no signs of wear at the
apex of the loop. What remains of the body - essentially just part
of one rivet base - is bent upwards. An incised circle is
discernible around the rivet hole, but the body metal is quite thin
(possibly as a result of the cleaning). There is patination within
the pitting on the surface of the object, but otherwise it displays
'bright' metal.
Maximum length: 23mm; maximum width of body: 10mm; maximum width of
loop: 7.5mm; internal diameter of loop: 8mm; height of scarp between
loop and body: 2mm; thickness of body: 0.25mm; diameter of rivet
hole: 2.5mm
7. 'Double spectacle' junction loop (type 3a: inv.no.75.1343). Fig.33.
The loop of this fitting has a very shallow U-shaped section and
meets the body in a flush junction on the upper face, whilst on the
lower there is a massive lug, and it may have been intended that the
strap should butt against this. The loop itself has fractured below
the apex before loss. The body flares in what are probably degraded
volutes between the loop and the first pair of rivet holes, the
bases of which are slightly recessed into the body of the fitting,
which is, in turn, almost convex. The rivet holes are surrounded by
an incised circle, with the left hand circle overlapping the right
(when seen with the loop at the top). There is no sign of burring on
the rear face of the rivet holes. Between this pair of rivet bases
and the next is a massive piece of moulded decoration, consisting of
a broad band with two pairs of narrow bands above and below it, the
outermost bands being slightly more pronounced than the inner ones.
The back of this moulding is hollow, and there is a steep scarp on
either side at the junction with the body. The body is terminated
with another dual rivet base, this time with the right-hand incised
circle overlapping the left. A small rectangular tongue of metal
finishes the item at the bottom. This item is heavily patinated with
quite heavy corrosion on the loop.
Maximum length: 52mm; maximum width of body: 19mm; maximum width of
loop: 10mm; length of moulding: 11mm; height of scarps between
moulding and body: 3mm; thickness of body: 1.75mm; thickness at
moulding: 6.5mm; diameter of rivet holes: 2mm
8. 'Double spectacle' strap fitting, probably a junction loop (type 3a?;
inv.no.75.1346). Fig.33. There is no loop surviving for this
fragment, but it is so similar to No.7 that it is probably from the
same hand. Details are virtually (but not exactly) the same as for
the previous example. The object has fractured across the upper pair
of rivet holes (which is a good reason for suspecting that it was
originally a junction loop) and is heavily patinated.
Maximum length: 29mm; maximum width of body: 18mm; length of
moulding: 10mm; height of scarps between moulding and body: 3mm;
thickness of body: 1.75mm; thickness at moulding: 6.5mm; diameter of
rivet holes: 2mm
9. Junction phalera (inv.no.75.1348). Fig.34. Concave (although flatter
towards the centre) copper alloy phalera of decorative type 5a, with
four rings on its rear face (functional type 6a). Within the central
dish of the phalera, a broad incised line surrounds the central
rivet hole. There are traces of silver foil adhering to the front
124
125
Fig.34: Junction phalera from Corbridge (No.9). Scale 1:1.
face of the object, but no remains of niello inlay can now be
discerned. The object appears to be cast, although it is not clear
whether the rings were formed as one with the piece or brazed on
after casting. The phalera is slightly bent about what is probably
the horizontal axis; otherwise, there is no sign of damage to the
object, although there is a small amount of corrosion under the
broad incised line, where the metal is thinner. On the rear face,
there are what may be file marks around the bases of the rings. The
object was marked '1996.' at some point subsequent to excavation.
Heavily patinated and slightly corroded.
Diameter: 77mm; diam. of central incised circle: 28mm; diam. of
central hole: 1.5mm; external widths of rings: 15mm; int. widths of
rings: 6-7mm, height of rings: 13-14mm; distance of rings from
perimeter: 9mm
The six spectacle type loops represented here are all individuals
and none appear to be from the same hand, let alone the same melt. As
such, they are similar in so far as they are 'in the tradition of'
spectacle type fittings. The double-spectacle fittings, on the other
hand, show a remarkable degree of similarity, although they are unlikely
to be from the same mould, since the cire-perdue method meant that each
casting was truly unique and the mould had to be broken to extract the
object.191 Objects may resemble each other when a series of castings
were made using the same master to form the mould (a method that has
been advocated by Oldenstein for the transmission, by copying, of
designs in the 2nd and 3rd centuries on the German and Raetian
limesl92). The differences between Nos.7 and 8 are small, but such that
they cannot be put down to idiosyncracies of the mould. It is, however,
possible that they are due to the finishing process (removing sprues,
filing off casting flashes). Nevertheless, it seems fairly certain that
these two pieces are the work of the same workshop and probably the same
craftsman (insofar as they probably derive from a common original used
to form their moulds). It is interesting to note in passing that none of
the pieces examined is intact.
There are two broad traditions of strap fitting represented by
these eight objects, but within these there are seven individual
designs. Such artefacts, seemingly so insignificant, nevertheless
exhibit individuality to such a degree that the notion that designs may
be characteristic of particular workshops cannot be ignored. 193 In that
sense, cavalry harness fittings might plausibly be diagnostic of the
unit that discarded them. Only detailed analysis of fittings like these
on a large scale will enable us to confirm this hypothesis.
126
APPENDIX 3: CATALOGUE AND TYPE SKETCHES
The catalogue of cavalry harness equipment presented here, along
with the type sketches which accompany it (Figs.35-56), can be used to
assist in the identification of archaeological finds. However, it is
necessary to stress that, whilst every effort has been made to make them
as comprehensive as possible, there are bound to be published examples
which have been missed, as well as a great many unpublished pieces. The
type sketches have been deliberately simplified: they are not intended
as accurate small finds drawings, but should be used simply as a guide
to form. Full references are given in the tables for each type
illustrated and the researcher is recommended to examine the original
publication if parallels are sought.
SADDLE HORNS
Type Site Component Height Reference
(mm)
1 Haltern Front 210.0 MAKW, 1905, Taf.XIX,8
1 Mainz Weisenau Rear Left 158.0 KESSLER, 1940, Abb.7,5
1 Mainz Weisenau Front 198.0 Ibid., Abb.7,6
1 Mainz Weisenau Rear Right 170.0 Ibid., Abb.7,7
1 Neuß Rear Right ? TAUCH, 1983, 10 (figure)
1 Newstead, Pit XXII Front 197.0 CURLE, 1911, Pl.XXXII
1 Newstead, Pit XXII Front 197.0 Loc. cit.
1 Newstead, Pit XXII Rear Left 185.0 Loc. cit.
1 Newstead, Pit XXII Rear Right 185.0 Loc. cit.
1 Newstead, Pit XXVII Front 200.0 Loc. cit.
1 Newstead, Pit XXVII Front 200.0 Loc. cit.
1 Newstead, Pit XXVII Rear Left 194.0 Loc. cit.
1 Newstead, Pit XXVII Rear Right 196.0 Loc. cit.
1 Rottweil Front 206.0 PLANCK, 1975, Taf.78,1
1 Rottweil Front 206.0 Ibid., Taf.78,2
1 Rottweil Rear Left 182.0 Ibid., Taf.77
1 Rottweil Rear Right 176.0 Ibid., Taf.76
1 Sheepen, region 3 Rear Right? 163.0 HAWKES & HULL,1947, Pl.CI,a; b
1? Sheepen, near A4 ? 175.0 Ibid., Pl.CI.c; d
TABLE 1
127
Fig.35: Type sketches for saddle horns.
(scale in mm)
Fig.36: Type sketches for girth buckles.
(scale in mm)
128
GIRTH BUCKLES
Table 2
129
(scale in mm)
Fig.37: Type sketches for saddle plates, Types 13.
130
SADDLE PLATES
Table 3
131
(scale in mm)
Fig.38: Type sketches for saddle plates, Types 46.
132
3b Doorwerth 84 101 Ibid., Nr.196
3b Doorwerth 120 80 Ibid., Nr.225
3b Doorwerth 122 132 Ibid., Nr.224
3b Doorwerth 122 79 Ibid., Nr.226
3b Doorwerth 127 135 Ibid., Nr.241
3b Neuß 73 101 LEHNER, 1904, Taf.XXXB,58
3c Castleford 114 125 Plate B: report forthcoming
Further fragments of type 6 plates are known from Moers-Asberg (BECHERT, 1974, Abb.71,6),
Wiesbaden (ORL Nr.31, Taf.X,58; 63), Hofheim (RITTERLING, 1904, Taf.III,17), Aislingen
(ULBERT, 1959, Taf.19,18), Rheingönheim (ULBERT, 1969, Taf.28,14; 16), and Vindonissa
(UNZ, 1973, Abb.11,118)
Table 3
133
JUNCTION RINGS
Site Ext. Int. No. of Reference
Diam. Diam. Loops
(mm) (mm)
Augsburg- 39 21 4 HÜBENER, 1973, Taf.13,1
Oberhausen
Birten 35 20 4 HINZ, 1972, Bild 12,16
Hüfingen 39 21 4 REVILLIO, 1929, Abb.4,2
Kempten 46 21 4 KRÄMER, 1957, Taf.16,23
Lincoln 30 17 4 WEBSTER, 1949, Pl.X,a
The Lunt 41 25 4 HOBLEY, 1969, Fig.20,3
Moers-Asberg 43 23 4 BECHERT, 1974, Abb.71,1 & 72
Nijmegen 38 20 4 BOGAERS & HAALEBOS, 1975, Fig.12
Strasbourg 45 24 4 FORRER, 1927, Taf.LXXVI,Q
Strasbourg 42 21 4 Ibid., Taf.LXXVII,l
Strasbourg 37 22 4 Ibid., Taf.LXXVII,2
Valkenburg 50 28 4 GLASBERGEN & GROENMAN-VAN
WAATERINGE,Pl.14,48
Table 4
134
Fig.39: Types sketches (decorative) for phalerae, Types 13.
