5 001 070 PDF

Soldiers and Settlers in Britain)
Fourth to Fifth Century

By SONIA CHADWICK HA WKES
St. Hilda's College, Oxford
With a Catalogue of
Animal-Ornamented Buckles and
Related Belt-Fittings
By S. C. HA WKES and G. C. DUNNING
Ins/icc/or of Ancicnt Monuments, Ministry of Works
Dorchester-on-Thames in the ye:u 1874 were discovered in the Dyke
K Hills a number of burials belonging, it seems, to an as yet only partially

explored late Roman cemetery outside the walls of the Roman town.
Only two of the burials were furnished with grave-goods, but these were of such
interest and importance as to be already well known before they received their
first full and detailed publication by the late E. T. Leeds and]oan Kirk in 1954.'
The attention paid to these two graves by Anglo-Saxon archaeologists has
perhaps tended to obscure their real character as late Roman burials, and they
have been somewhat neglected by students of the Romano-British period in
consequence. This is a pity, for these graves are in reality a rare and valuable
document for the history of events during the end of Roman rule in Britain.
Among their grave-goods is a group of bronzes which represents one of the last
recognizable phases of the art of the Roman Empire in the west, before its sub-
mergence under the Germanic invasions of the fifth century.
On the continent metalwork of this type and period has been the subject of
many studies, as we shall see below, but in this country it has scarcely been
noticed. The neglect has arisen mainly out of the fragmentary nature of the
archaeological evidence for the late fourth and early fifth centuries in this
country. This, which may be largely the result of the destruction of the upper
levels of Romano-British occupation in town and country by the industrial
and agricultural activities of the succeeding centuries, and of failure in more
recent times to locate and excavate late Roman cemeteries, has reduced the period
to the status of a kind of no-man's-land between the secure entrenchments
of known Roman and unambiguous Saxon. On the whole these problem-
riddled decades of transition probably commend themselves more readily to
I Kirk and Leeds (I 954). Cf. also E. T. Leeds, The Archaeology of the Anglo-Saxon Settlements (I 9 I3).
pp. 55-6. fig. H.
: it
j
-~
~' fiD ::~
10 d 0
::D
- 9
13
o 0
FIG. 1
DYKE HILLS. DORCHESTER, OXON.

i\'os. 1-13. objects from the military burial (grave I); nos. 14-16, objects from the woman's burial
(grave 2). Pp. 1 If. SC. ';
.Xftcr Ovoniensia, X\'II-XVIII, fig. 27, by courtesy
SOLDIERS AND SETTLERS IN BRITAIN 3
Anglo-Saxon archaeologists, who are already accustomed to interpreting history,
without the aid of inscriptions or coherent coinage, from what students of the
three previous centuries call small finds. This may be why most of the references
to metalwork of the type here discussed have appeared in works by Anglo-Saxon
specialists. But the volume of comment is in any case pathetically small,> and
has merely served to demonstrate how scattered and ill-documented most of
the material is, and how very little is known about it in this country. No previous
writer has given any idea of the quantity of these objects found in Britain, nor
of their wider significance within the setting of the late Roman Empire. We felt,
therefore, that the time had come to bring together 3 corpus of some of this, the
most distinctive metalwork of the late- and sub-Roman and earliest Anglo-
Saxon periods in Britain. If such objects have anything of importance to add to
our knowledge of this obscure epoch, it seemed that in this way they might be
made to disclose it.
We have not tried to assemble all the varied types of metal objects dating
from this time, but have made a selection on the basis of the Dorchester finds.
These give us our starting point in an associated group of objects-i-two sorts of
buckle, a strap-end, and various belt-attachments (FIG. I)~whose interest is
confirmed by numerous continental discoveries of similar sort. To the types
represented at Dorchester we have added other kinds of buckles which belong
to the same style-phase. The seven main types selected form, with their sub-types,
a natural group, at once numerous and distinctive. The volume of material
brought to light by the systematic scouring of publications and museum collections
has proved unexpectedly large. The search has been a thorough one, and we
are confident that not much can have escaped our notice. If anything has done
so, it is not likely to change either the pattern of distribution or the basis of the
classification. However much the individual pieces vary in detail, they all fall
into a few clearly recognizable types. This has enabled the catalogue to be
presented as a series of classified lists. For economy these have been set out in
tabular form, each piece with its provenience, present location, description, and
such details of date and association as have been recorded. Most important, too,
every object (bar one-the single case where permission to draw was refused)
is illustrated.
The different types (see Catalogue, pp. 41 ff, for detailed descriptions of the
variants) are, for ease of reference, numbered as follows:
I )
II I
III J' Buckles with confronted animal-heads.
IV
V Strap-ends.
VI Disc-attachments with suspension loops.
VII Tubular-sided attachment-plates.
a Baldwin Brown (1915), IV, 551-561, pIs. cl-clv; Tonnochy and Hawkes (1931); Bruce-Milford
(1954); Evison (1955), pp. 29-3 1; Hildyard (1957), pp. 243 ff.: Cook (1958), p. 73; Morris (1959), pp.
102-3; Boon (1959), pp. 80- 83.
MEDIEVAL ARCHAEOLOGY
THE MILITARY GRAVE FROM DORCHESTER AND ITS

CONTINENTAL BACKGROUND
The starting point for the corpus of bronze buckles and belt-fittings was
provided by the two late Roman burials in the Dyke Hills at Dorchester. One
of these must now lead us into the discussion.
b
a
d
FIG. 2
MILTON-next-SITTINGBOURNE, KENT
Group of belt-fittings of types v A, VI, VII (pp. 4 If). Sc. l
Grave I (FIG. I, nos. I-I3) contained the skeleton of a man, furnished with
a buckle (type III A, 2), by the shoulder; a strap-end (type v A, I); two disc-attach-
ments (type VI, 2-4), by the thighs; a tubular-sided attachment-plate (type VII,
I), found among the ribs together with a number of rectangular bronze plates;
a bone toggle; and an iron knife (not preserved). A very similar group of objects
from Milton-next-Sittingbourne in Kent is preserved in Maidstone Museum (FIG.
2). Here we have a strap-end (type V A, 3); two disc-attachments (type VI, 5-6);
and two tubular-sided attachment-plates (type VII, 2-3), of which the identically
flaky patina and fragile condition is an indication that they, too, were found in a
single grave.e A third burial of this type is suggested by the buckle (type III A, 8),
disc-attachments (type VI, 12- I 3) and attachment-plate (type VII, 5), from an
unknown site in Kent (PL. 1).4 No other burials of this period with this kind of
equipment are known from this country.
On the continent, on the other hand, graves with this type of furniture
are relatively common in the late Roman cemeteries of the north of Gaul and
the Rhineland. Invariably these graves are the burials of men-and men, more-
over, who were additionally equipped for death with one or more of the weapons
they bore in life. The similarity between the objects in these continental grave
groups and our own is most striking, and it is worth describing a few of them by
way of comparison:
Tournai, Belgium. Cimetiere du Parc, grave F.5

Burial of man: with throwing axe; spear; strap-end (type v); disc-attachment
(type VI); attachment-plate (type VII); and buckle.
Furfooz, near Namur, Belgium. Grave 3. 6

Burial of man: with pottery, bronze, and glass vessels; spear; throwing-axe; knife;
3 arrows; cross-bow brooch of developed type; buckle (type III A); chip-carved
strap-end with marginal animals (type v); belt-slide decorated with animal heads;
IO disc-headed rivets similar to type VI attachments; 2 attachment-plates (type VII);
bucket; and comb.
Vieuxville, near Liege, Belgium ,7

Burial of man: with pottery and glass vessels; sword, spear; axe; 6 arrows; 3
buckles (type III A), one with chip-carved plate, and one with very stylized animal
heads; small buckle; strap-end (related to type V but with rounded end); belt-slide;
5 disc-attachments (type VI); 3 chip-carved attachment-plates (type VII); pair of
scales; and 2 silver coins, of Constantine III (407- I I) and ]ovinus (411- I 3).
Vermand, Aisne, France. Chemin des Mortes cemetery, grave 284.8

Burial of man: with bronze bowl; spear; throwing axe; knife; bone comb; buckle
type III A) with stylized animal heads; strap-end of unusual form; 3 disc-attach-
ments (type VI); 2 attachment-plates (type VII), with several rectangular bronze
plates; and a number of studs and rivets.
J No details have been recorded about the discovery of these objects, but the extensive brick-earth
digging around Sittingbourne in the nineteenth century revealed Romano-British burials of all periods.
George Payne, Collectanea Cantiana (1893), pp. 23 fr.; V.C.H. Kent, III, 96 fr.
4 Nothing is known about the origin of these objects, but their preservation as a group very much
suggests that they originally came from a grave.
S Faider Feytmans (1951), p. 4.5. See also grave D and grave 2 of the same cemetery, which are
very similar.
6 Nenquin (1953), pp. 88-9, 54-7, etc., pis. vi, DIA-D, viii, D20. See also grave I.
7 Breuer and Roosens (1957), pp. 243 ff., figs. 3 I -4.

8 Eck (1891), pp. 84-5, 252-8, pis. xii, 9-10, xvii, 1-20. See also graves 190 (pp. 49-5 0), and 397 (pp,
104-5) with silver siliquae of Honorius (A.D. 395-423).
6 MEDIEVAL ARCHAEOLOGY
Monceau-Le-Neu], Aisne, France."

Burial of man: with glass cup; silver spoon; sword; knife; razor; 2 buckles (type
III A), one with chip-carved plate and loop; strap-end (type v}, with chip-carved
ornament and marginal animals; disc-attachment (type VI); attachment-plate
(type VII); tubular object; cross-bow brooch of developed type; bone comb;
whetstone.
lvfainz-Kostlzeim, Gcrmanv.t''
Burial of man: with sword; axe; buckle (type IV) with open-work decoration;
matching strap-end, similar in form to that from Vieuxville; 2 disc-attachments
(type VI); bronze neck-ring; tweezers; bead toggle.
It would be possible to list other graves of this general character, but these
six samples should serve our purpose if we bear in mind that they are merely
representatives of a larger group.
It will be noticed at once that the sets of metalwork from Dorchester, Milton,
and Kent repeat very closely the assemblages in the French, Belgian and German
graves. The lack of weapons and accessory vessels at Dorchester is the only factor
that distinguishes this grave from its continental counterparts. But it is probable
that weapons actually were found with the burial. The grave was found during
the levelling of part of the Dyke Hills and there is a record of pieces of iron being
thrown away by the labourers. The apparent lack of weapons, therefore, need
not disturb us unduly.
On the continent these 'warrior' graves with their characteristic equipment
have been the subject of much controversy, and to understand the significance
of the Dorchester burial it will be necessary to turn now to north Gaul and the
Rhineland. Two facts about these graves are generally accepted: (I) the burials
in question begin in the last half of the fourth century and extend into the early
fifth; and (2) they are the graves of soldiers in the late Roman army. The date is
established approximately by the occurrence of coins in a number of the grave-
groups (see above, and below p. 18 f). The military character of the burials is
attested primarily by their weapons and secondly by their distribution along the
military frontier-zone of northern Gaul and the Rhineland. Further, the attach-
ment-plate of type VII with its accompanying metal strips, may, as Leeds
suggested, have been the fittings of the protective leather apron which was an
essential feature of Roman military armour. In the continental graves the metal
strips sometimes occur, as at Vermand gr. 284, but very often the tubular-
sided plates are found alone, or in pairs. Werner has recently suggested that
these type VII plates were not apron-fittings but ends for the broad military
belt. According to his theory, the metal strips were used as stiffeners for these
belts. Whatever their function, the type VII plates are all very much alike, and
were certainly part of the uniform. The disc-attachments (type VI) are also very
standard in form and decoration and the Dorchester, Milton and Kent examples
exactly parallel innumerable continental finds. They are best interpreted as
9 Pilloy, III (1912), 99 ff., pI. iv, 1-12.
ro Behrens (1919), pp. 3 ff., pI. 1,2.
rr Kirk and Leeds (1954), p. 71, fig. 28.
t z Werner (1958), pp. 379 If.
ornamental rivets designed, with their loop and ring, to act as points of suspension
for leather straps, presumably to link the various belts and harness straps worn
by a soldier when carrying weapons and other gear of war. We can thus be
reasonably safe in assuming, with Leeds, that the occupant of grave I in the
Dyke Hills belonged to some late Roman military detachment stationed at
Dorchester.
There is less agreement about the origin of such burials, and the controversy
aroused by them began far back in the history of late Roman studies. The most
recent phase opened, in 1950, with an important paper by Werner on the warrior
graves of northern Gaul. In iU3 he suggested that the deposition of arms and
armour with the dead, and various other features of these graves, such as the
wearing of brooches, and the placing of rich burials among groups of poorly
furnished ones, were Germanic customs. The warrior graves, therefore, were to
be attributed to the German laeti who, according to the Notitia Dignitatum, were
settled in these parts of Gaul. After the disastrous incursions of the Franks and
Alamanni, which had laid waste and depopulated the frontier districts of Gaul in
the third quarter of the third century, the counter-measures of the emperor
Dioeletian and his successors provided not only for the construction of a chain of
forts along the Limes and the installation of half-barbarian limitanei, with their
families, as permanent garrisons, but also for the repopulation of the ravaged
land behind the frontier by colonies of land-hungry Franks, and Germans of
other tribal groups. These received grants of land in return for the hereditary
obligation of military service. The presence in north Gaul of these laeti-a
German word describing their semi-free social position-is recorded in a number of
panegyrics addressed to Roman emperors in the late third and early fourth
centuries, which mention their dual role as farmers and soldiers and speak also
of specific areas of settlement.i- Werner's contention is that by the middle of the
fourth century some of these laeti had been enabled by virtue of their military
importance to attain the high rank and social status evidenced by some of the
warrior graves. Already in 1891 Pilloy was attributing the most famous of the
warrior graves at Vermand IS to a Frankish chieftain holding high rank in the
Roman army. That there were many such by the fourth century is evident from
the historical sources which record their important role in the maintenance of
Roman military power in the West. I 6 These German noblemen usually had the
'3 'Werner (1950), pp. 23 ff.
q Panegyrici Latini, cd. A. Baehrens (191 I), no. VIII, 21, to Constantius Chlorus: 'Just as by thy
decision, 0 Maximian Augustus, the Frankish laetus, immediately on being granted legal and peaceful
status (A.D. 286), has brought into cultivation the waste-lying corn-lands of the Nervii and the Trevcri
(of Trier), so now through thy victories, invincible Caesar Constantius, the deserted areas of the lands of
the Ambiani (of Amiens), Bellovaci (of Beauvais), Tricasses (of Troyes) and Lingones (of Langres) are
growing green again under barbarian tillage.' (c. A.D. 293.)
ld., VI, 6, to Constantine the Great: 'What am I to say, then, of the tribes from the interior of the
territory of the Franks, uprooted, not merely from those districts previously invaded by the Romans,
but from their own original homelands and from the remotest shores of barbary, so that, settled III deserted
tracts of Gaul, they might maintain the peace of the Roman Empire by their agriculture, and her victories
by their military service (dilectu) ?'
'5 Pilloy, 1(1891),38-51 and figs., pI. opp. p. 53. Cf Eck (1891), pp. 22-3,121-134, pI. ii.
,6 Ammianus, iv, 10,8; xv, 5, 33, etc. Cf. .Juliian (1926), pp. 84-8; Lot, Pfister and Ganshof (1940),
pp. 18- 23.
status of chieftains of tribal groups of foederati. The foederati held a position
superior to that of the laeti in that they held land within the Roman frontiers by
a foedus, or treaty, which often confirmed the tenure of land that had originally
been taken by conquest. The federates kept their national identity and their
freedom, but their military obligations seem to have become similar to those of
the laeti. Breuer and Roosens have suggested that by the middle of the fourth
century the distinction betweenfoederati and laeti had become blurred, so that the
warrior graves, which make their appearance at this time, may logically be
attributed to either."?
We can now turn to the conclusion of Werner's argument. He stated that it
was the burial customs of these German soldiers in the Roman army which were
the foundation of similar Frankish customs which can be seen in cemeteries of
the sixth and seventh centuries. The bases of Werner's thesis were reviewed and
criticized by de Laet, Dhondt and Nenquin!" who advanced the opinion,
supported by much detailed and shrewd analysis, that the so-called Germanic
characteristics of the warrior graves were no more than a military fashion common
alike to German and Gallo-Roman soldiers. They stressed the fact that by the
middle of the fourth century the exchange of ideas between Gallo-Romans
and Germans in the racially-mixed northern regions of Gaul, in combination
with the late Roman material culture enjoyed by both peoples, make it impossible
to identify German racial characteristics with any certainty. In other words, the
majority of the laeti and foederati are archaeologically indistinguishable from
the Gallo-Romans.
However, there is an exception. This is the cemetery at Furfooz, in the valley
of the Lesse, some miles south of Namur.xf Here, in and around the bath-house
of the Roman fortress on the heights of Hauterecenne, were discovered 26
burials, and of these the eight male interments were all accompanied by weapons:
axes, spears and arrows. One of these graves has been described above, and many
of the others contain similar equipment. Coins were originally found in some of
the graves, and, although the associations were not preserved, the list, down to
Magnus Maximus (A.D. 383-8), suggests that the burials, which seem to have
begun in the middle of the fourth century, continued until the end of tha t century,
if not later. The grave-groups, although not so rich as those from Verrnand and
Monceau-Ie-Neuf, are very similar, and it is clear that the Furfooz people had an
almost identical material culture. The proportion of warrior graves, however,
is higher than in the other cemeteries and this as well as its situation on a Roman
fortified site, emphasizes the military character of this cemetery.
There are, in addition, various features of the cemetery which indicate that
the Furfooz people were not Gallo-Romans, but Germans. Nenquin has rightly
stressed that the burial of the dead among the hypocaust pillars of a perfectly
usable bath-house reflects a barbarian attitude to the amenities of civilization.
There is also the old woman in grave V who had been decapitated before burial
q Breuer and Roosens (1957), p. 294.
'7a De Laet, Dhondt and Nenquin (1952).
,8 Nenquin (1953); Bequet (1877).
and laid with her head between her knees-s-a custom that occurs later in Frankish
and Anglo-Saxon cemeteries-presumably to prevent the spirit of the aged
creature, who may have been regarded as a witch, from walking after death.
Finally, one locally-made pot from the cemetery has German prototypes. These
indications must mean that the Furfooz people were a group of Germanic laeti
settled with their wives and children in or near the fortress of Hauterecenne.
They can probably be identified with a section of the command of the praefectus
laetorum Actorum Epuso (Yvois-Carignan, near Sedan) Belgicae primae.i part of
whose duties seems to have been to guard the roads to Trier. Furfooz is close to
the Bavai-Dinant-Arlon-Trier road.
Here, then, at Furfooz, we have a group of military settlers whose material
culture is almost indistinguishable from that displayed in graves at Vermand
and other late Roman cemeteries in north Gaul, and who can be identified as
Germans because they left behind them unmistakable traces of their racial customs
and beliefs. That such customs survived at all is probably because the Furfooz
people remained a distinct racial group. At Vermand, Monceau-le-Neuf,
Abbeville-Homblieres.>v and Vert-la-Cravelle, on the other hand, the warrior
graves were in a proportion which suggests that the military section was in the
minority in these communities. Consequently, if these soldiers were originally
German, it is only to be expected that contact with the nominally Christian
civil population would have had the effect of eliminating many of their barbarian
characteristics.
That in fact they were German is overwhelmingly probable, despite all
arguments to the contrary, for reasons which we can now summarize briefly.
We know that the Roman army at the end of the fourth century recruited its
forces almost entirely from the barbarian peoples along the frontiers of the
Empire. We know that Germanic laeti andfoederati were settled injust the areas
where the warrior graves occur, and we know from Furfooz that the burials of
such German soldiers take the form of these warrior graves. Therefore we are
surely justified in concluding that Werner was correct when he said that the
warrior graves were the graves of Germans.
Confirmation of this is provided by the burials at Dorchester-on-Thames.
Close to the military grave was that of a woman, buried with a buckle (type
IE, 5), and two brooches of north German origin: 2 2 a cruciform brooch and the
backplate of an applied (komponierte) brooch (FIG. I, nos. 14-16). From a woman's
grave on the opposite side of Dorchester came a Roman key and bracelets,
two complete applied brooches and the backplate of a third.vs All five brooches
are, as Leeds has shown, unmistakable north German types, ancestors of the
brooches that came to England with the Anglo-Saxon settlers in the middle of
the fifth century. These Dorchester examples are the earliest examples of their
'9 Noiitia Dignitatum, pars occidcntis, chap. xlii.

zo Pilley, I (1886), 177 ff.
zr Lantier (1948).
z.a Kirk and Leeds (1954), pp. 67-9. 72-3, fig. 27, 14-16.
'3 Ibid., pp. 69-70, 73-4. figs. 29-30.
IO MEDIEVAL ARCHAEOLOGY
kind so far found in this country. The situation at Dorchester, therefore, is an

extremely interesting one. Here are two women, evidently immigrants from some
region of north Germany, very probably Frisia, one of whom was buried near a
man equipped in the characteristic fashion of the laeti of north Gaul and the
Rhineland. In the original account of the discovery the man's skeleton was
described as of 'enormous size', the height being later calculated at about 6 feet.
This certainly suggests that he too was a German;" one of a detachment of
German soldiers settled with their womenfolk in or near the Roman town of
Dorchester.o Moreover, they were evidently Germans from those 'remotest
shores of barbary' mentioned years earlier by the panegyrist of Constantine. I4
That troops were recruited from so far north is not as surprising as might be
supposed. The empire was desperately short of manpower, and north Germany
desperately overcrowded, so that the landless younger sons of the warlike
northern races would find service in the Roman army acceptable employment
for their fighting talents. In north Germany, in Frisia and Saxony, moreover,
finds of buckles, strap-ends, and disc-attachments-o show that some of these
soldiers eventually returned to their homelands; however, many left their bones
in the soil of the empire, as did the people at Dorchester.
IMPORTED CONTINENTAL METALWORK OF LATE ROMAN DATE:

BRONZE BUCKLES AND BELT-FITTINGS OF TYPES III A, IV, V A,
VI AND VII
Buckles of types III A and IV A and strap-ends of type V are of common
occurrence on the continent, and are often found buried together in sets of belt-
ornaments. The finest examples have geometric decoration executed in 'chip-
carved' work, as well as the characteristic animal heads. It is these handsome and
showy pieces which have received the most attention from archaeologists and
art-historians. In this country very few have been found, and this is not the place
to go into the details of their history, or to examine the full repertory of their
designs. It would in any case be impossible at present, in view of the large numbers
in western and central Europe and the lack of anything like a complete published
corpus of them. However, there have been several selective studies, and from
these we can obtain enough information to put our own finds in their proper
context. A long list of chip-carved metalwork was published with a bibliography
some thirty years ago,27 and this is still invaluable. Since that time there have
been few major contributions to the subject, except for the important work by
24 Such skeletal statistics as we have from late Roman Britain show that the population was slightly
built, unlike the Germans, who were remarkable at this period for their height and physique.
'5 There is every reason to suppose that only a small part of the Dorchester cemetery has been
excavated, and that further work there would uncover more burials.
26 Plettke (1921), pls, xiii, 18-2Ia, xv, 13-14, 20, 21 ane! 21a showing examples from Borgstedt,
Westerwanna, Hemmoor, Quelkhorn, Perlberg and Langen; Roeder (1933), pI. vi, 4, from Galgenberg
nr, Nesse; Tischler (1954), fig. 19, from Pritzien nr. Hagenow. See also the map and list of type V strap-
ends in Werner (1958), pp 410 IT., fig. 15, and his notes of warrior graves in north-west Germany, pp. 379 IT.
'7 Behrens (1930).
SOLDIERS AND SETTLERS IN BRITAIN I I
Forssander-f and the publication of the Furfooz ccmetery.tf Most of the British
pieces arc not highly decorated; on the continent, too, those with chip-carving
seem to be outnumbered by less ornamental examples. Yet these humbler brothers
of the fine chip-carved buckles and strap-ends have been so much neglected there
that their numbers are difficult to estimate. What is clear, none the less, is that
they are contemporary with the chip-carved pieces, have the same distribution,
and are an integral part of the same style-phase.
The chip-carving style is classical in origin, as are its principal motives and
designs-the pelta, the palmette, the rosette, and the vine-scroll. The use of
animal-head terminals is also a late antique fashion. Some of the type IV A
buckles are still very classical in style. The broken buckle-plate from Snodland
in Kent (PL. II, B; type IV A, 2), with its medallions and busts, is a good example,
and there are very similar pieces from Rome itself and Hungary.>v But the
great mass of the material, with its florid chip-carving and bizarre, stylized,
animal figures, is already barbaric in feeling. The move away from classical
naturalism towards a more abstract interpretation of ornament is quite pro-
nounced. The vine-scroll frequently loses its foliate character and becomes
simple spiral decoration, and the marginal animals and terminal heads become
progressively less naturalistic. The use of surface decoration as a whole becomes
less restrained-at once cruder and more striking. The resultant style has generally
been attributed to the influence of Germanic tastes in late Roman provincial
art, and to explain this we must place the buckles and belt-fittings in their
historical and cultural setting.
Nearly all the chip-carved metalwork is concentrated in a striking distribu-
tion along the frontiers of the late Roman Empire in the west (FIG. 3),30 from
Britain across Belgium and north France, up the Rhine from Cologne, then down
the Danube through Austria and Hungary, with outliers in Italy and Jugoslavia,
and ending in a thin scatter on the lower Danube in southern Roumania. Most
of this metalwork has been found on late Roman fortified sites, particularly at
such centres as Cologne, Mainz, and Tournai, and in their adjacent military
cemeteries. A great deal has also come from the graves of the laeti and federates
discussed above. The general context is thus a military one, and it is logical to
conclude that this type of metalwork was produced primarily for the use of the
late Roman army, more particularly for the limitanei of the frontier forts, and the
military settlers, laeti and foederati, established in the frontier zone. In time, of
course, the style must have become fashionable among the civilians in these
regions, and so we find chip-carved belt-sets in women's graves at Furfooz and
at Enns (Lauriacum) in Austria, and in men's graves without weapons at Mainz 32
and Chevincourt (Oise); There is little or nothing to show for it, however, in
the essentially non-military interior regions of Gaul. This metalwork, in fact,
zS Forssander (1937).
'9 Ricgl (1927), Eg. 86; Behrens (1930), fig. 8; Forssander (1937), fig. 19.
JO Behrens (1930), fig. I; brought up to date for north Gaul by Faider-Feytmans (1951), pI. viii.
J' Werner (1930), p. 59.
J' Bonner ]ahrb., ex LVII (1942), 249 ff.
JJ Bull. Soc. Hist. de Compiegne, VII (1888), 273 ff.
represents late Roman ornamental style in a distinctive version, limited to the

military districts. And the military forces there, as we have seen above-the
limitanei, laeti andfoederati, and the rest-were virtually all Germans. The military
style owes its character, in fact, to the Germanic taste of these forces, and their
womenfolk, who preferred Roman art in a Germanized interpretation. The chip-
FIG. 3
DISTRIBUTIO~ OF 'CHIP-CARVED' BELT-FITTINGS IN EUROPE (p. I I)
carved buckles and belt-fittings, classical in origin but barbarian in feeling, are
the outward expression of that interpretation and that preference. Moreover,
Germanic taste seems to have imposed itself even in high military circles, for
these florid chip-carved buckles, together with their cheaper and simpler counter-
parts, have all the appearances of being part of the official military uniform of
the period. Such an idea is perfectly justifiable-indeed almost inescapable-in
view of the amazing standardization of form and style in this metalwork, which
suggests that it must have been mass-produced by a highly organized and official
industry. Objects which were obviously made in the same workshop can be
found hundreds of miles apart along the great length of the frontier. For example,
the buckle of type IV A from Sucidava in south Roumania, the most easterly of
the finds,> is one of a series with one-piece rectangular plates that is found every-
where along the frontier, and as far west as Smithfield in London (PL. II, A; type
IV A, 3). Decorative details on the Sucidava buckle are repeated on other examples
34 Tudor (1945), fig. r a.

from Salona on the Dalmatian coast, and from Kreuznach, near the Rhine
below Mainz.w The ornament of the Smithfield buckle appears to have no exact
counterpart, but its coarse S-shaped scrolls are seen on a triangular plate, again
from Enns,> and the dense arrangement of its ornament is reminiscent of that
on a set of belt-fittings from Houdan, on the Seine, one of the most southerly of
the Gaulish pieces.sf The unusual design ofthe Richborough buckle (FIG. 2 I; type
IV A, I), with its pair of crouching animals at the top of the plate, is paralled on a
buckle from St. Polten in Austria.w and the fine buckle from Kent (PL. I, type
III A, 8) also has interesting continental counterparts. There is a very similar piece
at Mainz-v which has the same double tongue and animal-ornamented cross-
bar, and the same swastika-arm decoration on the loop. A third buckle of exactly
the same form was found at Herbergen, near Oldenburg in north-west Germany.
The fashion of placing animal heads or figures at either side of the base of the
tongue appears on yet another group of buckles of type III A, all obviously made
in a single workshop. These are the famous buckle from Sedan in north France-s
and its near-twin from Hungary.o together with a group of buckles and plates
from the cemeteries of Vermand and Misery.o They are made distinctive,
indeed, by the stylized engravings of animals that appear on their plates. Yet the
animals that flank their tongues are almost identical with those on the Mainz and
Herbergen buckles, and it is tempting to think that the same workshop produced
both groups.
Until the whole corpus of this chip-carved metalwork from the continent
has been assembled and studied, it will not be possible to guess at the number of
workshops engaged in manufacturing it for the Roman army. To locate them will
be even more difficult. Even if the industry was so organized that workshops
in the various headquarter towns and forts catered only for the forces in their
immediate area, the transfers of men and units to different parts of the frontier
will have scattered their products, obliterating any regional grouping that they
may originally have had. The retirement of ex-soldiers either to their own home-
lands, or to lands granted them within the frontiers, may have further contributed
to this dispersion. If we look again at the two groups of buckles, which we have
discussed immediately above, and if we decide from the little nucleus of them in
north-west France that they were made in that region, then we must envisage
soldiers from north Gaul being transferred as far away as Hungary on the one
side, and Britain on the other.
35 Riegl (1927), pl. xvii, 6.
36 Behrens (1930), fig. 3.
31 Riegl (1927), pl. xx, I. This heavily stylized version of scroll ornament was adopted also outside
the empire, at the beginning of the fifth century, on the equal-armed brooches of the Saxons in north-west
Germany.
3 8 c:. Roach Smith, Gollectanea Antiquo, IV, pl. xliii.
39 "Verner (1930), fig. 35, 2.
4' Baldwin Brown, IV (1915), pl. cl, 3; Behrens (1930), fig. 4.
4' Behrens (1930), pl. xxix, A; Forssander (1937), fig 23,2.
4' Salin (1904), fig. 338; Forssander (1937), fig. 24, I.
43 Riegl (1927). pl. xxii, 5; Forssander (1937), fig. 24, 2.
44 Eck (1891), pis. xv, r a, 2a and 3a, xvi, 8, and 9.
45 Rigollot (1850), pp. 216 ff., pls, x and xi; Werner (1958), pls. 82-3.
There can be little doubt that it was such troop movements which introduced
into Britain not only the chip-carved buckles from Kent (PL. I; type III A, 8), Rich-
borough, Snodland and Smithfield (PL. II, A; type IV A, 3), but also most of
the remaining buckles and belt-fittings of the types discussed in this section. Few
are of the high quality of the four buckles just mentioned, except for the buckle
and belt-suite (PL. III; type IV A, 4), unprovenienced but almost certainly from
this country, in the Mayer collection at Liverpool Museum.av and the buckle
from Oxford (FIG. 20, j; type III A, 7), which is now very worn. In addition to
these there are a few pieces with inferior and less interesting chip-carved ornament
such as the broken strap-ends from Richborough (FIG. 23, d and e; type V A, 4-5),
and that from Leicester (FIG. '23, a; type V A, 2), which, with its central scrolls
and marginal animals, is reminiscent of the much finer example from Furfooz.cz
The strap-end from Ixworth (FIG. 23, g; type V A, 10) with its exceptionally
shallow chip-carved ornament is somewhat unusual in other respects too.
Although its pelta-rosette is one of the variants that occurs with fair frequency
on the continental metalwork, its panel of cross-hatching is less easy to parallel.
Such cross-hatching seems to have played only a subordinate role in the con-
tinental repertory and examples of it are scattered. But it undoubtedly was a
favourite form of decoration on our buckle-plates of type I (see below, p. 24), and
these were an insular development. The Ixworth strap-end may therefore be
either an imported piece, or a British copy. Two other strap-ends also hold an
ambiguous place in the corpus. These, the stamp-decorated fragment from Rich-
borough (FIG. 23, b; type V A, 6) and the densely-stamped example from Ickling-
ham (FIG. 23, h; type V A, 9) with its worn marginal animals, have no exact
parallels abroad. Were they also local copies of imported pieces? It is difficult to
be certain. Stamped decoration of this type is common on both sides ofthe Channel
at this period, and on the whole it is perhaps best to assume that these pieces
were made abroad, since we have no real evidence to the contrary. The strap-
ends which have no ornament or else are very simply bordered with lines of stamps
are certainly continental. We have only four of these: those from Dorchester (FIG.
r, no. II; type V A, I) and Milton (FIG. 2, a; type V A, 3), and two others from
Anglo-Saxon graves at Cassington (FIG. 23, j; type V A, 7) and Croydon (FIG.
23, c; type v A, 8). Werner has recently made a list and distribution-map of
those found on the continent, and shown that they are numerous in north France,
Belgium, the Rhineland, and north Germany.sf
With the buckle from Catterick (FIG. 22; type IV B, I) and the buckles of
type III A, we have no problems. The Catterick one has been discussed at some
length by Hildyard.w who cites sufficient parallels to' show that it belongs to a
distinctive, if not numerous, continental class. Mostlikeit are buckles from Bingen
on the Rhine near Mainz 5 0 and Krefeld Gellep near Cologne. It was clearly
46 Baldwin Brown, IV (1915), pI. el, 3.
47 Nenquin (1953), pI. vi and fig. 13, no. DIA.
48 Werner (1958), pp. 410-11, fig. 15.
49 Hildyard (1957), pp. 243 ff., pI. ix, fig. 6, 12.
5 Ibid., fig. 6, 12, b; Kataloge Sudwest-deutscher arch. Sammlung, IV, pI. iii.
5' Hildyard (1957), fig. 6, 12C; Behrens (1930), pI. xxxi, 4.
SOLDIERS AND SETTLERS IN BRITAIN
made in the same workshop as these, probably somewhere in the Rhineland. For
the closest counterparts of our buckles of type III A we must look to the cemeteries
of Belgium and northern France. At Vermand, Aisne, we find a buckle with an
arcaded and tooled loop very similar to the one from Icklingham (FIG. 20, d; type
III A, 6) .52 The irregularly notched loop of the Holbury buckle (FIG. 20, c; III A, 3),
with its stylized animal-heads, is again very closely matched at Vcrrnand.se and
there is a somewhat similar example at Furfooz.s- Semi-circular plates like that
on the Dorchester piece are less common on this class of buckle than those of
rectangular form, but there are a few parallels, notably the fine silver-gilt buckles,
ornamented with niello and chip-carving, in the warrior's grave at Vermand.se and
another decorative piece in a grave at Abbeville-Homblieres, Aisne.so Presumably
the Dorchester buckle is a cheaper version of such exotic products. The other
type III A buckles from this country (FIG. 20, a, band e, type III A, 1,4-5) have no
plates. They have been found at Richborough and Bradwell, both 'Saxon-Shore'
forts. They are undecorated and their animal heads are stylized and flatly
rendered. There are several like them in the Belgian and north French cemeteries,
and among these the ones from Spontin-> and Molcnbeek-St.vjean.w both in
Belgium, provide the closest comparisons.
Sufficient has already been said about the disc-attachments and attachment-
plates to demonstrate their continental origin. Apart from those in the two graves
at Dorchester and Milton and those associated with the buckle from Kent (PL. I;
types VI, 12-13 and VII, 5), the only finds from Britain are those from Richborough
(FIG 24, C, d, g and h; types VI, 7-9, VII, 4), Croydon (FIG. 24, e; type VI, ro),
Croxton (FIG. 24,f; type VI, I I), and Caistor-by-Norwich (FIG. 24, b: type VI, I).
The only feature worth further comment is the simple running spiral decoration
on the Milton attachment-plates. Borders of this design are, of course, a simplified
linear version of the chip-carved tendril-scroll pattern, of which one could cite
many examples.as Incised running-spiral decoration comparable to that on the
Milton plates can be seen at this period at Verrnand.vv Monceau-Ie-Neuf.v-
Trier.v- and Frankfurt.es and at many other sites. It was a favourite form of
decoration in the late Roman Empire.
Here, then, we have several types of objects, of varying quality, but almost all
of continental manufacture, which have been found on sites in Britain (FIG. 4)
in one certain and two possible late Roman military burials (Dorchester, Milton
and Kent); on late Roman military sites (the 'Saxon-Shore' forts at Richborough
F Eck (l891), pl. xvi, 8.
53 Ibid., grave 284, pp. 252 ff., pl. xvii, 4a.
54 Nenquin (l95g), pI. vii, D6.
55 Eck (1891), pl. ii, 2 and 4.
56 Pilloy, 1 (1886), pl. v.
5i Dasnoy (1955), pI. i.
;8 de Loe (1937), pp. 251-3.
59 Eck (1891), pl. ii, 5-6; Riegl (1927), pls. xvii, 4-6.
60 Eck (1891), pl. xv, ga.
6, Pilloy, III (1912), pl. iv, 4.
6, Werner (1958), pl. lxxx, 2.
6J Behrens (1930), fig. 6.
fROid kOMAN 11m .. ..

IIIA riA IVI VA VI vn
Fml AN'LO lA~OH llHl

(HAH(f FINIS .. A
c: 0
e
~" .0 Mill!
....
0 10
!
~ JO 40
!
50 70
I
0 10 30 50 70 Kill.
bW w W I
FIG. 4
DISTRIBUTIO:'{ OF CONTINEl\'TAL METALWORK OF TYPES III A, IV, V A, VI and VII
IN BRITAIN (pp. 15-17)
and Bradwell); in late levels of Roman towns (London, Leicester, Caistor-by-
Norwich and Cattcrick); in a Roman villa with late occupation (Holbury,
Hants.); and as chance finds mostly in the east of England (Snodland, Kent;
Ixworth and Icklingham, Suffolk; Croxton in SW. Norfolk; and Oxford). The
two finds from Anglo-Saxon burials are almost certainly survivals from the late
fourth or early fifth century. Most, if not all, of these objects were made in work-
shops in Belgium, north France, and the Rhineland for the late Roman frontier
forces and their Germanic allies.
The predominantly eastern distribution of our own material (FIG. 4) suggests
that we are mainly concerned with th e forces of the Comes Litoris Saxonici: the
army charged with the defence of the eastern coast against Saxon raiders, and
based on the chain of 'Saxon-Shore' forts stretching from Brancaster in Norfolk
to Porchester in Hampshire. Apart from the names of the garrisons recorded in
the British sections of the Notitia Dignitatum, very little is known about the com-
position of this army in late Roman times, nor is much known about the forts
themselves. Many of the forts have been damaged by coast erosion, or by later
Anglo-Saxon and medieval building. It is the headquarters fort of Richborough,
the only one that has been extensively excavated in modern times, which has
given us the largest assemblage of late Roman military metalwork. Very little
has come to light from the other forts, and we must assume that their archaeology
was like that of Richborough. The pathetically small group of finds from Bradwell
hints as much. The military culture of Riehborough therefore should be regarded
not as a unique phenomenon, but as representative of that of the 'Saxon Shore'
as a whole.
The buckles and belt-fittings suggest that in the command of the Count of
the Saxon Shore in the late fourth century there were soldiers who had been
brought over from the continent, and who were probably of German stock. The
bronze objects are not the only testimony to this. A grave, found by chance
outside the defences at Richborough, contained the skeleton of a man, with long-
sword, spear, shield and pewter bowl (FIG. 5, a-c). 64 Bushe-Fox called it the
burial of a Saxon raider, but no band of Saxon raiders would have buried one
of their number is such careful fashion within sight of a Roman fort. In any
case, as we have seen above, this kind of equipment is typical of late Roman
military burials on the continent. German he may well have been, but this
warrior was certainly one of the Richborough garrison: one of the defenders
and not one of the attackers. This apparently solitary burial is likely to be part
of an as yet unexcavated cemetery of the 'Saxon-Shore' fort. The presence of
other soldiers with similar equipment is attested by the finds, in ditch fillings and
top soil, of a shield boss (FIG. 5, d), throwing axes, and spears, identical with those
of the continental warrior graves. 6s That the last defenders of the 'Saxon Shore'
included many half-barbarian troops like the limitanei, by now familiar, whose
graves and whose equipment cluster so thickly around the frontier towns and
64 Bushe Fox (1949), pI. lxiii, nos. 349-351. The sword, which apparently had a horn-mounted hilt.
was not preserved or illustrated.
65 Ibid., pls, lxiii, no. 352, lxi, nos. 341-2, and pI. lviii.
forts of north Gaul and the
Rhineland, is, as we shall
see, not in the least sur-
prising. More puzzling
perhaps is their presence in
inland towns of the civil
zone, at London, Leicester
and Dorchester. This ques-
tion we will leave for
b consideration in the next
section.
When, more exact-
ly, was all this taking
place? None of our mat-
erial is closely datable.
The best that the Rich-
borough finds, or the
Cattcrick buckle, can do
c is to indicate the late
a II
fourth or early fifth cen-
tury in general. For more
precise dating we must
return to the continent,
where some of the warrior
graves, and contemporary
graves with similar metal-
work, contained coins.
A glance over the
whole picture tells us at
once that the military
L style of chip-carved and
animal-ornamented buck-
u d les and belt-fittings, and