135
PHALERAE TYPES
Dec. Susp. Provenance Diam Reference
Type Type (mm)
la la Doorwerth 44.0 BROUWER, 1982, Nr.106
la la Doorwerth 43.0 Ibid., Nr.107
la la Doorwerth 45.0 Ibid., Nr.108
la la Doorwerth 45.0 Ibid., Nr.109
la la Doorwerth 45.0 Ibid., Nr.110
la la Doorwerth 40.0 Ibid., Nr.1ll
la la Doorwerth 44.0 Ibid., Nr.112
la la Doorwerth 42.0 Ibid., Nr.113
la la Doorwerth 44.0 Ibid., Nr.114
la la Doorwerth 45.0 Ibid., Nr.115
la la Doorwerth 40.0 Ibid., Nr.116
la la Doorwerth 45.0 Ibid., Nr.117
la la Doorwerth 43.0 Ibid., Nr.118
la la Doorwerth 45.0 Ibid., Nr.119
la la Doorwerth 43.0 Ibid., Nr.120
la la Doorwerth 43.0 Ibid., Nr.121
la la Doorwerth 45.0 Ibid., Nr.122
la la Doorwerth 45.0 Ibid., Nr.123
la la Doorwerth 44.0 Ibid., Nr.124
la la Doorwerth 44.0 Ibid., Nr.125
la la Doorwerth 42.0 Ibid., Nr.126
la la Doorwerth 42.0 Ibid., Nr.127
la la Doorwerth 43.0 Ibid., Nr.128
la la Doorwerth 45.0 Ibid., Nr.129
la la Doorwerth 45.0 Ibid., Nr.130
la la Doorwerth ? Ibid., Nr.131
lb lb Doorwerth ? BROUWER, 1982, Nr.132
lb lb Doorwerth 43.0 Ibid., Nr.133
lb lb Doorwerth 44.0 Ibid., Nr.134
lb lb Doorwerth 43.0 Ibid., Nr.135
lb lb Doorwerth 44.0 Ibid., Nr.136
lb lb Doorwerth 43.0 Ibid., Nr.137
lb la Xanten 43.0 JENKINS, 1985, NO.C26
lb la Xanten 43.0 Ibid., NO.C27
lb la Xanten 43.0 Ibid., NO.C28
lb la Xanten 43.0 Ibid., NO.C29
lb la Xanten 43.0 Ibid., NO.C30
lb la Xanten 44.0 Ibid., No.C31
lb la Xanten 40.5 Ibid., No.C32
lb la Xanten 40.5 Ibid., No.C33
lb la Xanten 41.5 Ibid., No.C34
lb la Xanten 41.0 Ibid., NO.C35
lb la Xanten 39.0 Ibid., No.C36
lb la Xanten 38.0 Ibid., No.C37
lb la Xanten 39.0 Ibid., NO.C38
lb la Xanten 39.0 Ibid., No.C39
1c 3c Doorwerth 89.0 BROUWER, 1982, Nr.176
1c 1d Doorwerth 85.0 Ibid., Nr.177
1c 1d Doorwerth 89.0 Ibid., Nr.178
1c ? Nawa 75.0 ABDUL-HAK, 1955, Pl.XI.l
1c 6b? Nawa 75.0 Ibid., Pl.XI.l
1c ? Nawa 75.0 Ibid., Pl.XI.l
1c 3a Xanten 100.5 JENKINS, 1985, No.Al
1c 4a Xanten 105.0 Ibid., No.A2
1c 4a Xanten 105.0 Ibid., No.A3
1c 6b Xanten 101.0 Ibid., No.A4
1d 2a Doorwerth 60.0 BROUWER, 1982, Nr.147
1d 2a Doorwerth 60.0 Ibid., Nr.148
1d 2a Doorwerth 60.0 Ibid., Nr.149
Table 5
136
Fig.40: Types sketches (decorative) for phalerae, Types 412.
137
1d 2a Doorwerth 60.0 Ibid., Nr.150
1d 2a Doorwerth 61.0 Ibid., Nr.151
1d 2a Doorwerth 60.0 Ibid., Nr.152
1d 2a Doorwerth 60.0 Ibid., Nr.153
1d 2a Doorwerth 60.0 Ibid., Nr.154
1d 2a Doorwerth 58.0 Ibid., Nr.155
1d 2a Doorwerth 57.0 Ibid., Nr.156
1d 2a Doorwerth 60.0 Ibid., Nr.157
1d 2a Doorwerth 63.0 Ibid., Nr.158
1d 2a Doorwerth 61.0 Ibid., Nr.159
1d 2a Doorwerth 61.0 Ibid., Nr.160
1d 2a Doorwerth 61.0 Ibid., Nr.161
1d 2a Doorwerth 61.0 Ibid., Nr.162
1d 2a Doorwerth 61.0 Ibid., Nr.163
1d 2a Doorwerth 60.0 Ibid., Nr.164
1d 2a Doorwerth 58.0 Ibid., Nr.165
1d 2a Doorwerth 60.0 Ibid., Nr.166
1d 2a Doorwerth 59.0 Ibid., Nr.167
1d 2a Doorwerth 58.0 Ibid., Nr.168
1d 2a Doorwerth 59.0 Ibid., Nr.169
1d 2a Doorwerth 60.0 Ibid., Nr.170
1d 2a Doorwerth 61.0 Ibid., Nr.171
1d 2a Doorwerth 59.0 Ibid., Nr.172
1d 2a Doorwerth 63.0 Ibid., Nr.173
1d 2a Doorwerth 62.0 Ibid., Nr.174
1d 2a Doorwerth 62.0 Ibid., Nr.175
1e 2b Doorwerth 38.0 Ibid., Nr.179
1e 2b Doorwerth 39.0 Ibid., Nr.180
le 2b Doorwerth 39.0 Ibid., Nr.181
le 2b Doorwerth 39.0 Ibid., Nr.182
le 2h Fremington Hagg 37.0 WEBSTER, 1971, No.24
le la Hod Hill 34.5 BRAILSFORD, 1962, Fig.5, A126
le 1d? Hüfingen 31.0 REVILLIO, 1929, Abb.2,2
le 3c Hüfingen 34.5 Ibid., Abb.2,6
le ? Lixus 28.0 BOUBE-PICCOT, 1964, No.14
le 2a Neuß 40.5 LEHNER, 1904, Taf.XXXIV,16
le 2b Newstead 70.0 CURLE, 1911, 300-1, Pl.LXXIV,l
le 2b Newstead 75.0 Ibid., 300-1, P1.LXXIV,2
le 2b Newstead 78.0 Ibid., 300-1, P1.LXXIV,3
le 2h? Sheepen 34.0 HAWKES & HULL, 1947, Pl.CII,30
le 2h? Sheepen 38.0 Ibid., Pl.CII,31
le 2d Wroxeter 40.0 BUSHE-FOX, 1916, Pl.XVIII,30
1f 6b? Bonn 56.0 AuhV 4, Taf.45,4
1f 3a Doorwerth 105.0 BROUWER, 1982, Nr.138
1f 3a Doorwerth 107.0 Ibid., Nr.139
1f 4a Doorwerth 108.0 Ibid., Nr.140
1f 4a Doorwerth 108.0 Ibid., Nr.141
1f 4a Doorwerth 105.0 Ibid., Nr.142
1f 4a Doorwerth 108.0 Ibid., Nr.143
1f 6a Doorwerth 106.0 Ibid., Nr.144
1f 6b Doorwerth 107.0 Ibid., Nr.145
1f 6a Doorwerth 105.0 Ibid., Nr.146
1f 2a Xanten 54.5 JENKINS, 1985, No.B5
1f 2a Xanten 55.0 Ibid., No.B6
1f 2a Xanten 53.0 Ibid., No.B7
1f 2a Xanten 53.0 Ibid., No.B8
1f 2a Xanten 52.5 Ibid., No.B9
1f 2a Xanten 55.0 Ibid., No.B10
1f 2a Xanten 54.5 Ibid., No.B11
1f 2a Xanten 55.0 Ibid., No.B12
1f 2a Xanten 55.0 Ibid., No.B15
1f 2a Xanten 54.5 Ibid., No.B16
1f 2a Xanten 56.0 Ibid., No.B18
1f 2a Xanten 54.5 Ibid., No.B20
1f 2a Xanten 55.0 Ibid., No.B21
Table 5
138
(scale in mm)
Fig.41: Type sketches (functional)for phalerae, Types 13.