the weapon-equipped
graves of which they are
RICHBOROUGH, KE:'oIT so characteristic a feature,
a-c, Spear-head, shield-boss and pewter bowl, from a burial N. of begin only in the second
'Saxon-shore' fort ; d, shield-boss from inner diteh of stone fort
(p. 17). Se. } half of the fourth cen-
tury. The woman's grave
at Enns in Austria, which
contained seven fine chip-carved belt-fittings.s- is dated after A.D. 360-3 by a
coin of Julian the Apostate. Two graves at Abbeville-Homblieres.vs which
contained almost identical buckles of type III A, had coins, one a corroded bronze
of either Constans (337-350) or Magnentius (350-3), the other a silver siliqua
66 Graves 4 and 67: ef. Pilloy, I (1886), 179-80, and 19 I -2.
of Valentinian I (364-75). The famous warrior grave at Vermand, with its chip-
carved silver buckles of type III A, was strongly suspeeted by Eck 1 5 to have
contained a gold solidus of Arcadius (395-408). It was in any case one of what
must have been a family groupo-all rieh burials with gold or silver coins, none
earlier than the three women's graves with coins of Valentinian 1.6 7 Yet another
grave at Vermand with a fine buckle of type III A and plate of type VII contained
two silver siliquae of Honorius (395-423).68 At Spontin, the stylized loop of type
III A, which we have already noticed as a close parallel to the Richborough
examples (p. 15), was buried with a gold solidus of Constantine III (407- I I), and
we must not forget the grave at Vieuxville (see above, p. 5), where an assemblage
of buckles and fittings of the types in question was buried together with coins
of Constantine III and Jovinus (411-13). At Furfooz, where no detailed coin
associations were preserved, the coin-list for the cemetery went down to Magnus
Maximus (383-8).69
Coins are admittedly deceptive. They may be preserved for many years
before burial, or they may be buried almost at once. They cannot therefore
be relied on to give an exact dating. In this case what they and the other grave
goods tell us is that the chip-carved and related metalwork cannot be dated much,
if at all, before the reign of Valentinian 1. We have seen that this metalwork
seems to have been produced in official workshops as part of the standard military
uniform of the frontier forces of the western empire, and it is only logical to
assume that at some point in time an official order was given to bring such a
state of affairs into being. Now Valentinian was first and foremost a soldier,
efficient and energetic when it came to military matters. He was chiefly concerned
with the frontier defences, and was the first emperor since Diocletian and
Maximian to undertake new fort building, both on the Rhine and in north Gaul.
He also strengthened his forces by recruiting Franks into his armies. Did he
also order the adoption of this distinctive new military uniform-this metalwork
which would not only have appealed to the tastes of his German soldiery but
by its standardization would also have given them a sense of esprit de corps? It is
an attractive theory and one that is not improbable.
During the first years of the reign of Valentinian, from 364 onwards, Britain
became increasingly subject to attacks from all quarters against her frontiers:
attacks from Saxons, Picts and Scots. In 367 the Roman armies in Britain were
defeated; the Commander of the Coastal Defences, the so-called 'Saxon Shore',
was killed and the General of the Field Army was routed. The province was
overrun and pillaged. The news reached Valentinian when he was in northern
Gaul attending, as ever, to the defences of the frontier. In 368 he sent Count
Theodosius across the Channel with a large force to restore order in Britain.
Theodosius found the country full of raiding bands of barbarians, and it was
not until 369 that he could begin to drive them out and restore peace. Except
that the Wall in the north was once more put in order and signal-stations were
6; Eek (1886), pp. 23-6.
68 Eek (1886), pp. 104-5, pI. xvi, 8-9.
69 :\'enquin (1953), pp. 20-21.
built as a northward extension of the 'Saxon-Shore' defences, we known very

little about his measures. But C. E. Stevens has made two very interesting
suggestions about the manning of the frontiers after 368,70 and these are of
importance to us. The first was that the Wall was from this time garrisoned by
local British militia and federates; the second was that the Milites Tungrecani
(of Tongres) and the Numerus Turnacensium (of Tournai), recorded by the Notitia
as the garrisons of the 'Saxon-Shore' forts of Dover and Lymne, were among
the troops brought over by Theodosius. From their names we can guess that
they were detachments which had been withdrawn from the northern frontier
of Gaul, that is troops from the very area which has produced so many of the
warrior graves and so much of the metalwork we have been discussing. There
is every reason to accept Stevens's view, and not only on documentary grounds.
The crisis of 367 had found Valentinian in this very area of Gaul, and in hurriedly
assembling an army to send to Britain he would naturally have found it quickest
to use the nearest available forces. The situation in Britain in 369 is thus an
interesting one. On the 'Saxon Shore' we have forces drafted in from northern
Gaul.:" whereas on the Wall we have only local British levies. This accords
perfectly well with the distribution of our imported buckles and belt-fittings,
which, as we have seen, are concentrated on or near the 'Saxon Shore', and
which do not occur at all on the Wall. We may, therefore, be justified in thinking
that the chip-carved and animal-ornamented military metalwork was first
introduced into Britain from north Gaul by the army of Theodosius in 368-9.
This was only a beginning, of course. As we can see from the continental
graves, buckles of types III A and IV and belt-fittings of types V A, VI and VII
continued in use throughout the latter half of the fourth century and into the
fifth. More of them must have come into Britain during that time. We must
remember, for example, that, still during the reign of Valentinian and presumably
after 368, at least one tribe of Germanic federates-Alamanni under their king
Fraomar-were settled somewhere in Britain.v- They may possibly have had this
kind of equipment. Then, in 383 Magnus Maximus took an army from Britain
to help him in his bid for the rule of the west. The next fifteen years are obscure,
but in 398-9 the general Stilicho had again to restore some sort of order in Britain,
and presumably he brought a continental army to do it. The Wall was never
refortified, but York, and perhaps Catterick too, must have been held. It is strange
that none of our metalwork has been found at York, but the two buckles at
Catterick do suggest the presence there of a military unit. One of them is a
barbarous locally-made piece of type I A, and it should belong to a date later
than 368-83. It was stratified in the same level as the fine continentally-made
buckle of type IV B and an illegible late fourth or early fifth century coin. Taken
together these objects suggest that the military unit was of Stilicho's rather than
of Count Theodosius' army, but of course it is impossible to be certain. Like
7' Stevens (1940).

7' We must assume that the garrisons of the other 'Saxon-Shore' forts were also brought up to strength
by new troops at this time.
7' Ammianus Marcellinus, Historiae, XXIX, 4.
Theodosius, Stilicho must have taken steps to consolidate the defences of the
'Saxon Shore'. We can see this from tiles at Pevensey which were stamped with the
name of Honorius, and from the late coins at Richborough. Undoubtedly some
of the metalwork from this fort belongs to the Stilicho period but it is nearly
impossible to say which. One could, however, point to the stylized buckles of
type III A from here and from Bradwell, for these seem, on the evidence of the
Spontin grave, to be relatively late in date.
In 407, a certain Constantine, another usurper set up by the British, removed
the army from Britain in order to take control of Gaul, which had been invaded
by barbarians. Defeated not long after, Constantine was later executed by
Honorius. His action had virtually stripped Britain of her armed forces and
brought to an end all effective Roman rule. We have no evidence that any
official Roman army ever returned after this date.zs We can be reasonably
certain, therefore, that the bulk of the imported military metalwork had arrived
in Britain before this time.
BRONZE BUCKLES MADE IN BRITAIN IN THE LATE AND SUB-

ROMAN PERIODS: TYPES I AND II
(excluding II c)
Types I and II are by far the most numerous among British finds of zoomor-
phic buckles. Type II A, with over twenty examples, is the largest of the sub-
groups. The complete form is illustrated by only one find, the composite
piece from Colchester (FIG. 17, e). The rest of the group is made up by finds of
detached loops and plates. From the scarcity of continental parallels it is clear
that this type is most unusual outside Britain. I have found few examples abroad;
there is a solitary arcaded plate in the museum at Worms, and another with
similar decoration but different hinge-mechanism from the cemetery of Krefeld-
Gellep. The best-known parallels are two complete buckles from the Vermand
cemetery is- one (FIG. 6, b) has a plate with two pierced round-headed arches,
which is nearly identical with an example from Caerwent (FIG. 17, a); the other
(FIG. 6, a) has similar arches in four opposed pairs-a design found only once over
here, on the example from Caistor-by-Norwich (FIG. 17, g). The loops, on the
other hand, are rather different. Admittedly they all have pairs of confronted
dolphins which, especially on the Colchester and Leicester (FIG. 17, i) loops, are
very like the Vermand examples, but they have additional features which do not
occur on the Vermand buckles. These are the characteristic involution of the loop
terminals, and the curled side wings of the tongues. The Vermand buckles
have the straight hinge-bar, which is more normal at this period, and straight,
double or single, tongue. There are, of course, many different types of buckles
with confronted dolphins on the loops, but the involuted loop is rare on the
73 Arguments, based on the ambiguous information in the Notitia Dignitatum, have been put foward
to prove that Britain was re-occupied after 407. These were discredited in Stevens (1940), and are no
longer tenable.
74 Eck (r891), pI. xv, 7 and xvi, ra.
continent at this period. I know of only one example, from a grave at Angclliers
(Aude) ;75 a problem piece, with a perfect example of a II A loop, but whose
openwork plate, decorated with a horse, looks more Merovingian than late
Roman.
The sparsity of comparative material on the continent puts the type II A
buckles in a different
category from the im-
ported types discussed
in the previous section.
Are we to believe that
the Colchester buckle
is the sole survivor of
a continental class, re-
lated closely to the
Vermand buckles,
which was brought to
Britain by the army of
Theodosius, and that
it was the only one out
of all the other im-
ported types to be
extensively copied in
the British workshops?
F or there can be no
reasonable doubt that
a b
FIG. 6
all the rest, with the
BUCKLES FROM THE CEMETERY OF VERMAND (AISNE), possible exception of
FRANCE (1'. 21). Sc. 1 the Leicester (FIG. 17,
i) and Caistor (FIG.
I 7, g) fragments, were
produced in this country. Their large numbers suggest this, and so also does the
fact that many of them are so degenerate in design and craftsmanship. On some
pieces the dolphins have lost their crests (FIGS. I7, c and I8,]), on others they
have been converted into strange eared creatures (FIG. I7, b and]), and on the
unique buckle from Sal tersford (FIG. I8, k) the original form has been changed
almost beyond recognition. The confronted dolphins remain to attest its origin,
but they have ears and perching birds where their crests should be, and instead
of the involuted terminals there are straight bars terminating in flatly treated,
full-face human masks, which with their bleak stare are extraordinarily evoca-
tive of the third-century stone heads from Corbridge, Northumberland.zv From
this we can see that the type II A buckles were popular in Britain, and that some
75 H. Zeiss, Die Grabfunde aus dem spanischen Westgoten Reich (1934), p. 115,1'1. 32, 9; E. Saiin, La
Civilisation mirovingienne, IV (1959), 160, table A, 2.
76 1. A. Richmond, 'Two Celtic stone heads from Corbridge, Northumberland,' Dark-age Britain:
studies [nesented to E. T. Leeds (1956), pr. I I - j 5, pI. iii.
were made by local craftsmen who were not closely familiar with the better pro-
ducts of the continental workshops. Was not the whole series in fact a British
development? Was not the Colchester buckle, which stands at the head of this
development, produced with others like it in some official workshop in Britain,
later to be copied and imitated at second or third hand as the demand for such
metalwork grew? It is difficult to be sure, but it is quite possible. It may even
be that the continental examples of the type
were British-made pieces, or copies of such,
that had travelled back to Gaul, perhaps
with the troops of Magnus Maximus in
383. Certainly it is a mistake to assume that
the traffic in ideas was entirely one-way.
Of the two buckles which have been
placed together in type II B the example
from Richborough (FIG. I9, a) is clearly
a degenerate one-piece copy of the com-
posite II A form. The Sleaford buckle (FIG.
I9, b) is artistically more impressive and far
better made, but it probably had a similar
ongm.
Buckles of types I A and I B are also
very numerous in Britain. Taken together
they make an impressive group numbering FIG. 7
over thirty pieces. They are by far the STRAP-END FROM BABENHAUSEN
(HESSEN), GERMANY. Sc. ;
smallest of all the buckles in the corpus, and
with their long narrow plates they are very
distinctive. They have no exact continental parallels and are indisputably of
British manufacture.z> Naturally certain of their features are taken from the
continental repertory. As we have seen, loops with straight hinge-bars and
confronted dolphins do occur on the continent, although they are not the
commonest form. Then again, the outward-facing horse-heads of the I B
loops are paralleled to some extent on contemporary continental metalwork.
Certain objects spring to mind at once, namely the famous strap-ends from
Babenhausen on the Rhine near Mainz (FIG. 7),78 whose horse-heads keep
company with panels of fine mosaic-style chip-carving and engraving, and a
whole series of similar but cruder strap-ends.tv Most of these come from the
frontier districts of Germany. From here the style was transmitted to the north,
where very similar horse-heads occur on nearly contemporary Scandinavian
metalwork.w The appearance of the style in Britain is probably again to be
explained by the activities of the Roman armies.
7i Werner (1958), p. 383, is certainly wrong when he says that buckles of this type are of north
German origin. He appears in any case to think that the type IE buckle from Dorchester is the only one
of its kind. The Frisian objects he compares it with bear only the slightest resemblance to it.
7 8 Salin (1904), fig. 335; Behrens (1930), fig. 10, no. 38; Forssander (1937), fig. 25, I.
79 Behrens (1930), fig. 12; Werner (1958), pp. 4II-I2, fig. IS.
80 Forssander (1937), fig. I.
3
Although no two of the type I B buckle-loops are exactly alike and some are
more stylized than others, they are all very similar. The few type I plates that
have survived are more varied, and one or two are very fine. The two most
interesting certainly came from the same workshop. These, from Duston in
Northamptonshire (FIG. IS, 0) and Cirencester in Gloucestershire (FIG. IS, n),
have engraved decoration composed of roundels alternating with cross-hatched
geometric panels, and enclosed between borders of
running scroll design. I know of one other object
which was made in the same workshop. This is the
strap-end from Tortworth, Gloucestershire (FIG. 8),
now in the Bristol Museum. Its flanking horse-heads
alone would have established its relationship with our
buckles, but it also has a decorated roundel and
cross-hatched triangle which are identical with those
on the Cirencester plate. There is a fragment of a
third buckle-plate from Silchester (FIG. IS, q), which
may have come from this workshop. Simpler cross-
hatched geometric panels and roundels occur on the
il l
I Dorchester example (FIG. I, no. 16), and cross-hatched
panels alone on the Popham plate (FIG. IS, p) and on
a second piece from Cirencester (FIG. 13, l). The long
plate on the buckle of type I A from Upper Upham
(FIG. 13, g) has only a single line of running scrolls.
Despite the varying quality of the workmanship on
these plates it can be seen that they form a distinctive
stylistic group, and it is a style for which it is difficult to
find an origin abroad. The running spirals, as we have
seen (p. IS), occur on both sides of the Channel. The
cross-hatching also has a limited existence on the con-
tinental metalwork, but it cannot be called a promi-
nent feature there.w Among finds of continental type
FIG. 8
STRAP-END FROM
the strap-end from Ixworth (see p. 14) provides the
TORTWORTH, GLOS. best analogy for this, but, as we have seen above, this
Sc. } object may conceivably have been made in Britain.
These are minor points, however. The combination
of forms and decoration which we see on the buckles
and plates of type I is a new development, confined to Britain.
The least characteristic but perhaps the most interesting of all these British
buckles is the buckle of type I B from Stanwick in Yorkshire (FIG. 15, m). This
has long been famous. On its plate are a pair of engraved and stylized peacocks
confronted on either side of a tree. This design has been compared to that on a
bronze nail-cleaner from Rivenhall in Essex 8z which is like it in subject if not in
8, It makes occasional appearances on the chip-carved metalwork: cf. Riegel (1927), pI. xxii, 5;
Forssander (1937), figs. 19, zd and 24, 2; Behrens (1930), fig. 8, 3; Eck (1891), pI. xv, 13b.
8, Tonnochy and Hawkes (1931).
treatment. This motive and others like it, such as birds or animals confronted
against vases, were introduccd into the Roman Empire from the eastern Mediter-
ranean and were characteristic of early Christian art. In the late Roman period
we find this style in western Europe on the 'Spangen' helmets and on Christian
sarcophagi.fc It was taken over by the Frankish settlers and employed on some
of their metalwork, notably a class of wire-inlaid iron buckles which have applied
repousse plates.s- A few of these have been found in England.s- and here we again
see the confronted peacocks, although now in an advanced state of stylization.
These buckles belong to the end of the fifth century. The decoration of the
Stanwick buckle, therefore, belonged to a style which was almost universal in
the late Roman Empire, and current in Gaul at least as late as the second half
of the fifth century, when the Frankish settlements were being established. There
is every reason to suppose that it was well known to the population of late and
sub-Roman Britain, especially to the Christian community. The survival of so
few examples here is probably only due to chance.
We must now consider the date of the buckles of type I and II. Buckles of
type II A have been found in several helpful contexts. There was a good example
in the late fourth-century filling of the theatre at St. Albans (FIG. 18, d). Another
rather more stylized example (FIG. I7,J) was found in the debris that sealed the
bath-block furnace of the villa at Lullingstone, in Kent. Stratified with it were
two coins of Valens, and this would seem to put the loss of the buckle somewhere
within the last quarter of the fourth century. At Lydney in Gloucestershire
another such buckle was found (FIG. 17, k) in the original make-up of the floor of
the temple cella. Wheeler dated the construction of this some time after 367.86
The most interesting finds are certainly those from North Wraxall (FIG. 18, b)
and Caistor-by-Norwich (FIG. 17, c and g). That the villa at North Wraxall met
with a violent end is evident from the broken masonry and corpses tumbled in
the well. It is one of many villas in the west which suffered a similar fate. The coin
series at North Wraxall ends with several of Gratian (367-383), and this, seeing
that the site is within striking distance of the Bristol Channel, most strongly
suggests that it was sacked by Irish pirates, probably in the years after 388 when
Britain, on the death of Maximus, was left without adequate defence.s> The
buckle of type II A may thus belong to this time or a little before. The situation
at Caistor-by-Norwich was similar and equally dramatic. Building 4, in which
parts of two buckles of type II A were found, was burnt down and its occupants
apparently put to the sword. The victims of the massacre were left unburied-in
one room no less than thirty-six skulls were found together with other human
8) Alfoldi (1934); Leeds (1936), pI. viia, fig. 4.
84 See especially the piece from Envermeu (Seine Infcrieurc) which has the tree flanked by peacocks:
Cochet (1854), pI. xii, 4; Leeds (1936), fig. 4; Werner (1953), pI. vi, 9. There is a more stylized rendering
of the same design on a buckle in the museum at Epernay.
85 There is an interesting example on a plate from Howletts, Kent: Smith (1923), fig. 37; also a
degenerate version on a buckle from Broadstairs, Kent: Evison (1958), p. 241, fig. I.
86 Wheeler (1932), p. 86.
8) This summary of the situation at North Wraxall has been taken from my husband, Professor
C. F. C. Hawkes's chapter on Roman Wiltshire, in the forthcoming archaeological volume of the Victoria
Counry History, Wiltshire.
bones. These events, which must surely have marked the end of Roman Caistor,
seem, from coin evidence, to have taken place not long after 400. Historically,
the most likely time for the disaster is the period immediately after the with-
drawal of the Roman army from Britain by the usurper Constantine, in 407.
With the 'Saxon Shore' now undefended by regular troops Caistor was exposed
to attack from the sea, and probably met its fate at the hands of Saxon raiders.
The two buckles were found on the latest occupation-level, their loss being
apparently contemporary with the destruction of building 4.
From this series of dates we can see that buckles of type II A may have come
into use soon after the military reoccupation of Count Theodosius, and we have
clear and direct evidence of their use down to the early years of the fifth century.
But from the numbers of devolved examples of the type, from their appearance in
three Anglo-Saxon graves, and from the fact that the Anglo-Saxon buckles of
type II c appear to be partly modelled on them, it is reasonable to suggest that
their manufacture and use extended into sub-Roman times, down to the middle
years of the fifth century.
The buckles of type I are less easy to date, since there are so few reliable
associations. The most that can be said is that the examples from Richborough
(FIG. IS, f and g), Dorchester (FIG. I, no. 16), Chichester (FIG IS, h) and
Catterick (FIG. IS, d) must have been lost or buried at the end of the fourth or
early in the fifth century. Several others have been found at Silchester (FIG.
IS,], i,j) and Cirencester (FIGS. IS, band l; IS, b, c and n) and these two towns
are among the few that have so far produced evidence of occupation extending
well into the fifth century. 88 Then again, the buckles of type I have occurred in an
unusually large number of Anglo-Saxon graves (FIGS. 14 and 16), and there is
also strong evidence to suggest that they influenced the style of one or two pieces
of early Anglo-Saxon metalwork.w This must imply that buckles of type I were
still being worn in the middle of the fifth century when the Anglo-Saxon settle-
ments of the south of England were being established. How early they were made
is uncertain, but probably they began only towards the end of the fourth century,
that is, somewhat after the other types we have been discussing. One thing is
definite: the type I and type II buckles represent the last recognizable phase of
provincial Roman metalwork in Britain.sv During the fifth century, little that
was new seems to have come in from Europe before the advent of Anglo-Saxons
bringing their own styles of pottery and metalwork. We must therefore imagine
that these last examples of Romano-British craftwork were precious and had a
long life.
It will be remembered that the distribution of the imported buckles and
88 For the late occupation of Silchester, see O'Neil (1944); and Boon (1957) and (1959). For infor-
mation about Cirencester I am grateful to Professor Donald Atkinson, who has recently been working on
the coins from the town.
89 Notably the strap-end from Chessell Down, Lo.W.: cr. Hillier (1856), fig. 65; Evison (1955),
pI. viii, a, no. 40. This has horse-head and cross-hatched decoration. Horse-heads and chip-carving occur
on the belt-slide from High Down, grave 12: cf Archaeologia, LlV (1895), pI. xxvii, 8; Evison (1955), pI.
viii, c and d. These are discussed in more detail in Chadwick Hawkes (1961), pp. 36-7 and 54-5, fig. 2,
pl. xvii, b.
QO Penannular brooches were made throughout this period, but they are a native Celtic type.
lA n INS I1A JII

FlOIII lOMAN 11m
...
FlOIiil ANGLO-IAXOU llTEl CJ 6.
Cl V'
(H~N([ FINDS A
10 20 3ll 40 50 .0 70 MILES
I
I
0 10 30 50 70 Klil.
l;;;U l;;;;1 !
L- --lL ~""__'___'
FIG. 9
DISTRIBUTIO:,\ OF BRITISH-MADE BUCKLES, TYPES I AND II (pp. 28 ff.)

belt-fittings discussed in the previous section was predominantly south-eastern,
except for the Dorchester grave-group, two other finds in the upper Thames
valley, and outliers at Leicester, Catteriek, and Holbury. The distribution of
the British-made buckles of types I and II is markedly different (FIG. 9), and since
they both occur in the same areas we will consider them together. There is a
small group at Riehborough, two of I B and one of II B, and a few strays in the
neighbouring Anglo-Saxon cemeteries. From north-west Kent there is the II A
buckle from Lullingstone; from Essex two examples, the II A prototype from
Colchester and the I B piece from Gestingthorpe; from Suffolk a single devolved
II A buckle. There is a I A buckle from London and from near by a II A example
from an Anglo-Saxon grave at Mitcham. There are two finds from the south,
both of type I A, from Chichester and from an Anglo-Saxon burial on Beddingham
Hill. Verulamium has produced one of the II A type. In the far north we have
the I B buckle from Stanwick and the I A piece from Catterick. These are outliers.
The majority of both types comes from towns and other sites in the west, the
south-west, and in the midlands. There are groups from Caerwent, Cirencester
and Silchester, more scattered finds from villas and villages in Wiltshire and
Gloucestershire, from the Roman temple at Lydney, from a river port near Bristol,
and from Dorchester. A few more have come from Anglo-Saxon graves in the
valleys of the upper Thames and the Warwickshire Avon. There is an important
group from in and around Water Newton, Hunts., and others at Duston,
Northants., Leicester, and Caistor-by-Norwich. Single finds occur at sites in
Rutland and south Lincolnshire. North of this there is nothing until we reach
Catterick.
This distribution is at once interesting and puzzling. If we assume that the
finds in the graves of Anglo-Saxons were in most cases plundered from Roman
sites near their settlements, we can see from the map (FIG. 9) that, apart from the
finds in Yorkshire, the buckles of types I and II lie behind or along a line extending
from the Bristol Channel in the west to the basin of the Wash in the east. It
will be remembered too that, once again with the exception of a find at Catterick,
the imported types of buckles and belt-fittings have not been found north of
this line either. What then is the significance of this pattern of distribution?
Any answer, however tentative, must depend on the answer to a second question.
Who wore these buckles? Were the British-made buckles also a military fashion,
manufactured for the army in Britain to bring them into line with the troops
on the continent-manufactured in Britain because the imported metalwork
was not sufficient to meet the demand? Or were they simply a civil fashion that
had been copied from the military?
Let us begin with a reminder about the function of buckles. It is all too easy
to think of them solely as ornamental metalwork and forget their real purpose,
which is to fasten belts or straps. Belts are so common a feature of modern dress,
especially that of women, that there is a tendency to forget that in antiquity
they were worn only where the type of dress really required them. In the Roman
world belts seem to have been the almost exclusive prerogative of the soldier.
We see them and their buckles as a prominent feature of the uniforms of the
soldiers of all ranks who are commemorated on the tombstones and monuments.
In Britain they are not well represented in the archaeology, but they are found
on sites dating from the earliest phases of the Roman occupation. At Camulo-
dunum, for example, there are mid first-century buckles of two main types:
buckles which in smaller, plainer, form exactly foreshadow our involuted buckles
of type II A; and buckles which, with their narrow D-shaped loops and rectangular
plates, are the simpler forerunners of our buckles of type 1. 91 They occur on
other military sites too, but rarely in civilian contexts. The later history of these
two types is difficult to trace, but there is no reason to suppose that versions
of them did not continue in use throughout the succeeding centuries. Certainly
our buckles of types I and II appear to hold them in memory. They seem, in fact,
to be hybrid types modelled both on these early military buckles, already known
in Britain, and the new 'chip-carved' and zoomorphic continental metalwork
that began to appear in Britain after 368. Thus there is a military background
to their production. They are found in associations that are military, too. We
have them at Richborough, the site which has produced the largest group of
continental military metalwork, and in the German woman's grave at Dorchester,
next to the soldier's grave. The example at Catterick was found in the same
occupation-level of the same building as the fine buckle of type IV E, which came
originally from the Rhineland, and was perhaps worn by a member of the late
Roman army in the north. The Stanwick piece, too, is most credibly explained
by a military context. Finally, at Holbury, Caistor and Leicester buckles of type
II A were found in areas that also produced an example of the continental military
metalwork. Surely this is significant.
But now, what about the west, where there is no imported metalwork? If
these buckles of types I and II are military, what are we doing with an army here?
We are almost in the dark, but there is one find that may shed a little light. One
buckle of type II A was found at the villa of North Wraxall, which, as we have
suggested above, seems to have been sacked and destroyed by Irish pirates.
North Wraxall produced another find which is of interest, namely the crescent-
shaped ornament, composed of a pair of boar tusks united by a decorated bronze
mount, which is paralleled by similar, more fragmentary, finds from Rich-
borough (FIG. IO).93 A complete example was found in the cemetery of Monceau-
le-Neuf, Aisne.v- where it was part of the furniture of one of those rich warrior
graves discussed above (FIG. I I), and yet another was found in a grave at Brumath
(Brocomagus) near Strasbourg, on the upper Rhine frontier.vs At an earlier
period bronze ornaments of similar form are known to have been used as horse-
trappings, since they are depicted on equestrian sculptures as pendants to the
9' Hawkes and Hull (1947), pp. 335 ff., pl. cii,
9' Wilts. Archaeol. Mag., VII (1862), 70-73, pl. vi, I I.
93 Roach Smith (1850), fig. on p. 110; Bushe-Fox (1949), p. 14', pl. xlvi, 173-4. There is another
fragmentary example among a group of miscellaneous bronzes and late Roman coins from a site at
Southery, in west Norfolk, now preserved in the British Museum, Reg. no. 1880, 11-24,60.
94 Pilloy, III ('9'2), pp. 115 ff., pl. v; Boulanger (1905), pis. x and xx; Werner (1949), pp. 248-257,
figs. 1-4.
95 Werner (1949), p. 252, fig. 5.
horses' breast-bands.ss It seems likely therefore that the boar-tusk ornaments had
a similar use among the half-barbarian cavalry of the late Roman army. Thus
the examples from Richborough are probably to be explained by those same
~)
-: /
/ /
/ /
/ /
I 1
/ I
/ I
1 I
I I
I I
I I
I \
\
\
\
\
-,
-,
<,
<,
/1
-:
/ I
/ I
/ \
I
I
I I
I I
I \
I \
\
\
\ \
\ -,
\ -,
<,
\
\
FIG. 10
BOAR-TUSK ORNAMENTS~FROM RICHBOROUGH, KENT (p. 29). Sc.
9 6 For first-century bronze examples see Hawkes and Hull (1947), pI. ciii, 17; E. Ritterling, 'Das
fruhrornische Lager bei Hofheim im Taunus,' Annalen des Vereinsfur Nassauische Altertumskunde und Geschichts-
forschung, XL (1913), pI. xiv. See also Germania, XII (1928), p. 24, fig. 3. There are sculptural representations
on Trajan's column at Rome, and occasionally on tombstones, e.g. Mainzer Zeitschrift, XI (1916), pI. x, 6.
Germanic soldiers who left behind their weapons and buckles (see p. 17). And
to return to North Wraxall, it seems justifiable to suggest that we have evidence
here for the presence of at least one
member of the late Roman army. The
villa may have been granted him on
retirement, or he may have been one
of a unit of troops billeted there. But
on the whole it is more probable that
he belonged to a detachment called in
to defend the place during the raids
that caused its destruction. The bodies
in the well are mute testimony that
some fighting took place. Such a situ-
ation could well account for the loss
not only of the boar tusks, but of the
buckle too. Finally, the situation at
Caistor calls for a similar explanation.
Here the defeated defenders seem to
have been decapitated. Their bodies
were no doubt first stripped of wea-
pons and ornaments and the best of
the loot carried off. The buckle-frag-
ments of type II A could be explained
as part of the equipment of the defend-
ing forces, broken perhaps in the fight-
ing, and thrown aside by the retreating
raiders before they set fire to the
building.
This is speculation of course, but
it is based on evidence that is difficult
to interpret in any other way, and
it leads us to further speculations
about the dispositions of the army in
the late fourth and fifth centuries. We
know something about the frontier
troops, especially of the 'Saxon Shore',
at the time of Count Theodosius's re-
organization. We know from the
FIG. II
Notitia that there was a field army,
MONCEAU-LE-NEUF (AISNE), FRANCE but we know little else, and we know
Military grave showing boar-tusk ornament in next to nothing about the provisions
dish near feet (p. 29) for defence made by Stilicho at the
After Boulanger (1905)
very end of the century. Is it possible
that our types I and II buckles can be
of assistance here? As we have seen, some of them have military associations. Does
the whole series reflect some otherwise unknown disposition of troops? It could
very well do so.
The destruction of North Wraxall together with many other villas in the
west during the late fourth century shows how dangerous was the situation for
the undefended rural estates which provided the food for the Diocese. The events
of 368 must have shown bow long was the reach of raiding bands once they
penetrated the outer ring of defences. No doubt the towns were safe enough
behind their walls, but the agricultural estates must have been easy prey to the
marauders, and, if the food supplies failed, the whole economy of Roman Britain
would risk collapse. What could be more logical or more likely than that the towns
in the danger zones would be given military garrisons.v" part of whose duty was
the defence of the villa estates within their districts? If we can believe in such a
system, and it is not without its precedents in Gaul and elsewhere, we have a
credible explanation of the buckles at Caerwent, Cirencester and Silchester, some
of the chief towns of the west. In the same way, the curiously rigid distribution
of such metalwork across tbe midlands could represent some form of inner
'frontier'-a reserve line of defence should the field army, presumably still based
on York, fail to hold the north. The Wall, it must be remembered, could no
longer be counted on as a secure defence even in 367. If we see this phantom
army as a semi-mobile force-a sort of yeomanry-then the occurrence of buckles
at villas and other sites outside the towns need not unduly disturb us. And if
we rely at all on the evidence provided at North WI'ax all we can see that the
system was operating, in part at least, in the time of Maximus. It could conceivably
have been instituted by him as a precaution before he took the regular troops
of the field army and the 'Saxon Shore' away to Gaul in 383, and would explain
the finds of late fourth-century continental military metalwork at towns like
London, Leicester, Caistor and Catterick. The system will have continued in
operation after his time, and down to and probably after the departure of Constan-
tine with a second army in 407. In 4IO the Britons were authorized by Honorius
to take measures for their own defence. What could be more likely than that they
maintained this force, thus accounting for the long life of the buckles of types
I and II after the imported metalwork had ceased to be used? The implications
are fascinating. Such troops, maintained throughout the first half of the fifth
century, and based on the towns of the west and midland regions of Britain, could
very easily have been the model for the forces which Ambrosius Aurelianus and
the dux bellorum Arthur used against the invading Saxons during the second half
of that century.
If we consider the other alternative, that is, that buckles of types I and II
were worn by civilians in imitation of military custom, we find little supporting
evidence. Belts and buckles are never shown on sculpture as part of civilian
dress, and buckles are never found in civil graves. The civilian costume ofthe later
9 6 a This idea is not entirely unsupported by other evidence. It is now known that, during the fourth
century, the defences of Romano-British walled towns were being adapted, with the provision of bastions
and wider ditches, to accommodate the use of ballistae (Philip Corder, 'The reorganization of the defences
of Romano-British towns in the fourth century,' Archaeol. ]., CXII (1956), 20-42). The introduction of
artillery defence implies some sort of military garrison.
Roman period seems to have consisted of a dress or tunic caught in at the waist
by a narrow girdle, apparently tied, and cloak or mantle fastened at the shoulder
by a pin or brooch. Brooches and pins are common finds on civil sites, but buckles,
even in the fourth century, are extremely rare except in military contexts. In
Gaul, for example, it is noticeable that buckles occur only in the military zone
and in the graves of the German laeti. 97 Therefore, we are forced to conclude that
if our buckles of types I and II were worn by Romano-British civilians, it was a
phenomenon peculiar to the western and midland parts of Britain at the end of
the fourth and during the first part of the fifth century. But the western region,
where our buckles cluster so thickly on the distribution-map, was the very area
in which was felt the main force of the Celtic revival that was taking place during
this same fifth century A.D. This movement seems primarily to have been a
political one, leading to the reappearance of Celtic chiefs and kinglets, with
their capitals in the old tribal centres, yet it must have had cultural repercussions
too. The penannular brooch was the traditional Celtic dress-fastener and it was
still in usc at this period. It is thus difficult to see what the late Roman buckles
were doing in this region unless they had some military association. In the present
state of knowledge, therefore, the military explanation is the best one.
At this point a last word must be said about the military grave at Dorchester.
So far I have not attempted to date the burial; deliberately, since the problem of
dating is a difficult one. As we have seen, graves of this general type begin in
the second half of the fourth century and run on into the fifth, though we do not
know how far. Such coin evidence as we have suggests very strongly that most
of the cemeteries of north Gaul, such as Furfooz, Vermand, and Abbeville-
Homblieres, which contained these graves, went out of use during the early part
of the fifth century. This is usually accounted for by the disastrous events of 407,
when the combined tribes of the Alans, Vandals and Suevi, having crossed the
frozen Rhine ncar Mainz on the last day of December 406, invaded and ravaged
Gaul from end to end. Stilicho had had to withdraw the field army to Italy in
41, and Gaul seems to have been without defenders except for her frontier
forces. Many walled cities fell before the onslaught and were sacked and burnt.
Among the casualties were most of the towns of north Gaul, Tournai, Amiens,
Arras, Reims and Trier. A contemporary poet tells us that 'the whole of Gaul
lay reeking on a single pyre'. ,)8 Such a catastrophe must have destroyed the
settled way of life of many of the inhabitants of north Gaul, and many of the
laeti and foederati may have thrown in their lot with the invaders. Others may
have been too impoverished to continue with the custom of furnished burial.
At any rate, we find few of their burials that can be securely dated after 407.
But one we have. The burial at Vieuxville (p. 5) cannot, from its coins,
have taken place before 4 I I, and perhaps not for some little time after. Thus
97 All the women's graves that have been found to contain buckles or belt-sets seem to have been the
graves of the womenfolk of these frontier troops. At Dorchester, too, we know from the brooches that
the woman was a Gcrman. That Germans of both sexes adopted the fashion for buckles at this period,
and continued to usc them, evolving new types, is well authenticated by the grave-finds of the fourth
to seventh centuries. This fashion was certainly due to their costume, which differed from that of the
Roman provincials.
9 8 Orientius, Commonitorium, II, lSI: 'Uno fumavit Gallia tota rogo'.
we have evidence here that the characteristic furniture of the continental military
graves, including some of the chip-carved material, did in places continue in usc
at least throughout the first quarter of the fifth century. In some remote parts
of the frontier it may have gone on a little later, especially if there was no other
style to supersede it. We have an illustration of this at Sucidava in Rournania,??
where a group of chip-carved belt-fittings seems to have continued in usc well
into the second quarter of the fifth century. In the western parts of the frontier
there is strong presumptive evidence that in the first half of the fifth century the
official workshops were producing a new style of buckle, type III E, to be dis-
cussed in more detail in the next section. What is of particular interest to us
here is that these fifth-century buckles are often found in association with plain
strap-ends of type v, similar to those from Dorchester and Milton.rvv While this
does not preclude such plain strap-ends from commencing at the end of the fourth
century, it certainly implies that some of them, at least, continued to be made
and used well into the fifth. Although neither the Dorchester nor the Milton
groups contained buckles of type III E, which would put them into the fifth
century without a doubt, their plain strap-ends make it dangerous to be too
categorical about their date. In a recent paper Werner has given his opinion that
the man buried at Dorchester was an early free Saxon settler.> But he must be
wrong. It is far too soon for such settlers, and in any case one would not expect
to find them so far inland in the early fifth century, nor buried so close to the
walls of a town like Dorchester. As we have said before, the Dorchester Saxon was
there in some military capacity, either as a member of a garrison billeted in
the town itself or as a federate settled near by. Two explanations are possible. He
may have come over with the forces of Stilicho, at the very end of the fourth
century, perhaps to remain behind when Constantine removed the main army in
407; or, in view of the suggested maintenance of a military force after 407, he
may have arrived only after 410, presumably as one of a group of mercenaries
"
accompanied by their own womenfolk. The employment of such half-barbarian
fighting men to act as 'watchdogs' must by this time have been a familiar idea.
Later still, we see this same tradition reflected in the employment of Hengest and
his followers by the British king Vortigern. The system had its uses and its dangers.
It was perhaps inevitable that the course of events in Britain should follow that
in Gaul, and that here also watchdog should become master.
BRONZES OF THE ANGLO-SAXON PERIOD:

TYPES II C, III B AND V B
We come now to a handful of objects, a few buckles and a strap-end, which
differ from the characteristic late Roman types, and which have beenfound in this
country only in Anglo-Saxon graves.
99 Tudor (1954), p. 5'9. They were found close to two coin-hoards in which the latest coins were of
Thcodosius II (408-50). Tudor has suggested that the deposition of the hoards and the loss of the belt-
fittings is to be connected with Attila's destruction of this part of the frontier in 443.
100 Attention is paid to them in Werner (1958), pp. 39 I ff., list on pp. 4 10- I I, map, fig. '5; see also
pI. lxvii, I.
,0< Ibid., p. 383.
The two buckles of type II c, one from an unrecorded grave at Bifrons, Kent,
(FIG. I9, c; type II C, I), and the other from grave 26 at High Down, Sussex
(FIG. I9, d; type II C, 2), were almost certainly made in the same workshop,
probably by an Anglo-Saxon. Certainly there is no exact parallel for this par-
ticular form. But the craftsman who made them had surely seen examples of
the British II A type, [or he has copied the involuted terminals of the loops and
preserved the distinctive side wings to the
tongue in vestigial form on the High Down
piece-the tongue of the Bifrons buckle is
missing. He has abandoned the confronted
dolphins, however, in favour of pairs of con-
fronted animal heads at the ends of the
unusual double hinge-bar. The heads are
very like those on the buckles of type III and
it is evident that the craftsman was familiar
with the late Roman continental style. FIG. 12
Presumably, therefore, these II C buckles BUCKLE FROM FRILFORD. BERKS.
were made very shortly after the establish- Sc. J
ment of the Anglo-Saxon settlements in
southern Britain, that is, early in the second
half of the fifth century. Certainly the High Down burial cannot have taken
place much, if at all, after 500. With the buckle was an annular brooch with
V-shaped pin-slit, ball-stop knobs, and stamped decoration, and this is a late
fifth-century type.tv- The little semi-circular-headed bow-brooch is also an early
form. The Bifrons buckle came from a collection of ungrouped material from
what must have been some of the earliest graves in the cemetery, or indeed in
all Kent. This collection also produced the type I B loop (FIG. I5, k; type I B,
I3). Wherever they were made, the fact that one II c buckle was found in Kent
and the other in Sussex is interesting. The existence of a cultural link between
the two kingdoms is evident from the distribution of another type of Anglo-
Saxon metalwork, also of fifth-century date.>
There is another curious buckle from either an Anglo-Saxon or a Roman
grave at Frilford, Berks (FIG. I2), ' 04 which has not been classified, but which
deserves mention here because it, too, seems to have been in some way derived
from the buckles of type II A. It appears to be unique, but whether it is of Anglo-
Saxon make it is impossible to say.
We have two examples of the buckles of type III B; one comes from a
woman's grave at Long Wittenham, Berks. (FIG. 20, g; type III B, I), the other
from an unknown grave at Sarre, Thanet (FIG. 20, h ; type III D, 2). They are
examples of a continental type which has recently been discussed by Werner.rve
'02 Chadwick Hawkes (1961), p. 46 f.
r03 Loc. cit. in note 102.
"4 Proc, Soc. Antiq. 2 ser., III (1864-7),136. Roman and Anglo-Saxon graves were found close together
at this site. No grave-groups were recorded.
lOS Breuer and Roosens (1957), appendix by Werner, pp. 320-3, pls. vi-viii; Werner (1958), pp.
389 fr. and 409 ff., pis. lxxvi, lxxvii and lxxxi , 2.
He lists thirty-nine examples, a total to which our hitherto unpublished Sarre
piece adds one morc. The continental distribution is interesting. The valley of
the Meuse has produccd no less than twelve examples, the Rhine below Mainz
eight, north Germany and north Holland eight, west Belgium two, north-west
France two, and the Trier region two. There are also two outliers in east and
south Germany, well away from the main groups. The dense concentration of
these buckles on the lower Rhine, in Belgium and north Germany, is of particular
interest to us here, for dearly it is from one of these regions that our two finds
must originally have come.
The III B buckles are a one-piece development from the hinged late Roman
III A form, but unlike them, they never occur in recognizably fourth-century
graves. They are a fifth-century version of the late fourth-century style, and,
according to Werner, they were being made in the workshops of the Meuse
valley throughout the first half of the fifth century, and perhaps even later.
The buckles found in the cemeteries of Krefeld-Gellep and the Namur region
show a variety of form that can only be the result of a comparatively long
period of manufacture. It is unfortunate that these cemeteries have never
been properly published, for they are crucial to the understanding of events
in this region,in the fifth century. Such famous and important cemeteries as
Samson and Epravevf were excavated none too well during the last century,
and many of their grave associations were unrecorded. Consequently the chrono-
logical sequence of the burials has long been a subject of controversy. Fortunately
one cemetery of this kind in the Namur region has been excavated in this century
and the recent publication is an excellent study in which, for the first time, a
coherent picture has emerged.v The earliest grave at Haillot, which belongs to
the second quarter of the fifth century, was that of a man wearing a buckle of
type III B very similar to the one from Long Wittenham. It was a rich grave, and
the other furniture consisted of pottery vessels, a glass cone-beaker, a francisca
(throwing axe), six arrows, a comb, a strap-end, an open-work disc-headed rivet,
and three fourth-century Roman coins, too early to be of use for dating.tvs
This is a warrior grave, but it is one which shows marked differences from those
fourth-century graves discussed in the earlier sections of this paper. There is
none of the characteristic late fourth-century metalwork. Instead we have objects
like the francisca and the strap-end, which are unmistakably the kind of thing
generally associated with early Frankish graves. The remaining sixteen graves in
the cemetery give us a sequence of burials extending to about the year 500.
Thus, as a whole, the cemetery extends from some time in the reign of the
emperor Valentinian III (425-50) down to a time within the reign of the Frankish
king Clovis (482-SII). It bridges the eventful years when Belgium and the north
of France were undergoing the transition from Roman province into Frankish
kingdom. There is every reason to suppose that when they are properly examined
the cemeteries of Samson and Eprave, which produced similar material, will
,,6 Del Marmol (1860); Bequet (1891).
'7 Breuer and Roosens (1957).
lOS Ibid., pp. 214 ff., figs. 12-13, grave I I.
tell the same story. That story is one of gradual change. Some of the objects in
the cemetery, principally the pottery, the glass and the buckles of type III B,
are still fundamentally late Roman. But others, the weapons at first and later
the metalwork too, are slowly changing, until in the latest graves we find some
early examples of cloisonne jewellery which we can definitely call truly Frankish.
Who were these people who were buried at Haillot? The fact that nine out
of the small total of seventeen graves contained weapons has led Breuer and Roo-
sens to conclude that Haillot was the cemetery of a group of Frankish laeti. That
they really were laeti is questionable. The settlement cannot have been estab-
lished long before the date of the earliest grave, that is before the period 425-50,
and whereas we have no reason to suppose that new groups of laeti were being
settled at this time, there is some documentary evidence that Salian Franks,
who had migrated from the district of Toxandria they had occupied since 358,
were pushing south into modern Belgium in the years before 446. In this year a
large body of Franks under the leadership of their chieftain Chlodio had reached
the region of Tournai, Arras and Cambrai, where their farther advance was
checked for a while by the Roman general Aetius, who defeated them in battle at
Vicus Helena (probably Helesrnes between Tournai and Cambraij.rvv Aetius
apparently found it expedient to leave the Franks in possession of the Tournai
region and confirm their territorial rights and limits by a foedus. Tournai thus
became the Frankish capital, and the Salian federates served Aetius well in 451
in the battle against Attila. According to Verlinden, the Frankish migration that
culminated in their settlement of Belgium did not begin until shortly before
446, and then their advance under Chlodio was rapid and without pause for
colonization along the route. We have no historical information to help us on
this point, but, on the whole, such a course of events seems unlikely. During the
ten or fifteen years before 446 Aetius was occupied in preventing or controlling
the inroads of the Visigoths, Burgundians and Ripuarian Franks. The Salians
may have seized their opportunity of moving from the inhospitable lands of
Toxandria during this time, while Aetius's attention was directed elsewhere.
The northern frontiers of Gaul always seem to have been remote and somewhat
inaccessible and the gradual movement and land-taking of the Salians in this
region may not have concerned Aetius unduly-indeed, if the land had been
badly devastated in 407 he may have seen it as desirable, as a means ofrepopulating
the area. The settlement of tribes of barbarians within the empire was one of
his policies. Possibly the Franks only became a danger to him, and forced him
to take counter-measures, when they showed signs of wishing to cross the Somme.
In view of this, Frankish settlements in Belgium before 446 are perfectly possible,
and cemeteries such as Haillot, Samson and Eprave in the Namur region are
most credibly interpreted as belonging to early Salian Frankish settlers, who
arrived perhaps as early as 430-40, and who were later given the status offederates.
These people became the inheritors of the surviving Gallo-Roman culture of the
"9 The account of this battle appears in Sidonius Apollinaris, Carmina, v, 210 ff The date 446 is
that given by Verlinden (1946). Gregory of Tours, Historia Francorum, II, 9, does not mention the battle,
but records the Frankish capture of Cambrai about this time.
Meuse valley, acquiring metalwork and glassware from the Gallo-Roman factories,
but evolving, as time wcnt on, fashions which are recognizably the earliest manifes-
tations of a distinctive and new Frankish culture.
The buckles of type III B which interest us here are clearly a product of the
transition period. These were the types of objects worn by the first Frankish
settlers before they began to produce their own metalwork. The distribution
suggests that these buckles were popular. We find them, as we have seen, not
only in the areas of Salian occupation, but at Krefeld and Rhenen in Ripuarian
Frankish graves and also further to the north in the original German homelands.
During the time that the Salians had been moving down into Gaul, the Ripuarian
Franks had occupied the valley of the lower Rhine about Cologne. Their material
culture in the fifth century was almost identical with that of the Salians, and
we find not only the III B buckles but many other objects which can be exactly
paralleled in the cemeteries of the Namur region. The best known of the burial-
groups from Krefeld is that in the warrior grave, no 43.rr The dead man was
buried with a small buckle of type III B, a strap-end very like that from the
Haillot grave described above, a spear, arrows and a sword. This sword had a
scabbard with mouthpiece with ovolo design and a chape terminating with the
upper part of a human figure with upraised arms. The little buckle finds its
closest parallels at Samson,: and the scabbard chape is almost identical with
others from Samson and Eprave.>
All this is of great importance to us if we are properly to understand the
meaning of our finds from England. As we have seen, the large and simple
buckle of type III B from Long Wittenham is very similar to the one in grave I I
at Haillot. It also resembles others from Tournai in Belgium, Bonn on the Rhine, II 3
and Rahmstorf near Hamburg in north Germany.n- The little buckle from
Sarre is of much the same size as those from Krefeld, grave 43, and its parallels
from Samson quoted above, but in its decoration it is more like one from Hamme
in west Belgium, and another from Ben Ahin, near Namur.rc They could there-
fore have come to England from anyone of these regions, although the lower
Rhine or north Belgium seem the more likely sources.
We can now turn to the strap-end found in an Anglo-Saxon grave at North
Luffenham in Rutland (FIG. 24, a; type v B, I), which is exactly paralleled by
another from Rhenen, grave 846.II6 The Rhenen strap-end formed part of a
complex belt-set preserved in position on a portion of the leather belt. The whole
set is in chip-carved work and is reminiscent of the late fourth-century belt-suits.
But the design of the strap-end, with the triangular panel in the centre of the butt
and the outward-facing hare-like creatures in the corners, recalls even more
no Steeger (1937), pp. 182-B.

i r r Werner (1953), pp. 41-2, pl. vii, 4 and 5; Breuer and Roosens (1957), pl. VI, 5.
r r z Werner (1953), pl. vii, 3b and 6-8.

"3 Breuer and Roosens (1957), pl. viii, I and 5.
"4 Werner (1958), pI. lxxxi, 3.
"; Breuer and Roosens (1957), pls, viii, 2, and vii, I.
,,6 Roes (1953), pp. 32 If.. pl. ii; P. Glazema, Kunst en schoonheid uit de vroege Middeleeuwen (1955), pis.
xiv-xv.
strongly the nearly identical design on the plates of two highly decorated buckles
of typc III B found at Samson.u> and these belong to the first half of the fifth
century. The likelihood that the Rhenen belt-set was made in the same workshop is
almost certain because its four decorative ring-attachments also find their
closest parallel at Samson. I I 8 It is overwhelmingly probable that this workshop
was situated in the Namur region. Thc North Luffenham strap-end, the twin of
the Rhenen piece, was obviously made here too.
We can now see that the buckles of type III B from Sarre and Long Witten-
ham and the strap-end from North Luffenham are examples of the metalwork
that was being produced in the workshops of the Meuse valley during the first
half of the fifth century. These workshops supplied both the Salian Frankish settlers
of Belgium and the Ripuarian Franks of the lower Rhine valley. Certain of
their less exotic products, that is some of the simpler III B buckles, were also
traded to the Germans of the north. We must now consider from which of these
regions, and by what means, these three objects came to England. The fact that
they were all found in Anglo-Saxon graves, and that nothing comparable to
them has turned up on late Roman sites, makes it certain that they were con-
nected with the Germanic settlement. The early date of the parallel material
on the continent renders it unlikely that they were traded goods. We are thus
entitled to suggest that they were brought over by settlers as personal possessions.
It is just possible that the two buckles came to Kent and Berkshire with their
wearers from Saxon homelands in north Germany. But buckles of this kind are
more numerous in Belgium and the lower Rhine district, and it seems more likely
that our two came from somewhere here, as did the North Luffenham strap-end.
They were not the only objects to reach England from this region. In grave 42
in the Anglo-Saxon cemetery at Abingdon, also in Berkshire, was found a sword
whose scabbard was furnished with a mouth-piece and chapeov exactly like
those from graves at Samson and Eprave and grave 43 at Krefeld Gellep (see
above, p. 38). From Petersfinger in Wiltshire, grave 21, has come a second sword
with such a scabbard mouthpiece.>v Both were certainly made in the workshop
which produced the continental examples. According to Wemer.w' this workshop
was one of those in the N amur region which also manufactured the buckles of type
III B. The swords are therefore part of the same story: and the story appears to be
a Frankish one.
With the sword in the Petersfinger grave was an iron buckle, with its plate
decorated with the early Frankish large-cell glass cloisonne work,> and a battle-
axe with symmetrically expanded cutting edge. I23 The latter is of a type which is
occasionally found in early Frankish graves, and there is an exact parallel in
grave 13 at Haillot.i The Petersfinger grave was certainly that of a Frankish
"i Breuer and Roosens (1957), pl. vi, 2-3.
,,8 Roes (1953), pis. v, 2 and vii, 4.
"9 Leeds and Harden (1936), pp. 33, 59-60, pl. ix, b. c, d.
r z n Leeds and Shortt (1953), pp. 16-17,53-4, pl. i.
r zr Werner (1953) and (1956).
no Leeds and Shortt (1953), pp. 17-3,49, pl. vi.
"3 Ibid., pp. 18,54, pl. i i, fig. 6.
"4 Breuer and Roosens (1957), fig. 15. 3.
4
warrior or his descendant, and this is also true of the Abingdon grave. There
are one or two other objects in the Abingdon cemetery which seem to be of
Frankish origin, including another of the cloisonne buckle-plates.o- Frankish
settlers provide the most credible explanation of the arrival of our buckles and
strap-end.
This is not the place to examine all the available evidence for early Frankish
settlement in England. The few graves and objects mentioned here have been
selected simply because they are most relevant to the subject in hand. But
perhaps a few more words should be said before we leave the subject entirely.
The largest group of early Frankish metalwork from a single cemetery has
recently been found at Lyminge in Kent. I2 6 Here there was clear indication that
a group of Franks settled one site at some time around the year 500. Lyminge
is exceptional. Most of our early Frankish finds are more scattered, in a way that
suggests that the Frankish element was only a small part of the whole. At
Abingdon, for example, there is early material of Saxon origin. The term Anglo-
Saxon is perhaps misleading when applied to the settlements of southern England.
There is much to suggest that the population was very mixed and drawn from
many different regions of the western and northern continent. It has been
suggested before that many of the Germanic settlers may have started out from
the mouth of the Rhine. Here, during the unsettled years of the migrations,
peoples of many different racial groups must have come together, and perhaps
banded together for the crossing to England, with much mixing of cultures.
Whether they were Salians or Ripuarians, the Franks who crossed, bringing their
swords and buckles with them, most certainly came by this route, in the initial
stages at any rate. Those who brought buckles of type III B and the strap-end of
type v B must have been amongst the earliest arrivals.
CONCLUSIONS
In Britain, information about the events of the late fourth and early fifth
century has in the past been derived almost solely from literary sources, from
the coin evidence which is limited in scope, and from the scanty results of excava-
tions on the sites of buildings. So fragmentary is such information that many
regard the period in question as one of irremediable obscurity. But not all sources
have yet been tapped. A most remarkable omission on our part is failure to
locate and excavate late Roman cemeteries, and in particular the military ones
that must exist. In this there has been a complete disregard of work done on the
continent. In France, Belgium and Germany, much of what is known of this
period has come from cemeteries, and from the study of objects that most fre-
quently occur as grave-goods. Weapons and small bronzes, for example, have
been used to good effect, and have been seen to fit into a historical pattern-a
pattern of a Romano-Germanic army, of Germanic laeti or foederati settled behind
the frontiers, and oftheir gradual replacement during the fifth century by invading
115 Leeds and Harden (1936), p. 55, fig. 8.
116 Warhurst (1955), pls, ix and x.
bodies of free Germans. In this paper, I have applied similar methods to a series of
corresponding bronze types found in Britain, and have tried to show, despite
the scattered and ill-documented state of the material, that here, too, a historical
pattern can be discerned, mitigating something of the previous obscurity.
Thus, in the first place, we have in the eastern parts of Britain some authentic
continental military metalwork, though in small amount. This seems to have
been brought over for the first time in the reign of Valentinian I by troops
from northern Gaul under the command of Count Theodosius. The soldiers
were probably Germans, and they appear to have been based on the 'Saxon
Shore'. Secondly, we have in southern Britain, and more especially in the west and
the midlands, two main classes of British-made versions of this foreign metalwork,
which point by their distribution to a hitherto unsuspected military force, possibly
a sort of yeomanry, based on the towns. The long life of these buckles in the fifth
century suggests that the force was maintained, perhaps with further recruitment of
German mercenaries, long after the year 410, when the British were empowered
to take measures for their own defence. Lastly, a handful of further metalwork,
of the kind found on the continent in very early Frankish graves, testifies to the
presence of some Franks in the initial phase of Anglo-Saxon settlement which
began in 443.
The buckles and belt-fittings, which have been the subject of this paper,
are only a small proportion of the mass of late Roman metalwork which lies
unstudied in our museums. My purpose has been to show that such 'unconsidered
trifles', when brought together and considered systematically, can be made
to yield information that is new to us, bringing to the passage between Roman
Britain and Anglo-Saxon England light that we had not looked for.
CATALOGUE
TYPE I A
Bronze buckles with sub-oval or D-shaped loop, and straight hinge-bar cast in
one piece with the loop. The curved side is formed by the flattened bodies of a pair
of confronted dolphins, with a pellet between their open jaws. The treatment of these
creatures varies greatly from buckle to buckle: on some they are executed in clear
relief with prominent, upstanding crests; on others they are degenerate and stylized,
sometimes portrayed merely by scored lines on an otherwise plain loop. The majority
fall somewhere between the two extremes. Surface decoration on the loops takes the
form of punched dots, stamped ornament of some kind, or transverse grooving that
often makes a collar around the animals' necks. Where the buckle-plate survives it is
generally a long, narrow strip of sheet bronze, doubled over the hinge-bar of the loop
and riveted. Decoration, where it occurs, usually takes the form of stamped ornament
and engraved geometric designs.
From Roman Sites

I. Caerwent. Mon. Loop only (FIG, 13,a): width 2'5 crrr, A very Romano-British town of
Newport Museum degenerate example. Dolphins survive only as Venta Silurum. Circum-
7.3, 1906 notched crests at sides, pellet as faceted knob stances of find not known.
at top, Tongue missing, hinge-bar damaged.
b
,\
~
,
I
k
\
,,
-,
,
' .... _--------
FIG. 13
Buckles of type 1 A, nos.

(pp. 21 IT., 41 IT.). Scv]
1-14, 16
(a=I; b-c=5-6; d=2; e:f=7-8; g=II; h=S; i=IO,o )=9; k=16; 1=4; m=12; n=14; O=IS)
2. Catterick, Yorks. Complete example (FIG. 13, d) : width 3' 9 cm. Romano-British town of
In the possession of Dolphins stylized, 3 bands of moulding for jaws, Calaraetonium. 1952 exca-
Mr. E. J. W. IIild- cyes set high in knob-like protuberance at top vations. Building I, room
yard, by whose per- of hatched band. Tongue has rectangular I. Found on floor below
mission we publish it. plate decorated with incised diagonal cross. occupation-level and da-
Plate damaged; rectangular sheet of bronze ted by stratified coin to
IIildyard (1957), p. with 2 rivet holes. late 4th or early 5th
246, fig. 6, I 3. century.
3 Chichester, Sussex. Loop only (FIG. 13, h) : width 28 em. Dolphins Romano-British town of
with prominent, notched, crests, broad groove Noviomagus Regnensium.
We are indebted to between jaws, circlet eyes, and hatched collar Found during 1960 exca-
Mr.]ohn Holmes and around necks. Sides of loop decorated with vations at County Hall
Mr. Alec Down for crescent stamps. Tongue missing. in top of drainage ditch
permission to publish beside Roman street, in
it. latest Roman level dated
late 4th or early 5th
century.
4 Cirencester, Glos. Loop and plate (FIG. 13, I): length 7' 5 cm. Romano-British town of
Dolphins with low, notched crests, triple Corinium Dobunnorum.
Corinium Museum, moulding for jaws, circlet eyes, collared necks. Circumstances of find not
Cirencester. Sides of loop decorated with crescent stamps. known.
Tongue missing. Plate folded in half and
Cripps Call. no. 291. riveted twice. Borders of crescents, the outer
forming a continuous arcade with dots at
points of junction. Cluster of four crescents in
centre; at either end irregular, lightly cross-
hatched triangles.
5 Cirencester, Glos. Loop and tongue (FIG. 13, b): width 2' 2 cm. As no. 4.
Dolphins with prominent, notched crests.
British Museum, Heads simplified; eyes missing. Tongue of
London. simple type with back folded over hinge-bar.
Reg. no. FL. 522.
6. Holbury, West Dean, Loop only (FIG. 13, c): width 3 em. Dolphins Romano-British villa
Hants. without crests, suggested by circlet eyes, crossed Among surface finds near
Salisbury, South grooves for pellet, and stamped crescents at late Roman building, to-
Wilts. and Blackmore sides of loop. Tongue missing. gether with other buckles,
Museum nos. II A, 8 and III A, 3, and
late 4th-century coins.
. Wilts. Archaeol. Mag.,
( 187 2),33,276.
XIII
7 London (City oj), Loop only (FIG. 13, e): width 4 cm. Dolphins Romano-British town of
Lothbury. with low, notched crests, circlet eyes, and deep Londinium. Chance find.
London Museum. grooves marking outline of jaws. Loop
thickened at ends of hinge-bar and decorated
at either side by circlet. Tongue missing.
8. Silchester, Hants. Loop only (FIG. 13,j): width 4'4 cm. Stylized Romano-British town of
Reading Museum. dolphins with high, upstanding crests, hatched CallevaAtrebatum. Circum-
Boon (1959), p. 80, vertically, and slight traces of eyes. Slight stances of find not known.
pI. iii, A, 10. thickening ofloop at ends of hinge bar. Tongue
missing.
9 Silchester, Hants. Loop and plate (FIG. I s.i length 3' 8 em. As no. 8.
Reading Museum. Dolphins suggested by engraved lines and
Boon (1959), p. 80, notching for crests, punched dots for eyes.
pI. iii, A, 6. Tongue missing. Plate undecorated and
damaged.
10. Silchester, Hants. Complete example (FIG. 13, i): length 5' 7 cm. As no. 8.
Reading Museum. Dolphins suggested by engraved lines and
Boon (1959), p. 80, hatching for crests, collared necks. Plate
pI. iii, A,S. damaged; decorative grooving on fold, and
remains of border of repousse dots. Originally
2 rivets at end.
I
+ a
r
l
~
\::1 1
I
a FIG. 14
Anglo-Saxon grave-groups from (a) Blewburton Hill, Berks. (grave 2) and (b) Reading (grave 13)
containing buckles of type 1 A, nos. 15 and 17 respectively (pp. 26, 45). Sc. f, except as marked.
II. Upper Upham, Complete example (FIG. 13, g): length 9 em. Romano-British village.
Aldbourne, Wilts. This differs from others in the series in that the Chance find.
Ashmolean Museum, hinge-bar is a separate piece inserted through
Oxford. the pierced terminals of loop. Dolphins stylized,
with no crests, circlet eyes, and broad groove
for jaws. Second pair of stylized heads ncar
loop terminals. Tongue has stylized head at tip.
Plate unusually long and narrow, border of
cross-tooled grooves and lines of stamped
crescents, down centre single line of running
scroll ornament, formed by linked, punched
rings. Plate damaged at edges, folded double
and riveted at end.
12. Water Neioton, Hunts. Loop only (FIG. 13, m): damaged, width 3.6 em. Romano-British town of
Peterborough Dolphins with vestigial crests, punched dot eyes, Durobriuae. Circumstances
Museum. and large oval pellet between jaws. Hinge-bar of find not known.
and tongue missing.
From Anglo-Saxon Sites