139
1f 2a Xanten 54.5 Ibid., No.B22
1f 2a Xanten 50.5 Ibid., No.B23
1g 3c Fremington Hagg 73.0 WEBSTER, 1971, No.3
1g 3d Fremington Hagg 71.0 Ibid., No.16;
CRADDOCK et al., 1973, Fig.9
1g 3d Fremington Hagg 54.0 WEBSTER, 1971, No. 17;
CRADDOCK et al., 1973, Fig.10
1g 2h Seven Sisters 53.0 DAVIES & SPRATLING, 1976, No.2
140
Table 5
5a la Newstead 31.0 CURLE, 1911, 300, Pl.LXXIII,1
5a 2b Newstead 67.0 Ibid., 300-1, P1.LXXIV,8
5a 3e Oberstimm 72.0 SCHONBERGER, 1978, Taf.24,B203
5a la Rißitissen 34.0 ULBERT, 1970, Taf.1,18
5a 2e Tanger 42.0 BOUBE-PICCOT, 1964, No.18
5a 2d Tanger 40.0 Ibid., No.19
5a 2e Thamusida 38.0 Ibid., No.13
5a 2f Vindonissa 28.0 UNZ, 1973, Nr.146
5b 2g Fremington Hagg 34.5 WEBSTER, 1971, No.23
5b 3c Hüfingen 32.0 REVILLIO, 1929, Abb.2,5
5b 5a? Hüfingen 66.5 Ibid., Abb.2,8
5b 2b Volubilis 45.0 BOUBE-PICCOT, 1964, No.l
5c 6a Buciumi 74.0 CHIRILA et al., 1972, Taf.LXXX
(scale in mm)
Fig.42: Type sketches (functional) for phalerae, Types 4-7
141
Fig.43: Type sketches for pendants, Types 1at.
142
PENDANTS
Type Site Width( Height( Reference
mm) mm)
la Doorwerth 120.0 145.0 BROUWER, 1982, Nr.138
la Doorwerth 120.0 139.0 Ibid., Nr.139
la Doorwerth 73.0 89.0 Ibid., Nr.140
la Doorwerth 70.0 91.0 Ibid., Nr.141
la Doorwerth 70.0 93.0 Ibid., Nr.142
la Doorwerth 70.0 94.0 Ibid., Nr.143
la Doorwerth 70.0 94.0 Ibid., Nr.144
la Doorwerth 74.0 94.0 Ibid., Nr.145
la Doorwerth 74.0 92.0 Ibid., Nr.146
la Doorwerth 71.0 88.0 Ibid., Nr.147
la Doorwerth 73.0 88.0 Ibid., Nr.148
la Doorwerth 70.0 90.0 Ibid., Nr.149
la Doorwerth 70.0 88.0 Ibid., Nr.150
la Doorwerth 72.0 80.0 Ibid., Nr.151
la Doorwerth 74.0 86.0 Ibid., Nr. 152
la Doorwerth 72.0 88.0 Ibid., Nr.153
la Doorwerth 72.0 89.0 Ibid., Nr.154
la Doorwerth 73.0 87.0 Ibid., Nr.155
la Doorwerth 73.0 87.0 Ibid., Nr.156
la Doorwerth 73.0 82.0 Ibid., Nr.157
la Doorwerth 71.0 87.0 Ibid., Nr.158
la Doorwerth 72.0 85.0 Ibid., Nr.159
la Doorwerth 72.0 91.0 Ibid., Nr.160
la Doorwerth 73.0 81.0 Ibid., Nr.161
la Doorwerth 75.0 86.0 Ibid., Nr.162
la Doorwerth 73.0 80.0 Ibid., Nr.163
la Doorwerth 74.0 84.0 Ibid., Nr.164
la Doorwerth 72.0 81.0 Ibid., Nr.165
la Doorwerth 70.0 88.0 Ibid., Nr.166
la Doorwerth 73.0 84.0 Ibid., Nr.167
la Doorwerth 71.0 90.0 Ibid., Nr.168
la Doorwerth 74.0 88.0 Ibid., Nr.169
la Doorwerth 73.0 79.0 Ibid., Nr.170
la Doorwerth 73.0 83.0 Ibid., Nr.171
la Doorwerth 70.0 88.0 Ibid., Nr.172
la Xanten 108.5 128.0 JENKINS, 1985, No.Al
lb Doorwerth 110.0 134.0 BROUWER, 1982, Nr.176
lb Lincoln 52.5 52.0 WEBSTER, 1949, Pl.X,b
(height damaged)
lb The Lunt 56.5 50.0 HOBLEY, 1973, Fig.23,30 (height damaged)
lb Xanten 62.0 82.0 JENKINS, 985, No.B5
lb Xanten 62.0 80.0 Ibid., No.B6
lb Xanten 54.0 58.5 Ibid., No.B7 (height damaged)
lb Xanten 58.5 ? Ibid., N0.B8 (height damaged)
lb Xanten 57.0 77.5 Ibid., No.B9
lb Xanten 62.5 81.5 Ibid., No.B10
lb Xanten 64.0 80.5 Ibid., No.Bll
lb Xanten 62.0 67.0 Ibid., No.B13 (height damaged)
lb Xanten 62 65.5 Ibid., No.B14 (height damaged)
lb Xanten 60.0 62.0 Ibid., No.B15 (height damaged)
lb Xanten 60.5 63.0 No.B17 (height damaged)
lb Xanten 62.0 62.0 Ibid., No.B19 (height damaged)
lb Xanten 60.0 ? Ibid., No.B20 (height damaged)
lb Xanten 61 79.0 Ibid., No.B21
lb Xanten 60.0 61.5 Ibid., No.B24 (damaged)
lb Xanten 60.0 63.5 Ibid., No.B25 (damaged)
1c Doorwerth 46.0 53.0 BROUWER, 1982, Nr.179
1c Doorwerth 46.0 54.0 Ibid., Nr.180
1c Doorwerth 45.0 53.0 Ibid., Nr.181
Table 6
143
1c? Heddernheim 45.0 38.5 FISCHER, 1973, Abb.26,4 (height damaged)
1d Doorwerth 80.0 101.0 BROUWER, 1982, Nr.183
1e Banasa 42.0 55.0 BOUBE-PICCOT, 1964, No.4
le Colchester 55.5 61.5 WEBSTER, 1960, Fig.4,58
le Hofheim ? ? Unpublished (Wiesbaden Stadtmuseum)
le Richborough 47.0 48.0 BUSHE-FOX, 1949, Pl.LI,180 (height
damaged)
1e Wiesbaden 49.0 51.0 ORL Nr.31, Taf.X,20
1f Sea Mills 31.0 63.0 WEBSTER, 1960, Fig.7,181 (damaged)
1g Aislingen 38.0 55.5 ULBERT, 1959, Taf.20,1
1g Aislingen 36.0 41.0 Ibid., Taf.20,2 (damaged)
1g Chester 24.0 46.0 Unpublished (Grosvenor Museum
186.R.1976) (damaged)
lg Corbridge 29.0 36.0 Allason-Jones in BISHOP & DORE,
forthcoming, No.197 (damaged)
1g Corbridge 42.0 44.0 Ibid., No.198 (damaged)
1g Dura-Europos ? ? ROSTOVTZEFF et al., 19 , Pl.VI,88
1g Heddernheim 44.5 47.0 FISCHER, 1973, Abb.19,2 (damaged)
1g Hüfingen 40.5 41.5 REVILLIO, 1929, Abb.2,5 (height
obscured)
1g Hüfingen 69.0 69.0 Ibid., Abb.2,9 (height damaged)
1g Hüfingen 45.0 55.5 ORL Nr.62a, Taf.XI,73 (height damaged)
1g Kempten 33.0 48.0 MACKENSEN, 1987, Abb.66,13 (damaged)
1g Mainz 45.5 55.0 BEHRENS, 1912, Abb.4,18 (height
damaged)
1g Neuß 60.0 74.0 LEHNER, 1904, Taf.XXXIV,6 (damaged)
1g Newstead 75.0 82.0 CURLE, 1911, P1.LXXIV,5 (damaged)
1g Newstead 66.0 77.0 Ibid., P1.LXXIV,7 (damaged)
1g Oberstimm 43.0 49.0 SCHONBERGER, 1978, Taf.24,B206 (height
damaged)
1g Rißltissen 37.0 35.0 ULBERT, 1970, Taf.1,15 (height damaged)
1g Rißtissen 31.0 33.0 Ibid., Taf.23,346 (damaged)
1g Rottweil 36.0 46.0 PLANCK, 1975, Taf.35,1
1g Rottweil 43.0 51.0 Ibid., Taf.73,1 (height damaged)
1g Rottweil 33.0 43.5 Ibid., Taf.73,2 (height damaged)
1g Sisek 37.0 49.0 HOFFILLER, 1912, SI.45
1g Thamusida 46.0 40.0 BOUBE-PICCOT, 1964, No.5
lg Vindonissa 33.0 41.0 UNZ, 1973, Abb.12,146
1g Volubilis 66.0 45.0 Ibid., No.6
1g Volubilis 48.0 63.0 Ibid., No.1
1g Wroxeter 39.0 40.0 WEBSTER, 1960, Fig.8,260 (height
damaged)
1h Hod Hill 46.0 55.5 RICHMOND, 1968, Fig.31,"Pit 15b
Secondary"
1h Hüfingen 48.0 49.5 ORL Nr.62a, Taf.XI,72 (height obscured)
1h Sheepen 34.0 36.0 HAWKES & HULL, 1947, P1.CIII.22 (height
obscured)
1i Brecon 46.0 49.0 WHEELER, 1926, Fig.57,2 (damaged)
1i Newstead 60.5 64.0 CURLE, 1911, Pl.LXXIII,3 (height
damaged)
1j Newstead 55.0 55.0 Ibid., P1.LXXIII,2 (height damaged)
lj Newstead 54.0 54.0 Ibid., P1.LXXIII,4 (height damaged)
1k Fremington Hagg 55.5 72.5 WEBSTER, 1971, No.l
1l Fremington Hagg 37.0 48.0 Ibid., No.2
1l Greensforge 42.0 51.0 Id., 1981, Fig.34 (height damaged)
1l Hofheim 38.0 43.0 RITTERLING, 1913, Taf.XII,37 (height
damaged)
1l Hüfingen 46.5 46.0 REVILLIO, 1929, Abb.2,6 (damaged &
obscured)
1l Neuß 44.0 40.0 LEHNER, 1904, Taf.XXXIV,16 (height
Table 6
144
Fig.44: Type sketches for pendants, Types 1u3.
145
obscured
1l Oberstimm 37.0 44.0 ULBERT, 1957, Abb.2,4 (height damaged)
11 Rißtissen 36.0 32.0 Id., 1959, Taf.63,19 (height damaged)
11 Valkenburg 40.0 47.0 GLASBERGEN & GROENMAN-VAN WAATERINGE, 1974,
Pl.16,4
11 Volubilis 40.0 34.0 BOUBE-PICCOT, 1964, No.12
11 Volubilis 30.0 39.0 Ibid., No.8
11 Wiesbaden 45.0 51.0 ORL Nr.31, Taf.X,22
1m Fremington Hagg 41.0 44.0 WEBSTER, 1971, No.18
1m Tanger 52.0 71.0 BOUBE-PICCOT, 1964, No.10
1n Hofheim ? ? Unpublished (Wiesbaden Stadtmuseum)
1n The Lunt 57.5 62.5 HOBLEY, 1973, Fig.22,20
1n Seven Sisters 50.0 53.5 DAVIES & SPRATLING, 1976, No.3 (height
damaged)
1n Wroxeter 5S.0 62.0 BUSHE-FOX, 1915, Fig.7,22
1n Wroxeter 46.5 43.0 Ibid., Pl.XVIII,30 (height damaged)
1o Neuß 97.0 94.0 LEHNER, 1904, Taf.XXXIV,17
1o Wiesbaden 42.0 44.0 ORL Nr.31, Taf.X,21
1p Wiesbaden 40.0 44.0 Ibid., Taf.X,24 (height damaged)
1q Neuß 57.0 71.5 LEHNER, 1904, Taf.XXXIV,18 (height damaged)
1r Neuß 48.5 53.0 Ibid., Taf.XXXIV,19 (damaged)
1s Thamusida 50.0 45.0 BOUBE-PICCOT, 1964, No.3
1t Banasa 25.0 28.0 Ibid., No.9
1u Volubilis 63.0 52.0 Ibid., No.11
1v Cirencester 65.0 75.0 WACHER & MCWHIRR, 1982, Fig.35,98 (height
damaged)
1w Nawa ? ? BOUBE-PICCOT, 1964, Pl.III
1x Southwark ? ? HAMMERSON & SHELDON, 1987, Pl.2
1y Wroxeter 59.0 86.0 West Midlands Archaeological News Sheet
No.17, 1974, front cover
Table 6
146
Fig.45: Type sketches for pendants, Types 45.
147
2b Rißtissen 31.0 34.5 Ibid., Taf.23,348 (height damaged)
2c Baden 38.0 57.0 UNZ, 1971, Abb.6,51
2d Canterbury 65.0 47.0 HASSALL, 1980, Fig.4,A (height damaged)
2d Canterbury 71.0 84.0 Ibid., Fig.4,C
2e Chester 50.0 49.0 Unpublished (Grosvenor Museum 185.R.1976)
(height damaged)
Table 6
148
Fig.46: Type sketches for pendants, Types 68e.
149
5e Chester 23.0 37.0 Ibid., Pl.VII,10
5e Chester 30.0 50.0 Ibid., Pl.VII,12(height damaged)
5e Hofheim 23.0 41.0 RITTERLING, 1904, Taf.III,31
5e Hofheim 38.0 69.0 Ibid., Taf.III,32 (damaged)
5e Hüfingen 42.0 57.0 ORL Nr.62a, Taf.XI,78
5e Hüfingen 64.5 94.5 Ibid., Taf.XI,79 (damaged)
5e Hüfingen 39.0 60.0 Ibid., Taf.XI,80
5e Hüfingen 36.0 60.0 Ibid., Taf.XI,83
5e Neuß 77.0 115.0 LEHNER, 1904, Taf.XXXIV,9
5e Rottweil 65.0 97.5 PLANCK, 1975, Taf.35,6
5e Rottweil 54.0 71.0 Ibid., Taf.73,4
5e Rottweil 54.0 79.5 Ibid., Taf.73,6
5e Rottweil 81.0 109.5 Ibid., Taf.73,9
5e Rottweil 33.0 51.0 Ibid., Taf.73,8
5e Strasbourg 28.5 52.5 FORRER, 1927, Taf.LXXVII,14
5e Unknown 49.0 76.0 LAWSON, 1978, Taf.53,4
5e Vindonissa 39.0 60.0 UNZ, 1973, Abb.11,128 (width damaged)
5e Vindonissa 40.0 65.0 Ibid., Abb.11,131
5e Wiesbaden 40.0 62.0 ORL Nr.31, Taf.X,37
5e Wiesbaden 66.0 91.0 Ibid., Taf.X,40
5f Lorenzberg 42.0 81.0 ULBERT, 1965, Taf.2,14
5g Heddernheim 42.0 66.0 FISCHER, 1973, Abb.17,1
Table 6
150
7a Wroxeter 82.5 40.0 WEBSTER, 1960, Fig.8,263 (height damaged)
7b Aislingen 36.0 46.5 ULBERT, 1959, Taf.20,3 (width damaged)
7b Baden 50.0 44.0 UNZ, 1971, Abb.7,68 (height damaged)
7b Baden 53.0 48.0 Ibid., Abb.7,69
7b Baden 60.0 59.0 Id., 1974, Abb.1,10 (height damaged)
7b Baden 46.0 42.0 Ibid., Abb.1,11 (height damaged)
7b Bern 50.0 39.0 MÜLLER-BECK & ETTLINGER, 1963, Taf.5 7,4
(height damaged)
7b Bern 51.0 51.0 Ibid., Taf.57,5
7b Besançon 102.5 78.5 FEUGÈRE, 1983, Fig.25,a
7b Bregenz 51.0 52.5 MACKENSEN, 1987, Abb.64,7 (width damaged)
7b Bregenz 64.5 51.0 Ibid., Abb.64,8
7b Chichester 50.0 46.5 DOWN, 1978, Fig.10.32,42 (width damaged)
7b Chichester 51.0 43.5 Ibid., Fig.10.32,43 (height damaged)
7b Colchester 38.0 47.5 CRUMMY, 1983, No.4233 (width damaged)
7b Harlow 30.0 39.0 WICKENDEN, this vol., Fig.6,39 (width damaged)
7b Hod Hill 45.0 45.0 BRAILSFORD, 1962, Fig.3,A40
7b Hofheim 52.0 38.0 RITTERLING, 1913, Taf.XIV,7 (damaged)
7b Hofheim 42.0 36.0 Ibid., Taf.XIV,9 (damaged)
7b Hofheim 38.0 35.0 Ibid., Taf.XIV,10 (height damaged)
7b Hüfingen 39.0 42.0 ORL Nr.62a, Taf.XI,87 (damaged)
7b London 55.5 46.5 WEBSTER, 1960, Fig.6,144
7b Longthorpe 52.5 50.5 FRERE & ST.JOSEPH, 1974, Fig.30,62
7b Luxeuil 58.0 51.0 FEUGÈRE, 1983, Fig.25,b
7b Mainz BEHRENS & BRENNER, 1911, Abb.26,104
55.0 54.0 (height damaged)
7b Neuß 78.0 68.0 LEHNER, 1904, Taf.XXXIV,10
7b Oberstimm 74.0 66.0 ULBERT, 1957, Abb.2,6 (height damaged)
7b Rheingönheim 48.0 48.0 ULBERT, 1969, Taf.36,1 (width damaged)
7b Rheingönheim 44.0 42.0 Ibid., Taf.36,2
7b Rheingönheim 52.5 41.0 Ibid., Taf.36,4 (damaged)
7b Rißtissen 38.0 47.0 Id., 1970, Taf.33,507 (width damaged)
7b Rißtissen 50.0 52.0 Ibid., Taf.33,508 (width damaged)
7b Roanne 41.0 40.5 FEUGÈRE, 1983, Fig.24
7b Sheepen 63.0 64.0 HAWKES & HULL, 1947, Pl.CIII,12
7b Strasbourg 52.5 40.5 FORRER, 1927, Taf.LXXVI1,16 (damaged)
7b Unknown LAWSON, 1978, Taf.53,2
118.0 95.0
7b Valkenburg GLASBERGEN & GROENMAN-VAN WAATERINGE, 1974,
74.0 66.0 Pl.14,52
7b Valkenburg 44.0 43.0 Ibid., Pl.16,6
7b Vechten 40.0 38.0 FEUGÈRE, 1983, Fig.25,e (damaged)
7b Vechten 39.5 48.0 Ibid., Fig.25,f (damaged)
7b Verulamium 63.0 55.0 Ibid., Fig.25,c (width damaged)
7b Vindonissa 44.0 44.0 UNZ, 1973, Abb.11,127 (width damaged)
7b Vindonissa 51.0 45.0 Ibid., Abb.11,129
7b Xanten 72.0 105.0 HINZ, 1971, Abb.33,8 (height damaged)
7c London 69.0 61.5 WEBSTER, 1960, Fig.6,161 (height damaged)
7d Hofheim 53.0 51.0 RITTERLING, 1904, Taf.III,23
7d Sheepen HAWKES & HULL, 1947, Pl.CIII,ll (height
48.0 45.0 damaged)
7e Cirencester 182.0 128.0 WACHER & MCWHIRR, 1982, Fig.36,100
7e Moers-Asberg 126.0 125.0 BECHERT, 1974, Abb.73,9
Decorated suspension necks for 7e pendants are known from Chelmsford (WICKENDEN, this
vol., Fig.3,2); Colchester (WEBSTER, 1960, Fig.4,69); London (Ibid., Fig.6,141); Wroxeter
(Ibid., Fig.8,256; Kingsholm (LYSONS, 1817, Pl.XV,10); Aislingen (ULBERT, 1959,
Taf.21,17); Kempten (MACKENSEN, 1987, Abb.66,14); Oberstimm (SCHÖNBERGER, 1978,
Taf.23,B189)
Table 6
151
Fig.47: Type sketches for pendants, Types 8f9i.