13. Ash, Kent. Loop only (FIG. 13,0): width 3.8 em. Dolphins Anglo-Saxon cemetery.
Royal Museum, without crests, punched dot eyes, oval pellet Circumstances of find not
Canterbury. between jaws. Stylized zoomorphic heads at known.
C. Roach Smith, ends of hinge-bar. Scales suggested by furrowing
Collectanea Antigua, II, of sides of loop. Tongue missing.
pI. xxxvii, 8.
14 Beddingham Hill, Loop only (FIG. 13, n): width 3'4 cm. Stylized Barrow with Anglo-Saxon
Sussex. dolphins with high, upstanding crests. Slight burial(s). Associations: 3
British Museum, thickening ofloop at ends of hinge-bar. Tongue disc brooches, and 2
London. missing. buckles of simple, early
Reg. no. 1853, 4-12, type.
58.
V.C.H. Sussex, I, 337.
15 Blewburton Hill, Loop with iron tongue (FIG. 14, a): width 3 em. Anglo-Saxon cemetery,
Berks. Dolphins stylized with vestigial crests and grave 2, of woman. Asso-
Reading Museum. circlet eyes. Sides of loop decorated with ciations: amber beads,
Berks. Archaeol.]., LIII transverse grooves and stamped crescents. applied brooch.
(I95 2-3),5 I,fig.I9,5
16. Broadway, Wares. Loop and tongue (FIG. 13, k): width 2' 4 em. Anglo-Saxon cemetery,
Cook (1958), PI'. 62 Dolphins with high, upstanding crests, heads grave I, of woman. Asso-
and 73, pI. xi, a, stylized, open jaws and lozenge-shaped pellet. ciations: glass beads,
fig. 4, 2. bronze and silver wire
rings, pair of cast bronze
saucer-brooches with
whirligig of legs design.
Reading, Berks. Loop only (FIG. 14, b): width 2'9 em. Dolphins Anglo-Saxon cemetery,
Reading Museum. with low, notched crest, open jaws, oval pellet, grave 13. Associations:
]. Brit. Archaeol. punched dot eyes. Tongue missing. bronze tubular object,
Assoc., L (1894), 150 If. strap-end of early type,
bronze and iron rings,
Roman coin, and pot
with pedestal foot.
TYPE IE
Bronze buckles, basically similar to those of type I A, on which the crests of the
dolphins have been developed into outward-facing horse-heads. These generally have
clearly marked eyes, ears, and hatched manes. On some examples the dolphin heads
are still distinguishable, with eye, jaw and pellet, but on others the horse-heads have
become the main decorative feature and the dolphins have disappeared. Transitional
a
H
~g m
~~
II
~~
I
~~
\
~~
I
II
i~
I
,,' ~
~~, ~'
q
n p
FIG. 15
a-m. Buckles of type IB, nos. 1-4,6-13, 15 (pp. 21 ff., 45 ff.)
(a-g=I-8; h-i=II-I2;j=9; k=I3; I=I5; m=IO)
n-q. Buckle-plates of type 1 A/B, nos. 1-4 respectively (pp. 21 ff., 50). Sc.]
stages in the development occur also. The plates associated with this type are similar
to those of r A.
From Roman Sites
I. Alwalton, Hunts. Loop only (FIC. 15, a): width 3'3 cm. Fine Romano-British occupa-
Peterborough example, little worn. Dolphins wcll defined tion site. Circumstances
Museum. with triple moulding for jaws, dot eyes, trans- of find not known.
V.G.H. Hunts., I, verscly grooved collars, slight thickenings that
24 3-9, fig. 7. may reprcsent tails. Horse-heads clear, open
Proc. Soc. Antiq., XXlII, jaws, punchcd eyes, and notched manes. One
4 13-4 has pair of grooves on nose that may represent
(wrongly part of harness. Tongue missing.
provenienccd) .
2. Cirencester, Glos. Loop only (FIG. 15, b): width 26 cm. Worn Romano-British town of
Corinium Museum. example. Stylized dolphins with punched dot Corinium Dobunnorum. Cir-
Cirencester. eyes. Horse-heads clear, dot eyes, notched cumstances of find not
known.
Cripps Col!. no. 231. manes. Line of crescents along junction of necks
wi th loop. Broken fragment of bronze tongue.
3 Cirencester, Glos. Loop only (FIG. 15, c): width 2' I ern. Devolved As no. 2.
Corinium Museum, example on which dolphins have disappeared
Cirencester. and the horse-heads become two featureless
Cripps ColI. no. 234. lumps with circlet survivals of eyes. Sides ofloop
decorated with incised double zig-zag. Tongue
missing.
4 Clipsham, Rutland. Loop only (FIG. 15, d): width 2' 3 cm. Dolphins Romano-British villa.
Oakham School stylized, triple moulding for jaws, prominent Circumstances of find not
Museum. blobs for heads with circlet eyes, collared known.
Reg. no. D.J. 69, necks. Horse-heads well moulded, open jaws
A.L.T. and deep groove for manes. Running S-shaped
scroll decoration at sides ofloop. Hinge-bar and
tongue missing.
5 Dorchester, Oxon. Complete example (FIG. I, no. 16): length 3 From the late Roman
Ashmolean Museum, cm. Well-defined dolphin-heads on loop with cemetery outside walls of
Oxford. circlet eyes and mane, represented by line of Roman town of Dor-
Kirk and Leeds punched dots. Sides of loop cross-hatched. chester, grave 2, burial of
(1954), pp. 63 s., Long rectangular plate secured to loop by pro- woman, apparently of N.
fig. 27, no. 16, pI. iv, jecting tabs, doubled over hinge-bar and rivet- German culture. Asso-
B. ed. Secured to belt by two rivets at other end. ciations: very early type
Border of punched dots and running S-shaped of cruciform brooch, back-
scrolls. Central decoration of three engraved plate of applied brooch.
roundels made of hatched outer ring with For man's grave and
central dot, alternating with two cross-hatched other burials see Kirk and
lozenges. Stamped crescents occur at corners of Leeds (1954), pp. 63 fr.,
lozenges and sides of roundels. and above, pp. I ff.
6. Gestingihorpe Loop only (FIG. 15, e): width 26 em. Dolphins Romano-British settle-
(Hill Farm), stylized, 3 bands of moulding for jaws, triple ment with late 4th to
Halstead, Essex. collars at necks. Horse-heads flat and feature- early 5th century occu-
In possession of Mr. less except for notched name. Unfinished, with pation. Buckle found in
H. Cooper, by whose casting ridges on inside. Little sign of wear. destruction level of small
courtesy we publish masonry building of late
it. 4th century date. Close by
a ditch with late 4th to
5th century coins, and late
pottery. The site pro-
duced abundant evidence
of bronze-working, and,
being unfinished, the ob-
ject was perhaps made on
site.
7 Richborough, Kent. Loop only (FIG. 15, f): width 2' 25 ern. 'Saxon-Shore' fort of Rutu-
Richborough Castle Dolphins very stylized. Horse-heads clear, piae. From latest levels
Museum. circlet eyes, notched manes. Hinge-bar and but exact find-spot not
tongue missing. recorded.
b
J-~
~ --~ ---
c
d
FIG. 16
a. Buckle of type 1 B, no. 14 (p. 49). bog. Othcr early Anglo-Saxon finds from the same grave (no. 70 ) in
the Anglo-Saxon cemetery at Stratford-an-Avon, Warwicks, Sc. ]
8. Richborough, Kent. Loop only (FIG. 15, g): width 2'25 cm. 'Saxon-Shore' fort of
Richborough Castle Dolphins have disappeared. Horse-heads Rutupiae. Surface find Il1
Museum. marked off from loop by lines of billeting; SW. sector of interior of
Reg. no. 4477. circlet cycs and lines of triangular stamps for fort.
manes. Sides of loop decorated by tooling.
Tongue missing.
g. Spoonley Wood, Glos. Complete cxample (FIG. 15, j): width 2' 9 cm. Romano-British villa. Cir-
Sudeley Castle. Dolphins have disappeared. Horse-heads with cumstances of find not
Antig.]., XVII (1937), prominent ears, notched manes, dot eyes. Sides known.
447, fig. b."7 of loop decorated by groups of transverse
grooves. Simple tongue. Small rectangular
plate with parallel grooves at base. Tongue and
plate appear to have been replaced in antiquity.
10. Stanwick, Yorks. Complete example (FIG. 15, m): length 10 cm. Circumstances of find no t
British Museum, Dolphins stylized, circlet eyes and collared known, but probably con-
London. necks. Horse-heads separated from loop by line nected with Roman re-
Reg. no. 47, 2-8, 82. of stamped crescents; circlet eyes and tooled occupation of iron-age
Smith (1923), p. 40, manes. Sides of loop decorated by crescents. fortified site.
fig. 108. Tongue bronze. Long rectangular plate folded
Tonnochy & Hawkes over hinge-bar to a third its length and riveted.
(1931), pp. 12 3-8. Border of parallel, tooled grooves; central
fig. 2. decoration oflightly-engraved pair of peacocks,
back-turned, and with small tree between them.
Extremely stylized. Birds' heads crowned by
curious antennae, bodies divided into hatched
panels, long tails represented by engraved lines
and supplementary dots, rings, and crescents.
Edges and end of plate damaged.
II. Water Ncicton, Loop only (FIG. 15, h): width 28 cm. Worn Romano-British town of
Hunts. example; dolphins with well-moulded heads, Durobriuae. Circumstances
Peterborough circlet eyes. Horse-heads much worn. Tongue of find not known.
Museum. missing.
12. Water Neioton, Loop only (FIG. 15, i): width 3' I em. Dolphins As no. 11.
Hunts. flatly treated, circlet eyes. Horse-heads with
Peterborough notched manes and circlet eyes. Incised zig-zag
Museum decoration on loop. Hinge-bar and tongue
Reg. no. L. 340. missing.
55/4

13. Bijrons, Loop and tongue (FIG. 15, k): width 3 em. A Anglo-Saxon cemetery,
Patrixbourne, Kent. worn example. Dolphins with incised, slanting grave associations un-
Maidstone Museum. eyes. collared necks. Horse-heads divided from known. In Conyngham
Baldwin Brown loop by lines of punched dots, very worn. CoIl., amongst material
(1915), nr. pl. lxx, 6; from some of earliest
Leeds (1936), pl. xii. graves in cemetery. See
also p. 59, II C, I.
14 Straiford-on-Auon. Complete example (FIG. 16, a): width 3 cm. Anglo-Saxon cemetery,
Warwicks. Dolphin heads stylized, dot eyes. Horse-heads grave 70, of woman.
Stratford New Place moulded in relief, circlet eyes, grooved manes. Associations : bronze
MUSelUTI. Bronze tongue. Rectangular plate ohscured by strap-end with silver
textile remains. plating, beads, 3 bronze
needles on ring, and pair
of cruciform brooches
(Aberg, class II).
"7 There is some doubt whether this is the same buckle which was found at Spoonley Wood and was
published by J. H. Middleton in the Winchcombe and Sudeley Record, IV (1893), 39-48. If it is, as seems most
probable since it is in the Sudeley Castle collection, we must assume that Middleton's drawing was
inaccurate. The illustration (FIG. 15, j) is after Waterhouse in Antiq. ]., loco cit.
Unknown Provenience
15. Cambridge Loop only (FIG. 15, I): width 3 cm. Worn
University Museum example. Dolphins stylized. Horse-heads divi-
of Archaeology and ded from loop by lincs of punched dots, much
Ethnology. worn with punched eye dots. Lines of dots at
Reg. no. 83.53.2. sides of loop. Tongue and hinge-bar missing.
TYPE I AlB
Bronze plates which belong to this type, now detached from their buckles.
From Roman Sites
I. Cirencester, Glos. Plate only (FIG. 15, n): length 7' 7 ern. Long rect- Romano-British town of
Corinium Museum, angular plate, folded double, and riveted: Corinium Dobunnorum. Cir-
Cirencester. broken at fold. A fine example with border of cumstances of find not
Bathurst Call. running S-shaped scrolls. Central design of row known.
no. 376. of 4 roundels, each composed of 2 concentric
circles around a central rosette which consists of
four circlets and central ring. Roundels linked by
diagonal, incised crosses that form pairs of sub-
triangular panels filled with cross-hatching.
Circlets at either side of junction of triangles,
and in each corner of plate. Last roundel
masked by large rivet.
Duston, Northants, Plate only (nG. 15, 0): length 7.8 cm. Long Romano-British village.
Northampton rectangualr plate, folded double and riveted. Circumstances of find not
Central Museum. Fine example with double border of running known.
S-shaped scrolls. Central design of three
roundels composed of 2 concentric, hatched
circles around a four-petal marguerite, with
central ring and dotted petals. Similar
detached petals as fill-up in field. At two
ends, adjacent to outer roundels, are cross-
hatched triangles, and in spaces between
the roundels cross-hatched rectangles. Between
rectangles and roundels, on each side, groups of
3 spotted petals, and on remaining sides of
rectangles crescent stamps. Dome-headed rivet
pierces central roundel, and two others at end
of plate.
3 Popham, Plate only (FIG. 15, p): length 5' 4 cm. Rect- Romano-British building.
(College Wood), angular plate folded double and riveted at end. Surface find.
Hants. Border of cross-tooled grooves. Central decora-
British Museum, tion of three engraved cross-hatched lozenges
London. with pair of punched rings at each angle.
Reg. no. 1953, 7-9, I.
Bruce-Mitford (1954),
pp. 75-6, pI. xvii, b.
4 Silchester, Hants. Fragment of plate (FIG. 15, q): length 3' I cm. Romano-British town of
Reading Museum. Surviving portion probably half the original, Calleva Atrebatum. Circum-
Boon (1957), p. 80, and end shaped to fit buckle. Outer border, stances of find not known.
fig. 9. row of crescent stamps; inner, crosses in incised
Boon (1959), p. 80, frame. Central decoration of 2 (originally 3 or
pI. iv, A, II. 4) roundels composed of 2 concentric circles
with crescent stamps between, around a cross
with circlet stamps in the angles. Broken roun-
del, pierced by rivet-hole. Between surviving
roundels a pair of cross-hatched triangles with
circlets at angles, and, between them and the
the roundels. transverse rows of hatched ovals.
TYPE II A
Bronze buckles made up of separate loop, tongue, and plate, held together by a
bolt. The loop is formed by a pair of confronted dolphins, similar in type, decoration,
and in varying degree of stylization, to those on buckles of type I A, but with tails making
involuted terminals. The tongue, which rarely survives, has an expanded back and
down-curling projections at the sides that interlock with the loop terminals. At the
bottom of the tongue and the two loop terminals are pierced cylindrical attachments
that fit into others (normally 4 in number) at the top of the plate. They are usually
decoratively grooved, and are locked together by a long pin or bolt, which secures
loop, tongue and plate together. The plate itself is normally decorated by an openwork
arcade design usually consisting of three or four rectangular openings with pierced
circles at top and bottom, or of some variant of this. Sometimes there is some form of
edge decoration. Almost invariably there are rivets or rivet-holes at each corner of the
plate. Surface decoration on the plate, and sometimes on the tongue and loop too, takes
the form of punched dots or circlets.
From Roman Sites

I. Caerwent, Mon. Plate and loop fragment (FIG. 17, a): max. Romano-British town of
Newport Museum. length 4.6 em. Loop terminals still held in posi- Venia Silurum. Circum-
tion by hinge-bar. Tongue missing. Plate pierced stances of find not known.
by 2 round-headed arches. Ornamental excres-
cences at bottom and sides. Surface decoration
of circlets.
Caerwent, Mon. Fragment of loop (FIG. 17, b): max. width 4' 3 As no. I.
Newport Museum. em. Stylized dolphins with ears instead of crests,
oval depression for pellet, punched dot eyes.
One terminal broken. tongue missing.
3 Caistor-by-Norwich, Fragment ofloop (FIG. 17, c). Stylized dolphin Romano-British town of
Norfolk. with no crest, incised line marking jaw. circlet Venia Ieenorum. 1930 exca-
This, and no. 4, are eye, and triple-collared neck. vations, Building 4, des-
in the possession of troyed by fire. Dismem-
Professor Donald At- bered and unburied human
kinson, by whose per- rem-iins associated with
mission we publish Theodosian coins suggest
them. destruction after 400 by
Saxon raiders. This and
no. 4 were on the latest
occupation level in the
room next to that con-
taining the corpses.
]. Rom. Stud., XXI (1931),
232-3; Arehaeol. .T, CVI
(1949).65,
Caistor-by-Nonoich, Plate only (FIG. 17. g): length 4' 5 em. Pierced As no. 3.
Norfolk. by eight openwork arches whose rounded
heads are partly divided off by ornamental
projections from the intervening pillars-in
four opposed pairs with the rounded heads at
top and bottom. Decorative notching at bottom
and sides. Rivet-holes in corners. Hinge com-
ponents grooved. One, broken in antiquity,
was replaced by a patch, grooved to match.
Chedioorth, Loop only (FIG. 17, d): width 3' 3 ern, Clear Romano-British villa.
Yanworth, Glos. dolphins with open jaws, lozenge-shaped pellet. Surface find to east of
Chedworth circlet eyes. and notched manes. Tongue buildings in outer court
Villa Museum. missing. of villa.
Reg. no. 83. B.
6. Colchester, Essex. Complete example \ FIG. 17, e): length 6 em. Romano- British town of
Colchester and Essex Dolphins well-executed, with prominent notch- Camulodunum. Circum-
Museum. ed jaws. round pellet, circlet eyes. Tongue has stances of find not known.
stylized animal head at tip and at ends of curled
side-pieces. Large circlet at base. Hinge com-
ponents grooved. Plate pierced by 4 rectangles
with circles at top and bottom. Bands of

diagonal hatching at top and bottom of plate,
and ornamental excrescences at base. Rivet-holes
in each corner. Surface decoration of punched
dots and circlets on all parts of buckle and
plate.
7 Duston, Northants. Fragment of loop (FIG. I 7,f): max. width 3.8 Romano-British village.
Northampton em. Dolphins stylized with cars in place of Circumstance of find not
Central Museum. crests, semi-circular openings under jaws, known.
leaving square pellet, punched dot eyes. One
terminal broken, stylized head with circlet eye
at tip of other. Tongue missing.
8. Hotbury, West Dean, Plate only (FIG. 17, h): width 3 cm. Crude Romano-British villa. A-
Hants, example with triple arcade in which the pierced mong surface finds ncar
Salisbury, South rectangles and roundels have been run to- late Roman building to-
Wilts. and Blackmore gether. Irregular notching at base. Rivet-holes gether with nos. I A, 6 and
Museum. at corners, iron incrustation. Hinge attach- III A, 3 and late 4th cen-
ments broken. tury coins. Wilts. Archaeol,
Mag., XIII (1872), 33, 276.
9 Leicester. Fragment ofloop (FIG. 17, i): max. width 5 ern. Romano-British town of
Leicester Museum. Well executed dolphin with hatched crest, open RataeCoritanorum. Circum-
Reg. no. 3 I. 195 I. jaws, round pellet, circlet eye. One half of loop stances of find not known.
broken off, tongue missing.
10. Leicester. Plate only (FIG. 17, I): length 4'7 cm. Triple Romano-British town of
(.Jewry Wall), arcade similar to no. 6. Transverse grooves at Ratae Coritanorum. From
Leicester Museum. base and ornamental excrescences at base and disturbed levels which
Reg. no . .l.W. 845. sides. Rivet-holes at corners. Surface decoration produced other late Ro-
Kenyon (1948), p. of circlets. Hinge components grooved. man finds, and the strap-
256, fig. 84, no. 5. end, type v A, 2.
II. Lullingstone, Kent. Loop only (FIG. I7,j): width 28 em. Dolphins Romano-British villa.
In the possession of stylized with prominent crests, lozenge-shaped From the latest occupa-
Lt.-Col. G. W. pellet, oval eyes, and collared necks. One tional debris sealing Bath
Meates, by whose terminal broken, tongue missing. block furnace which went
permission we pub- out of use c. 380. This
lish it. level contained two coins
of Valens (364-378).
12. Lydney Park, Glos. Loop and tongue (FIG. 17, k): width 3 em. Late Romano-British tem-
Wheeler (1932), p. Dolphins with ear-like crests. Back of tongue ple. From original
86, fig. 101. pierced. Hinge-bar and two hinge-components make-up of cella floor,
from plate surviving. dated (by coins) after
36 4-7.
Lydney Park, Glos. Plate with fragments of loop and tongue (FIG. Late Romano-Bri tish
Wheeler (1932), pl. 18, a): length of plate 4' 9 em. Plate with un- temple. 1805 excavations.
xxviii, 130. usual openwork decoration, consisting of a Exact find-spot not
rough central roundel flanked by irregular known.
rectangles. At the bottom three rectangles, at
the top curvilinear patterns. Ornamental pro-
jections at bottom and sides, notching on the
hinge-components. Rivet-holes in the corners.
North Wraxall, Wilts. Loop only (FIG. 18, b) : width 4' 4 em. Dolphins Romano-British villa.
Devizes Museum. with hatched crests, open jaws, rectangular Exact find-spot not
Reg. no. 399. pellet, dot eyes. Stylized head at end of terminal. known, but certainly from
One terminal and tongue missing. a late level. Villa came to
violent end in late 4th
century (corpses and
broken masonry in well).
Wilts. Archaeol. Mag., VII
(1862),59-75
//-' .._-"''''',
/
/ " ----- ......... ,", '
//
/ I \
I I
I I
I I
I I
I I
-, \
" ---'d
/
f I
,// /
/
I
I /
I I
I I
\
;"':~',
\.
, \
' .... _- -_ .... / ':,
-,
,
. . . _--. r r :
I
I~-
_I
k
FIG. 17
Buckles of type II A, (pp. 21 fl., 51 If.). Scv]

nos. 1-12
(a-c=I-3; d-j=5-7; g=4; h-i=8-9;j-k=cIl-I2; I=IO)
54
~
/
/
I
/
/
"1
- ,-, \
I \
\ \
\ I /
\ ' ..... - "" I
\ /
-, /
..... .., ,.,.
I I
b LJ
FIG. IS If 51 ff.). Sc. i