152
7f Strasbourg 108.0 108.0 FORRER, 1927, Fig.394,E
7f Straubing 75.0 79.5 WALKE, 1965, Taf.98,38
7g Neuß ? ? LEHNER, 1904, Taf.XXXIV,14
7g Unknown 39.0 37.0 AuhV 3:5, Taf.5,6 (damaged)
7g Valkenburg 42.0 44.0 GLASBERGEN S, GROENMAN-VAN WAATERINGE, 1974,
Pl.16,8
7h Fingringhoe ? ? WEBSTER, 1981, 133 fig.
Wick
Table 6
153
Fig.48: Type sketches for pendants, Types 9j10h.
154
9e Strasbourg 46.5 60.0 FORRER, 1927, Taf.LXXVII,15
9f Burum 45.0 39.0 ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977,
Pl.30,9
9f Ferwerd 50.0 43.0 Ibid., Pl.30,8
9f Piaam 46.0 40.0 Ibid., Pl.30,10
9f Unknown 37.0 40.0 Ibid., Pl.30,1
9f Unknown 47.0 41.0 Ibid., Pl.30,6
9f Wiesbaden 54.0 45.0 ORL Nr.31, Taf.X,41
9g Castleford 54.0 57.0 Report forthcoming
9g Castleford 48.0 43.0 Report forthcoming (height damaged)
9g Corbridge 50.0 59.0 Allason-Jones in BISHOP & DORE, forthcoming.
No.121 (damaged)
9g Hüfingen 49.5 48.0 ORL Nr.62a, Taf.XI,81 (height damaged)
9g Nijmegen 44.0 48.0 ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977,
Pl.33,36
9g Nijmegen 43.0 71.5 Ibid., Pl.34,37 (width damaged)
9g Nijmegen? 56.0 56.0 Ibid., Pl.34,38
9g Vindonissa 60.0 63.0 UNZ, 1973, Abb.13,158
9g Vindonissa 31.0 41.0 Ibid., Abb.13,164
9h Dormagen 54.0 66.0 MULLER, 1979, Taf.79,12
9h Straubing 50.0 57.0 WALKE, 1965, Taf.98,28
9h Unknown 40.0 41.0 ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977,
Pl.32,23
9h Unknown 30.0 39.0 Ibid., Pl.34,40
9h Unknown 72.0 67.0 LAWSON, 1978, Taf.53,3
9h Vechten 29.0 38.0 ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977,
Pl.33,33
9h Waddon Hill 26.0 37.0 WEBSTER, 1960b, Fig.6,1
9i Baden 36.0 54.0 UNZ, 1971, Abb.5,48 (height damaged)
9i Neuß 37.0 57.0 LEHNER, 1904, Taf.XXXIV,23
9j Nijmegen 39.0 46.0 ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977,
Pl.31,15 (height damaged)
9j Vindonissa 31.0 47.0 Id., 1973, Abb.13,165
9k Haltern 63.0 71.5 MAKW 2, 120 Abb.
91 Haltern 25.0 24.0 Ibid., 120 Abb.
9m Vechten 57.0 59.5 ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977,
Pl.30,4
9m Voorburg 35.0 44.0 Ibid., Pl.34,44
9n Friedberg 71.0 121.0 ORL Nr.26, Taf.V,3 (width damaged)
9o Colchester 53.0 72.0 DUNNETT, 1971, Fig.11,19
9p Carnuntum 57.0 50.0 RLO XXXII, Taf.66,17 (height damaged)
9q Rheinberg 51.0 51.0 BINDING, 1968, Abb.17 (height damaged?)
9r Unknown 85.0 88.5 LAWSON, 1978, Taf.53,1 (height damaged)
9s Dangstetten 150.0 138.0 FINGERLIN, 1981, Abb.5
Table 6
155
Table 6
10m Nijmegen 79.0 50.0 ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977, Pl.32,18
10m Vindonissa 76.0 54.0 UNZ, 1973, Abb.13,160
l0n Vindonissa 72.0 71.0 Ibid., Abb.13,163
l0o Valkenburg 81.0 62.0 GLASBERGEN & GROENMAN-VAN WAATERINGE, 1974,
Pl.14,50 (height damaged)
l0p Valkenburg 64.0 87.Q Ibid., Pl.15,9
10q Augsburg- 60.0 59.0 HÜBENER, 1973, Taf.14,3
Oberhausen
l0r Augsburg- 44.0 68,0 Ibid., Taf.14,10
Oberhausen
10s Nijmegen 76.0 67.0 BOGAERS & HAALEBOS, 1975, Fig.11,5
l0t Nijmegen 97.0 64.0 Ibid., Fig.11,6
156
Fig.50: Type sketches for junction loops, Types 15.
157
JUNCTION LOOPS
Table 7
158
1l Lincoln 55.0 16.5 WEBSTER, 1949, Pl.X,a
1l Lincoln 57.0 16.5 Ibid., Pl.X,a
1l Lincoln 52.5 17.5 Ibid., Pl.X,a
1l Lincoln 55.5 17.0 Ibid., Pl.X,a
1m Colchester 37.0 12.5 CRUMMY, 1983, No.4200
1m Colchester 34.0 11.0 Ibid., No.4201
1m Kempten 28.0 10.0 KRÄMER, 1957, Taf.B,23
1m Kempten 28.0 11.0 Ibid., Taf.B,23
1m Kempten 29.0 11.0 Ibid., Taf.B,23
1n Heddernheim 58.0 11.0 FISCHER, 1973, Abb.16,2
1n Heddernheim 61.0 10.5 Ibid., Abb.16,2
1n Moers-Asberg 62.0 13.0 BECHERT, 1974, Abb.71,2
1o Longthorpe 75.5 17.0 FRERE & ST.JOSEPH, 1974, Fig.30,58
1o Vindonissa 49.0 17.0 UNZ, 1973, Abb.9,94 (length damaged)
1p Gallanach 56.0 15.0 RITCHIE, 1974, Fig.2,8 (length damaged
1p Oberstimm 56.0 13.5 SCHÖNBERGER, 1978, Taf.23,B182 (length damaged)
1q Oberstimm 46.5 14.0 Ibid., Taf.23,B186
Table 7
159
3d Hüfingen 46.5 16.5 REVILLIO, 1929, Abb.4,1
3d Hüfingen 51.0 18.0 Ibid., Abb.4,1
3d Hüfingen 43.5 16.5 Ibid., Abb.4,1 (length damaged)
3d Hüfingen 51.0 16.5 Ibid., Abb.4,2
3d Hüfingen 52.5 18.0 Ibid., Abb.4,2
3d Hüfingen 54.0 18.0 Ibid., Abb.4,2
3d Hüfingen 55.5 18.0 Ibid., Abb.4,2
3e Seven Sisters 40.5 15.0 DAVIES & SPRATLING, 1976, Fig.3,4 (damaged)
3f Hofheim 32.0 12.0 RITTERLING, 1904, Taf.III,14
3g Castleford 35.0 11.0 Report forthcoming (length damaged)
3g Kingsholm 58.0 18.0 HURST, 1985, Fig.11,7 (length damaged)
Table 7
160
Fig.51: Type sketches for junction loops, Types 610.