. f~~ 13
-I ,21- 2 4 (pp .. 21 0"_0 )
Buckles of typ e II A, nos. _ I 8-
9 ._.
2 I ; g-h l-k=~2 -4
(a-e= I,-I?, 19;
OldSorum, Wilts. Plate only (FIG. 18, c): length 3' 7 em. Eccentric Romano-British settle-
Salisbury, South example with only two attachments for loop, ment of Soruiodunum. Cir-
Wilts. and Blackmore and long tongue-like projection at bottom. cumstances of find not
Museum. Threc key-hale-shaped openings with circle known.
Reg. no. 05/A8. above.
Archaeol.]., CIV
(1947), 136 (pub-
lisbcd as medieval).
16. St. Albans, Herts. Loop only (FIG. 18, d) : width 28 em. Dolphins Romano-British town of
Verulamium with small crests, open jaws, lozenge-shaped Verulamium. From late
Museum. pellet, circlet eyes. Sides of loop have punched 4th century filling of the
Archaeologia, LXXXIV dot decoration. Tongue missing. Theatre.
(1934), 259, fig. 12,
16.
17 Sea Mills, Bristol. Plate only (FIG. 18, e): length 4' 7 ern. Origin- Romano-British port.
Bristol City Museum. ally of doubled sheet-bronze, the folded end cut Chance find.
We are indebted to to form the four grooved hinge-components:
Mr. G. C. Boon for iron bolt in position still retaining portion of
drawing our a tten- bronze tongue. Openwork decoration consists of
tion to this piece. variant of triple arcade: rectangles with arches
at either end partially divided off. Incised lines
and cross-tooling around each opening. Back
part subsequently broken, and repaired by
insertion of sheet of bronze between the two
halves of plate and secured by four rivets. Very
worn.
18. Silchester, Hants. Loop only (FIG. 18, g) : width 3' 2 ern. Dolphins Romano-British town of
Reading Museum. with low notched crests, open jaws, circlet on Calleva Atrebatum. Circum-
Boon (1959), p. 80, pellet, circlet eyes. Tongue missing. stances of find not known.
pI. iii, A, 9.
19 Water Newton, Hunts. Loop only (FIG. 18, h): width 2' 4 cm. Dolphins Romano-British town of
British Museum, without crests, circlet eyes, collared necks. One Durobrivae. Circumstances
London. terminal broken, tongue missing. of find not known.
Reg. no. 1882,6-21,
69

20. Douer, Kent. Loop with iron tongue. Dolphins very stylized, Anglo-Saxon cemetery,
(Buckland estate). surviving only as hatched crests at sides, and grave 48, of woman.
British Museum, rectangular protuberances for heads. Tongue Associations: saucer-
London. may perhaps be a later replacement. brooch with five-spiral
We are indebted to decoration, button-
Miss V. 1. Evison for brooch, beads, and knife.
permission to include
this in advance of her
forthcoming excava-
tion report.
21. Mitcham, Surrcy. Loop only (FIG. 18,!): width 4'4 ern. Dolphins Anglo-Saxon cemetery,
Cambridge without crests, open jaws, lozenge-shaped grave 38. Associations:
U ni versity !VIuseum pellet with hollow centre, circlet eyes, collar of bronze finger-ring.
of Archaeology and five rows of slashes round necks. Similar slashes
Ethnology. on terminals and stylized, open-jawed heads.
Morris (1959), pp. Tongue missing.
102-3, pI. xiii, 38.
:22. Sarre, Thanet, Kcnt. Plate only (FIG. 18, i): length 4'3 ern. Three Anglo-Saxon cemetery,
Maidstone Museum. sub-rectangular openings with heart-shaped grave 94. Associations:
Archaeol. Caniiana, VI piercings at either end. Decorative notching at silver-gilt circular brooch
( 1864),174. bottom. One side broken. with central garnet and
nielloed border, fragments
of gold braid, miscel-
laneous objects, iron knife
and key, beads of amber
and glass.
5
I Ii
I
I I 'I
'_J II
I,
~J
c
FIG. 19
a-b. Buckles of type II B, nos. 1-2; cod. Buckles of type II C, nos. 1-2 (pp. 34, 57 f.). Sc. }
,,,
a b
FIG. 19 bis
a. Fragment of buckle-plate, type II A, no. 25; b. Buckle, type III B, no. 3 (postscript, p. 68). Sc. t
Chance Finds
23. Lakenheath, Suffolk. Loop only (FIG. IS,}): width 56 cm. Dolphins Circumstance of find un-
Cambridge without crests, nat jaws with groove belween, known.
University Museum circlet eyes, collared necks. Line of punched
of Archaeology and dols along sides of loop. Terminals in form of
Ethnology. small, open-jawed heads, with notched crests
Baldwin Brown and dol eyes. Tongue missing.
(1915), IV, pI. cliv, 2.
Saltersford, Loop only (FIG. 18, k): max. width 5 em. An Found on bank above
Grantham, Lines, eccentric example. Confronted heads without filter beds of Grantham
Hull Museum crests, with open jaws, lozenge-shaped pellet, waterworks.
Publications, no. circlet eyes and pricked ears. Sides and base of
lOS, Quarterly Record loop decorated with incised zig-zag decoration.
of Additions, LiV On curving outside edge of loop, at each side, a
(1916), 1-4, fig. I. perching bird, apparently of pigeon family,
Antiquary, June 1914, cast in one with loop. Base of loop straight,
PP 2 07- 8. pierced lobes at outside angle. (Apparently to
Archaeol. ]., XCI attach buckle to belt.) Pierced attachments of
(1934), 149, pI. usual type on under side. Terminals in form of
xxviii, B full-face human masks. These have vertical
hatching for hair, wide-set oval eyes, wedge-
shaped noses, and straight mouths. Tongue
missing. Unique, but apparently a variant of
this general type.
TYPE II B
Bronze buckles, basically similar to those of type II A, but on which the openwork
plate is cast in one with the loop, resulting in the loss of the in curved terminals.
From Roman Site

I. Richborough, Kent. Loop and plate (FIG. 19, a): width 28 em. 'Saxon-Shore' fort of
Richborough Stylized dolphins with upstanding crests, jaws Rutupiae. Surface find.
Castle Museum. indicated by single slashes, punched dot eyes.
Top of plate pierced by 3 openwork roundels, 2
rivet-holes, and the tops of two originally
rectangular openings. Bottom of plate and
tongue missing.
From Anglo-Saxon Site

2. Sleaford, Lines, Loop and plate (FIG. 19, b): length 5 em. Anglo-Saxon cemetery.
Museum of the Dolphins well-executed with upstanding, Associations not known.
Spalding notched crests, circlet eyes, openwork roundel
Gentlemen's below jaws, and lozenge-shaped pellet. Plate
Society. with openwork design of four triangles around
a central lozenge, notched edges, grooves at
sides, and surface decoration of circlet stamps.
Rivet-holes at corners and square hole for
tongue, now missing.
TYPE II C
Bronze buckles which appear to have been derived ultimately from the involuted
II A type, but which show considerable differences in construction and ornament.
The interlocking hinge-components are missing, and instead there are two hinge-bars,
around one of which was folded the back of the tongue, around the other the end
of the belt. There is no evidence to suggest that these buckles ever had plates. The con-
fronted dolphins in the middle of the II A loops are here missing, and instead we find
a pair of confronted animal-heads at either side. These spring from the bottom of the
loop and the ends of the outer hinge-bar. The tongue survives on one example only, and
is attached to the inner hinge-bar. It has projecting wings at the sides-a simplified
f
FIG. 20
a-f. Buckles of type nos. 1,3-7 (pp. IO ff., 59 f.). (a=I; b=4; c= 3; d=6; e= 5; 1=7)
III A,
g-h. Buckles of type III n, nos. 1-2 (pp. 34 ff., 60). Sc.}
version of the II A form. The edges of the loop are decorated with panels of billeting,
and there are simplified scrolls on the involuted terminals and on the tongue.
I. Bifrons, Loop only (FIG. 19, c): width 3'9 em. Examplc Anglo-Saxon cemetery,
Patrixbournc, with well-executed animal-heads, with ears and grave associations un-
Kent. open jaws. Rather worn. known. In Conyngham
Maidstone Museum. coIl., material from some
Baldwin Brown of earliest graves in ceme-
(1915), III, pI. lxx, 9; tery. See also p. 49,
Leeds (1936), pl. xii, 1 B, 13.
2. High Down, Sussex. Loop and tongue (FIG. 19, d): width 3.8 em. Anglo-Saxon cemetery,
Worthing Museum. Example with stylized animal heads. Rather grave 26, of woman.
Reg. no. 3436. worn. Associations: pair of
Archaeologia, LIV bronze bow-brooches,
(1895),377, pI. xxvii, annular brooch, and iron
5 ring.
TYPE III A
Bronze buckles with semi-circular loops terminating in open-jawed animal heads
confronted across the hinge-bars. The loop may be plain or decorated with chip-carved
work, and incised or stamped designs. The heads vary in degree of naturalism as on
the other buckle types. but they are intended to represent some sort of quadruped's
head, perhaps lion or leopard. The plates may be cast or cut from sheet metal, folded
double over the hinge-bar of the loops, and are semi-circular or a broad rectangle in
form. Most are plain except for some kind of edge decoration.
From Roman Sites
I. Bradwell, Essex. Loop only (FIG. 20, a): width 6 5 cm. Loop 'Saxon-Shore' fort of
Colchester and plain, heads stylized but clear. Tongue missing. Othona. Unstratified find.
Essex Museum. Site has yielded high pro-
portion oflate 4th century
coins and a few 5th-6th
century Anglo-Saxon ob-
jects. Cf. Report of the
Colchester and Essex Museum
April I947-Mareh 1948,
pp. 25-28, pl. ix, 1-2.
2. Dorchester, axon. Complete example (FIG. I, no. I): length 7' 2 Late Roman cemetery
(Dyke Hills). cm. Loop plain and worn, heads stylized and outside walls of Roman
Ashmolean Museum, indistinct, circlet eyes. Tongue curved at tip, town of Dorchester, grave
Oxford. bevelled edges and transverse grooves at back. I, burial of man. Associa-
Kirk and Leeds Plate semi-circular with border of ovolo decora- tions: strap-end (see p.
(1954), pp. 63 ff., fig. tion, 3 rivets; repaired in antiquity with patch. 63, no. v A, I), 3 disc
27, I, pl. iv, A. attachments (see p. 66,
nos. VI, 2-4), metal plates
from sporran (see p. 66,
no. VII, I), bone toggle
and metal plates.
3. Holbury, West Dean, Complete example (FIG. 20, c): width 4' 2 em. Romano-British villa.
Wilts. Loop has grooved surface and notched edge, Among surface finds near
Salisbury, South heads stylized. Plate rectangular, plain but for late Roman building, to-
Wilts. and Blackmore serrated edge, 2 rivets. gether with other buckles,
Museum. I A, 6 and II A, 8, pp. 43,
52, and late 4th century
coins. Wilts. Archaeol,
Mag., XIII (1872),33,276.
4. Richborough, Kent. Loop only (FIG. 20, b): width 6 2 em. Loop 'Saxon-Shore' fort of
Richborough plain, heads stylized. Tongue missing. Rutupiae. From middle
Castle Museum. layer of inner stone-fort
Bushe-Fox (1949), pl. ditch, in association with
xxxii, no. 67. coins up to house of
Theodosius.
5 Richborough, Kent. Loop only (FIG. 20, e): width 5' 9 cm. Loop As no. 4.
Richborough plain. Heads with clearly defined cars and
Castle Museum. eyes and punched dots along jaws. Tongue
Bushc-Fox (1949),1'1. missing.
xxxii, no. 68.
Chance Finds
6. Icklingham, Suffolk. Complete example (FIG. 20, d); length 7 em. Exact find-spot not
Moyses Hall Well-executed heads with circlet eyes. Edge known. Perhaps from the
Museum, Bury St. of loop decorated by band of diagonal tooling, Roman villa or the Anglo-
Edmunds. uppcr surface by arcade design of contiguous Saxon cemetery on
crescentic stamps with dots at points of Mitchell's Hill.
junction. Tongue waisted, with curved tip and
rectangular panel on back decorated with 3
transverse grooves. Plate folded double and
roughly cut, plain, with 3 rivets.
7 Oxford. Loop and tongue (FIG. 20,1) : width 6 ern. Loop Circumstances of find
Ashmolean Museum, worn, crouching animals in chip-carved work unknown.
Oxford. still visible around edges, circlet stamps on
uppcr surface, well-executed heads. Tongue
waisted with curved tip and rectangular panel
on back.
8. Unknown Complete example (PL. I, A); width I I' 7 ern. Apparently found with
provenience, Kent. A finc example in 'chip-carved' work. The loop the attachment-plate (I"
British Museum, has clear animal-head terminals, a border of 68, VII,S) and disc
London. triangles, and on the upper surface a panel of attachments (I'. 66, VI,
Reg. no. 1942, 10-7, geometric motives like single arms of swastikas. 12- 13) .
5 There are two tongues with stylized animal-
heads at the tip, which are linked across the
back by a transverse bar decorated with pro-
jecting animal-heads; a cross-within-triangle
motive at the base of each tongue, and a
stamped circlet between. The plate has a milled
edge, and a central panel of swastikas and
triangles, bordered by a line of annulets. 2
rivets.
TYPE III B
Bronze buckles of similar form to type III A, but with plates cast in one piece with
the loops.

J. Long Wittenham, Complete example (FIG. 20, g): width 7' 2 ern. Anglo-Saxon cemetery,
Berks. Loop plain with well-executed animal heads, grave 57, of woman.
British Museum, eyes oval. Rectangular plate with billeted Associations: pottery jar
London. border and vertical grooving at top. Two rivets. with bossed and stamped
Reg. no. 47, 2-8, 82. decoration.
Archaeologia, XXXVJJJ
(1860) 333,pl.xix, 10.
2. Sarre, Thanet, Kent. Complete miruature example (FIG. 20, h): Anglo-Saxon cemetery.
Maidstone Museum. width 3 cm. Well-executed animal heads with Grave not known.
circlet eyes. Plate rectangular, with notched
edges and two rivets. Circlet stamps on loop
and upper part of plate.
TYPE IV A
Bronze buckles with loops basically of type III A, but attached inside one- or two-
piece rectangular plates, which are decorated with 'chip-carved' ornament and have one
tubular side decorated with ribbing.
FIG. 21
Buckle, type IV A, no. I, from Richborough, Kent (p. 62). Sc.
FIG. 22
Buckle, type IV B, no. I, from Catterick, Yorks. (p. 62). Sc.J
From Roman Sites

I. Richborough, Kent. Completc except for tongue (FIG. 2 r }: max. 'Saxon-Shore' fort of
Liverpool Museum. length 5' 4 cm. Loop with well-executed animal Rutupiae. Unstratified.
Reg. no. M.6g62. heads, cast in fixed position inside the one-piece
Roach Smith (1850), plate. This is decorated at onc cnd with a pair
pl. v, fig. 2. of sty liz cd confronted animals with back-turned
heads. The remaining surfaces are occupied by
crude geometric 'chip-carving'. The loop and
tubular end are both ornamented with an
incised zig-zag pattern with punched rings in
the angles. Rivet-holes in corners of plate.
Snodland, Kent. Broken half of one-piece(?) plate (PL. II B): Found near a Roman
British Museum, max. width 7 em. It contains the oval aperture structure on site of Snod-
London. for the missing buckle, and is decorated in semi- land gas-works.
Reg. no. 1928, classical style with tooled roundels containing
5-1 I, I. engraved human busts on a background of
Antiq. J., VII (1927), niello. Between the roundels are 'chip-carved'
522 ff. scrolls. The plate is bordered by a line of bronze
squares, in reserve against a nielloed field, each
decorated by a circle of punched dots. Rivet-
holes in corners.
Chance Finds
3. Smithfield, London. Complete example (PL. II A): max. length 9' 5 Circumstances of find
British Museum, cm. Loop, with stylized animal heads, and unknown.
London. plain tongue, both swivel on hinge-attachments.
Reg. no. 1856, 7-1, One-piece plate enclosed in beaded frame and
147 0. decorated by arrangements of coarse 'chip-
Smith (1923), fig. 4. carved' scrolls, paired or enclosed in beaded
triangles and lozenges. Rivet-holes in corners.
4a. Unknown Complete example, with two-piece plate (PL.
provenience, but III A): max. length 8 7 cm. Loop, with well-
perhaps from Kent. executed animal-heads, and tongue, both swivel
Liverpool Museum. on hinge-attachments. 'Chip-carved' decora-
Reg. no. M.6922. tion of running S-shaped scrolls on one half of
plate, and scroll-rosettes within roundels on the
other. Rivet-holes in corners.
4b-d. As no. 4a. Set of three belt-plates (PL. III, B-D), two tri-
Liverpool Museum. angular and one rectangular, which accom-
Reg. nos. M.6916- panied the buckle. Decoration of scroll and
69 2 1. geometric patterns. Rivet-holes in corners.
TYPE IV B
Bronze loop basically of type III A, set in openwork frame.
From Roman Site
I. Catterick, Yorks. Complete example (FIG. 22): width 9.6 em. Romano-British town a
Recently acquired by Semi-circular loop with open-jawed heads, con- Cataractonium. 1952 excava-
the British Museum, fronted at ends of hinge-bar, which have large tions. Building I, Room I.
London. circlet eyes, ears, and hatched manes. Tongue On floor below occupa-
Hildyard (1957), pp. with stylized animal head at tip. Loop enclosed tion level, dated (by
243 f., fig. 6, 12, on three sides in an ornamental frame or plate stratified coin) late 4th
pl. ix. partially separated from it by openwork panels. or 5th century. Same
This plate consists of a roughly straight bar, level as buckle no. I A, 2,
attached to centre of curved side of loop, which P43
is decorated by ornamental notching in chip-
carved style on its outer edge and by stamped
circlets, large and small, on upper surface. Two
openwork roundels on either side of where tip of
tongue rests served as means of attachment to
belt. At either side of loop, and joined to the
bar of the plate, are realistic dolphins with
open-jawed heads and circlet eyes, flat bellies,
tapering backs, out-curved tails, and surface
decoration of punched dots.
TYPE V A
Strap-ends with pear-shaped front and splayed, square-ended, split butt, into which
the belt end was inserted and then secured by one or more rivets. The ornament varies.
Some are decorated in 'chip-carved' work with scrolls and marginal animals, others by
stamped decoration. Many are plain.
From Roman Sites
I. Dorchester, axon. (FIG. I, no. 11): length 5' I em. Butt slightly Late Roman burials,
Ashmolean Museum, bevelled, with 2 rivets. Pair of incised grooves grave 1, of man. Associa-
Oxford. divide off pear-shaped front. tions: zoomorphic buckle
Kirk & Leeds (1954), (p. 59, no. III A, 2), disc-
p. 66, pl. iv, A, fig. attachments (p. 66, no.
27, 11. VI, 2-4), metal plates (p.
66, no. VII, I), bone
toggle, misc. bronze
plates.
2. Leicester (FIG. 23, a): length 7 em. Two rivets. Incised Romano-British town of
(Jewry Wall site). groove marks off pear-shaped front which is Ratae Coritanorum. From
Leicester Museum. decorated in chip-carved work. Inside a double disturbed levels, which
Reg. no.J.W. 84.15. grooved frame is a design of seven spiral scrolls, prod uced other late
Kenyon (1948), p. 3 in triangular cluster at top, and 4, downward Roman finds and the
255, fig. 84, 15 curling, in pairs at bottom. On lower edges a buckles (p. 52, nos. II A,
pair of stylized, crouching, marginal animals. 9 and 10).
3 A1ilton-next- (FIG. 2, a): length I I' 5 em., but front damaged. Circumstances of find not
Sittingbourne, Kent. Two rivets on butt. Plain except for transverse known. Probably from
Maidstone Museum. moulding between butt and front. late Roman burial. Asso-
ciated with 2 disc-attach-
ments (p. 66, no. VI, 5-6)
and 2 attachment-plates
(p. 68, nos. VII, 2-3).
4 Richborough, Kent. (FIG. 23, e): length 3' 9 cm., tip broken. Single 'Saxon-Shore' fort of
Richborough rivet in butt, which has notched edges and Rutupiae. From disturbed
Castle Museum. surface decoration of roughly scored lines. On levels.
Reg. no. 335. front, a 'chip-carved' design ofa single S-shaped
Bushe Fox (1928), scroll enclosed in a tooled frame.
pI. xxi, 48.
5 Richborough, Kent. (FIG. 23, d) : butt only, width 2' 5 em. 2 rivets. 'Saxon-Shore' fort of
Richborough Decorated in centre with crudely 'chip-carved' Rutupiae. From inner ditch
Castle Museum. seven-pointed star inside a grooved and tooled of stone fort at depth of
Reg. no. 1631. frame. Edges notched. 5 ft. in filling; layer con-
Bushe Fox (1949), taining Theodosian coins.
pl. liii, 207.
6. Richborough, Kent. (FIG. 23, b): butt missmg, length 4' 5 em. 'Saxon-Shore' fort of
Richborough Decorated down centre with 3 large circlet Ruiupiae. Details of find
Castle Museum. stamps, border and one transverse band of not known.
Reg. no. 4647. pellet within triangle stamps, and three more of
these stamps in line up from tip.