161
5e Newstead 70.5 21.0 Ibid., Pl.LXXII,16
5f Hofheim 46.5 10.5 RITTERLING, 1913, Taf.XII,17
5g Augsburg- 64.0 11.0 HÜBENER, 1973, Taf.13,5
Oberhausen
5g Lorenzberg 84.0 16.5 ULBERT, 1965, Taf.2,11 (length damaged)
5g Neuß 80.0 17.0 LEHNER, 1904, Taf.XXXA,16
5g Strasbourg 78.0 15.0 FORRER, 1927, Taf.LXXVI1,18
5h Mainz 60.0 ? BEHRENS, 1912, Abb.3,6
5i Augsburg- 46.0 16.0 HÜBENER, 1973, Taf.13,1 (damaged)
Oberhausen
5i Augsburg- 60.0 13.0 Ibid., Taf.13,1 (damaged)
Oberhausen
5i Augsburg- 44.0 15.0 Ibid., Taf.13,1 (damaged)
Oberhausen
5i Augsburg- 50.0 13.0 Ibid., Taf.13,1 (damaged)
Oberhausen
5i Valkenburg 73.0 17.0 GLASBERGEN & GROENMAN-VAN WAATERINGE,1974,
Pl.14,48
5i Valkenburg 73.0 18.0 Ibid., Pl.14,48
5i Valkenburg 70.0 18.0 Ibid., Pl.14,48
5i Valkenburg 75.0 17.0 Ibid., Pl.14,48
162
6a Xanten 43.0 0.0 Ibid., No.E65 (damaged)
6a Xanten 35.5 0.0 Ibid., No.E66 (damaged)
6a Xanten 32.5 0.0 Ibid., No.E67 (damaged)
6a Xanten 17.5 0.0 Ibid., No.E68 (length damaged)
6a Xanten 28.0 0.0 Ibid., No.E69 (damaged)
6a Xanten 31.0 12.0 Ibid., No.E70 (length damaged)
6b Doorwerth 92.0 23.0 BROUWER, 1982, Nr.144
6c Aislingen 55.5 10.5 ULBERT, 1959, Taf.18,5
6d RiBtissen 40.5 10.5 Ibid., Taf.62,17
Table 7
163
Fig.52: Type sketches for strap terminals, Types 183.
164
STRAP TERMINALS
Type Site Length Width Reference
(mm) (mm)
Table 8
165
6a Colchester 35.0 10.5 Ibid., Fig.4,72 (length damaged)
6b Doorwerth 71.0 16.0 BROUWER, 1982, Nr.133
6b Doorwerth 63.0 17.0 Ibid., Nr.134
6b Doorwerth 73.0 18.0 Ibid., Nr.135
6b Doorwerth 69.0 16.0 Ibid., Nr.136
6c Aislingen 46.0 9.0 ULBERT, 1959, Taf.18,10
6c Chester 58.0 10.5 Unpublished (384.R.1976)
6c Ham Hill 53.0 12.0 GRAY, 1926, Pl.XIV,E21
6c Longthorpe 59.0 13.0 FRERE & ST.JOSEPH, 1974, Fig.28,45
6c Mildenhall 40.0 11.0 GRIFFITHS, 1983, No.9
6c Neuß 50.0 13.0 LEHNER, 1904, Taf.XXXA,12
6c Oberstimm 50.0 12.0 SCHÖNBERGER, 1978, Taf.22,B154
6c Rißtissen 43.0 8.0 ULBERT, 1970, Taf.23,353 (length damaged)
6c Sheepen 52.0 12.0 HAWKES & HULL, 1947, P1.CII,24
6c Sheepen 51.0 12.0 Ibid., P1.CII,25
6c Sheepen 55.5 14.0 NIBLETT, 1985, Fig.72,105
6c Thamusida 45.0 7.0 BOUBE-PICCOT, 1964, No.45
6c Volubilis 36.0 9.0 Ibid., No.46
6d Sheepen 48.0 16.0 HAWKES & HULL, 1947, Pl.CII,27
6e Hufingen 51.0 13.5 ORL Nr.62a, Taf.XI,6
6e Verulamium 49.5 14.0 FRERE, 1984, Fig.11,80
6f Moers-Asberg 36.0 9.0 BECHERT, 1974, Abb.67,10
6f Newstead 44.0 9.5 CURLE, 1911, Pl.LXXVI,4
6f Newstead 62.5 14.5 Ibid., P1.LXXVI,7
6f Newstead 54.0 9.5 Ibid., P1.LXXVI,8
6f? Corbridge 52.0 10.0 Allason-Jones in BISHOP & DORE,forthcoming,
No.114
6g Wiesbaden 48.0 18.0 ORL Nr.31, Taf.X,50
Table 8
166
Fig.53: Type sketches for strap terminals, Types 8f10.
Fig.54: Type sketches for strap fasteners (female).
167
STRAP FASTENERS (FEMALE)
Table 9
168
STRAP FASTENERS (MALE)
Table 10
169
Fig.56: Type sketches for strap mounts.
170
STRAP MOUNTS
Type Site Length Width Reference
(mm) (mm)
la Sheepen 70.0 26.0 HAWKES & HULL, 1947, Pl.CIII,21
lb Gellep 53.0 16.0 PAAR & RUGER, 1971, Bild 23,24
1o Rißtissen 55.0 13.0 ULBERT, 1959, Taf.62,14
1d Hofheim 70.0 15.0 RITTERLING, 1913, Taf.XIII,17
1d LincoIn 66.0 14.0 Unpublished
1d Neuß 64.0 13.0 LEHNER, 1904, Taf.XXXA,7
1e Hüfingen 51.0 13.5 ORL Nr.62a, Taf.XI,4
1f Mainz 63.0 10.0 BEHRENS, 1918, Abb.8,9
1g Mainz 53.0 14.0 Ibid., Abb.8,10
1h Unknown 62.0 12.0 BOUBE-PICCOT, 1964, No.44
1i Vindonissa 58.0 13.0 UNZ, 1973, Abb.10,112
1j Dormagen 31.0 14.0 MULLER, 1979, Taf.79,21
1k Manton Down 69.0 12.0 GRIFFITHS, 1983, No.15
Table 11
171
6b Kempten 38.0 10.0 MACKENSEN, 1987, Abb.66,7
6c Aislingen 61.5 18.0 ULBERT, 1959, Taf.18,2
6c Doorwerth 112.0 36.0 BROUWER, 1982, Nr.256
6c Doorwerth 109.0 35.0 Ibid., Nr.257
6c Doorwerth 107.0 34.0 Ibid., Nr.258
6c Doorwerth 108.0 33.0 Ibid., Nr.259
6c Doorwerth 109.0 35.0 Ibid., Nr.260
6c Doorwerth 104.0 32.0 Ibid., Nr.261
6c Doorwerth 104.0 33.0 Ibid., Nr.262
6c Doorwerth 103.0 35.0 Ibid., Nr.263
6c Doorwerth 101.0 33.0 Ibid., Nr.264
6c Doorwerth 101.0 32.0 Ibid., Nr.265
6c Doorwerth 102.0 33.0 Ibid., Nr.266
6c Doorwerth 100.0 34.0 Ibid., Nr.267
6c Doorwerth 99.0 34.0 Ibid., Nr.268
6c Doorwerth 97.0 34.0 Ibid., Nr.269
6c Doorwerth 96.0 33.0 Ibid., Nr.270
6c Fremington Hagg 56.5 20.5 WEBSTER, 1971, Fig.14,59 (length damaged)
6c Fremington Hagg 73.0 22.0 Ibid., Fig.14,63
6c Hofheim 58.5 17.0 RITTERLING, 1913, Taf.XII,21
6c Hofheim 58.5 13.5 Ibid., Taf.XV,94
6c Newstead 66.5 15.5 CORLE, 1911, P1.XCII,2
6c Rheingönheim 71.0 18.0 ULBERT, 1969, Taf.28,24
6c Richborough 43.0 20.0 CUNLIFFE, 1968, Pl.XXXVII,126 (length
damaged)
6c Rißtissen 50.0 20.0 ULBERT, 1959, Taf.62,1 (length damaged)
6c Rißtissen 63.0 20.0 Ibid., Taf.62,2
6c Sheepen 67.0 17.0 HAWKES & HULL, 1947, Pl.CIII,20
6c Thamusida 84.0 20.0 BOUBE-PICCOT, 1964, No.38
6d Aislingen 61.5 18.0 ULBERT, 1959, Taf.18,1
6d Aislingen 60.0 18.0 Ibid., Taf.18,3 (length damaged)
6d Chichester 66.5 16.5 DOWN & RULE, 1971, Fig.3.18,4
6d Doorwerth 89.0 18.0 BROUWER, 1982, Nr.272
6d Doorwerth 85.0 18.0 Ibid., Nr.273
6d Doorwerth 84.0 19.0 Ibid., Nr.274
6d Fremington Hagg 54.0 13.0 WEBSTER, 1971, Fig.14,57
6d Fremington Hagg 71.5 14.5 Ibid., Fig.14,62
6d Hüfingen 69.0 15.0 ORL Nr.62a, Taf.XI,18
6d Hüfingen 75.0 15.0 Ibid, Taf.XI,26
6d Neuß 74.0 20.0 LEHNER, 1904, Taf.XXXA,l
6d Rheingönheim 51.0 13.5 ULBERT, 1969, Taf.28,25
6d Rheingönheim 80.0 18.0 Ibid., Taf.28,26
6d Rißtissen 60.0 13.5 Id., 1959, Taf.62,4
6d Rißtissen 78.0 17.0 Id., 1970, Taf.1,14 (width damaged)
6d Sheepen? 66.0 15.5 CRUMMY, 1983, No.4219
6d Waddon Hill 41.0 13.0 WEBSTER, 1979, Fig.28,40 (damaged)
6d Wiesbaden 75.0 20.0 ORL Nr.31, Taf.X,52 (length damaged)
6d Xanten 100.0 13.0 JENKINS, 1985, No.D40
6d Xanten 102.5 25.0 Ibid., No.D42
6d Xanten 89.0 24.0 Ibid., NO.D43
6d Xanten 100.0 25.0 Ibid., No.D44
6d Xanten 57.0 24.5 Ibid., No.D45 (length damaged)
6d Xanten ? 23.0 Ibid., No.D46 (length damaged)
6d Xanten ? ? Ibid., No.D54 (damaged)
6d Xanten ? 24.5 Ibid., No.D55 (length damaged)
6d Xanten ? 24.0 Ibid., No.D56 (length damaged)
6d Xanten 44.0 ? Ibid., No.D58 (damaged)
6d Xanten 34.0 11.0 Ibid., No.D59 (damaged)
6e Doorwerth 59.0 17.0 BROUWER, 1982, Nr.279
Table 11
172
6e Fremington Hagg 53.0 9.0 WEBSTER, 1971, Fig.14,56
6e Harlow 54.0 12.5 FRANCE & GOBEL, 1985, Fig.46,121
6e Rheingönheim 52.5 9.0 ULBERT, 1969, Taf.28,21
6f Doorwerth 87.0 33.0 BROUWER, 1982, Nr.275
6f Hüfingen 48.0 24.0 REVILLIO, 1929, Abb.4,16 (damaged)
6f Hüfingen 52.5 27.0 Ibid., Abb.4,17 (damaged)
6g Baden 47.0 10.0 UNZ, 1971, Abb.5,35
6g Banasa 46.0 11.0 BOUBE-PICCOT, 1964, No.43
6g Fremington Hagg 41.5 7.5 WEBSTER, 1971, Fig.14,66
6g Fremington Hagg 46.0 10.0 Ibid., Fig.14,68
6g Hofheim 48.0 9.0 RITTERLING, 1913, Taf.XV,88
6g Rheingönheim 45.0 13.5 ULBERT, 1969, Taf.28,23 (length damaged)
6g Rißtissen 40.0 9.0 Id., 1959, Taf.62,8
6g Rißtissen 38.0 9.0 Id., 1970, Taf.23,352
6g Thamusida 46.0 8.0 BOUBE-PICCOT, 1964, No.42
6g Verulamium 38.0 11.0 FRERE, 1972, Fig.32,39 (length damaged)
6g Verulamium 47.5 12.5 FRERE, 1984, Fig.12,82
6g Volubilis 45.0 9.0 BOUBE-PICCOT, 1964, No.41
6g Woodcock Hall 47.0 11.0 BROWN, 1986, No.207
6h Rheingönheim 57.0 17.0 ULBERT, 1969, Taf.28,20 (length damaged)
6i Rißtissen 49.5 16.5 ULBERT, 1959, Taf.62,19
6j Stockton 58.0 10.0 GRIFFITHS, 1983, No.21
Table 11
173
NOTES
1. This paper was originally submitted in 1982 for inclusion in the
projected volume Roman Military Studies 1. When this still had not
appeared in January 1986, contributors withdrew their papers for
publication elsewhere, and this has allowed a complete revision of
the text and illustrations in line with more recent information
about Roman horse equipment.
2. Throughout this paper, the term 'cavalry' has been used, although
the writer fully realises that material from horses belonging to
officers, and even civilians, may be included in the
archaeological record. Furthermore, in its narrowest sense, the
term would refer to horse troops of the alae and cohortes
equitatae, although the equites legionis can conveniently be
included without implying that the tactical role of the latter was
as cavalry per se.
5. SCHLEIERMACHER, 1984.
174
horse equipment.
16. SCHLEIERMACHER, 1984, Nr.77; RIB 1172. Pers. obs., July 1986.
17. SCHLEIERMACHER, 1984, Nr.79; RIB 121. Pers. obs., May 1983.
18. SCHLEIERMACHER, 1984, Nr.75; RIB 109. Pers. obs., May 1983.
27. ESP.6455.
175
40. Ibid. Nr.46; ESP.6018. Pers. obs., September 1982.
48. Ibid. pl.18; pl.44 harnachements IVa, IVb, IVc; selles IVa, Ivb,
IVc.
62. See GREEN, 1986, 171-5 on Celtic depictions of the horse (including
pipeclay figurines) and the role of Epona.
176
67. In the case of stones like those of T. Flavius Bassus or C.
Romanius Capito, not only the phalerae and pendants are close to
those found in the archaeological record, but the equipment of the
rider himself, such as the helmet (cf. KLUMBACH, 1974, Nr.32,
Taf.32) and the lorica hamata fastening hook (ROBINSON, 1975,
Pl.480).
69. Leather adhering: CURLE, 1911, 177-8; HAWKES & HULL, 1947, 339.
70. Newstead: CURLE, 1911, 177 & Fig.17; Neuß: LAWSON, 1978, Taf.52,1;
Moers-Asberg: HORN, 1977, 83.
74. For one suggestion on the method of fastening the fixed part of
the girth strap to the saddle, see CONNOLLY, 1987, 11.
79. Cf. the analyses by CRADDOCK et al., 1973 and Craddock and Lambert
in JENKINS, 1985, 162-4.
81. See the comments by BOUBE-PICCOT, 1964, 175 with reference to his
no. 13.
177
to the phallic 'pendants', so it may belong with vehicle harness;
similarly, the studs on the rear suggest that it was attached to a
backing of some sort. In addition, the inlaid decoration is
fairly crude.
87. The association of Type 1 and 2 pendants in both the Doorwerth and
Fremington Hagg assemblages would seem to be strongly suggestive
of contemporaneity, a notion supported by their common decorative
style.
88. Similar objects were also used to decorate tool sheaths (AuhV 5,
Taf.10,167; JUNKELMANN, 1986, Taf.70c).
90. Augustan period: Haltern (MAKW 2, 120 Abb.); Rodgen (SCHÖNBERGER &
SIMON, 1976, Taf.41,6); second century: CHIRILA et al., 1972,
Taf.LXXI,28.
94. Only Colchester Sheepen, the Lunt, and Canterbury (see catalogue)
produce fittings of this type.
95. In Britain, with the exception of a piece from Newstead, they are
mostly confined to the south (see catalogue).
97. Their association with pendants of Types 6 and 10 has already been
noted.
98. Nijmegen: BOGAERS & HAALEBOS, 1975, 156 & pl.LII,la-d; Arlaines:
BOUSQUET, 1977, pl.XLII,Ar.76-37.
178
99. This need not, of itself, mean that ring junctions were more
common than those using phalerae - for it could be argued that the
wear on ring junction fittings was responsible for a higher
attrition rate amongst fittings.
100. For discussion of the reasons why artefacts are deposited in the
archaeological record, see BISHOP, 1985; 1986.
101. 'Harness clip': NIBLETT, 1985, 3:C1; 'baldric clip': loc. cit.
no.101; 'harness loop': FRERE & ST.JOSEPH, 1974, 56-8; 'baldric
mount or apron terminal': HURST, 1985, 28 no.7; 'strap loop':
CRUMMY, 1983, 132 Nos.4199-4201.
104. Since it would have been easier this way. In fact, this form of
fastening is very specialised, as it is not adjustable (one of the
reasons why such fastenings would be of little use for a baldric,
where a buckle is better suited to the task).
105. These rivet heads usually take the form of domed or dished studs
with niello inlay.
118. Cf. ibid. 108 plate. Most modern martingales join the reins at
some point and are designed to keep the horse under the control of
the bit (ibid. 60-1). The Roman martingale is more like that found
179
in modern driving harness, where it stops the breastplate riding
up and choking the horse (RICHARDSON, 1982, 131).
125. See the discussion of the Gundestrup cauldron above; for phalerae
in Halstatt period harness, see KOSSACK, 1954, 116-7 with Abb.l.
127. ZADOKS-JOSEPHUS JITTA & WITTEVEEN, 1977, 173; the usual symbol for
the sun was a spoked wheel (cf. GREEN, 1986, 39-55).
132. Such rivet holes are often found on Type 7a pendants - cf. HURST
1985, Fig.11,1 & 3.
136. The essence of control with the bridle are the seven pressure
points (RICHARDSON, 1982, 30-1) and Celto-Roman harness was able to
exploit all of these.
180
139. As in the case of the scale bards from Dura-Europos (ROBINSON,1975,
Pls.529-30), where a hole is left for the saddle.
143. BROUWER, 1982, 157, citing a relief from Auxerre (ESP.2885). This
clearly shows a shabracque with square patterning all over it,
which Brouwer suggests as the probable explanation for the 'saddle
plates' on the Rhineland tombstones. However, the objects on the
Rhineland tombstones are very clearly attached to triplet straps,
and cannot be explained in this way. Similarly, the saddle on the
north-east face of the arch at Orange (which Brouwer does not
cite) has no shabracque under it and again depicts rectangular
plates fastened to straps (see Fig.19,3).
150. Size: NOBIS, 1973, 232-40; age: ibid., 232; sex: loc. cit.
155. Vercundi: ibid. Fig.9; T. Capitoni Marian: ibid. Fig.4; for the
decoration on the face of Al, see ibid. Fig.2, and for that of the
B series pendants ibid. Fig.7; note the style of the tendrils and
181
leaves in each case. However, see. now SADDINGTON, 1987, where it
is suggested that the reading should be T(urma) Capitoni Marian(i)
- 'belonging to the turma of Capitonius Marianus'.
165. In the first instance, he apparently did not intend going into
battle with the cavalry himself (Bell. Iud. V,52), but evidently
learned from this expertise and utilised them (V,288) as a strike-
force to save the day at least once.
171. See, for example, the Reiter tombstone of the tribune L. Pompeius
182
Marcellinus (SCHLEIERMACHER, 1984, Nr.122).