7. Cassington, axon. (FIG. 23, f): length 6 em. Two rivet-holes in Anglo-Saxon cemetery,
Purwell Farm site. butt. Plain except for incised grooves at junc- grave 2, of woman. Re-
Ashmolean Museum, tion of butt and front. used as charm or keep-
Oxford. sake. Threaded on wire
Oxoniensia, VII (1942), ring with 2 bone discs and
65-6, fig. 15, pl. v, c. bronze tag, and placed in
purse that also contained
2 boar tusks and a frag-
ment of glass. Other
associations : pair of
saucer brooches with 5-
spiral decoration, amber
and crystal beads, and a
knife.
\ --------/
\ .- " I
\ 1/ I .. ' I
\ I
\ I
\ I
b II
,
, e
,,
I
,
I
" 9
FIG. 23 6 If) SC.
If., 3 .. 11
Strap-ends a fty~ e -8'
V A, nos. 2, 4- ~ O', h=9)
10 (
pp. . g-I
d=5; e=4;!=7,
10
(a=2; b=6, c- ,
8. Croydon, Surrey. (FIG. 23, c): length 7 em. Tip broken. 2 rivets. Circumstances of find not
Croydon Public Butt divided from front by transverse ridge. known. Published with
Library. All edges decorated by row of circlets inside unassociated finds from
Proc, Soc. Antiq., xv panel of billeting. Immediatcly below butt an Anglo-Saxon graves.
(1895), 328 ff. engraved pendant triangle, divided diagonally
V.C.H. Surrey, I, pI. into a series of lozenges filled with punched
opp. p. 257, no. 7 dots.
Chance Finds
9. Icklingham, Suffolk. (FIG. 23, h): length 9.6 em. 2 rivets. Tooled Circumstances of find un-
Cambridge grooves divide butt from shoe-shaped end known.
University Museum which is decorated at lower edges by worn
of Archaeology and crouching animals with openwork roundels
Ethnology. below neck. Four large stamped circlets in
no. 32. 346. middle surrounded by dense surface decoration
of small circlets. Around the edges, and around
each large circlet, are rows of crescent-shaped
stamps. Below grooves at top are five tooled,
pendent triangles.
10. Ixworth, Suffolk. (FIG. 23, g): tip broken, length 4 em. Single As no. 9.
Ashmolean Museum, rivet at top of butt, which has notched upper
Oxford. edge and a fill-up of cross-hatched decoration
no. 1909.442. bordered by double line of punched dots.
Similar border to the remaining part of pear-
shaped front and central rosette made up of
four peltas in low relief. At the broken edge of
the tails of a pair of crouching marginal
animals just visible.
TYPE VB
Bronze strap-end, in form a broader version of type v A. Decoration florid.
From Anglo-Saxon Site

I. North Luffenham, (FIG. 24, a): length 4'2 em. Decorated in Anglo-Saxon cemetery.
Rutland. chip-carved work. Butt has triangular panel Grave associations not
Oakham School containing a foliate design, animals crouching known.
Museum. back-to-back in corners. 2 rivets. Front in
Assoc. Arch. Sacs. open-work. Central lozenge contains stylized
Reports and Papers, human mask and pair of legs. At sides, crouch-
XXVII (1903), pI. iv, ing marginal animals, facing towards tip. Very
7; V.C.H. Rutland, I, worn.
frontispiece, fig. 5.
TYPE VI
Bronze disc-attachments. These are composed of a decorated disc with a suspension-
loop which projects down the back as a long tapering tang. In most cases a stout cast
bronze ring has been attached to the loop. Disc and tang are gripped together by a
strong rivet, the head of which appears in the centre of the disc. The whole device
was clearly used to link together two or more leather belts or straps. The discs are more
or less standardized in their decoration, which is in 'chip-carved' work, and usually
consists of concentric circles around the central rivet, regular notching around the
edge, and transverse grooves at the base of the suspension-loop.
From Roman Sites

Caistor-by-Norwich, (FIG. 24, b): example with ring. Frilled border Romano-British town of
Norfolk. decorated with deeply carved triangles. Venta Icenorum. 1931 -3 ex-
In the possession of cavations on site of the
Professor Donald At- two fora. Unstratified, but
kinson, by whose in area which produced
permission we pub- ample evidence of late
lish it. occupation.
Dorchester. Oxon. (FIG. I, nos. 5, 7 and 8) : 3 examples, one very Late Roman cemetery
(Dyke Hills). worn. All with cross-tooling on the concentric outside walls of Roman
Ashmolean Museum, circles. 2 had bronze tags attached to the rings. town of Dorchester, grave
Oxford. They arc decorated with transverse lines. I, of man. Associations:
Kirk and Leeds strap-end (p. 63, no. v A.
(1954), pp. 63 fT., I), buckle (p. 59, no. III A,
fig. 27, 5, 7, 8, pl. 2), metal plates (p. 66, no.
iv, A. VII, I), bone toggle, etc.
5-6. Milton-next- (FIC~. 2, b and c): pair, one without ring. Cross- Circumstances of find not
Sittingbourne, Kent. tooling on the concentric circles. known. Probably from
Maidstone Museum. late Roman burial. Asso-
ciations: strap-end (p.
63, no. v A, 3), 2 attach-
ment-plates (p. 68, no.
VII, 2-3).
7-9. Richborough, Kent. (FIG. 24, z, d and g) : 3 examples, very similar, 2 'Saxon-Shore' fort of
Riehborough with rings, 1 without. Rutupiae. From inner stone
Castle Museum. fort ditch, in association
Bushe-Fox (1949), with Theodosian coins.
pp. 123, 145, pis.
xxxii, 70, lii, 187-8.
[0. Croydon, Surrey. (FIG. 24, e): example with unusually long tang. Circumstances of find not
Croydon Public Cross-tooling on some of the concentric circles. known. Published with
Library. un associated material
Proc. Soc. Antiq., xv from Anglo-Saxon graves.
(1895),328 ff.
V.C.H. Surrey, I, 260,
pI. opp.p. 257,no. 9.
Chance Finds
1 I. Croxton, (FIG. 24, I); example with ring. Border of Surface find near Mickle
SW. Norfolk. ovolo decoration and stamped circlets. Hill round barrow.
Norwich Castle
Museum.
Reg. no. 13.07.
Naif. Antiq. Misc., 2
ser. II (1907),1 -4, and
plate; Nor]. Archaeol.,
XXVII (1940), 175,
183, 237; R. R.
Clarke, East Anglia
(1960), p. 129.
12-13. Unknown (PL. I, B, c): pair with stamped triangles and Apparently found with
provenience, Kent. punched dots around central rivet on disc. Both the fine chip-carved
British Museum, with rings. buckle and attachment-
London. plate (pp. 60, 68, nos. III A,
Reg. no. 1942, 10-7,5. 8 and VII, 5).
TYPE VII
Bronze tubular-sided attachment-plates. These are usually composed of a rolled
cylinder of sheet metal, decorated by ribbing, whose edges grip the long side of a rect-
angular plate. Presumably some sort of solder was used to strengthen the joint. The
plate was secured to the belt by '2 rivets. These plates seem to have been either the
attachments for a military sporran, or else the ends of the broad military belt.
From Roman Sites
I. Dorchester, Oxon. (FiG. I, no. 3) ; single plain example, found with Late Roman cemetery
Ashmolean Museum, a series of rectangular bronze plates which seem outside the walls of
Oxford. to have also been part of the belt furniture. Roman town of Dor-
Kirk and Leeds chester, grave I, of man.
(1954), pp. 66-7, Associations: buckle (p.
7 1-2, figs. 27, 3, 59, III A, 2), strap-end
and 28. (p. 63, V A, I), disc-
attachments (p. 66, VI,
2-4), bone toggle.
b
a
FIG. 24
a. Strap-end, type v B, no. I (pp. 38 f., 63).

b-g. Disc-attachments of type VI, nos. I, 7-1 I (p. 66). (b=I,o c-d=7-8,o e-j= IO-II,o g=9).
h. Attachment-plate, type VII, no. 4 (p. 68). Sc. f
Milton-next- (FIG. 2, d and e): matching pair with border of Circumstances of find not
Sittingbourne, Kent. incised running-spiral decoration along edges known. Probably from
Maidstonc Museum. of plates. late Roman burial. Asso-
ciations: strap-end (p.
63, no. v A, 3:), dise-
attachments (p. 66, no. VI,
5-6).
+ Richborough,Kent. (Fl(;. 24, h): a one-piece imitation in sheet 'Saxon-Shore' fort of

Riehborough bronze of the more usual composite form. Rutupiae. From middle
Castle Museum. Decorated by 3 inscribed circlets enclosed in layer of outer ditch of
Reg. no. 2793. panels with border of tooling in herring-bone stone fort, a level which
pattern. produced many Theodo-
sian coins.
Chance Finds
5. Unknown (PL. I, D): fine example in 'chip-carved' work Apparently found with
provenience, Kent. with milled edges. In the centre are three square the fine chip-carved
British Museum, panels filled with a composite pelta design, and buckle (p. 60, no. III A,
London. around them is a border of running tendrilled 8) and the disc-a ttaeh-
Reg. no. 1942, 10-7, scrolls. ments (p. 66, nos. VI,
5 12- 13).
POSTSCRIPT
After this paper had gone to press two more buckles were discovered:
Type II A
25. Luton, Beds. Fragment of plate only (FIG. 19 bis). It may Anglo-Saxon cemetery,
Luton Museum, originally have been decorated with three open- grave 8, of a woman.
Reg. no. L/50/33' work rectangles with roundels at top and Associations: pair of
Antiq. ]., VIII bottom. bronze disc brooches with
(1928), 187, pI. engraved circles and
xxxii,8. stamped decoration. An-
tiq. ]., VIII (1928), pI.
xxvii, 1-2.
TypeIIIB
3. Unknown Tongue missing (FIG. 19 bis): width 5' 7 em.
provenience but Loop plain with well-executed animal heads
probably Kent. with oval eyes and collared necks; decorated
Royal Museum, with tooling. Rectangular plate with grooved
Canterbury. and tooled border, two rows of stamped circlets,
Reg. no. 2187. and vertical grooving at top. Two iron rivets.
The Luton buckle-plate adds one more to the already numerous group of type II A
buckles. It was buried in worn and broken condition in a cemetery which has produced
other very early material.
The Canterbury Museum buckle is the third known example of the type III B
buckle found in this country. Presumably it came from an Anglo-Saxon cemetery.
Like the other two, it is in origin a continental piece, and most resembles examples from
Tongres (Belgium) and Mainz (Germany).128
BIBLIOGRAPHY
Alfoldi, A. (1934) 'Eine spatromische Helmform und ihre Schicksale im Germanisehe-Romanisehen
Mittelalter,' ActaArchaeologica, v (1934), 99 If.
Baldwin Brown, G. (1915) The Arts in Early England (1915)'
Behrens, G. (1919) 'Germanisehe Kriegergraber der 4. bis 7. Jahrhunderts in Stadtischen Altertums-
museum zu Mainz,' Mainzer Zeitschrijt, XIV (1919), 3 ff.
- - - (1930) 'Spatrornische Kerbschnittschnallen,' Schumacher Festschrift (1930), ppo 285 ff.
Bequet, A, (1877) 'La forteresse de Furfooz,' Annqles Soc. archeol. Namur, XIV (1877), 399 ff.
- - - (1891) 'Les cimetieres de la forteresse d'Eprave-La Croix Rouge,' Annales Soc. archeol, Namur, XIX
(1891),435 ff.
128 Breuer and Roosens (1957), ppo 321-2, pis. vii, 4-5 and viii, 40
Boon, G. C. (1957) Roman Silchester (1957)
--~ (1959) 'Thc latest objects from Silchester, Hants.,' Med. Archaeol., III (1959), 79 fT.
Boulanger, C. (1905) Le Mobilierfuneraire gallo-romain etfranc en Picardie et en Artois (1905).
Breuer, J., and Roosens, H. (1957) 'Le Cimetiere franc de Haillot,' Archaeologica Belgica, XXXIV (1957)
(Extrait dcs Annales Soc. archeol. Namur, XLVIII (1956), 171 ff.).
Bruce-Mitford, R. L. S. (1954) 'A late or sub-Roman buckle-plate from College Wood near Winchester,'
Antiq. ]., XXXIV (1954), 75 fro
Bushc-Fox, J. P. (1928) Richborough, II (Res. Rept. Soc. Antiq. Lond., VII, 1928).
- - - (1949) Richborough, IV (Res. Rept. Soc. Antiq. Lond., XVI, 1949)'
Chadwick Hawkes, Sonia (1961) 'The Jutish Style A. A study of Germanic animal art in southern England
in the fifth century A.D.,' Archaeologia, XCVIII (1961), 29 ff.
Cachet,]. B. D. (1864) La Normandiesouterraine (1854)'
Cook, Jean M. (1953) 'An Anglo-Saxon cemetery at Broadway Hill, Broadway, Worcestershire,' Antiq.
]., XXXVIII (1958), 58 ff.
Dasnoy, A. (1955) 'Quelques tombes de la region namuroise datces par des monnaies,' Annales Soc. archeol.
Namur, XLVIII (1955), 5 ff.
De Laet, S. J., Dhondt, J., and Nenquin, J. (1952) 'Les Laeti du Namurois et I'origine de la civilisation
merovingicnne,' Etudes d'histoire et d'archeoiogie dediees aFerdinand Courtoy (1952), 149 ff.
De Loe, A. (1937) Belgique Ancienne, III (1937).
Del Marrnol, E. (1860) 'Fouilles dans un cimeticrc de l'epoque franque a Samson,' Annales Soc. archeol:
Namur, VI (1860), 345 ff.
Eck, T. (1891) Les deux cimetieres gallo-romains de Vermand et St. Q.uentin (1891).
Evison, Vera 1. (1955) 'Early Anglo-Saxon inlaid metalwork,' Antiq,]., xxxv \1955) 20 ff
- - - (1958) 'Further Anglo-Saxon inlay,' Antiq.]., XXXVIII (1958), 240 ff.
Faider Feytmans, G. (1951) 'Sepultures du IVe. siecle a Tournai,' Latomus, X (1951), 29 fT.
Forssander, .1. (1937) 'Provinzialromisches und Germanisches,' Meddel. fran Lunds Universitetets Hist.
Museum, VII (1937), II ff.
Hawkes, C. F. G, and Hull, M. R. (1947) Camulodunum (Res. Rept. Soc. Antiq. Lond., XIV, 1947).
Hildyard, E . .1. W. (1957) 'Cataractonium, fort and town,' Yorks. Archaeol.]., XXXIX (1957), 243 ff.
Hillier, G. (1856) The History and Antiquities of the Isle of Wight (1856).
Jullian, C. (1926) Histoire de la Gaule, VII (1926).
Kenyon, Kathleen (1948) Excavations at theJewry Wall site, Leicester (Res. Rept. Soc. Antiq, Lond., XV, 1948).
Kirk, .1. R., and Leeds, E. T. (1954) 'Three early Saxon graves from Dorchester, Oxon.,' Oxoniensia,
XVII-XVIII (1954), 63 ff.
Lantier, R. (1948) 'Un cimetiere du IVe siecle au "Mont-Ange" (Vert-Ia Gravelle, Marne),' L'Antiquite
Classique, XVII (1948), 373 ff.
Leeds, E. T. (1936) Earlv Anglo-Saxon Art and Archaeology (1936).
- - - , and Harden, D. B. (1936) The Anglo-Saxon Cemetery at Abingdon, Berkshire (1936).
- - - , and Shortt, H. de S. (1953) An Anglo-Saxon Cemetery at Petersfinger, near Salisbury (1953).
Lot, F., Pfister, C., and Ganshof, F. (1940) Histoire du Moyen Age. Les Desiinees de l'Empire en Occident de
3.95 a 768 (194 0).
Morris, J., and Bidder, H. F. (1959) 'The Anglo-Saxon cemetery at Mitcham,' Surrey Archaeol. Coli., LVI
(1959),51 ff.
Nenquin,J. (1953) La Necropole de Furfoo z (Soc. archeol. Namur, 1953).
O'Neil, B. H. St. John (1944) 'The Silchester region in the 5th and 6th centuries A.D.,' Antiquity, XVIII
(1944), 133 ff.
Pilloy,J. (1886, 1891, 1912) Eiudes sur d'nnciens lieux de sepultures dans l'Aisne, 3 volumes (1886, 1891, 1912).
Plettke, A. (192 I) Ursprung und Ausbreitung der Angeln und Sachsen : Die Urnenfriedhife in Niedersachsen, III (1921).
Riegl, A. (1927) Spatromische Kunstindustrie (1927).
Rigollot, Dr. (1850) 'Recherches historiques sur les peuples de la race teutonique qui envahirent les
Gaules au 5e siecle ... ,' Memoires Soc. Antiquaires de Picardie, X (1850), 121 ff.
Roach Smith, C. (1850) The Antiquities of Richborough, Reculver and Lymne (1850).
Roeder, F. (1933) 'Neue Funde auf Kontinental-Sachsischen Friedhofen der Viilkerwanderungszeit,'
Anglia, LVII (1933), I ff.
Roes, Anna (1953) 'A travers les collections archcologiques de la Hollande,' Berichten van de Rijksdienst
voor het Oudheidkundig Bodemonderzoek, IV (1953), 26 ff.
Salin, B. (1904) DieAltgermanische Thierornamentik (1904.).
Smith, Reginald (1923) Guide to the Anglo-Saxon and Foreign Teutonic Antiquities (British Museum, 1923).
Steeger, A. (1937) Germanische Funde der Volkencanderung reit aus Krifeld (1937).
Stevens, C. E. (1940) 'The British sections of the Notitia Dignitatum,' Archaeol. ]., XCVII (1940), 125 ff.
Tischler, F. (1954) 'Der Stand der Sachsenforschung, archaologisch gesehen,' Bericht der Romisch-
Germanischen Kommission, xxxv (1954), 2 I ff.
Tonnochy, A. B., and Hawkes, C. F. C. (1931) 'The sacred tree motive on a Roman bronze from Essex,'
Antiq.]., XI (1931),123 ff.
Tudor, D. (1945) 'Spatromischc Gurtclbcschlage aus Sudrumanien,' Dacia, IX-X (1945), 513 ff.
Verlinden, C. (1946) 'De Franken en Aetius", Bijdragen voor Geschiedenis der Nederlanden, I (1946), I ff.
Warhurst, A. (1955) 'The Jutish cemetery at Lyminge,' Arch. Cantiana, LXIX (1955), I ff.
7
Werrn:r , J. (1930) 'Spatrornische Gur tclgarnituren in Keilschnitt Technik,' }ahreshefte des Osterreichischen
Archiiologischen Insiituts in Wien, XXVI (1930), 53 ff.
- - - (1949) 'Eberzicr von Monceau-Ie-Ncuf (Dcp. Aisnc),' "leta Archaeologica, xx (1949), 248 ff.
--- (1950) 'Zur Entstehung del' Rcihcngraberzivilisation;' Archaeologica Geographiea, I (1950),23 IT.
(1953) 'Zu frankischcn Schwertcrn des 5. Jahrhunderts,' Gormania, XXXI (1953), 38 ff.
(1956) 'Frankische Schwerter des 5. Jahrhunderts aus Samson und Petersfingcr,' Germanic, XXXIV
(1956), 156 ff.
-- (1958) 'Kricgcrgrabcr aus del' ersten Halftc des 5. Jahrhundcrts zwischen Schelde und Weser.'
Bonner }ahrbucher, CLVIII (1958), 372 ff
Wheeler, R. E. M., and T.V. (1932) Report on the excavation of the Prehistoric, Roman, and post-Roman site ill
Lydney Park, Gloucestershire (Res. Rept. Soc. Antiq. Lond., IX, 1932).
ACK:"OWLEDGEMENTS
The bringing together for drawing and study of so many widely dispersed objects has only been
achieved with the kindly cooperation of the many museums and individuals whose names appear in the
Catalogue. We offer them our gratitude and appreciation of all the trouble they have taken on our
behalf. We should also like to thank Professor Donald Atkinson, Mr. George Boon, the late Dr. Philip
Corder, Mr. Charles Green, Mr. Jack Lindsay, Mr. P. Storer Peberdy, Mr. J. N. Taylor and Mr. Hugh
Thompson, all of whom gave us information or answered queries. We are indebted to His Grace the
Duke of Wellington for permission to publish the material from Silchester, and to the Rt. Hon. the
Viscount Bledisloe, Q.c., for kindly allowing the use of the material from Lydney Park.
With the exception of FIGS. I, 3-4, 6-7, 9 and I I, all the drawings used in this paper are by Miss
Elizabeth Meikle (now Mrs. Fry-Stone).

5 001 070 PDF

Uploaded by

Copyright:

Available Formats

5 001 070 PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

5 001 070 PDF

Uploaded by

Copyright:

Available Formats

Soldiers and Settlers in Britain)

Fourth to Fifth Century

Dorchester-on-Thames in the ye:u 1874 were discovered in the Dyke

K Hills a number of burials belonging, it seems, to an as yet only partially

DYKE HILLS. DORCHESTER, OXON.

THE MILITARY GRAVE FROM DORCHESTER AND ITS

Tournai, Belgium. Cimetiere du Parc, grave F.5

Furfooz, near Namur, Belgium. Grave 3. 6

Vieuxville, near Liege, Belgium ,7

Vermand, Aisne, France. Chemin des Mortes cemetery, grave 284.8

7 Breuer and Roosens (1957), pp. 243 ff., figs. 3 I -4.

Monceau-Le-Neu], Aisne, France."

'9 Noiitia Dignitatum, pars occidcntis, chap. xlii.

kind so far found in this country. The situation at Dorchester, therefore, is an

IMPORTED CONTINENTAL METALWORK OF LATE ROMAN DATE:

represents late Roman ornamental style in a distinctive version, limited to the

34 Tudor (1945), fig. r a.

fROid kOMAN 11m .. ..

Fml AN'LO lA~OH llHl

u d les and belt-fittings, and

built as a northward extension of the 'Saxon-Shore' defences, we known very

7' Stevens (1940).

BRONZE BUCKLES MADE IN BRITAIN IN THE LATE AND SUB-

lA n INS I1A JII

DISTRIBUTIO:,\ OF BRITISH-MADE BUCKLES, TYPES I AND II (pp. 28 ff.)

BOAR-TUSK ORNAMENTS~FROM RICHBOROUGH, KENT (p. 29). Sc.

BRONZES OF THE ANGLO-SAXON PERIOD:

no Steeger (1937), pp. 182-B.

r r z Werner (1953), pl. vii, 3b and 6-8.

From Roman Sites

Buckles of type 1 A, nos.

From Anglo-Saxon Sites

From Anglo-Saxon Sites

From Roman Sites

with circles at top and bottom. Bands of

Buckles of type II A, (pp. 21 fl., 51 If.). Scv]

FIG. IS If 51 ff.). Sc. i

From Anglo-Saxon Sites

From Roman Site

From Anglo-Saxon Site

From Anglo-Saxon Sites

From Roman Sites

From Anglo-Saxon Sites

From Anglo-Saxon Site

From Roman Sites

a. Strap-end, type v B, no. I (pp. 38 f., 63).

+ Richborough,Kent. (Fl(;. 24, h): a one-piece imitation in sheet 'Saxon-Shore' fort of

You might also like