172. It has been suggested that one phalera from Xanten may have
belonged to the Elder Pliny (JENKINS, 1985, 154), but a number of
alternative interpretations of the inscription in question have
been suggested (ibid. 155-6).
173. There would probably have been under 200 sets of harness belonging
to a legion, even allowing for the equites legionis, officers'
horses, and the mounts of centurions. The provision of remounts
would, naturally, mean there would be more animals than this bare
minimum (officers probably had more than one horse and there must
have been a general unit 'pool' of spare animals; cf. DAVIES,
1962, 429 and n.3).
177. Cf. BROUWER, 1982, Abb.4. For details of the silvering process,
see JENKINS, 1985, 145.
180. Although the burials with riding horses at Inota are presumably
civilian, their nature suggests considerable social status
(PALAGYI, 1986).
181. Cf. FRERE & ST.JOSEPH, 1974; NIBLETT, 1985; BROWN, 1986.
183. E.g. WEBSTER, 1971 where the illustrations are captioned as 1:2,
but are reproduced at 1:1.
183
188. Whilst the difference between ancient and modern damage is usually
readily apparent on the object itself, it is seldom possible to
convey this by means of an archaeological illustration - cf.
LEAHY, 1980.
184
BIBLIOGRAPHY
AMY et al. 1962: R. Amy, G.-Ch. Picard, J.-J. Hatt, P.-M. Duval, Ch.
Picard, & A. Piganiol, 'L'arc d'Orange', Gallia, Suppl.15, 1962
ANDERSON & WACHER 1980: A.S. Anderson & J.S. Wacher, 'Excavations at
Wanborough, Wiltshire: an interim report', Britannia, XI, 1980,
115-26
BEHRENS 1912: G. Behrens, 'Neue Funde aus dem Kastell Mainz', Mainzer
Zeitschrift, 7, 1912, 82-109
BEHRENS 1914: G. Behrens, 'Dritter Bericht über Funde aus dem Kastell
Mainz', Mainzer Zeitschrift, 8-9, 1913-14, 65-93
BEHRENS 1918: G. Behrens, 'Dritter Bericht über Funde aus dem Kastell
Mainz', Mainzer Zeitschrift, 12-13, 1917-18, 21-46
BISHOP 1985b: M.C. Bishop, 'The military fabrica and the production of
185
arms in the early principate', in BISHOP, 1985a, 1-42
BISHOP & DORE forthcoming: M.C. Bishop & J.N. Dore, Corbridge
Excavations 1947-1980. Archaeological Investigations of the Roman
Fort and Town at Corbridge, Northumberland (HBMCE Archaeological
Report), (London forthcoming)
BOGAERS & HAALEBOS 1975: J.E. Bogaers & J.K. Haalebos, 'Problemen rond
het Kops Plateau', Oudheidkundige Mededelingen uit het Rijksmuseum
van Oudheden te Leiden (nieuwe reeks), 56, 1975, 127-78
BRAILSFORD 1962: J.W. Brailsford, Hod Hill 1; Antiquities from Hod Hill
in the Durden Collection, (London 1962)
BREEZE 1969: D.J. Breeze, 'The organisation of the legion: the first
cohort and the equites legionis', Journal of Roman Studies, 59,
1969, 50-55
BROWN 1986: R.A. Brown, 'The Iron Age and Romano-British settlement at
Woodcock Hall, Saham Toney, Norfolk', Britannia, XVII, 1986, 1-58
186
Richborough, Kent (Reports of the Research Committee of the Society
of Antiquaries of London 16), (Oxford 1949)
CHEESMAN 1914: G.L. Cheesman, The Auxilia of the Roman Imperial Army,
(Oxford 1914)
CURLE 1911: J. Curie, A Roman Frontier Post and its People. The Fort of
Newstead in the Parish of Melrose, (Glasgow 1911)
DAVIES 1962: R.W. Davies, 'The supply of animals to the Roman army and
the remount system', Latomus, 28, 1962, 429-59
DAVIES & SPRATLING 1976: J.L. Davies & M.G. Spratling, 'The Seven
Sisters hoard: a centenary study', in G.C. Boon & J.M. Lewis
(eds.), Welsh Antiquity: Essays Mainly on Prehistoric Topics
Presented to H.N. Savory upon his Retirement as Keeper of
Archaeology, (Cardiff 1976), 121-47
DOWN & RULE 1971: A. Down & M. Rule, Chichester Excavations Vol.1,
(Oxford 1971)
187
VAN DRIEL-MURRAY 1985: C. van Driel-Murray, 'The production and supply
of military leatherwork in the first and second centuries A.D.: a
review of the archaeological evidence', in BISHOP, 1985a, 43-81
DRINKWATER 1979: J.F. Drinkwater, 'A note on local careers in the Three
Gauls under the early empire', Britannia, X, 1979, 89-100
FRERE & ST.JOSEPH 1974: S.S. Frere & J.K.S. St.Joseph, 'The Roman
fortress at Longthorpe', Britannia, V, 1-129
188
Verarbeitung im römischen Legionslager Vindonissa
(Veröffentlichungen der Gesellschaft Pro Vindonissa 1), (Basel
1942)
GRAY 1923: H.St.G. Gray, ' Archaeological remains, Ham Hill, south
Somerset', Proceedings of the Somerset Archaeological & Natural
History Society, 69, 1923, 49-53
GRAY 1924: H.St.G. Gray, 'Excavations at Ham Hill, south Somerset (part
1)' , Proceedings of the Somerset Archaeological & Natural History
Society, 70, 1924 (1925), 104-16
GRAY 1926: H.St.G. Gray, 'Excavations at Ham Hill, south Somerset (part
3)', Proceedings of the Somerset Archaeological & Natural History
Society, 72, 1926 (1927), 55-68
GREEN 1986: M.J. Green, The Gods of the Celts, (Gloucester 1986)
HAMMERSON & SHELDON 1987: M. Hammerson & H. Sheldon, 'Evidence for the
Roman army in Southwark', in DAWSON, 1987, 167-74
HAWKES & HULL 1947: C.F.C. Hawkes & M.R. Hull, Camulodunum. First
Report on the Excavations at Colchester 1930-1939 (Reports of the
Research Committee of the Society of Antiquaries of London, 14),
(Oxford 1947)
HINZ 1971: H. Hinz, '4. Bericht über Ausgrabungen in der Colonia Ulpia
Traiana bei Xanten', in Beiträge des römischen Rheinlands II
(Rheinische Ausgrabungen 10), (Düsseldorf 1971), 96-199
189
HOFFILLER 1912: V. Hoffiller, 'Oprema rimskoga vojnika u prvo doba
carstva', Vjesnik Hrvatskoga Arheoloskoga Drustva (Zagreb), N.S.
12, 1912, 16-123
HOLWERDA 1931: J.H. Holwerda, 'Een vondst uit den Rijn bij Doorwerth',
Oudheidkundige Mededelingen uit het Rijksmuseum van Oudheden te
Leiden, 12 suppl., 1931, 1-26
KEPPIE 1984: L.J.F. Keppie, The Making of the Roman Army from Republic
to Empire, (London 1984)
190
Jahrbuch des Römisch-Germanischen Zentralmuseums Mainz, 25, 1978,
131-72
LEAHY 1980: K. Leahy, 'A first Century military pendant from Normanby,
South Humberside', Lincolnshire History and Archaeology, 15, 1980,
83-4
191
PAAR & RÜGER 1971: I. Paar & C.B. Rüger, 'Kastell Gelduba: Forschungs-
und Grabungsgeschichte bis 1969', in Beiträge des römischen
Rheinlands II (Rheinische Ausgrabungen 10), (Düsseldorf 1971),
242-339
RABEISEN & MENU 1985: E. Rabeisen & M. Menu, 'Métaux et alliages des
bronziers d'Alesia', in Laboratoire de recherches des musées de
France, Recherches gallo-romaines I (Notes et documents des musées
de France 9), (Paris 1985), 143-81
RITCHIE 1974: J.N.G. Ritchie, 'Iron Age finds from Dùn an Fheurarin
Gallanach, Argyll', Proceedings of the Society of Antiquaries
Scotland, 103, 1974, 100-12
192
RITTERLING 1913: E. Ritterling, 'Das frührömische Lager bei Hofheim
i.T.', Annalen des Vereins für nassauische Altertumskunde und
Geschichtsforschung, 40, 1913, 1-416
SWANN 1970: V.G. Swann, 'An unpublished early Roman bronze from
Nettleton, Wiltshire', Wiltshire Archaeological and Natural History
Magazine, 65, 1970, 195-8
193
ULBERT 1957: G. Ulbert, 'Zum claudischen Kastell Oberstimm (Ldkr.
Ingolstadt)', Germania, 35, 1957, 318-27
UNZ 1986: C. Unz (ed.), Studien zu den Militärgrenzen Roms III. 13.
International Limeskongreß Aalen 1983 Vorträge, (Stuttgart 1986)
WACHER & MCWHIRR 1982: J.S. Wacher & A.D. McWhirr, Cirencester
Excavations 1: Early Roman Occupation at Cirencester, (Cirencester
1982)
194
in Roman Yorkshire - Essays to Commemorate the Nineteenth Century
of the Foundation of York, (Leicester 1971), 107-25
WEBSTER 1985: G. Webster, The Roman Imperial Army of the First and
Second Centuries A.D., ed.3, (London 1985)
195